article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
presently the origin still remains elusive for the observed galactic and extragalactic magnetic fields of about a few @xmath5 g @xcite .
it is generally believed that the so - called galactic dynamo @xcite mechanism must be involved to amplify a magnetic seed field @xmath6 g on a comoving scale larger than mpc .
since the universe is in a highly conducting plasma state through the most of its history , the ratio of the magnetic energy density and the thermal background , @xmath7 , remains constant along the cosmic evolution .
the required @xmath8 is then translated into @xmath9 .
depending on the correlation scale of the magnetic field , the seed fields can be produced inside or outside the hubble sphere .
various astrophysical hypotheses , _
e.g. _ the biermann battery @xcite , and cosmological scenarios , _ e.g. _
harrison s vorticity model @xcite , may generate large magnetic fields locally . even if magnetohydrodynamical inverse cascades are invoked , however , the correlation scales of the resultant fields will not extend beyond 100 pc @xcite .
apparently these processes are incapable of explaining the observed cosmic magnetic fields which are coherent over much larger scales .
on the other hand , the correlation scale of primordial seed fields produced during inflation in the early universe is much greater than the hubble radius . as we know ,
the de sitter like expansion is better described by the spatially flat friedmann - robertson - walker ( frw ) metric which conformally preserves the standard electromagnetic ( em ) action @xcite .
therefore , a viable scenario of an inflationary magnetogenesis must exploit some mechanism to break the conformal invariance of the em field .
there are several popular methods to serve the purpose . for examples
, one can consider the em field tensor @xmath2 coupling to the gravitational curvature scalar @xmath10 in the forms @xcite of @xmath11 , @xmath12 , @xmath13 , where @xmath14 and @xmath15 is the vector potential ; to scalar inflatons or dilations @xcite ; to pseudo - scalar fields like axion @xcite; ... _etc .
_ however , most of them do not produce satisfactory outcomes @xcite .
motivated by the mechanism of axial couplings @xcite , a scenario connecting the primordial magnetic field ( pmf ) to the dark energy has been developed @xcite . assuming that an evolving pseudo scalar field is responsible for the late time acceleration of the universe @xcite
, a sufficiently large pmf can be produced inside the hubble sphere via the spinodal instability .
furthermore , the spinodal effect is capable of generating long - wavelength fluctuations such that the created magnetic fields coherently correlate themselves over a large distance in the order of 10 mpc . in a previous work @xcite
, we extended the axial coupling mechanism , _
i.e. _ assuming that the em field couples to a pseudo scalar inflaton , to explore the possibility of generating a significant pmf outside the hubble radius .
once again we found that the spinodal instability provides a robust mechanism for generating pmf in the inflationary epoch .
however , a pre - inflationary fast roll stage must be involved in order to increase the efficiency of pmf production .
the correlation length of the resultant seed fields is consistent to that produced by the same mechanism inside the hubble sphere , _ i.e. _ at a scale around 10 mpc .
unfortunately , a further check on the energy constraint required by the inflaton field driving the de sitter like expansion showed that such an axial coupling mechanism , when applying to the magnetogenesis process outside the hubble sphere , would produce too much magnetic power to become a reasonable source for pmf . in this work
, we will generalize the previous work by combining both the spinodal instability in the axial coupling mechanism and the dilaton electromagnetism .
this is different from the recent work that has just added em helicity to inflationary magnetogenesis based on dilaton electromagnetism @xcite .
in fact , the pmf generation in our work mainly relies on the spinodal instability .
this paper is organized as follows . in the next section
, we present our formulation for the axial coupling of the em field to a pseudo scalar field in the context of dilaton electromagnetism .
we will then summarize the difficulties of inflationary magnetogenesis that makes use of dilaton and axion fields . in sec .
iii , we employ the chaotic inflation scenario as a background evolution to carry out the numerical calculation and show how the pmf can be created in our model . finally , we conclude our findings with some discussions .
we consider @xmath16 , \label{emaction}\ ] ] where @xmath17 is the em field strength tensor and @xmath18 is its dual .
note that @xmath19 is added to the dual tensor because @xmath20 is a tensor density of weight @xmath21 .
the pseudoscalar @xmath22 is an axion - like field , @xmath23 is an energy scale , and @xmath24 is the fine structure constant .
here we assume a spatially flat frw metric : @xmath25 where @xmath26 is the cosmic scale factor and @xmath27 is the conformal time related to the cosmic time by @xmath28 .
the expansion rate is governed by the hubble parameter @xmath29 , where the dot and the prime denote derivatives with respect to @xmath30 and @xmath27 , respectively .
after rescaling the vector potential by the coupling constant , @xmath31 , the action ( [ emaction ] ) becomes @xmath32 .
\label{emaction2}\ ] ] now we consider a time - dependent coupling constant described by a time funciton @xmath33 which is driven by a time - evolving dilaton field @xmath34 . in the axion sector , we assume that the axion mean field is @xmath35 .
in addition , inspired by extra - dimension theories , we assume that the energy scale is related to the reduced planck mass by @xmath36 .
for example , this relation has a simple geometrical origin in many string theory constructions with the factor @xmath37 determined by a combination of the size of the compactification manifold , the string length , and the string coupling @xcite .
this factor @xmath37 may be time dependent due to a possible temporal evolution of extra dimensions and string parameters .
we define a time function @xmath38 to better monitor the axial effect on the processes of magnetogenesis .
subsequently , in terms of the pair of coupling functions @xmath39 and @xmath40 , the action can be recast in the form of @xmath41 .
\label{ijaction}\ ] ] to derive the em wave equation , we chose the temporal gauge , i.e. @xmath42 , and the coulomb gauge , i.e. @xmath43 . under these gauges and the metric ( [ frw ] ) , the physical electric and magnetic fields are respectively given by @xmath44 from the action ( [ ijaction ] ) , we obtain the wave equation for the vector potential , @xmath45 to proceed , we decompose the gauge field @xmath46 as @xmath47 , \label{decomposition}\ ] ] where the annihilation and creation operators obey @xmath48 = \delta_{\lambda\lambda'}\delta ( { \vec k}-{\vec k}'),\ ] ] in which @xmath49 are normalized circular polarization vectors satisfying @xmath50 , @xmath51 , @xmath52 , and @xmath53 . inserting the decomposition ( [ decomposition ] ) into eq .
( [ wave ] ) , we obtain the equation of motion for the mode functions , @xmath54 a_{\pm}(\tau , k ) = 0 .
\label{f_v}\ ] ] then , the energy density of the produced em fields is governed by the vacuum expectation value , @xmath55 where @xmath56 terms characterize the electric and the magnetic energy density of the pmf as @xmath57
as we have mentioned in sec . [ intro ] , there have been many attempts to generate large - scale pmf in inflation by invoking a breaking of the conformal invariance of the em field . the main hurdle that one has to overcome can be easily understood by a simple scaling argument as follows .
we need to produce pmf with magnetic energy density at the end of inflation of order @xmath9 .
not only the produced magnetic field strength has to be big enough , but also the wavelength of the magnetic fields is of large scales . as such ,
em @xmath58-modes are thought to be amplified and leave the horizon at about @xmath59 e - foldings before inflation ends . because of the @xmath60 dilution factor in @xmath61 ( see eq .
( [ benergy ] ) ) , the pmf energy density is about @xmath62 times larger at the time of production than that at the end of inflation . if the reheating process is rapid and efficient , the energy density of the thermal background @xmath63 will amount to that stored in the inflaton field @xmath64 during inflation , i.e. @xmath65 right after inflation .
therefore , at the time pmf are produced , @xmath66 , which exceeds the inflaton energy by @xmath67 times .
this huge discrepancy already poses a serious challenge to the inflationary magnetogenesis .
let us go back to the action ( [ ijaction ] ) .
when @xmath68 and @xmath69 , it resumes maxwellian electromagnetism , in which em mode functions @xmath70 in eq .
( [ f_v ] ) have simply plane - wave solutions , as dictated by the conformal invariance of the action .
this invariance prohibits the growth of em fields in inflation . in the following
, we will briefly review the efficiency of each mechanism to break the conformal invariance of the em field for the generation of pmf in inflation .
when @xmath71 and @xmath69 , it is the so - called dilaton electromagnetism . in the strong em coupling case ,
i.e. @xmath72 , strong enough pmf can be produced when the effective coupling constant , which is inversely proportional to i , is extremely large in the beginning of inflation and becomes of the order of one at the end of inflation .
but one has to bear with an extremely strongly coupled em theory that should not be trusted at all . in the weak em coupling case , i.e.
@xmath73 , no sizable pmf can be produced @xcite .
when @xmath68 and @xmath74 , where the axion is identified as the inflaton @xmath75 , the mode equation ( [ f_v ] ) becomes @xmath76 a_{\pm}(\tau , k ) = 0 , \hspace{5 mm } \xi \equiv \frac{\alpha \dot{\phi}}{2 f h}\ , .
\label{axionxi}\ ] ] it is well known that either one of the two modes exhibits a spinoidal instability as long as the modes satisfy the condition , @xmath77 , where @xmath78 is nearly a constant for a slow - roll inflation . when the inflaton rolls down the potential , these unstable modes grow exponentially by consuming the inflaton kinetic energy .
it has been found that the generation of helical magnetic fields during the single - field slow - roll inflation due to this axial coupling of em field to the inflaton leads to a blue spectrum of magnetic fields .
although the helical magnetic fields can further undergo , during the radiation and matter dominated eras , a process of inverse cascade that transfers spectral power from small to large scales , the magnetic fields generated by such an axial coupling would not lead to a strong enough field strength on cosmological scales @xcite . in ref .
@xcite , a pre - inflationary fast roll stage is involved in order to enhance the efficiency of helical magnetic field production .
unfortunately , when applying to the magnetogenesis process outside the hubble sphere , it would produce too much @xmath79 that imposes too strong a backreaction on @xmath64 to become a reasonable source for pmf .
recently , the authors in ref .
@xcite have discussed the weak em coupling case in which a new parity violating term is introduced , given by the @xmath40 term with @xmath80 .
this parity violating term allows more freedom in tuning the amplitude of the produced helical magnetic field at the end of inflation .
however , the model still relies on the inverse cascade processes to amplify a helical magnetic field at large scales during the radiation dominated epoch . as a consequence , a magnetic seed field @xmath81
g can be produced if inflation occurs at an energy scale ranging from @xmath82 to @xmath83 gev .
another difficulty is that the electric field density would be larger than the total energy density of the universe if a sufficient amount of pmf is produced .
suppose the helicity @xmath84-mode grows efficiently as @xmath85 , where @xmath86 is treated approximately time independent and @xmath87 .
then , from eqs .
( [ emenergy ] ) and ( [ benergy ] ) we have the ratio of the electric and magnetic spectral energy densities as @xmath88 the @xmath58 values for large - scale pmf are about @xmath89 , so this ratio is as small as @xmath90 .
going back to the end of the magnetogenesis phase that is assumed to occur at @xmath91 , the energy densities of the pmf and the electric fields are given by , respectively , @xmath92 hence , at the end of the magnetogenesis phase we have @xmath93 showing that the electric energy density is larger than the total energy density of the universe , which contradicts the conservation of energy .
in the present work , we assume an inflationary background driven by the inflaton field @xmath75 . we employ the chaotic inflation scenario @xcite to unravel the inflationary magnetogenesis by full numerical calculations .
in such a universe , the standard slow - roll inflationary background is driven by the potential @xmath94 where the parameter @xmath95 characterizes the mass of the inflaton field .
on the other hand , in order to solve for the ratio of the magnetic energy density to the energy density of the thermal background @xmath63 at the end of inflation , we mimic the evolution of the thermal background by tracing out changes in the energy density of the radiation component @xmath63 .
accordingly , the background dynamics are governed by @xmath96 where @xmath97 , and all other dynamical variables have been properly rescaled by the reduced planck mass , _ e.g. _
@xmath98 , @xmath99 , @xmath100 , _ etc .
_ in eq .
( [ varphieq ] ) the interaction of the inflaton with other light fields is modeled by a damping term @xmath101 , which is not important in the slow - roll regime where @xmath102 is dominant .
the hubble parameter @xmath102 that measures the expansion rate of the universe is characterized by @xmath103 since the accelerated expansion would dilute everything that may have existed prior to the inflation , the initial value for @xmath63 does not matter to our calculations .
the initial value of the hubble parameter at @xmath104 is determined by values of @xmath105 and @xmath106 through @xmath107 .
we take @xmath108 gev for the inflaton mass , @xmath109 , and @xmath110 , @xmath111 respectively for the initial position and the initial velocity of the inflaton @xmath112 .
the e - folding since the beginning of inflation is defined by @xmath113 and we find that our inflation ends at @xmath114 . since we control the production of pmf during inflation such that @xmath115 , the backreaction of em field production on the inflationary background can be safely neglected . to make our scenario of inflationary magnetogenesis more general , we do not specify any model for the dynamics of dilaton or axion during inflation .
instead , we assume that @xmath116^n,\quad\quad j'(\tau)=c\,h(\tau ) a(\tau)^p\ , , \label{ijform}\ ] ] where @xmath117 , @xmath118 , and @xmath119 are constant parameters within time intervals , and @xmath120 denotes the time at the end of inflation .
moreover , we presuppose that @xmath68 and @xmath121 for @xmath122 , so the em theory becomes maxwellian electromagnetism after inflation .
now we insert @xmath39 and @xmath123 in eq .
( [ ijform ] ) into the mode equation ( [ f_v ] ) . to solve for the mode functions , we set the initial conditions at @xmath104 as @xmath124 then
, we calculate the spectral magnetic energy density , @xmath125 and its ratio to the energy density of the thermal background at the end of inflation , @xmath126_{\tau=\tau_f}.\ ] ] here we are not going to extract that of the parameter space in our model for a successful inflationary magnetogenesis .
rather , we will give some simple examples to illuminate how and why axion dynamics can greatly facilitate the pmf production during inflation .
the mode equation ( [ f_v ] ) then becomes @xmath127{\dot a}_{\pm}+\left[\frac{k^2}{a^2 } \mp \frac{ckh a^{n+p-2}}{a(\tau_f)^n}\right]a_{\pm}=0 .
\label{f2_v}\ ] ] as the first example , we just take @xmath128 to efface the friction term . as long as @xmath129 ,
the mode function @xmath130 becomes unstable and grows exponentially with time .
the mode growth rate can be estimated by @xmath131 , where @xmath86 is the growth index given by @xmath132^{1/2}$ ] . during this weak em coupling regime ,
the electric energy density dominates over the magnetic energy density . specifying @xmath133 and @xmath134 leads both the electric and magnetic energy densities to grow exponentially near @xmath135 . to avoid over - production
, we have imposed a strong em coupling regime with @xmath136 after @xmath135 and introduced a wavenumber cutoff by assuming that @xmath137 for @xmath138 .
the strong em coupling damps out the electric fields while freezing the magnetic fields .
the wavenumber cutoff signifies the effect of backreactions due to the rapid production of high @xmath58-modes .
here we simply ignore these modes .
it would be interesting to take into account the backreaction in the mode equation to study how the production of high @xmath58-modes can be self - regulated by the backreaction .
figure [ fig1 ] shows the time evolution of the electric and magnetic energy densities with @xmath139 .
also shown is the evolution of @xmath140 that monitors the condition of a small backreaction . in fig .
[ fig2 ] , the ratio @xmath141 is plotted against @xmath142 , where the length scale is set up such that @xmath143 corresponds to the mode that leaves the horizon about @xmath59 e - foldings by the end of inflation , i.e. its wavelength is comparable to the size of the present universe .
the spectral magnetic energy density of the pmf shows a peak around @xmath144 , which corresponds to a comoving scale of order mpc , because of the wavenumber cutoff . integrating this
spectrum amounts to @xmath9 at the end of inflation .
if the backreaction is included in the pmf production , the production of high @xmath58-modes should be effectively reduced to a level comparable to the peak height of the spectrum . .
the time is counted with e - foldings @xmath139 from the beginning of inflation .
the dashed ( dotted ) line represents the electric ( magnetic ) component .
the horizontal dotted - dashed line signifies the conservation of energy.,scaledwidth=80.0% ] .
the mode with @xmath145 corresponds to the size of the present universe while the spectrum peaks at @xmath146 on a comoving scale larger than mpc.,scaledwidth=80.0% ] let us return to the axion inflation discussed in sec .
[ diff ] , where @xmath68 and @xmath74 , and the growth condition is @xmath77 with @xmath78 a nearly constant parameter .
the axion inflation indeed corresponds to the case with @xmath147 and @xmath148 , or to the case with @xmath128 and @xmath149 as far as the mode growing behavior is concerned . as a consequence
, we see that the mode growth index @xmath86 with @xmath149 is suppressed by the scale factor @xmath150 for the modes of cosmologically interesting scales , say @xmath151 , towards the end of inflation .
now it is transparent that we need at least @xmath152 to boost the generation of pmf .
figure [ fig1 ] shows that the @xmath40 term with @xmath134 allows the modes of comologically interesting scales to grow sufficiently fast at a much later time ( @xmath153 ) , thus circumventing the difficulty for the inflationary magnetogenesis , which has assumed that the comologically interesting modes get amplified and leave the horizon at about @xmath154 , as discussed in sec .
[ diff ] . we have also tried two different evolution histories .
the first one is in the strong em coupling regime with @xmath117 changing from @xmath155 to @xmath156 at @xmath157 .
the constant parameter is @xmath158 and the parameter @xmath119 is time dependent in such a way that @xmath159 for @xmath160 and otherwise @xmath137 .
the second one is in the weak em coupling regime with @xmath117 changing from @xmath147 to @xmath161 at @xmath162 , @xmath149 , and @xmath119 changing from @xmath137 to @xmath163 at @xmath164 .
the time evolutions of the electromagnetic energy densities for the first and the second cases are shown in fig .
[ fig3 ] and fig .
[ fig5 ] , respectively .
the spectra of the generated pmf for both cases are shown in fig .
[ fig4 ] and fig .
[ fig6 ] , respectively .
once again , we have seen how the @xmath40 term allows the modes of comologically interesting scales to grow sufficiently fast at a much later time . however , in the weak em coupling regime , the generated pmf is still too small to be the seed fields .
but in the strong em coupling regime.,scaledwidth=80.0% ] many papers have studied the problem of magnetogenesis during inflation spoiling the cosmological density perturbations @xcite and generating primordial tensor perturbations ( ref . @xcite and references therein ) .
the studies most relevant to the present consideration are the constraints from measurements of the power spectrum and bispectrum of the cosmic microwave background @xcite , in which the authors have considered the above mentioned model @xmath68 and @xmath74 as shown in eq .
( [ axionxi ] ) and derived an upper limit on a combined parameter : @xmath165 .
this limit translates into the condition that during inflation @xmath166 . in fig .
[ fig1 ] and fig .
[ fig3 ] , where we have successful magnetogenesis , @xmath167 is much less than @xmath168 for the time when em fields are generated during inflation . when the modes of comologically interesting scales grow at @xmath169 , @xmath170 in fig .
[ fig1 ] and @xmath171 in fig .
it is worth to study in details the effects of magnetogenesis in the present work on the cosmic microwave background .
but in the strong em coupling regime.,scaledwidth=80.0% ] but in the weak em coupling regime.,scaledwidth=80.0% ] but in the weak em coupling regime.,scaledwidth=80.0% ]
it is amazing that the universe has been so good a conductor that a seed magnetic field can hardly be generated in it .
but once seed magnetic fields were generated by a whatsoever mechanism in the early universe , they were kept frozen and preserved in the conducting plasma state of the universe .
there has been a lot of proposals for generating magnetic fields in the cosmic plasma via astrophysical magnetogenesis .
however , it was thought that the best way to avoid the high conductance is to generate primordial magnetic fields from photon quantum fluctuations during inflation when the universe was in a vacuum state .
unfortunately , this thought failed due to the conformal invariance of electromagnetism .
so far , most common models for inflationary magnetogenesis involve the breaking of conformal invariance in the context of dilaton electromagnetism . however , a successful model would require a strong electromagnetic coupling regime that can not be trusted physically .
these models have been extended by including an axion - photon coupling .
this coupling can generate and add helicity to a blue spectrum of magnetic fields , so they still rely on the inverse cascade to convert shorter - wavelength magnetic fields to cosmologically interesting ones . in the present work ,
we have pointed out that the main obstacle in inflationary magnetogenesis is the weakening of the generated large - scale magnetic fields by a hugh conformal factor due to the inflationary dilution . in light of this , we have generalized the axion dynamics to a time dependent coupling @xmath1 .
we have found that in order to generate large - scale magnetic fields during inflation , the @xmath4 function must be able to delay their growth to a later stage of inflation .
thus we have given a recipe of the forms of the @xmath172 and @xmath4 functions for which a successful inflationary magnetogenesis can be made , though we admit that it is rather contrived .
it seems that the axion dynamics may involve a specific form of the axion potential or even the evolution of extra dimensions in string - inspired theories .
then it would be interesting to build a particle and field model to realize the idea of using @xmath173-@xmath174 functions proposed in this work .
furthermore , a complete analysis including the backreaction and observational constraints is in order .
this work was supported in part by the national science council , taiwan , roc under the grant no .
nsc101 - 2112-m-001 - 010-my3 ( k.w.n . ) and the office of research and development , national taiwan normal university , taiwan , roc under the grant no .
102a05 ( w.l . ) .
e. n. parker , _ astrophys .
j. _ * 163 * , 255 ( 1971 ) ; _ cosmical magnetic fields _ ( clarendon , oxford , england , 1979 ) ; ya .
b. zeldovich , a. a. ruzmaikin , and d. d. sokoloff , _ magnetic fields in astrophysics _ ( gordon and breach , new york , 1983 ) .
b. ratra , _ astrophys .
j. _ * 391 * ( 1992 ) l1 ; m. gasperini , m. giovannini and g. veneziano , _ phys .
lett . _ * 75 * , 3796 ( 1995 ) ; m. giovannini , _ phys .
d _ * 64 * , 061301 ( 2001 ) ; k. bamba and j. yokoyama , _ phys .
d _ * 69 * , 043507 ( 2004 ) ; _ phys .
d _ * 70 * , 083508 ( 2004 ) ; k. bamba and m. sasaki , _ jcap _ * 02 * , 030 ( 2007 ) .
n. barnaby , r. namba , and m. peloso , _ phys . rev .
d _ * 85 * , 123523 ( 2012 ) ; m. giovannini , _ phys .
d _ * 87 * , 083004 ( 2013 ) ; t. fujita and s. yokoyama , _ jcap _ * 03 * , 013 ( 2014 ) ; r. j.z .
ferreira , r. k. jain , and m. s. sloth , _ jcap _ * 06 * , 053 ( 2014 ) . | we discuss the generation of primordial magnetic fields during inflation in the dilaton - axion electromagnetism , in which the dilaton and axion dynamics are introduced in terms of two time dependent functions of the cosmic scale factor , @xmath0 and @xmath1 , respectively , where @xmath2 is the electromagnetic field strength and @xmath3 is its dual .
we study the form of @xmath4 that can generate a large seed magnetic field .
although the @xmath4 function is model dependent , the axion - photon coupling may open up a new window for a successful inflationary magnetogenesis . |
simulating finite density qcd is one of the outstanding problems in lattice gauge theory @xcite . because of the sign problem
no reliable numerical simulations of finite density qcd with three colors exist even in the strong coupling limit @xcite .
this is a rather frustrating situation in view of the current intense interest in finite density qcd fueled by the phenomenology of heavy ion collisions , neutron stars , early universe and color superconductivity .
therefore even a qualitative description of strongly coupled qcd at finite density using field theoretical methods is quite welcome .
strongly coupled qcd at finite quark chemical potential @xmath0 has previously been studied analytically both in the euclidean @xcite and in the hamiltonian @xcite formulations . except for @xcite ,
the consensus is that at zero or low temperatures strong coupling qcd undergoes a first order chiral phase transition from the broken symmetry phase below a critical chemical potential @xmath1 to a chirally symmetric phase above @xmath1 . in this work
we calculate the equation of state of strongly coupled qcd at finite @xmath0 using wilson fermions and show that in the phase with spontaneously broken chiral symmetry the quark fermi sea has negative pressure indicating its mechanical instability .
this result is in qualitative agreement with those obtained using continuum field theory models with four point interactions @xcite .
we begin by defining our effective hamiltonian for strongly coupled qcd which was first derived by smit @xcite .
henceforth we adopt the notation of smit @xcite , set the lattice spacing to unity ( @xmath2 ) and work in momentum space .
the charge conjugation symmetric form of smit s effective hamiltonian with a chemical potential @xmath3 in momentum space is @xmath4 \biggl [ \bigl(\psi^\dagger_{a\alpha}\bigr)_\sigma(\vec{p}\ , ) , \bigl(\psi_{a\alpha}\bigr)_\nu(-\vec{p}\ , ) \biggr]_- \nonumber \\ & & -\frac{k}{8n_{\rm c } } \sum_{\vec{p}_1,\ldots,\vec{p}_4 } \sum_{l=1}^3 \delta_{\vec{p}_1+\cdot\cdot\cdot+\vec{p}_4,\vec{0 } } \biggl [ e^{i((\vec{p}_1+\vec{p}_2)\cdot \hat{n}_l ) } + e^{i((\vec{p}_3+\vec{p}_4)\cdot \hat{n}_l ) } \biggr ] \nonumber \\ & & \!\!\!\!\!\!\!\!\!\!\!\!\!\!\ ! \otimes \biggl [ \left ( \sigma_l \right)_{\sigma\nu } \bigl(\psi^\dagger_{a\alpha}\bigr)_\sigma(\vec{p}_1\ , ) \bigl ( \psi_{b\alpha}\bigr)_\nu(\vec{p}_2\ , ) - \left ( \sigma_l \right)^\dagger_{\sigma\nu } \bigl(\psi_{a\alpha}\bigr)_\nu(\vec{p}_1\ , ) \bigl ( \psi^\dagger_{b\alpha}\bigr)_\sigma(\vec{p}_2\ , ) \biggr ] \nonumber \\ & & \!\!\!\!\!\!\!\!\!\!\!\!\!\!\ ! \otimes \biggl [ \left ( \sigma_l \right)^\dagger_{\gamma\delta } \bigl(\psi^\dagger_{b\beta}\bigr)_\gamma(\vec{p}_3\ , ) \bigl(\psi_{a\beta}\bigr)_\delta(\vec{p}_4\ , ) - \left ( \sigma_l \right)_{\gamma\delta } \bigl(\psi_{b\beta}\bigr)_\delta(\vec{p}_3\ , ) \bigl ( \psi^\dagger_{a\beta}\bigr)_\gamma(\vec{p}_4\ , ) \biggr ] \label{eq : heffp}\end{aligned}\ ] ] where @xmath5 with the wilson parameter @xmath6 taking on values between 0 and 1 . in the above hamiltonian color , flavor and dirac indices
are denoted by @xmath7 , @xmath8 and @xmath9 , respectively , and summation convention is implied .
three of the four parameters in this theory are the wilson parameter @xmath6 , the current quark mass @xmath10 and the effective coupling constant @xmath11 which behaves as @xmath12 with @xmath13 being the qcd coupling constant .
the fourth parameter @xmath3 is in general _ not _ the total chemical potential @xmath14 of the interacting many body system .
as shown below the interaction will induce a contribution to @xmath14 which is momentum dependent .
we shall therefore refer @xmath3 as the `` bare '' chemical potential .
our strategy is to construct an ansatz which obeys the equation of motion corresponding to @xmath15 at finite @xmath0 and use it to calculate the equation of state of cold and dense quark matter . in free space
this ansatz has the same structure as the free lattice dirac field and obeys the free lattice dirac equation with a mass term which is interpreted as the dynamical quark mass @xcite .
temporarily dropping color and flavor indices , this ansatz is given by @xmath16 the annihilation operators for particles , @xmath17 , and anti
particles , @xmath18 , annihilate an interacting vacuum state and obey the free fermion anti commutation relations .
the properties of the spinors @xmath19 and @xmath20 are given in @xcite .
the equation of motion for a free lattice dirac field fixes the excitation energy @xmath21 to be @xmath22 where @xmath23 plays the role of the dynamical quark mass or , equivalently , the chiral gap .
we now introduce temperature @xmath24 and @xmath0 simulatneously into our formalism and take the @xmath25 limit to construct an ansatz for finite @xmath0 .
this is accomplished by subjecting the @xmath17 and @xmath18 operators in eq .
( [ eq : psip ] ) to a generalized thermal bogoliubov transformation as in thermal field dynamics @xcite @xmath26 the thermal field operators @xmath27 and @xmath28 annihilate a quasi particle and create a quasi hole at finite @xmath24 and @xmath0 , respectively , while @xmath29 and @xmath30 are the annihilation operator for a quasi anti particle and creation opertor for a quasi anti hole , respectively .
the thermal annihilation operators annihilate the interacting thermal vacuum state @xmath31 _ for each @xmath24 and @xmath0_. @xmath32 the thermal doubling of the hilbert space accompanying the thermal bogoliubov transformation is implicit in eqs .
( [ eq : avac1 ] ) and ( [ eq : avac2 ] ) where a vacuum state which is annihilated by thermal operators @xmath27 , @xmath33 , @xmath29 and @xmath34 is defined .
since we shall be working only in the space of quantum field operators it is not necessary to specify the structure of @xmath31 .
the thermal operators satisfy the fermion anti
commutation relations just as the @xmath17 and @xmath18 operators in the free space ansatz .
the coefficients of the transformation are @xmath35 , @xmath36 , @xmath37 and @xmath38 , where @xmath39^{-1}$ ] are the fermi distribution functions for particles and anti particles .
they are chosen so that the total particle number densities are given by latexmath:[\ ] ] from eq .
( [ eq : ved ] ) we can evaluate the difference of the vacuum energy densities in the wigner
weyl ( @xmath68 ) and nambu goldstone ( @xmath69 ) phases of the theory .
we find @xmath70 and therefore the true ground state of our interacting many body system is in the phase with broken chiral symmetry .
having solved the gap equation for the dynamical quark mass we proceed to determine the equation of state of cold and dense quark matter .
we present results obtained with a vanishing current quark mass and wilson fermions for @xmath6 = 0.25 to leading order in @xmath56 since in this case we observe both first and second order chiral phase transitions .
the pressure of the cold and dense quark matter is obtained by numerically evaluating the thermodynamic potential density . in figure [ fig2 ]
we plot the pressure as functions of quark number density for @xmath11 = 0.8 and 0.9 . in both cases
we find the pressure to be negative in the broken symmetry phase and therefore the quark fermi sea is mechanically unstable with an imaginary speed of sound .
our conclusion regarding the quark matter stability at finite density is consistent with results obtained using the nambu jona
lasinio model @xcite and the effective instanton induced t hooft interaction model @xcite . in both @xcite and @xcite mean field calculations show that cold and dense quark matter is unstable in the phase with spontaneously broken chiral symmetry and lead the authors to speculate the formation of quark droplets reminiscent of the mit bag model .
in this work we studied strongly coupled qcd in the hamiltonian lattice formalism at finite density using wilson fermions . starting from an effective hamiltonian we constructed and an ansatz which exactly diagonalizes the hamiltonian to second order in field operators for all densities .
this ansatz obeys the free lattice dirac equation with a chemical potential term and a mass term which plays the role of the dynamical quark mass .
this mass and the total chemical potential of the interacting many body system were determined to lowest order in @xmath56 by solving a coupled set of equations obtained from the equation of motion .
we find a first order chiral phase transition with naive fermions while the order of the phase transition can be either first or second for wilson fermions . as in earlier studies using continuum four fermion interaction models
we find that the quark fermi sea is mechanically unstable in the phase with broken chiral symmetry .
our results for the equation of state of cold and dense quark matter should certainly be verified in any future numerical simulation of finite density qcd .
m. creutz , _ lattice gauge theory : a retrospective _ , los alamos archives arxiv : hep - lat/0010047 .
r. aloisio , v. azcoiti , g. di carlo , a. galante and a.f .
grillo , nucl .
b * 564 * ( 2000 ) 489 . j.b .
kogut , d.k .
sinclair , s.j .
hands and s.e .
morrison , _ two colour qcd at non zero quark number density _ ,
arxiv : hep - lat/0105026 .
damgaard , d. hochberg , n. kawamoto , phys . letts . b * 158 * ( 1985 ) 239 .
ilgenfritz and j. kripfganz , z. phys .
c * 29 * ( 1985 ) 79 .
n. bili , k. demeterfi and b. petersson , nucl .
b * 377 * ( 1992 ) 651 .
a. le yaouanc , l. oliver , o. pne and j.c .
raynal , m. jarfi and o. lazrak , phys .
d * 37 * ( 1986 ) 3691 ; * 37 * ( 1986 ) 3702 .
gregory , s.h .
guo , h. krger and x.q .
luo , phys .
d * 62 * ( 2000 ) 054508 .
m. buballa , nucl .
phys . a * 611 * ( 1996 ) 393 .
m. alford , k. rajagopal and f. wilczek , phys .
letts . b * 422 * ( 1998 ) 247 .
j. smit , nucl .
phys . b * 175 * ( 1980 ) 307 .
y. umino , phys .
letts . b * 492 * ( 2000 ) 385 .
h. umezawa , h. matsumoto and m. tachiki , _ thermo field dynamics and condensed states _ , ( north holland , amsterdam , 1982 ) ; h. umezawa , _ advanced field theory _ , ( aip , new york , 1992 ) .
y. nambu and g. jona lasinio , phys . rev . * 122 * ( 1961 ) 345 ; * 124 * ( 1961 ) 246 .
m. asakawa and k. yazaki , nucl .
phys . a * 504 * ( 1989 ) 668 .
also see the review article by s.p .
klevansky , rev .
* 64 * ( 1992 ) 649 . | using an effective strongly coupled lattice qcd hamiltonian and wilson fermions we calculate the equation of state for cold and dense quark matter by constructing an ansatz which exactly diagonalizes the hamiltonian to second order in field operators for all densities .
this ansatz obeys the free lattice dirac equation with a chemical potential term and a mass term which is interpreted as the dynamical quark mass .
we find that the order of chiral phase transition depends on the values of input parameters . in the phase with spontaneously broken chiral symmetry the quark fermi sea has negative pressure indicating its mechanical instability .
this result is in qualitative agreement with those obtained using continuum field theory models with four point interactions . |
x - ray binaries ( xbs ) represent the end - points of stellar evolution , and the xb population of a galaxy is informed by the star formation history of that galaxy .
one way of characterizing such populations is the distribution of their orbital periods ; however , the estimation orbital periods for extra - galactic xbs is extremely challenging , unless they happen to be at high enough inclinations to exhibit dips and/or eclipses .
the most accessible extra - galactic population for period studies is the m31 x - ray transient group .
these are expected to be low mass x - ray binaries ( lmxbs ) with black hole ( bh ) primaries , since the galactic bh lmxbs are nearly all transient @xcite ; transient neutron star lmxbs also exist , but the majority of transients contain bh primaries ( see e.g. * ? ? ? * ) .
x - ray and optical observations of these transients allow us to estimate their size , and therefore period , using the methods described below .
comparing the period distributions of the m31 and galactic bh transient lmxbs may shed light on possible differences in the star formation and binary evolution histories of these galaxies .
* hereafter vpm94 ) found the absolute v band magnitudes of 18 galactic lmxbs to vary between @xmath45 and + 5 , depending on the size of the accretion disk , and x - ray luminosity . to do this
, they required a sample with well known distances , and therefore limited the sample to systems exhibiting x - ray bursts with radius expansion , soft x - ray transients and z - sources .
the eddington limit was assumed to be 2.5@xmath5 erg s@xmath6 .
they found that the optical counterparts to bursters were fainter than those of non - bursters ; this was a first clue that the optical and x - ray luminosities were linked , because bursters tend to have low x - ray luminosities , while the non - bursting z - sources operate near the eddington limit .
furthermore , previous work suggested that @xmath7 was proportional to the orbital period ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) ; this led vpm94 to postulate that the relation between x - ray and optical luminosities ( @xmath8 and @xmath9 respectively ) was caused by x - ray reprocessing in the accretion disk .
the vpm94 toy model suggested that @xmath9 @xmath10 @xmath11 , where @xmath12 is the binary separation .
one might expect @xmath9 to be proportional to @xmath13 , since that is related to the area of the disk assuming that the disc radius is proportional to @xmath12 .
however , the disk temperature decreases with radius : @xmath14 @xmath10 @xmath15 , and @xmath9 @xmath10 @xmath13 @xmath16 where @xmath17 @xmath02 for the observed temperature range in lmxbs ( vpm94 ) .
kepler s third law gives @xmath12
@xmath10 @xmath18 ; vpm94 simplified this by assuming that the masses of the two stars were similar , and defined @xmath19 as @xmath20 .
their empirical relation @xmath21 provided a good match to observations over 3 orders of magnitude in @xmath19 .
however , their sample includes black hole and neutron star accretors , and a mix of different x - ray spectral states .
more recently @xcite examined the correlations between optical / infra - red and x - ray luminosities for neutron star and black hole x - ray binaries separately , also factoring in the masses of the accretor and donor stars ( i.e. @xmath22 and @xmath23 ) .
they found that the broadband bvri luminosity ( @xmath24 ) correlated well with @xmath25 over 8 orders of magnitude in @xmath8 for black hole binaries , and also neutron star binaries in the hard state ; the black hole normalization was @xmath010 times higher than the neutron star normalization , consistent with the expectation that black hole systems have larger disks for a given orbital period because @xmath26 is larger for black holes than for neutron stars .
moss et al .
( 2012 , in prep , hereafter m12 ) focused on the @xmath7 vs. @xmath19 relation for black hole binaries in the high state at the peak of their outbursts , with the aim of providing the best period estimates for x - ray transients . they used x - ray and optical data for transients near their outburst maximum .
their new empirical relation @xmath27 tends to result in a shorter period than the relation described in equation 1 .
we note that the only significant difference between equations 1 and 2 is in the normalization , again caused by the larger disks in the black hole systems .
we have conducted a 12 year campaign to monitor the central region of m31 for bright x - ray transients .
over 130 chandra observations were made over this time , averaging one per month ; no observations are made when m31 is too close to the sun .
bright new transients are followed up with hst acs wfc observations using the f435w filter that approximates the b band .
one observation was made shortly after the transient outburst , and another observation was made @xmath06 months later , if there were no serendipitous coverage in an earlier observation ( see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and nooraee et al . 2012 ) . in this work
, we present our analysis of 12 x - ray transients observed with chandra acis and hst .
herein we present the discovery of 2 new optical counterparts , and provide revised period estimates for all 12 systems .
x - ray transient behaviour appears to be governed by thermal - viscous instability in the accretion disc , with hot and cold stable phases , along with an unstable transitional phase ( see e.g. * ? ? ? * and references within ) .
the outbursts occur when the accretion disc enters the hot stable phase ; however , the resulting x - ray illumination prevents the disc from returning to the cold state until the x - ray luminosity is sufficiently diminished ( * ? ? ? * hereafter kr98 ) .
kr98 showed that we may expect exponential decay lightcurves so long as the the whole disc is irradiated with x - rays ; otherwise , linear decay is expected . for long transient outbursts ,
kr98 expect the decay to switch from exponential to linear after a few e - folding times .
@xcite analyzed the decay lightcurves of various transients , and found good agreement with the theory of kr98 ; furthermore , they found that systems with multiple outbursts tended to exhibit the same type of decay during each outburst .
we provide a journal of hst observations in table [ journ ] . for each hst acs observation
we provide a number , the observation i d , the date , the exposure time , the pointing , and the target ( labeled t1t12 in order of increasing ra ) ; we also provide the closest acis observation , along with the time of observation with respect to the hst observation .
if the transient is in a region that we have already observed , then a dedicated `` off '' observation is unnecessary .
we have examined 90 chandra acis observations of the central region of m31 , spaced over @xmath012 years ; these observations comprise a monitoring programme that is designed to find transient x - ray sources in the central region of m31 .
the roll - angle of each observation was unconstrained , resulting in an approximately circular field of view with 20@xmath28 radius .
we determined the position of each source from a merged 0.37 kev acis image , using the iraf tool imcentroid .
the merged acis image was registered to the field 5 b band image from the local group galaxy survey observations of m31 @xcite , using 27 x - ray bright globular clusters .
we were unable to register the chandra positions with the hst observations directly because only a few of the hst fields contained x - ray bright globular clusters .
after refining the positions of the globular cluster in the chandra and lgs b band images with imcentroid , we found the sky co - ordinates of the new positions with xy2sky , then registered the x - ray positions to the lgs b band image using the iraf task ccmap .
the mean r.m.s .
uncertainty in position due to registration was 0.11@xmath29 in r.a . and 0.09@xmath29 in dec .
we extracted source and background lightcurves and spectra , along with the appropriate response files , for each observation of all 12 sources .
source and background regions were circular , with the same radius ; the radius was optimized for crowding and off - axis angle , and ranged over @xmath0230@xmath29 .
the location of each background region was chosen to be near to the source region , but not so close that it might be contaminated by source flux .
we created long term , calibrated lightcurves for each source , obtaining a 0.310 kev unabsorbed luminosity for each observation .
we used these lightcurves to determine the overall variability of each x - ray source , and the x - ray luminosity at the times closest to each hst observation .
since we expect our transients to be black hole x - ray binaries in the high / soft state , we initially estimated the conversion from flux to luminosity by assuming a disk blackbody emission spectrum with inner disk temperature k@xmath30 = 1 kev , and @xmath31 = 6.7@xmath32 atom cm@xmath6 ( the galactic line of sight column density ) , then determining the unabsorbed 0.310 kev luminosity equivalent to 1 count s@xmath6 .
after correcting for the exposure , vignetting and background , multiplying the source intensity by this conversion factor gave the source luminosity .
source spectra with @xmath34200 net counts were freely fitted with disk blackbody models . when good spectral fits were found for a particular observation of a source
, the parameters of these fits replaced the default parameters when estimating the source luminosity for observations of the source with @xmath2200 photons . for each acs observation , we examined two types of images provided by the stsci .
each observation had contained one drizzled ( drz ) image and 48 flat - fielded ( flt ) images .
the drizzled image is the final product : the flat fielded images are combined , the sky background is subtracted , and cosmic rays are removed ; the image is scaled by intensity ( electron s@xmath6 ) .
the flt images were corrected for instrumental effects ( dark , bias , and flat ) ; however , the sky was not subtracted , and cosmic rays were not removed ; furthermore , the image was scaled to the total number of electrons received over the observation .
we used the drz images to create the difference images described below , but the flt files for photometry since daophot requires an accurate measurement of the sky background .
each drizzled hst observation was registered to the lgs b band image , using the method described in the previous section .
we then made a difference image for each pair of on / off observations . to achieve this
, we first used the iraf task wregister to ensure that the pixels in each image had the same co - ordinate system and orientation .
we then used the farith ftool to create the difference image .
we examined the difference image for each observed variable x - ray source , looking for significant changes within a 3@xmath1 radius , including the uncertainty in the centroid position , and the r.m.s uncertainty in registration of the chandra image to the lgs b band image ; the r.m.s uncertainties in registering the hst images to the lgs image were negligible .
if no counterpart was found in the difference image , then we estimated the 4@xmath1 upper limit to the b magnitude of an undetected star . to do this
, we summed the counts in each of the flt images , within a circle with 2 pixel radius at the position of the x - ray source , for a total of @xmath35 counts over @xmath36 seconds .
the maximum intensity of the counterpart is then 4@xmath37 , which we convert to vega b magnitude via @xmath38 + 25.76.\ ] ] in order to ensure that our b magnitude limits were believable , we ensured that daophot was able to detect several stars with similar or fainter magnitudes in the flt images . for images where
the sky level varied significantly , we restricted the search for similarly faint stars to areas with similar sky levels as the target . unfortunately , we did not have the simultaneous bvri observations to utilize the relation found by @xcite ; however , we did obtain period estimates using the vpm94 and m12 methods . both the vpm94 and m12 methods for estimating the orbital period require the absolute magnitude in the v band .
we assumed that our f435w flux represented the b band flux , and used the relation b@xmath4v = 0.09 that was empirically derived for lmxbs @xcite .
systematic studies of x - ray scattering halos around galactic lmxbs have shown that @xmath39 @xmath0 @xmath31 / 1.8@xmath40 atom @xmath41 @xcite while @xmath39/e@xmath42 @xmath03 ( see e.g. * ? ? ?
* ) ; hence a@xmath43 @xmath0 @xmath44
. finally , the distance modulus for m31 is 24.47 @xcite .
hence @xmath45 when calculating the periods derived from the m12 relation , we were able to include uncertainties in the relation .
we achieved this by randomly altering the @xmath7 of every galactic black hole binary high state in the moss et al ( 2012 ) sample ten thousand times , drawn from a normal distribution scaled by the uncertainties , and obtaining a new @xmath7 vs. @xmath46 @xmath19 relation for each iteration ; these @xmath7 distributions had means and spreads that were consistent with the data .
we then obtained the @xmath46 @xmath19 corresponding to the @xmath7 of each target in every iteration , and derived the mean log @xmath19 and uncertainties for each target by fitting a gaussian to the resulting log @xmath19 distribution .
unfortunately , we did not have access to sufficient information to include uncertainties in the vpm94 relation ; instead , we propagated the uncertainties in @xmath7 and @xmath8 assuming that the relation was fixed .
figure [ btrans ] shows the locations of the 12 transients on a log - scaled b - band image of the central region of m31 , provided by the lgs ( massey et al . , 2006 ) .
table [ translist ] gives the positions of each target , and summarizes the results discussed later in this section ; we provide the 0.310 kev luminosity during the closest acis observation to the hst on observation , our estimated 0.310 kev luminosity at the time of the hst on observation , the b magnitude ( or 4 @xmath1 upper limit ) for the counterpart , and period estimates using the empirical relations of vpm94 and m12 .
we present the @xmath012 year , 0.310 kev unabsorbed luminosity lightcurves of the 12 transients in fig .
. only four of the targets had @xmath34200 photons in the chandra observation that was closest to the hst on exposure : t1 , t2 , t5 , and t9 .
spectral fit parameters are provided in table [ spectab ] for absorbed disk blackbody models .
the t9 spectrum was piled up ; we account for this in noorae et al .
( 2012 , submitted ) .
only t1 apparently experienced greater than galactic absorption , with @xmath31 = [email protected]@xmath40 atom @xmath41 ( uncertainties are quoted at the 90% confidence level ) .
we find that @xmath47 = [email protected] for t1 , and 0.50 for the remaining transients .
vpm94 obtained x - ray luminosities from the ariel v x - ray catalog , and ariel v employed the 210 kev band @xcite .
hence we estimated the 210 kev luminosity of each target at the time of the hst observation .
for those transients with freely fitted spectra , we used the @xmath48 factors provided in table [ spectab ] . for transients with no freely fitted spectra , we assume k@xmath30 = 1 kev , giving @xmath48 = 0.48
. however , it is possible that certain transients were in their low / hard state at the time of the hst on observation ; the transition from low state to high state during the rise of the outburst occurs at a luminosity @xmath4910% eddington @xcite , while the transition from high state to low state during the decay occurs at @xmath02% eddington due to hysteresis ( see e.g. * ? ? ?
* and references within )
. the low state may be characterized by a power law with photon index @xmath50 @xmath0 1.7 ; assuming a power law with @xmath50 = 1.7 rather than a disk blackbody with k@xmath30 = 1 kev results in 210 kev luminosities that are @xmath075% higher , and periods that are @xmath035% shorter .
since 8 of the 12 transients only yielded upper limits to their periods , we present the more conservative upper limits obtained by assuming high state spectra .
we present the difference images for t1t12 in fig .
[ difim ] , showing the difference in intensity between the drizzled images of the on and off observations ; white is positive , and black is negative . the 1@xmath1 and 3@xmath1 uncertainty ellipses are shown , including uncertainties in x - ray position , and x - ray registration to the lgs b image .
north is up , east is left .
each image shows a 5@xmath29@xmath515@xmath29 field .
the background level varied considerably , with the nuclear region being drastically more crowded than the outer regions ; as a result , 4@xmath1 detection limits ranged over b@xmath02629 .
[ [ t1 ] ] t1 ^^ chandra , swift , hst and einstein observations of t1 are discussed in detail by @xcite .
t1 has no acs off image with the f435w filter , because the wfc was not in use at the time .
instead , we used a serendipitous 81 s v band acs / wfc observation ( i d j92ga5naq , 2005 - 02 - 17 , pi crotts ) .
there are many black stars in the difference image , because they are brighter in the v band than in the b band ; however , there are no white stars to signify the transient , and we estimate that b @xmath3428.7 .
the sky background was extremely low , and we detected several stars over the image with b @xmath5228.7 at the 4@xmath1 level in one flt image .
since there was no b band off image , we examined the 8 individual flt images to determine whether there were any detections of stars within the 3@xmath1 ellipse ; this allowed us to discriminate between stars and e.g. cosmic rays .
we found no stellar candidates that were detected in @xmath343 out of 8 flt images , and conclude that there were no bright b band stars within the 3@xmath1 ellipse . for t1 ,
the closest acis observation to the hst on image was obs 7139 ( day @xmath02500 in fig .
2 ) , taken 27 days previously ; however , t1 was not covered by any acis observations for @xmath34100 days either side of the 2006 outburst . in a later outburst ( day @xmath03900 ) ,
the 0.310 kev luminosity went from 8.6@xmath5110@xmath53 to 0.4@xmath5110@xmath53 erg s@xmath6 over 27 days . assuming exponential decay ,
the e - folding time was 8.8 days .
we applied the same exponential decay to the 2006 outburst , and found a 0.310 kev luminosity of 0.5@xmath5110@xmath53 erg s@xmath6 at the time of the hst on observation . for the peak spectrum
, we obtained a best fit k@xmath30 of [email protected] kev ; as a result , @xmath48 = 0.55 .
[ [ t2 ] ] t2 ^^ the t2 outburst ( day @xmath02550 ) lasted @xmath0150 days .
the x - ray peak was during obs 7140 , 32 days prior to the hst on observation .
fitting the lightcurve with exponential decay yielded an e - folding time of 38 days ; however , the residuals were large , and the @xmath54/dof was unacceptable ( 47/3 ) .
we also fitted the lightcurve with power law and straight line models ; the best fit values had worse @xmath54/dof than the exponential fit .
there was insufficient data to estimate the intensity for a model where the decay went from exponential to linear , hence the exponential decay model provided the best estimate of the 0.310 kev luminosity of t2 at the time of the hst on observation . using the best fit e - folding time of @xmath040 days , we obtained a 0.310 kev luminosity of @xmath06@xmath5110@xmath53 erg s@xmath6 for the hst on observation .
the best fit disk blackbody model for the peak spectrum yielded k@xmath30 = 0.54 @xmath30.04 kev ; this results in a 210 kev luminosity of just 1.5@xmath55 erg s@xmath6 .
the positioning of the t2 on observation was off by @xmath341@xmath28 , due to the misidentification of the guide star . as a result , it was impossible to accurately register the flt images to the drizzled image , and this limited our photometric accuracy .
the difference image shows one particularly variable star ; its b magnitude changed from @xmath024.7 to @xmath025.6 between the on and off observations .
we therefore consider this star to be the most likely counterpart . since the our optical counterpart to t2 had a b magnitude @xmath025.6 even after the x - rays had disappeared , we removed this component when calculating the period .
[ [ t3 ] ] t3
^^ t3 was first studied by @xcite , who found no counterpart and concluded that b @xmath3425.5 . however , our method of estimating the magnitude limit for the counterpart yielded b @xmath3427.2 .
t3 exhibited two outbursts , @xmath02000 days apart ; the hst observations were triggered by the first outburst .
this outburst had an observed peak 0.310 kev luminosity of [email protected]@xmath5110@xmath53 erg s@xmath6 , 22 days before the hst on observation ( day @xmath01700 ) ; however , the next observation of t3 was @xmath0150 days afterward , where the 0.3 - 10 kev luminosity had fallen to [email protected] @xmath5110@xmath53 erg s@xmath6 .
if we assume that these observations cover a single exponential decay , then the e - folding time is 68 days .
the 0.310 kev luminosity would then be @xmath02.73@xmath5110@xmath53 erg s@xmath6 at the time of the hst on observation .
[ [ t4 ] ] t4 ^^ t4 was mostly quiescent , but exhibited a @xmath01500 day period of x - ray activity , starting on day @xmath01500 where the intensity varied by a factor @xmath020 .
the hst on observation was triggered at day @xmath02500 .
the 0.310 kev luminosity of t4 was @xmath561.5@xmath55 erg s@xmath6 for the chandra observations closest to the hst on observation , and @xmath00.4@xmath55 erg s@xmath6 at the time of the off observation .
we found no evidence for stellar variability in the difference image of t4 ; we set @xmath57 @xmath3428.9 . [ [ t5 ] ] t5 ^^ t5 was first discussed by @xcite , who found no counterpart , and set a magnitude limit of b @xmath34 24.7 due to the high sky background caused by proximity to the nuclear region .
our careful analysis of the flt images from this observation with daophot improved the magnitude limit to b @xmath34 26.6 .
t5 exhibited a single outburst that was observed for @xmath080 days , starting at day @xmath01750 .
exponential fits to the decay lightcurve were rejected ( @xmath54/dof @xmath5863/3 ) ; however the lightcurve was well described by linear decay with slope @xmath40.06 ( @xmath54/dof = 3.5/3 ) .
the hst on observation occurred 29 days after the observed peak , hence we estimate the 0.310 kev luminosity to be 4.0@xmath55 erg s@xmath6 .
the brightest observation yielded a spectrum with k@xmath30 = [email protected] , yielding a 210 kev luminosity of 2.5@xmath55 erg s@xmath6 .
[ [ t6 ] ] t6 ^^ we discovered from our long - term acis lightcurve of t6 that it is not a transient , but does vary by a factor @xmath5610 .
fortunately , t6 was observed 6 times in our survey , and we were able to make a difference image from the two observations with the highest contrast in x - ray luminosity ( observations 3 and 4 in table 1 ) .
we found no evidence for variability in the difference image within the 3@xmath1 ellipse ; instead , we estimate b to be @xmath3426.3 .
we estimate that the 0.310 kev luminosity of t6 varied by a factor @xmath04 between hst observations 3 and 4 ( as described in table 1 ) , from @xmath00.9@xmath5110@xmath53 to @xmath00.2@xmath5110@xmath53 erg s@xmath6 .
[ [ t7 ] ] t7
^^ t7 exhibited a slow , faint outburst that lasted @xmath0250 days , starting at day @xmath02300 .
the shape of the outburst was unlike the classical fast rise followed by exponential decay ( fred ) ; instead , the lightcurve was approximately flat for the four observations either side of the hst on image , with a 0.310 kev luminosity of @xmath0 0.6@xmath5110@xmath53 erg s@xmath6 .
since the 0.310 kev luminosity of t7 did not significantly vary between its on and off observations , we can not expect to detect optical variability in the counterpart . as a result
, we must set our magnitude limit to that of the brightest star within the 3@xmath1 position uncertainty ellipse , giving b @xmath3425.0 .
[ [ t8 ] ] t8 ^^ t8 was initially quiescent , turned on around day 1400 , and has been active ever since . the 0.310 kev luminosity was @xmath00.4@xmath5110@xmath53 erg s@xmath6 at the time of the hst on observation , and also at the the time of the off observation .
as with t7 , there was no significant difference between the x - ray luminosity of t8 at the times of the hst on and off observations .
hence we set our magnitude limit to be fainter than the brightest star within 3@xmath1 of the x - ray position , i.e. b @xmath3423.9 .
[ [ t9 ] ] t9 ^^ t9 was brightest x - ray source ever seen in m31 , and is discussed in detail in @xcite and noorae et al .
the difference image shows a clear counterpart within the 1@xmath1 region .
b = [email protected] .
no counterpart was observed in the `` off '' image , and b @xmath3427.8 at the 4@xmath1 level ; hence , we assume that all of the optical light comes from reprocessed x - rays .
the closest acis observation was 14 days after the hst on image .
however , the observed peak occurred in a hrc observation , 34 days before the hst on observation .
the decay lightcurve appears to transition from exponential decay to linear decay ( see nooraee et al . , 2012 )
; we estimate the 0.310 kev luminosity to have been @xmath050@xmath55 erg s@xmath6 at the time of the hst on observation . using a disk blackbody emission model , and
correcting for pile - up effects , yielded k@xmath36 @xmath00.8 kev ; as a result , we estimate the 210 kev luminosity to have been @xmath018.6@xmath55 erg s@xmath6 at the time of the hst on observation .
t10 is discussed in detail by @xcite , who used daophot and allstar to disentangle the counterpart from its bright neighbor .
the counterpart dominated the blend for the on observation , but was significantly fainter in the off observation , allowing @xcite to obtain b = 24.90 for the other star . assuming the magnitude of the other star was 24.90 during the on observation yielded b = [email protected] for the counterpart during the on observation , and [email protected] during the off observation .
as with t2 , we subtracted the `` off '' component of the optical emission when calculating the orbital period for t10 .
the t10 outburst was captured in 5 acis observations over @xmath0180 days ; the shape is flat - topped , rather than the typical fast rise followed by exponential decay .
the closest acis observation occurred 1 day before the hst on observation , with 0.310 kev luminosity [email protected]@xmath5110@xmath53 erg s@xmath6 ; given the slow evolution of the outburst , we assume the same luminosity at the time of the hst on observation . the t11 outburst was significantly detected in 5 acis observations over 54 days .
the decay was well described by an exponential with an e - folding time of 33 days ( @xmath54/dof = 2/5 ) .
we calculated the 0.310 kev luminosity to be @xmath00.34@xmath5110@xmath53 erg s@xmath6 .
the t11 difference image revealed a likely counterpart within the 1@xmath1 ellipse .
we found that b = [email protected] for the on observation .
we found no trace of a counterpart in the `` off '' observation ; b @xmath3428.2 at the 4@xmath1 level .
hence , we assume that all of the optical light comes from reprocessed x - rays .
t12 was first discussed by @xcite , who identify a counterpart despite the fact that it is brighter in the off observation than in the on observation ; it appears as a black star in fig .
[ difim ] .
we disagree with this interpretation , since there is no scenario where the disc brightens during decay , and instead impose a 4@xmath1 upper limit of b @xmath34 27.9 .
t12 exhibited an apparently double - peaked outburst over @xmath0200 days , with the second peak brighter than the first .
however , the previous observation to include t12 was @xmath0150 days earlier , meaning that the true peak was not observed .
the 0.310 kev luminosity fell from [email protected]@xmath55 to [email protected]@xmath55 erg s@xmath6 over 61.5 days ; if we assume exponential decay , then the 0.310 kev luminosity was @xmath00.9@xmath55 erg s@xmath6 at the time of the hst on observation .
since the e - folding time for transient decay is @xmath02.2 times longer in the optical than in the x - ray lightcurves , the vpm94 and m12 relations only truly apply to the peak of the outburst , and get more inaccurate as the burst evolves .
we made two sets of period calculations for each source : one set estimating the x - ray luminosity of each transient at the time of the hst observation , and one set estimating the optical luminosity at the time of the x - ray peak .
comparison of the two sets of periods indicated the robustness of these estimates . for t1
, the ns periods agreed within to within 30% , and the bh periods varied by 40% ; the e - folding time was particularly short for this system . for t9
there was a @xmath030% discrepancy between the periods .
for t12 , the periods differed by a factor @xmath03 ; this large discrepancy may be due to the double - peaked nature of the t12 outburst .
the two sets of periods agreed to within @xmath010% for all targets not previously specified .
these results suggest that the observed peaks were generally close enough to the true peaks for the vpm94 and m12 relations to hold .
the periods listed in table [ translist ] are those obtained by estimating the x - ray luminosity at the time of the hst on observation ; we provide 1@xmath1 uncertainties for systems with identified counterparts , and 3@xmath1 upper limits to the periods for systems with magnitude limits .
we note that some of the transients are rather x - ray faint , suggesting that they are more likely to be in the low / hard state , rather than the thermal high state ; the bolometric luminosities of these sources would then be somewhat higher , leading to shorter periods .
however , most of these systems only have upper limits to the period already ; t11 is the only x - ray faint source with a counterpart .
we note that @xcite fitted the acis spectra of t10 despite their low quality ( @xmath0120 net source counts ) .
they favor power law emission models with n@xmath59 @xmath02@xmath40 atom @xmath41 ; they dismissed disk blackbody emission models because the @xmath31 was below the line - of - sight value instead of fixing @xmath31 . as a result , they found l@xmath60 = 6@xmath55 erg s@xmath6 , and @xmath7 = @[email protected] , both a factor @xmath02 brighter than our values ( see table [ translist ] ) ; these results give a @xmath0130 hr period , considerably longer than the @xmath050 hr period that we obtained .
however , we checked these spectra and found @xmath31 to be consistent with the galactic line - of - sight absorption , and we prefer the results from our assumed spectral model . in fig .
[ perdist ] we order the transients by increasing bh period ; the y axis serves to separate them , but does not reflect any physical parameter .
circles and stars represent ns and bh periods respectively , for systems with counterparts ; otherwise , we give the 3@xmath1 upper limits , represented by arrows .
we show the period distribution of 15 milky way bh binaries listed in @xcite for comparison , represented by squares ; the three persistently bright bh hmxbs are excluded .
our period distribution for the m31 transients is similar to that of the galactic bh lmxbs ; however the m31 transients may tend towards shorter periods .
we have not included the period for t8 since its x - ray magnitude did not vary between the hst on and off observations , meaning that our b band magnitude limit was set by the brightest star within the 3@xmath1 positional uncertainty .
this limit is brighter than most of the observed counterparts , and the resulting period limit is not useful . while the x - ray luminosity of t7 did not vary between the hst on and off observations either , the b magnitude limit was fainter than any of the observed counterparts so we have included t7 .
we tested the possibility that the periods of the transients in m31 are consistent with those in the milky way using the tests for univariate data implemented in the asurv package ; this method allows us to compare a censored population , containing some systems with upper limits only , with an uncensored population , where all the values are known , within uncertainties ( see e.g. * ? ? ?
* ; * ? ? ?
we find that the two distributions are inconsistent at the @xmath09298% confidence level . in order to assess the effect of uncertainties on the orbital periods of the m31 systems we repeated the test several times by randomly sampling periods assuming that they follow a normal distribution with the mean and standard deviation calculated from the values in table 2 . in all cases the results give very similar confidence levels .
our sample of m31 transients differs from the mw transients at the @xmath02@xmath1 level ; increasing the m31 transient population should amplify this result .
@xcite discovered an enhanced population of x - ray binaries within @xmath0100@xmath29 of the m31 nucleus .
the radial distribution of `` primordial '' , i.e. expected , lmxbs follows the distribution of stellar mass in the galaxy @xcite ; however , the surplus population appears to follow the @xmath61 distribution expected of dynamically formed binaries @xcite .
dynamically formed lmxbs are normally very rare outside globular clusters ; however , @xcite propose that the bulge of m31 is sufficiently large and dense to allow the formation of a significant number of dynamically formed binaries .
however , the stellar velocities in the m31 bulge are @xmath0510 times higher than in globular clusters ; hence @xcite expect only short period binaries to survive , with most of the survivors containing black holes .
since the number of dynamically formed black binaries is strongly dependent on the distance from the center ( @xmath62 ) , one might expect to observe some correlation between orbital period and radial distance for our transients . in the top panel of fig .
[ rvp ] we plot @xmath62 vs. @xmath63 for each transient , excluding t8 .
the bottom panel of fig .
[ rvp ] shows f435w ( @xmath0b band ) detection limit for each transient ; t7 lies in a diagonally - shaded region because it showed no x - ray variability between the hst on and off observations , hence the b magnitude was limited by the brightest star within the 3@xmath1 location ellipse rather than the local background .
the sensitivity increased steadily with increasing @xmath62 , varying by 2.5 magnitudes ( i.e. a factor 10 ) .
we employed the cox hazard test and the kendal tau test , both implemented in asurv to deal with bivariate data , to compare the population located 20@xmath29 @xmath2 @xmath62 @xmath2150@xmath29 with the population at 150@xmath29 @xmath2 @xmath62 @xmath2350@xmath29 ; the probability that these two populations were drawn from the same population ( @xmath64 ) is @xmath00.12 . limiting @xmath62 to 300@xmath29 , thereby removing t2 , and re - running the cox hazard and kendal tau tests produces the same result .
by contrast , extending @xmath62 to 400@xmath29 for the second group , thereby including t1 , eradicates any trace of a correlation .
therefore , a more substantial transient population is required to confirm this suspected correlation .
x - ray transients are expected to exhibit exponential or linear decay , depending on whether or not the disc is fully ionized ; this depends on the orbital period and peak x - ray luminosity @xcite .
hence we used fig . 1 of that paper to determine whether the decays of our targets should be exponential or linear . with two exceptions ,
all of our targets are expected to exhibit exponential decay ; t7 is consistent with either linear or exponential decay , while t11 should decay linearly according to @xcite . interestingly
the decay of t11 was definitely more exponential ( @xmath54/dof = 2/5 ) than linear ( @xmath54/dof = 93/5 ) ; we suggest that the true peak luminosity of t11 was considerably higher than the observed peak 0.310 kev peak of @xmath04@xmath5110@xmath65 erg s@xmath6 . the true orbital period of t11 could then be significantly shorter than our estimate of 150@xmath330 hr .
furthermore , the t5 outburst appears to have decayed linearly , even though exponential decay is expected ; no observations were made in the @xmath0170 days prior to the observed outburst , hence we may have caught the linear tail of a longer and brighter outburst .
a brighter peak x - ray luminosity for t5 would mean a shorter period ; however , this would not affect our results , since we only have an upper limit for t5 .
we report new orbital period estimates for 12 x - ray transients observed by chandra and hst . we have identified counterparts to 4 x - ray sources , 2 of those are new to this work ;
we have estimated 4@xmath1 upper limits to the b magnitudes of the remaining 8 systems .
period estimates were previously published for 5 sources ( t1 , t3 , t5 , t10 , t12 ) , using the vpm94 empirical relation @xcite , and noorae et al .
( 2012 ) have already applied the m12 relation to t9 .
we obtained periods that ranged from @xmath20.17 hours to @xmath2175 hours for black hole lmxbs using the m12 relation , and @xmath20.4490@xmath390 hours for neutron star lmxbs using the vpm94 relation .
the m31 transients that we have followed up appear to favor shorter periods than the galactic black hole transients ; if all of our transients contain bhs , then the probability for the m31 and mw binaries being drawn from the same population is @xmath20.08 .
this result may be explained by an excess of x - ray binaries dynamically formed in the high density bulge of m31 ; such systems require short periods to survive @xcite .
however , some of the transients with the shortest periods may contain neutron stars .
we searched for a correlation between the period and distance from the nucleus that would further support the hypothesis of @xcite .
we found a hint of a correlation , at the @xmath090% confidence level .
further observations of m31 transients with chandra and hst will add significance to these relations .
we thank the referee for their thoughtful comments that helped clarify this work . this research has made use of data obtained from the chandra data archive , and software provided by the chandra x - ray center ( cxc ) .
is funded by chandra grant go9 - 0100x and hst grant go-11013 .
m.r.g . and s.s.m are both partially supported by nasa contract nas8 - 03060 .
, m. , isobe , t. , & feigelson , e. 1992 , in astronomical society of the pacific conference series , vol . 25 , astronomical data analysis software and systems i , ed .
d. m. worrall , c. biemesderfer , & j. barnes , 245 cccccccccccc obs & i d & date & exp & ra & dec & target & chan obs & sep
. + 1 & j8vp02010 & 2003 - 12 - 25 & 2200 & 00:42:56.03 & + 41:12:19.0 & t10 on & 4680 & + 2 days + 2 & j8vp03010 & 2004 - 01 - 23 & 2200 & 00:42:43.86 & + 41:16:30.1 & t6 on & 4681 & @xmath668 days + 3 & j8vp04010 & 2004 - 06 - 14 & 2200 & 00:42:33.39 & + 41:17:43.4 & t3 on & 4682 & @xmath422 days + 4 & j8vp05010 & 2004 - 08 - 15 & 2200 & 00:42:41.50 & + 41:17:00.0 & t5 on & 4719 & + 29 days + 5 & j8vp07010 & 2004 - 10 - 02 & 2200 & 00:42:56.04 & + 41:12:19.1 & t10 off & 4721 & + 2 days + 6 & j8vp08010 & 2004 - 11 - 02 & 2200 & 00:43:09.93 & + 41:23:32.6 & t12 on & 4721 & @xmath429 days + 7 & j8vp09010 & 2004 - 11 - 22 & 2200 & 00:42:41.49 & + 41:17:05.2 & t5 off & 4723 & + 13 days + 8 & j8vp10010 & 2005 - 01 - 01 & 2200 & 00:43:09.91 & + 41:23:32.5 & t12 off & 4723 & @xmath427 days + 9 & j9ju01010 & 2006 - 02 - 10 & 4360 & 00:42:52.38 & + 41:16:49.2 & t8 on & 7136 & @xmath435 days + 10 & j9ju02010 & 2006 - 07 - 09 & 4360 & 00:42:33.89 & + 41:23:31.1 & t4 on & 7139 & + 22 days + 11 & j9ju03010 & 2006 - 07 - 15 & 4352 & 00:42:51.17 & + 41:19:17.9 & t7 on & 7139 & @xmath416 days + 12 & j9ju04010 & 2006 - 08 - 27 & 4344 & 00:42:16.08 & + 41:19:26.1 & t1 on & 7139 & @xmath427 days + 13 & j9ju05010 & 2006 - 10 - 25 & 2148 & 00:42:17.34 & + 41:15:37.0 & t2 on & 7140 & @xmath432 days + 14 & j9ju06010 & 2007 - 01 - 10 & 4672 & 00:42:52.38 & + 41:16:49.2 & t8 off & 8183 & + 4 days + 15 & j9ju07010 & 2007 - 01 - 12 & 4672 & 00:42:33.88 & + 41:23:31.2 & t4 off & 8183 & + 2 days + 16 & j9ju08010 & 2007 - 01 - 12 & 4672 & 00:42:51.14 & + 41:19:18.0 & t7 off & 8183 & + 2 days + 17 & j9ud17010 & 2009 - 08 - 25 & 4360 & 00:42:17.31 & + 41:15:37.5 & t2 off & 10555 & @xmath455 days + 18 & jb9d15010 & 2010 - 01 - 21 & 4360 & 00:42:53.15 & + 41:14:22.9 & t9 on & 11278 & + 14 days + 19 & jb9d20010 & 2010 - 07 - 20 & 4360 & 00:42:53.15 & + 41:14:22.9 & t9 off & 11840 & @xmath43 days + 20 & jbdl01010 & 2010 - 12 - 09 & 5472 & 00:42:57.00 & + 41:20:05.0 & t11 on & 12162 & @xmath41 day + 21 & j9ju08010 & 2011 - 07 - 20 & 4672 & 00:42:51.17 & + 41:10:17.9 & t11 off & 12972 & @xmath45 days + cccccccccccc targ & ra & @xmath67 & dec & @xmath68 & @xmath69 & @xmath70 & @xmath57 & @xmath7 & @xmath71 & @xmath72 + 1 & 00:42:16.106 & 0.015 & + 41:19:26.36 & 0.27 & [email protected] & 0.3 & @xmath3428.7 & 2.45 & @xmath210.6 & @xmath211.6 + 2 & 00:42:17.330 & 0.012 & + 41:15:37.48 & 0.11 & [email protected] & 2.89 & [email protected] & @xmath40.2 & 120@xmath35 & 16@xmath73 + 3 & 00:42:33.426 & 0.012 & + 41:17:03.56 & 0.14 & [email protected] & 1.59 & @xmath3427.2 & 2.3 & @xmath23.6 & @xmath24.5 + 4 & 00:42:33.883 & 0.016 & + 41:23:31.49 & 0.19 & [email protected] & 0.88 & @xmath3428.9 & 4 & @xmath20.4 & @xmath20.97 + 5 & 00:42:41.805 & 0.010 & + 41:16:35.94 & 0.10 & [email protected] & 2.52 & @xmath3426.6 & 1.7 & @xmath26.0 & @xmath25.6 + 6 & 00:42:43.834 & 0.011 & + 41:16:30.60 & 0.11 & [email protected] & 0.53 & @xmath3426.3 & 1.4 & @xmath235 & @xmath231 + 7 & 00:42:51.199 & 0.012 & + 41:19:17.81 & 0.13 & [email protected] & 0.35 & @xmath3425 & 2.9 & @xmath2380 & @xmath2175 + 8 & 00:42:52.433 & 0.010 & + 41:16:48.75 & 0.10 & [email protected] & 0.29 & @xmath7423.9 & @xmath41 & @xmath22400 & @xmath21000 + 9 & 00:42:53.181 & 0.010 & + 41:14:22.83 & 0.10 & 38@xmath39 & 18.62 & [email protected] & @xmath40.99 & 61@xmath311 & 18@xmath75 + 10 & 00:42:56.049 & 0.012 & + 41:12:18.64 & 0.13 & [email protected] & 1.77 & [email protected] & @xmath40.38 & 163@xmath38 & 9@xmath76 + 11 & 00:42:56.991 & 0.017 & + 41:20:05.35 & 0.31 & [email protected] & 0.2 & [email protected] & @xmath40.03 & 490@xmath390 & 140@xmath77 + 12 & 00:43:09.963 & 0.030 & + 41:23:32.56 & 0.43 & [email protected] & 0.53 & @xmath34 27.9 & 3 & @xmath22.8 & @xmath24.8 + cccccccccccc object & net source counts & @xmath31 / 10@xmath78 atom @xmath41 & k@xmath36 / kev & @xmath54/dof & @xmath79 + t1 & 374 & [email protected] & [email protected] & 10/15 & 0.55 + t2 & 756 & 0.67 & [email protected] & 35/31 & 0.25 + t5 & 330 & 0.67 & [email protected] & 4/12 & 0.59 + t9 & 1135 & 0.67 & 0.8 & 64/49 & 37 + | almost all galactic black hole binaries with low mass donor stars are transient x - ray sources ; we expect most of the x - ray transients observed in external galaxies to be black hole binaries also . obtaining period estimates for extra - galactic transients is challenging , but the resulting period distribution is an important tool for modeling the evolution history of the host galaxy . we have obtained periods , or upper limits , for 12 transients in m31 , using an updated relation between the optical and x - ray luminosities . we have monitored the central region of m31 with chandra for the last @xmath012 years , and followed up promising transients with hst ; 4@xmath1 b magnitude limits for optical counterparts are 2629 , depending on crowding .
we obtain period estimates for each transient for both neutron star and black hole accretors .
periods range from @xmath20.4 to 490@xmath390 hours ( @xmath20.97 to @xmath2175 hrs if all are bh systems ) .
these m31 transients appear to be somewhat skewed towards shorter periods than the milky way ( mw ) transients ; indeed , comparing the m31 and mw transients with survival analysis techniques used to account for some data with only upper limits yield probabilities of @xmath00.020.08 that the two populations are drawn from the same distribution .
we also checked for a correlation between orbital period and distance from the nucleus , finding a 12% probability of no correlation .
further observations of m31 transients will strengthen these results . |
vortices have fascinated the minds of scientists throughout history . beginning with the angular momentum eigenstates of the hydrogen atom , quantum - mechanical vortices emerged as key entities characterizing quantum liquids such as superconductors and superfluid helium @xcite .
more recently , quantized vortices have been observed e.g. in bose - einstein condensates @xcite , quantum degenerate fermi gases @xcite , and in coherent optical @xcite and acoustic fields @xcite .
quantized vortices , known as coherence vortices , have since been discovered in the cross - spectral density and related coherence functions of partially - coherent classical - optical fields @xcite . here
we show that coherence vortices , which are screw - type singular phase defects in glauber s second order degree of quantum coherence , may exist even in systems with only one spatial dimension where orbital angular momentum and conventional quantized vortices can not be defined .
we exhibit a one - dimensional model system in which decoherence @xcite is accompanied by a proliferation of quantized phase vortices in the coherence function associated with the resulting statistical mixture . in a space of such low dimensionality ,
the presence of a statistical mixture is a necessary condition for the existence of coherence vortices .
we outline proposals for experimentally creating and observing coherence vortices and vortex lattices in spatially one - dimensional systems , and speculate the possible connection between coherence vortices and decoherence .
vortical flows , associated with velocity fields @xmath0 , have non - zero circulation @xmath1 here , @xmath2 denotes a smooth closed contour which is traversed once in a positive sense at a fixed time @xmath3 , @xmath4 is an infinitesimal arc length vector pointing along the direction in which @xmath2 is traversed , and @xmath5 is position . for single - valued differentiable pure - state complex scalar wave - functions @xmath6 with phase @xmath7 $ ] we can define a velocity field @xmath8 where @xmath9 is the mass of a particle .
the corresponding circulation integral is quantized in units of @xmath10 : @xmath11 where the integer @xmath12 is termed the topological charge .
this quantization of circulation is a consequence of the single - valued and continuous nature of the function @xmath6 which may describe e.g. an optical or a matter - wave field . the integer @xmath12 denotes the number of circulation quanta and is related to the orbital angular momentum of the particles which is only meaningful in two or more spatial dimensions .
let us now shift attention , from pure states to mixed states .
if one has a mixed - state wave - function , describing an ergodic stochastic quantum system via an ensemble of realizations each of which carry a specified statistical weight , vortical flows may be studied in terms of the circulation integral in eq.([spatialvortexclassicalcirculation ] ) applied to the velocity field @xmath13 , where angular brackets denote ensemble averaging ( obtained , for example , using the density - matrix formalism ) .
we assume there to be no explicit time - dependence in the external parameters and thus the time - independence of @xmath14 follows from the ergodicity of the stochastic process describing the field . the circulation of the averaged velocity field @xmath15 is not quantized in general , even though circulation of the velocity field is quantized for individual members of the ensemble .
however , as shown below , quantized circulations exist for _ coherence functions _ associated with stochastic quantum fields .
moreover , such quantized coherence circulation may be meaningfully defined for fields in one spatial dimension , even though this dimensionality is too low to permit spatial vortical flows .
lastly , we show that , for the one - dimensional model system studied here , a proliferation in the number of coherence vortices and their associated coherence - function zeros is associated with the process of decoherence .
for an ergodic stochastic one - dimensional complex quantum field @xmath16 the normalized two - point equal - time correlation function is given by the following special case of glauber s degree of quantum coherence @xcite @xmath17 where the angular brackets now denote both quantum mechanical and statistical average .
note that @xmath18 is independent of time , on account of the assumed ergodicity of the stochastic process . as a consequence of the single - valued and continuous nature of each complex field in the ensemble
, it follows that @xmath18 is also a continuous single - valued complex function of its arguments .
hence , for any simple smooth positively - traversed closed loop @xmath2 in the two - dimensional space coordinatized by @xmath19 , which is such that @xmath20 is strictly positive for all @xmath21 , the coherence - function circulation integral will be given by @xmath22 \cdot { \bf dl}_\perp= 2 \pi p,\ ] ] where @xmath23 denotes two - dimensional operators and vectors , and @xmath24 is an integer denoting topological charge .
suppose that , for a given closed contour @xmath2 in the @xmath25@xmath26 plane shown in fig .
[ fig1]a , @xmath24 in eq .
( [ coherencevortexquantizedcirculation ] ) has a non - zero value .
then , there exists at least one coordinate pair @xmath27 in the interior of @xmath2 , at which @xmath18 vanishes .
this statement may be proved by contradiction , as follows : the continuity of @xmath18 implies that @xmath28 $ ] will be defined and continuous for every coordinate pair @xmath19 inside and on @xmath2 .
now continuously contract the contour @xmath2 to a point , which coincides with some coordinate pair @xmath29 in the the interior of @xmath2 , via the infinite sequence : @xmath30 which is such that all `` intermediate contours '' @xmath31 etc .
are fully contained within the region bounded by @xmath2 . since @xmath32 can be continuously contracted to a point without changing the value of circulation between adjacent members of the contour sequence in eq .
( [ contractomega ] ) , the non - zero circulation about @xmath2 ( i.e. @xmath33 , for some non - zero integer @xmath24 ) is equal to the zero circulation about an infinitesimally - small contour wrapped around @xmath34 ( this latter circulation vanishes , since @xmath32 is continuous at and in the infinitesimal vicinity of @xmath35 ) .
this logical contradiction implies the falsity of the initial supposition that @xmath36 for every coordinate pair @xmath19 inside @xmath2 .
hence @xmath20 vanishes for at least one coordinate pair @xmath37 inside @xmath2 .
non - zero coherence current circulation @xmath33 is associated with quantized coherence vortices , as a low - dimensional special case of the concept of coherence vortices formulated in the theory of partially - coherent classical - optics fields , as quantified through the mutual coherence function or cross - spectral density @xcite . in light of the theorem proved above , non
- zero @xmath24 is associated with at least one pair of points @xmath37 for which @xmath18 vanishes .
notwithstanding that the term `` coherence vortex '' is employed in the literature , we prefer the term `` decoherence vortex '' since @xmath37 is a pair of points at which the two - point correlation vanishes an absence of coherence indicating a topologically - inevitable zero in the correlation between two spatially - distinct points in a stochastic quantum field .
having noted this preference , we will retain the term coherence vortex to generically refer to screw - type defects in the phase of complex coherence functions such as glauber s two - point correlation function , the mutual coherence function or the cross - spectral density .
the physical meaning of a coherence vortex may be further clarified with reference to fig .
the phase of the complex function @xmath18 is denoted by color in fig .
[ fig1](a ) where the core of the coherence vortex at @xmath38 is labeled by @xmath39 . four locations @xmath40 , and @xmath41 corresponding to pairs of points are marked by crosses along the path @xmath2 , which encircles the phase singularity at point @xmath39 . figure [ fig1](b ) shows a schematic of the interference fringes observed in a young - type double - slit experiment where the two slits are placed at points determined by @xmath40 , and @xmath41 and are illuminated by a spherical wave of light sourced from the corresponding locations @xmath25 and @xmath26 .
the background intensity @xmath42 produced by a completely incoherent source field has been subtracted from the total intensity @xmath43 .
the resulting interference fringes are seen to ` ratchet ' as the coherence vortex core is encircled .
that is , the troughs and peaks of the interference pattern move continuously in one direction ( within the overall fringe envelope ) when a loop around the coherence vortex is traversed . if the two slits are placed at the location of the coherence vortex core @xmath44 the intensity of the fringes disappears , as illustrated in fig .
[ fig1](c ) .
having discussed the elements of coherence vortices in one spatial dimension , we now propose experiments to observe them .
suppose we have a monoenergetic beam of unpolarized photons propagating from negative infinity @xmath45 through vacuum and into a block of birefringent material placed at @xmath46 which can be viewed as a potential step whose height @xmath47 depends on the state of the incoming photons .
the beam is composed of two unentangled polarization states of light @xmath48 and @xmath49 and can be described as a statistical mixture of two pure state wavefunctions @xmath50 and @xmath51 . due to the polarization - dependent propagation in the medium the two components experience different phase evolution and travel through the material with different wave vectors .
the interference of the different wave components facilitates the development of coherence vortex structures .
notice that the light source itself must _ not _ be strictly coherent , as a necessary condition for the existence of coherence vortices in one spatial dimension . by performing a set of interference measurements by varying the locations @xmath25 and @xmath26 of the interferometer arms the phase map of the two - point coherence function @xmath18 may be generated , from which the loci of the coherence vortices may be read off . the photon field
could also be replaced with sound waves to observe acoustic coherence vortices in one spatial dimension .
similarly , we may also consider a matter - wave analog of the electromagnetic field scattering experiment , where the beam of photons would be replaced by , e.g. , an electron beam or a stream of cold atoms such as those in an atom laser beam sourced from a bose - einstein condensate . in the case of electrons ,
the spin - up @xmath52 and spin - down @xmath53 states of the electrons provide the source of the required mixed states and an external magnetic field could be used for creating the polarization dependent potential . in the case of an atomic beam two or more internal hyperfine spin states
would facilitate the mixing of states and the required potential could be obtained using optical or magnetic coupling to the atomic dipole moments . in both cases the measurement of coherence vortices could be achieved via a two - point interference experiment . turning to the theoretical description of the above and related physical systems ,
consider a mixed quantum state composed of two ensemble members such that @xmath54 less normalization . for incoming plane - wave states ,
see fig .
[ fig4 ] , which scatter from a potential step of height @xmath47 transparent to the first species @xmath55 , solving the schrdinger equation we obtain @xmath56 where @xmath57 and @xmath58 are the wavevectors on each side of the potential step located at @xmath46 and @xmath59 are the transmission and reflection coefficients , respectively .
the energy @xmath60 of the incoming particles of mass @xmath9 and the height of the potential step @xmath47 are linked to the wave vectors by the formulae @xmath61 phase singularities can be produced by an interference of three or more plane waves @xcite .
guided by this fact and using complex phasors @xcite we solve for the zeroes of @xmath18 in regions @xmath62 , to find the locations of the quantized coherence vortices .
we obtain two interlaced lattices determined by lattice vectors @xmath63 : @xmath64 where @xmath65 and @xmath66 are integers , @xmath67 and @xmath68 and the cartesian shifts are determined by the angles @xmath69 these two lattices correspond to , respectively , coherence vortices and coherence antivortices , determined by the sign of the coherence circulation @xmath70 . in fig .
[ fig5](a ) we have plotted the coherence density @xmath71 scaled by its maximum value . the coherence vortices and antivortices have been marked by circles and crosses , respectively , at locations determined by eq .
( [ hila ] ) .
figure [ fig5](b ) shows the corresponding phase plot @xmath72 $ ] from which it can be verified that the loci of phase singularities correspond to the cores of the coherence vortex lattice .
the hermitian property of @xmath18 is inherited by the coherence vortex lattice .
numerically , it is possible to construct the function @xmath18 for an arbitrary potential shape and to find the resulting coherence vortices and antivortices . for
such general potential landscapes , the coherence vortices and antivortices will not lie on a regular lattice .
we emphasize that if the density matrix for one - dimensional systems can be written in a pure - state form , coherence vortices are absent .
this follows from the fact that we need a minimum of three phasors to produce phase singularities via plane - wave interference .
conventionally , quantized vortices have been studied as phase singularities in the medium " supporting them , e.g. superconducting order parameter , complex electromagnetic field , or macroscopic wavefunction describing a matterwave of degenerate quantum liquids .
here we have considered quantized vortices emerging in the coherence or correlation function derived from the quantum states describing the system .
such phase singularities , known as coherence vortices , are topologically enforced zeroes in the two - point coherence function .
they appear as a field interference and hence may emerge even when there are no conventional quantized vortices in the field itself . in particular , coherence vortices can exist in systems with one spatial dimension only , as shown here , where conventional vorticity is manifestly absent .
we have described the physics of coherence vortices and have proposed experiments to observe them .
we have shown using an analytically soluble model how a light or matter - wave field scattering from a step potential creates a stationary coherence vortex and antivortex lattice .
decoherence is considered to be a mechanism for the way classical realism emerges from a quantum coherent system description in terms of mixed - state density matrices .
we speculate it to be possible that decoherence in a quantum system leaves a topological imprint in terms of coherence vortices , which could be used for detecting the amount of decoherence in a system .
this possibility is suggested by the fact that , in the model one - dimensional system studied here , coherence vortices are absent from the pure - state systems and only emerge when one has at least the two - member statistical mixture given by eq .
( [ eqg1 ] ) .
further work is also required to clarify whether the theory of coherence vortices may be connected with the proliferation of thermally activated vortex - antivortex pairs @xcite and quasi - coherence in the observed superfluid to normal transition in cold quasi - two - dimensional bose gases @xcite .
finally , the presence of coherence vortices might signal some new physical material property in analogy with the way conventional quantized vortices are inherently linked together with the phenomena of superfluidity and superconductivity .
90 r. j. donnelly , _ quantized vortices in helium ii _ , cambridge university press ( cambridge 1991 ) .
m. r. matthews , b. p. anderson , p. c. haljan , d. s. hall , c. e. wieman , and e. a. cornell , phys .
lett . * 83 * , 2498 ( 1999 ) . m. w. zwierlein , j. r. abo - shaeer , a. schirotzek , c. h. schunck , and w. ketterle , nature * 435 * , 1047 ( 2005 ) . c. tamm and c.o .
weiss , j. opt .
b * 7 * , 1034 ( 1990 ) .
b. t. hefner and p. l. marston , j. acoust .
* 106 * , 3313 ( 1999 ) .
h. f. schouten , g. gbur , t. d. visser , and e. wolf , opt . lett . * 28 * , 968 ( 2003 ) .
g. gbur and t. d. visser , opt . commun . * 222 * , 117 ( 2003 ) .
g. gbur and t. d. visser , opt
. commun . * 259 * , 428 ( 2006 ) .
w. wang , z. duan , s. g. hanson , y. miyamoto , and m. takeda , phys .
lett . * 96 * , 073902 ( 2006 ) .
w. wang and m. takeda , phys .
rev . lett . * 96 * , 223904 ( 2006 ) .
g. gbur , t. d. visser , and e. wolf , opt .
* 239 * , 15 ( 2009 ) .
y. gu and g. gbur , opt
. commun . * 282 * , 709 ( 2009 ) .
m. l. marasinghe , m. premaratne , and d. m. paganin , opt
. expr . * 18 * , 6628 ( 2010 ) .
m. l. marasinghe , d. m. paganin and m. premaratne , opt . lett . * 36 * , 936 ( 2011 ) .
w. h. zurek , rev .
75 * , 715 ( 2003 ) . c. h. bennett and d. p. divincenzo , nature * 404 * , 247 ( 2000 ) .
r. j. glauber , phys .
rev . * 130 * , 2529 ( 1963 ) .
k. w. nicholls and j. f. nye , j. phys .
a : math . gen . * 20 * 4673 ( 1987 ) .
j. masajada and b. dubik , opt .
commun . * 198 * , 21 ( 2001 ) .
k. oholleran , m. j. padgett , and m.r .
dennis , opt .
* 14 * , 3039 ( 2011 ) .
d. m. paganin , _ coherent x - ray optics _ , oxford university press , ( oxford 2006 ) .
g. ruben , d. m. paganin , and m. j. morgan , phys . rev .
a * 78 * , 013631 ( 2008 ) .
g. ruben , m. j. morgan , and d. m. paganin , phys .
lett . * 105 * , 220402 ( 2010 ) .
t. p. simula and p. b. blakie , phys .
lett . * 96 * , 020404 ( 2006 ) .
z. hadzibabic , p. krger , m. cheneau , b. battelier , and j. dalibard , nature * 441 * , 1118 ( 2006 ) .
p. clad , c. ryu , a. ramanathan , k. helmerson , and w. d. phillips , phys .
* 102 * , 170401 ( 2009 )
. s. tung , g. lamporesi , d. lobser , l. xia , and e. a. cornell , phys .
lett . * 105 * , 230408 ( 2010 ) .
. hung , x. zhang , n. gemelke , and c. chin , nature * 470 * , 236 ( 2011 ) . | coherence vortices are screw - type topological defects in the phase of glauber s two - point degree of quantum coherence , associated with pairs of spatial points at which an ensemble - averaged stochastic quantum field is uncorrelated .
coherence vortices may be present in systems whose dimensionality is too low to support spatial vortices .
we exhibit lattices of such quantum - coherence phase defects for a one - dimensional model quantum system .
we discuss the physical meaning of coherence vortices and propose how they may be realized experimentally . |
mira ( o ceti ) is the prototype of the class of mira - type variable stars , characterized observationally by large visual photometric pulsation amplitudes and regular periods between 100 and 1000 days .
mira - type variables represent one of the late stages of evolution for 1 - 2 solar mass stars , with luminosities of several thousand times solar and mass loss rates of order @xmath2 solar masses per year .
the winds from mira - type variables are relatively slow ( typically @xmath310 kms@xmath4 ) , which means that the mass lost in these winds can be readily accreted onto any stellar companions .
it is known that 57% of the progenitors of mira - type variables , g and f dwarfs , have a companion with a mass more than a tenth of the primary mass @xcite .
however , most of the companions around mira - type primaries remain undetected because of the overwhelming flux of the mira - type primary and the difficulty in making precise radial velocity measurements due to the stellar pulsation .
the most reliable ways to detect companions are through the signs of wind accretion : either a short - wavelength ( e.g. ultraviolet ) excess or rapid photometric variability @xcite .
mira b was first discovered visually in 1923 , when it was at a separation of 0.9 @xcite , and has since been imaged in x - rays , ultraviolet and the radio @xcite .
observations of the mira ab system with the infrared spatial interferometer also offered a tentative detection of flux from near the position of mira b at 11.17@xmath1 m , interpreted as a clump of dust @xcite .
previous attempts at deconvolution of mid - infrared images @xcite showed a coma - like elongation in the direction of mira b and a feature at @xmath0twice the mira ab separation , but quantitative interpretation of these data was difficult due to the difficulties in de - convolving seeing - limited images .
the nature of mira b has been and remains controversial , due to the radiation originating from mira b being dominated by accretion . according to the arguments originally made by @xcite and reinforced by @xcite , mira b must be a low - mass main sequence star , as the x - ray and total luminosity of mira b is far too weak for the companion to be a white dwarf .
this argument relies on measurements of the mira b wind , reasonable values for the true ( i.e. non projected ) separation of mira ab and the bondi - hoyle accretion rate being correct within a factor of 10 .
this paper presents novel non - redundant interferometry observations of mira ab , enabling robust detection of mid - infrared flux from the vicinity of mira b. these and other traditional imaging observations are presented in detail in section [ sectobservations ] .
an examination of possible models for this emission is presented in section [ sectmodelling ] and the implications of the preferred ( disk ) model presented in section [ sectdiscussion ] .
in the mid - infrared , observations on a 10 m class of telescope can be nearly diffraction limited in good seeing conditions . with a 15 cm atmospheric coherence length at 550 nm ,
a 10 m aperture is two coherence lengths wide at 10@xmath1 m . in this regime
, the image of a point source usually has a single bright core , but deconvolution is difficult due to changes in seeing and changes in low order aberrations with changing elevations .
we first overcame these difficulties by individually re - pointing and re - phasing 4 sets of 6 segments out of the 36 separate segments that made up the primary mirror of the keck i telescope .
as part of an observing program designed to resolve the dusty winds from mira variables , these `` segment tilting '' observations of @xmath5 ceti were made at 8 , 9.9 , 10.7 and 12.5@xmath1 m .
these four sets of segments formed 4 interferograms on the long - wavelength spectrometer ( lws ) camera , each of which gave 15 visibility amplitude and 20 closure phase ( 10 independent ) measurements on the source . this technique is described in @xcite and @xcite . at 10.7@xmath1 m , data was also taken in a configuration with 3 sets of 7 segments with non - redundant spacings .
calibration observations were made of @xmath6 cet and @xmath6 tau to calibrate visibility amplitude and subtract out any instrumental closure - phase .
the macim algorithm : a monte - carlo technique for imaging from closure - phase and squared visibility @xcite was used to create images at all wavelengths . as part of this imaging process
, we incorporated prior knowledge that a set fraction of the flux comes from a compact central source ( i.e. the central star mira a ) with a 50mas diameter .
this size comes from the longer - baseline observations of @xcite .
we estimated the fraction of light from this 50mas source by assuming simple black - body emission , based on the source containing 75% of the 8@xmath1 m flux .
this step had little effect on flux near the location of mira b , but affected the deconvolution of the central source and dust shell , due to the documented tendency for techniques in the maximum - entropy family to spread flux over the image at the expense of the compact central source .
unambiguous detections of flux near the location of mira b were made at 12.5 and 10.7 microns , with a lower signal - to - noise detection at 9.9 microns .
the 10.7 and 12.5 micron images are shown in figure [ figkeckims ] .
the separation and position angle of mira b with respect to mira a was obtained for the 10.7 and 12.5@xmath1 m data by searching for a peak in the difference image between the reconstructed image and the azimuthally - averaged reconstructed image with mira b masked - out .
flux measurements of mira b were obtained from aperture - photometry on this differenced image .
we also attempted to simultaneously reconstruct a dust shell image and perform a binary fit to the fourier data using macim , obtaining consistent results and an independent estimation of errors on the binary parameters directly from macim s markov chain .
the average separation and position angle from these fits is @xmath7and @xmath8 degrees .
we independently calibrated the camera plate scale and orientation by observing binaries xy per and @xmath6 her , giving results in agreement with previous calibrations @xcite .
plate scale uncertainties are included in the above error estimates . to obtain comparable astrometry for the ultraviolet position of mira b , we extrapolate based on a linear fit between @xcite and the 2004 february hst stis o8wy02040 data sets .
projected to jd2453248 , we find a separation of @xmath9 at @xmath10 degrees . in comparing the ultraviolet and mid - infrared detections ,
the position angles are found to be consistent . however , the difference in separation is significant , so the thermal ir and uv emission from mira b must arise from different physical regions . to confirm this result
, we obtained additional data using an aperture - mask in the t - recs camera on gemini south ( program i d gs-2006b - q-15 ) .
the mask is installed in the cold lyot stop wheel of t - recs , decreasing the background significantly from the keck experiment , where the full telescope pupil contributed to the background in all interferograms .
this mask has 7 holes in a non - redundant configuration , amounting to 21 simultaneous visibility amplitudes and 35 closure - phases ( 15 independent ) .
baseline lengths projected on the primary mirror range from 1.78 to 6.43 m , and sky rotation is used to increase the fourier coverage while maintaining the same relationship between positions on the mask and positions on the primary mirror .
the masking observations of mira ab used @xmath11 eri , @xmath6 ceti and @xmath6 tau as interferometric calibrators .
these data were obtained in november 2006 at 7.7@xmath1 m , 9.7@xmath1 m , 12.3@xmath1 m and 18.3@xmath1 m . due to a greater instrumental stability ,
these data were of better quality than the keck data , although the resolution was slightly less and as mira a was near minimum , the contrast between mira a and the dust shell was lower .
images reconstructed using macim are shown in figure [ figgeminiims ] - again showing a clear feature in the vicinity of mira b in three out of four filters .
astrometry and flux estimates were obtained in a similar way to the keck data , with fits summarized in table [ tabbinaryfits ] .
flux ratios and astrometry were consistent between the keck and gemini data .
in addition to these mid - infrared aperture - masking observations , we analyzed earlier keck / lws observations of @xmath5 ceti from september 1999 and august 2002 made without the use of a mask or segment - tilting . rotating the images 180 degrees and
subtracting enabled most of the primary psf to be removed .
again , a bright feature in the vicinity of mira b was detected at a high level of confidence .
the contrast ratios obtained from these data were [email protected] in september 1999 ( jd 2451423.6 ) and @xmath13 in august 2002 ( jd 2452506.6 ) .
the errors were estimated by applying the same analysis method to 6 calibrator star observations .
other observations at wavelengths between 0.97 microns ( hst , nicmos data set n4rk02otq ) 3.08 microns ( aperture - masking ) and 4.67 microns ( direct imaging ) did not detect the companion , but were able to place 2-@xmath14 upper limits on the contrast ratio of 500:1 , 700:1 and 700:1 respectively .
all observations reported in this paper as well as previously published hst observations , are summarized in table 1 , with the spectral - energy distribution shown in figure [ figsed ] .
in addition to these observations , the paper of @xcite provided model - fits to infrared spatial interferometer ( isi ) observations with a dust ` clump ' near the location of mira b. they reported the separation and position angle of this clump to be 0.55@xmath15 and 120 degrees , but with no estimate of uncertainty due to the limited uv coverage of the array .
the flux ratios of the clump were not given , although the dust has an optical - depth near unity and a linear size of 17 au .
lllrrrrrr 0.346 & foc & 2450063.3 & 0.70 & [email protected] & 578@xmath122 & [email protected] & 10.14 m & 10.23 m + 0.501 & foc & 2450063.3 & 0.70 & [email protected] & - & - & 10.48 m & 11.30 m + 0.374 & stis & 2453051.8 & 0.65 & [email protected] & 546@xmath123 & [email protected] & 11.07 m & 11.95 m + 0.97 & nicmos & 2451037.3 & 0.62 & @xmath30.002 & - & - & - & - + 3.08 & nirc & 2453252 & 0.24 & @xmath30.0014 & - & - & - & - + 4.67 & nirc2 & 2453909.5 & 0.22 & @xmath30.0014 & - & - & - & - + 18.75 & lws & 2451423.6 & 0.77 & [email protected] & 476@xmath1220 & 99@xmath122 & 1050j & 107@xmath1228j + 17.65 & lws & 2452506.6 & 0.02 & [email protected] & 511@xmath1234 & 96@xmath124 & 1020j & 112@xmath1235j + 10.7 & lws & 2453248 & 0.24 & [email protected] & 470@xmath1214 & [email protected] & 4190j & 92j + 12.5 & lws & 2453248 & 0.24 & [email protected] & 489@xmath1214 & [email protected] & 2450j & 51j + 7.9 & t - recs & 2454053.6 & 0.65 & @xmath160.005 & - & - & 1780j & - + 9.7 & t - recs & 2454053.6 & 0.65 & [email protected] & 480@xmath1230 & 101@xmath123 & 2040j & 51j + 12.3 & t - recs & 2454053.6 & 0.65 & [email protected] & 467@xmath1230 & 104@xmath123 & 1410j & 38@xmath1213j + 18.3 & t - recs & 2454055.6 & 0.65 & [email protected] & 487@xmath1230 & 105@xmath123 & 1690j & 110@xmath1235j + + [ tabbinaryfits ]
in order to find a preferred model for the mid - infrared emission coming from the region around mira b , we must explain the time - dependent astrometry and the total flux ( i.e. the energetics ) of the emission .
we will now examine several hypotheses in turn .
if the radiation near mira b were re - processed radiation from mira a , then the flux would change with the mira variability cycle .
there is a decrease in flux at the most temperature - sensitive wavelength of 10@xmath1 m for the feature near mira b occurring over the interval between the keck ( lws ) and gemini ( t - recs ) observations .
this decrease co - incides with a decrease in the flux from mira a due to the mira variability cycle ( see table [ tabbinaryfits ] ) , consistent with the re - processsed radiation hypothesis .
this is further supported by the modeling of @xcite , where a constant amount of dust was used to model observations at several epochs .
one possible origin is therefore re - radiated flux from an unusually dense clump in the wind from mira a. the likelihood of a clump of dust in the wind causing the mid - infrared emission is not great , as an optical - depth of order unity is required to produce enough emission . in turn , this requires a wind over - density of at least 25 ( based on full silicate condensation at solar metallicity )
. we can be slightly more sophisticated and model this dust clump as a uniform slab , assuming that the dust is made from a combination of fully - condensed corundum and olivine at solar metallicity .
we take the optical constants for olivine from @xcite and those of corundum from @xcite and calculate dust opacities as in @xcite . from three - color photometry
, we can derive a column density , a temperature and an angular extent as the radius of an equivalent uniformly - illuminated disk . for the near - maximum lws data , we obtain 0.28g@xmath17 , 310k and a disk radius of 59mas , and for the near - minimum t - recs data we obtain 0.2g@xmath17 , 240k and 96mas .
these temperatures are also consistent with the assumption of heating of the dust by mira a for moderate ( roughly 45 degree ) angles between the clump to the line - of - sight .
these column densities can be compared to those of a 100au deep column in the time - averaged wind at the @xmath050au projected separation , which totals only 0.007g@xmath17 .
these calculations are somewhat dependent on the assumed and unknown dust opacities , but they serve to provide plausible physical model parameters for the emission .
the large required column densities are due to the emission in the 12@xmath1 m bandpasses being brighter than expected from optically - thin emission in the 10 and 18@xmath1 m bands .
alternatively , one can think of the emission at 10 and 18@xmath1 m as being suppressed due to the dust being optically - thick .
the modeling of the fourier data in @xcite put a dust ` clump ' near the location of mira b between 1990 and 1994 .
the position angle of 120 degrees and separation of 0.55@xmath15 places this emission inside the early-1990s position of mira b but outside the position of the dust clump reported in this paper .
therefore , not only is this clump of dust inexplicably over - dense , but its apparent motion is in the opposite direction to the outflow from mira a , which should move outwards at 14mas per year according to the outflow measurements of @xcite with an assumed distance of 107pc .
this would mean that if the dust ` clump ' seen by keck and gemini moved with the mira a outflow , it would have been at @xmath18 during the @xcite epoch . despite the simple modeling in that paper , it is not plausible that the clump separation was overestimated by a factor of 1.8 , or that in addition to the 0.55@xmath15 clump , a ( brighter ) @xmath00.31@xmath15 clump was missed in the analysis . therefore , due to both the anomalously high optical depth of dust near mira
b and the motion of the emission counter to to the prevailing wind from mira a , we conclude that the mid - infrared emission is inconsistent with a clump in the wind from mira a. although the variability of the mid - infrared flux near mira b is consistent with re - processed radiation from mira b , the phase coverage and signal - to - noise of the data in table [ tabbinaryfits ] is clearly inadequate to exclude processes independent of mira a s radiation field .
therefore , we will check if the energy involved in the wind interaction region is enough to explain the mid - infrared emission .
@xcite model the early 2004 outflow from mira b as a 450kms@xmath4 wind losing @xmath19yr@xmath4 .
this wind is variable , with evidence that it was weaker in the late 1990s and stronger in the iue era @xcite .
as this wind encounters the more dense wind from mira a , it will lose it s energy in a shock front .
this shock luminosity ( @xmath20 ) will be @xmath21 , which is a tiny fraction of the mid - infrared luminosity coming from near mira b , which is @xmath22 .
thus , if the mid - infrared emission were to originate from the wind - interaction region , it would have to be due to a quasi - static increase in the density of mira a s wind due to the wind from mira b. according to the model of section [ sectclump ] , this region has to have at least @xmath23 solar - masses of material , over a length scale of @xmath24au .
20au is roughly the maximum distance over which the flux would still be largely unresolved in our images , once projection effects are taken into account . the simplest geometry in which to consider the outflow from mira b is one where the wind from mira b is largely perpendicular to the wind from mira a. in this geometry , the mass excess due to the slowing of mira a s wind by mira b s wind is of order @xmath25 over length scales @xmath26 , which is @xmath27 , or 6 orders of magnitude too little material .
the outflow from mira b may be nearly spherically - symmetric , including a component as just described , as well as a region where the two winds collide nearly head - on , possibly forming a stagnation point .
calculation of the stagnation point position is a simple balance of dynamic ( ram ) pressures @xmath28 , where @xmath29 is the wind density and @xmath30 the wind speed ( for either mira a or mira b ) .
this stagnation point is significantly influenced by the gravitational field from mira b : both by an change in density @xmath29 of mira a s wind and an increase in velocity .
we have taken these effects into account using newtonian physics , and calculated the stagnation point position .
with mira b wind parameters of @xmath31@xmath32 at 450kms@xmath4 , mira a wind parameters of @xmath33@xmath32 at 6.7kms@xmath4 @xcite a mira ab distance of 90au and a mira b mass of 0.7m@xmath34 ( see section [ sectdisk ] ) , this stagnation point occurs only 1au from mira b ( projection effects not taken into account ) . an upper limit for this distance is the 3.7au calculated in @xcite for the stronger iue era wind , with different assumed wind and orbital parameters .
therefore , a large build - up of dust at this stagnation point could not form a region geometrically large enough to provide the mid - infrared flux .
t tauri stars are well known to have significant mid - infrared excesses . as accretion onto mira
b occurs through a disk , we can also ask if the emission is consistent with expectations from known and well - characterized accretion disks . by comparing the emission of the mira b system to t tauri stars ( i.e. systems with significant accretion luminosity ) as in figure
[ figsed ] , it is clear that although the blue and ultraviolet accretion luminosity is typical , the very large mid - infrared flux from mira b is anomalous by more than a factor of 50 .
therefore , this emission must be due to a mechanism atypical of isolated systems with a disk and outflow . furthermore , the problem of the in - phase variation with the mira a primary presents an additional stumbling block to this model .
we will now examine a model of the mid - infrared emission near mira b that is caused by illumination of an accretion disk around mira b by radiation from mira a. in this section we will not discuss the possibilities of this disk being around a white dwarf or a main sequence star . this disk will be supposed to be in the plane of the mira ab orbit , so that one edge is illuminated and the opposite edge is in shadow .
this co - planar geometry is expected if angular momentum is preferentially accreted in the plane of the orbit .
it is illustrated in figure [ figquickdiagram ] .
first , we shall review the theory of wind accretion as it applies to mira ab .
the accretion of the wind from mira a onto mira b is given by the formulae of bondi - hoyle - lyttleton accretion ( e.g. * ? ? ?
* ) . in its simplest form
, the wind is gravitationally focused into a wake behind the accretor , and the wake is accreted wherever it is gravitationally bound to the accretor .
this gives an accretion radius , which is the radius of a circular area that corresponds to the effective cross - section of the accretor : @xmath35 here @xmath36 is the accretor ( i.e. mira b ) mass , @xmath30 the wind velocity and @xmath37 the orbital velocity .
this formula is most applicable to highly supersonic flows , which is the case for mira b accreting the wind from mira a. this accretion radius can be used to find an accretion rate : @xmath38 the accretion luminosity is then given by the simple formula @xmath39 , where @xmath40 is the radius of the companion .
this formula assumes an equilibrium situation , which may not be the case if material is accumulating within the disk . in order to test hypotheses
that the accretion disk around mira b has been directly observed , we should first examine some basic parameters of the system .
the best estimate for @xmath41 is @xmath42 per year , with @xmath43kms@xmath4 @xcite .
this mass - loss rate is based on a co to h@xmath44 ratio of @xmath45 which is in turn based on a near - solar metallicity .
if mira a has sub - solar metallicity as expected from its kinematic association with the thick - disk population , then this mass - loss rate will be underestimated .
the orbit of mira b is poorly known due to it s long period , but if we assume a circular orbit and fix the total system mass to 1.5@xmath46 , then the astrometry of @xcite combined with the most recent hst epoch ( in table [ tabbinaryfits ] ) gives a crude orbital radius of @xmath47 au , an inclination is 63 degrees and a period of 610 years . changing the assumed total system mass to 2.0@xmath46 only changes the orbital radius to 85 au and the inclination to 66 degrees . for the total mass 1.5@xmath46 orbit
, we have an orbital velocity @xmath48kms@xmath4 and an accretion radius in au of 32@xmath49 according to equation [ eqnbhradius ] .
the maximum size of the accretion disk around mira b will be approximately the bondi - hoyle accretion radius , as beyond this radius the wind from mira a will have enough energy to both escape from the gravity of mira b and transfer some momentum onto material orbiting around mira b. conversely , the minimum accretion radius is the size of the disk measured from the difference in astrometry between mira b and the mid - infrared emission . in the plane of the sky , the mid - infrared emission is separated from mira b by @xmath50au .
the error in this astrometric difference is difficult to quantify in individual images , particularly given the possibility that asymmetries in the atmosphere of mira a could move its centroid around by of order 1au .
however , even if we do not count the @xcite result as a single measurement , then we have 6 independent measurements of the astrometric difference in table [ tabbinaryfits ] that give the same sign and a similar magnitude .
determination of the average separation between mira b ( from the uv ) and the mid - infrared emission therefore appears statistically robust with reliable errors .
the true separation of mira b and the mid - infrared emisson is then @xmath51au , given that the vector between mira a and b is at 38 degrees to the line - of - sight from our preferred orbit .
this is consistent with a disk @xmath52 au in size once the effect of the curved geometry of the disk edge is taken into account .
this disk size is is consistent with bondi - hoyle accretion theory , predicting an accretion radius range from 16 to 22au for @xmath36 between 0.5 and 0.7 @xmath53 .
importantly , the disk size shows that the accretion radius can not be under - predicted using bondi - hoyle accretion theory by more than a factor of @xmath02 .
we must also establish that the amount of mid - infrared emission seen from the edge of the disk around mira b is consistent with the observed emission . already in section [ sectclump ]
we have seen that the emission is consistent with a region of cross - sectional area about 100 square au at temperature 310k ( from the slab model with a radius of 59mas ) .
if mira b is on the far side of mira a as viewed from earth , we see the edge of the disk that is directly illuminated by mira a. if the disk edge has a temperature of 310k , then its hydrostatic equilibrium density distribution is gaussian with full - width half - maximum ( fwhm ) of 1.5au at a 10au radius .
this disk of diameter 20au would only give a 100 square au cross - sectional area if the disk remained optically - thick to radiation from mira a out to several times the density fwhm .
however , given the significant uncertainties and simplifications in the slab model ( including the opacities themselves ) , this emission is certainly consistent with an origin in a side - illuminated accretion disk . note also that the spectral energy distribution shown in figure [ figsed ] is not consistent with the optical depth of unity temperature being the same throughout the mid - infrared : the sillicate emission peak and the upper limit at 7.9@xmath1 m ( from the near - minimum gemini epoch ) demonstrate that the outer surface of the disk has to be heated .
this is an essential part of any physical model of the disk , where radiative energy flows inwards .
given that we have established that mira b has a @xmath010au radius dusty disk around it , we should establish whether the term ` protoplanetary ' can be applied to it .
this term implies that there is or will be enough material in the disk to form planets .
according to the prescription of @xcite , the minimum - mass solar nebula had 2 jupiter masses out to a radius of 10au .
the total mass accreted onto mira b during the lifetime of mira a is of order 7 jupiter - masses , assuming that mira b has 0.7 solar - masses and mira a loses 0.6 solar masses while on the asymptotic giant branch .
the ability for such a disk to form planets clearly depends on a number of factors , such as the role of viscosity in the disk and the maximum accretion rate that will occur as mira a s wind transitions to a radiatively driven wind with a @xmath020 times increase in mass - loss rate . according to the viscous evolution prescription of @xcite with their ` typical ' disk parameters ,
the current equilibrium disk mass around mira b is only 0.26@xmath54 based on a @xmath550.7@xmath46 , an accretion rate given by equation [ eqnbhacc ] and a disk radius of 10au .
the time scale for viscous evolution of the disk is @xmath56 years . as the time scale for mira a s evolution
is based on the time between thermal pulses which is of order @xmath57 years @xcite , the disk will likely maintain a near - equilibrium mass . at the time mira
a becomes a white dwarf , and the mass - loss rate from mira a is @xmath58 solar - masses per year , the equilibium disk mass will have increased to several jupiter - masses , certainly enough to justify the title ` protoplanetary disk ' .
now that we have established that mira b has a @xmath59au accretion disk around it , we should re - examine the idea that mira b is a main - sequence star and not a white dwarf as often identified in the literature .
the mira ab system is quite different to symbiotics such as mwc 560 , where the accretion luminosity is 1000 instead of 0.1@xmath60 .
however , accretion luminosity should vary by several orders of magnitude for symbiotics due to differing accretion rates , and as mira ab is unique in its relatively well - known geometry and wind properties , it is difficult to make quantitative comparisons .
a second general property of symbiotics such as mwc 560 is that there are spectral features with wind absorption and emission profiles with widths that are one to several thousand kms@xmath4 @xcite , comparable to or greater than the escape velocity from the surface of a white dwarf . despite the large number of lines observed for mira b ( e.g. * ? ? ?
* ) , there are no lines with half - widths of several thousand kms@xmath4 as one would expect if mira b were a white dwarf .
the analysis of section [ sectdisk ] demonstrated that the observed size of the accretion disk around mira b is consistent with bondi - hoyle accretion , and that this accretion theory can not underestimate the accretion rate onto mira b by more than a factor of a few . on the other hand
if we assume that mira b is a 0.6@xmath46 white dwarf , then the same analysis gives an accretion rate of @xmath61/yr and an accretion luminosity of 9.4l@xmath62 .
this is inconsistent with the measured accretion luminosity of mira b , which is @xmath63 , depending on what fraction of the flux is from accretion and what the extinction is .
the value of @xmath64 is an upper limit based on all the measured luminosity at @xmath65 m being due to accretion and @xmath66 , the extinction calculated from full silicate condensation in mira a s wind .
there is a possibility that the accretion luminosity is lower than expected for mira b due to a non - equilibrium situation where mass is accreted into a disk but not onto mira b itself .
the disk evolution time - scale discussed in section [ sectdisk ] makes this a highly unlikely scenario .
if on the other hand , we assume that mira b is a 0.7@xmath46 main - sequence star , then the accretion luminosity is 0.2@xmath67 , completely consistent with the measured luminosity of mira b. finally , if mira b were a main - sequence star , we could expect to see spectral features that indicate the presence of a main - sequence star , although these features could be very weak .
the weak features are due to significant veiling by the accretion flux : a good example of this is for t tauri stars can be found in the spectrum of bp tau , which is significantly veiled for all wavelengths shorter than 0.5@xmath1 m @xcite .
the faint object camera ( foc ) spectra partially published in @xcite do indeed show weak ( i.e. veiled ) spectral features of an early - mid k dwarf .
we have examined the full wavelength range of this spectrum ( including that not published ) , derived from hst data set x31g010at , and displayed the long - wavelength end of it in figure [ figkdwarf ] .
we took the wavelength scale calibration from chapter 8 of the foc manual , referencing the zero point to the mg ii doublet near 2800prominent in the mira b spectrum .
prominent spectral features at 4250and 5100are both detected at @xmath68 , matching a veiled model @xmath04000k dwarf spectrum well .
the @xmath060% veiling in v - band places the photospheric spectrum of mira b at an absolute v magnitude of @xmath69 , also consistent with a 0.7@xmath46 k5 dwarf ( with @xmath70 mag of extinction at v band ) .
all remaining possibilities for the nature of mira b that could influence our conclusions consist of some kind of exotic geometries .
an example could be a triple system with a radial velocity signature less than @xmath71kms@xmath4 ( so that it is not betrayed by ultraviolet spectral shifts ) and a close - system separation less than @xmath72au . in the absence of any data to support such a possibility
, we will eliminate these possibilities with an appeal to occam s razor .
therefore , we conclude that mira b is indeed a main - sequence star , with a most probable mass of roughly 0.7@xmath46 . from the standpoint of population synthesis , we can examine how common we expect mira b like systems to be .
our current galactic location in - between spiral arms means that the column - integrated star formation rate is about a quarter of what it was when mira a s progenitor was born @xcite . in order to accrete at least a jupiter mass via wind - accretion ,
a wind - accreting binary should have log(p ) between 4 and @xmath06.5 in days ( the upper limit for p is dependent on primary and secondary mass ) . according to @xcite , @xmath023% of solar - mass stars have companions with log(p ) in this range .
these results , taken together , imply that the birth - rate of mira b - like disks around companions to 1 - 2@xmath53 stars should be similar to the local birth - rate of 1 - 2@xmath53 stars themselves .
this immediately raises the question : where are the disks that formed by wind - accretion ?
most debris disks in the solar neighborhood are around stars likely belonging to young associations @xcite
. however , these debris disks are also generally cool and large , while a wind - accretion disk will be truncated by the dynamical pressure of the wind at roughly the bondi - hoyle accretion radius .
the disks around secondaries in wind - accretion binaries should grow viscously after the primary dwindles into a white - dwarf , but clearly this process must leave debris disks much less often than the large primordial disks around young stars .
the expected abundance of disks around companions to young white dwarfs is a clearly testable hypothesis of the bondi - hoyle truncation model of wind - accretion disks presented here .
however , in order to test it , a sample of very young white dwarfs in binaries would have to be chosen .
there are some such systems discovered through uv excesses : @xcite discusses many of them and some of the difficulties in forming a near - complete sample .
other systems are discovered through infrared excesses to white dwarfs : many of these have been imaged with hst @xcite .
disk fraction around the main sequence companions should approach unity for appropriate separations and ages less than several times the disk viscous evolution time - scale , i.e. a few @xmath73 years .
this paper has presented evidence that mira b is a @xmath00.7@xmath46 main sequence star surrounded by a @xmath59au radius disk . the new observations on which this is based were made possible by the high - contrast imaging capability of non - redundant interferometry on single - aperture telescopes .
mira ab is unique amongst wind accretion binaries in its well known binary and wind parameters , but this class of system is predicted to be relatively common as the end product of roughly 1 in 5 star systems with a solar - type primary . systems like mira ab should produce the clear observational signatures of an accretion disk around the secondary for at least a few @xmath73 years after the primary becomes a white dwarf .
we gratefully acknowledge the support of charles townes , the assistance of marc kassis in making the keck observations , the assistance of adwin boogert in planning the gemini observations and helpful discussions with nuria calvet and klaus pontoppidan .
m.i . would like to acknowledge michelson fellowship support from the michelson science center and the nasa navigator program .
this research has made use of the simbad database , the ines principle center and the iraf point source catalog .
j.d.m . acknowledges support from the grant nasa - jpl 1267021 .
some of the data presented herein were obtained at the w.m .
keck observatory , which is operated as a scientific partnership among the california institute of technology , the university of california and the national aeronautics and space administration .
the observatory was made possible by the generous financial support of the w.m .
keck foundation .
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 . based in part on observations obtained at the gemini observatory , which is operated by the association of universities for research in astronomy , inc .
, under a cooperative agreement with the nsf on behalf of the gemini partnership : the national science foundation ( united states ) , the particle physics and astronomy research council ( united kingdom ) , the national research council ( canada ) , conicyt ( chile ) , the australian research council ( australia ) , cnpq ( brazil ) and conicet ( argentina ) . , b. , danchi , w. c. , bester , m. , hale , d. d. s. , lipman , e. a. , monnier , j. d. , tuthill , p. g. , townes , c. h. , degiacomi , c. g. , geballe , t. r. , greenhill , l. j. , cruzalebes , p. , lefevre , j. , mekarina , d. , mattei , j. a. , nishimoto , d. , & kervin , p. w. 1997 , , 488 , 807 | the mira ab system is a nearby ( @xmath0107pc ) example of a wind accreting binary star system . in this class of system
, the wind from a mass - losing red giant star ( mira a ) is accreted onto a companion ( mira b ) , as indicated by an accretion shock signature in spectra at ultraviolet and x - ray wavelengths . using novel imaging techniques , we report the detection of emission at mid - infrared wavelengths between 9.7 and 18.3@xmath1 m from the vicinity of mira b but with a peak at a radial position about 10 au closer to the primary mira a. we interpret the mid - infrared emission as the edge of an optically - thick accretion disk heated by mira a. the discovery of this new class of accretion disk fed by m - giant mass loss implies a potential population of young planetary systems in white - dwarf binaries which has been little explored , despite being relatively common in the solar neighborhood . |
this work is supported in part by the national science foundation of china ( no . 11075139 and no.10705024 ) .
g.z is also supported in part by the fundamental research funds for the central universities .
t. aaltonen _ et al . _
[ cdf collaboration ] , [ arxiv:1104.0699 [ hep - ex ] ] .
a. e. nelson , t. okui and t. s. roy , [ arxiv:1104.2030 [ hep - ph ] ]
. m. r. buckley , d. hooper , j. kopp , e. neil , [ arxiv:1103.6035 [ hep - ph ] ] .
m. beneke , g. buchalla , m. neubert and c. t. sachrajda , phys .
lett . * 83 * , 1914 ( 1999 ) [ arxiv : hep - ph/9905312 ] .
m. beneke , g. buchalla , m. neubert and c. t. sachrajda , nucl .
b * 591 * , 313 ( 2000 ) [ arxiv : hep - ph/0006124 ] .
m. beneke and m. neubert , nucl .
b * 675 * , 333 ( 2003 ) [ arxiv : hep - ph/0308039 ] .
y. y. keum , h. n. li and a. i. sanda , phys .
b * 504 * , 6 ( 2001 ) [ arxiv : hep - ph/0004004 ] .
y. y. keum , h. n. li and a. i. sanda , phys .
d * 63 * , 054008 ( 2001 ) [ arxiv : hep - ph/0004173 ] . c. d. lu , k. ukai and m. z. yang , phys .
d * 63 * , 074009 ( 2001 ) [ arxiv : hep - ph/0004213 ] . c. w. bauer , s. fleming , d. pirjol and i. w. stewart , phys .
d * 63 * , 114020 ( 2001 ) [ arxiv : hep - ph/0011336 ] . c. w. bauer , d. pirjol and i. w. stewart , phys .
lett . *
87 * , 201806 ( 2001 ) [ arxiv : hep - ph/0107002 ] . c. w. bauer , d. pirjol , i. z. rothstein and i. w. stewart , phys .
d * 70 * , 054015 ( 2004 ) [ arxiv : hep - ph/0401188 ] .
p. ball , r. zwicky , phys .
* d71 * ( 2005 ) 014015 .
[ hep - ph/0406232 ] .
k. nakamura [ particle data group ] , j. phys .
g * 37 * , 075021 ( 2010 ) .
j. charles _ et al . _ [ ckmfitter group ] , eur .
j. c * 41 * , 1 ( 2005 ) [ arxiv : hep - ph/0406184 ] ; and updated results from http://ckmfitter.in2p3.fr .
d. -s .
du , j. -f .
sun , d. -s .
yang , g. -h .
zhu , phys .
* d67 * ( 2003 ) 014023 .
[ hep - ph/0209233 ] . | recently nelson _ et al . _ proposed an interesting flavor symmetric model to account for the top quark forward - backward asymmetry and the dijet anomaly at cdf simultaneously with just three parameters : a coupling constant of order one , and two scalar masses of 160 gev and 220 gev .
however these fiducial values of the parameters lead to the branching ratio of a almost pure penguin @xmath0 decay about one hundred times larger than the experimental results .
consider also the precision electroweak constraints , the scalar masses should be at least around 500 gev . actually with the coupling constant larger than one , it is impossible to explain either of the two cdf measurements consistently in this model .
but one may raise the charged scalar mass to , for example , @xmath1 gev and reduce the coupling strength to @xmath2 to meet the b physics constraints . with this parameter set
, the wjj cross section is found to be in the right range .
but due to the scalar mass splitting , its correction to t - parameter is about @xmath3 away from the precision electroweak constraints .
in addition , the top quark forward - backward asymmetry should be well below @xmath4 with this small coupling constant .
the cdf collaboration has recently updated the measurements on the forward - backward asymmetry in top quark pair production with a larger data sample about @xmath5 fb@xmath6@xcite .
interestingly , deviations from the standard model ( sm ) predictions are observed in the total forward - backward asymmetry both in the semi - leptonic @xmath7 data and in the di - lepton channel .
in addition , a distributional measurement found that @xmath8 in the @xmath9 rest frame , which deviates from the qcd correction prediction @xmath10 by @xmath11 .
the cdf collaboration has also reported another @xmath12 anomaly in the @xmath13 gev range of the invariant dijet mass distribution in association with a w boson @xcite .
a flavor symmetric model was proposed in @xcite to explain simultaneously the @xmath7 asymmetry and wjj excess at cdf .
gauge boson , though the predicted total cross section of @xmath9 production seems to be too small at the tevatron ( see , e.g. , @xcite ) ] a @xmath14 triplet of complex scalar fields @xmath15 is introduced in @xcite .
these color - singlet weak - doublet scalars respect the flavor symmetry : @xmath16 where @xmath17 and @xmath18 have charge @xmath19 under @xmath20 and @xmath21 , respectively , while @xmath22 is in a fundamental representation of @xmath23 .
this flavor symmetry is also preserved in the sm without yukawa interactions . in this model
the interaction of the scalars @xmath24 with the sm quarks are completely determined by the flavor symmetry with a universal coupling strength .
the @xmath25 anomaly can then be interpreted as @xmath26 via a s - channel @xmath27 exchange , and @xmath28 decays subsequently to a jet pair with its mass to be around @xmath29 gev .
the top quark forward - backward asymmetry can be explained by @xmath30 via a t - channel @xmath31 exchange and @xmath32 via a t - channel @xmath33 exchange . at first glance
, this seems to be in contradiction with the observation of @xcite that t - channel exchange of a color - singlet scalar has great difficulty to produce a large positive contribution to the top quark forward - backward asymmetry .
however a closer look at fig .
2 of @xcite reveals that there does have a narrow window with the scalar mass lighter than @xmath1 gev
. however this flavor symmetry model also contributes to hadronic b decays .
although there is no new cp phase introduced , we will show in the following that the effective operator @xmath34 via an exchange of such a light @xmath24 is constrained severely by the penguin dominant processes , such as @xmath0 decays . in this flavor symmetry model ,
the color - singlet weak - doublet scalars @xmath24 are charged @xmath35 under @xmath36 and singlets under @xmath37 .
the interaction between @xmath38 and the sm quarks @xcite @xmath39 is completely determined by the flavor symmetry in which @xmath38 ( @xmath40 ) are charged as @xmath41 under @xmath42 , and charged as @xmath43 under @xmath44 .
then the only free parameters are the coupling constant @xmath45 and the scalar masses @xmath46 and @xmath47 . to interpret
the cdf anomalies of wjj and forward - backward asymmetry of top quark , @xmath48 , @xmath49 gev and @xmath50 gev have been chosen in @xcite as `` fiducial '' values .
however in the mass basis , eq .
( [ eq : interaction ] ) also generates effective four fermion operators , among which contains @xmath51 as noticed in @xcite , this operator contributes to the charmless process @xmath52 in comparison to the relevant effective hamiltonian of the sm ( where electroweak penguin operators have been neglected ) @xcite @xmath53 with @xmath54 since eq .
( [ eq : bsuu ] ) is obtained at tree level , we will also consider the wilson coefficients in the sm at leading order .
matching the effective operators to the full theory at @xmath55 , one finds @xmath56 and other wilson coefficients to be zero at leader order in the sm .
but the flavor symmetry model contributes to @xmath57 as @xmath58 which is even comparable to @xmath59 in the magnitude . running the scale down from @xmath60 to @xmath61 , one finds in the sm @xmath62 but
when the new scalar contributions are included , the wilson coefficients @xmath63(@xmath64 ) are changed to be @xmath65 one may easily notice that @xmath66 is surprisingly large in this flavor symmetry model .
even considering the theoretical uncertainties on hadronic b decays , it will lead to too large branching ratios on the penguin dominant decays , such as @xmath0 channels as we will show immediately .
for charmless b decays , there are three factorization approaches being widely used : qcd factorization @xcite , the perturbative qcd method ( pqcd ) @xcite and soft collinear effective theory ( scet ) @xcite . here
we will adopt qcd factorization method .
notice that the new physics amplitude is calculated at tree level , correspondingly the wilson coefficients are calculated at leading logarithm . to be consistent ,
the decay amplitudes of qcd factorization are also evaluated at leading order of @xmath67 .
let s consider the almost pure penguin process @xmath68 decay .
taking @xmath69 mev , the form factor @xmath70 @xcite , the current quark mass @xmath71 mev @xcite and the relevant ckm parameters @xcite @xmath72 , @xmath73 , we obtain @xmath74 which is about one hundred times larger than the experimental measurement @xmath75 @xcite .
therefore the fiducial values of @xmath48 , @xmath50 gev taken by nelson _
et al . _
@xcite are apparently inconsistent with the penguin dominant b decays .
branching ratio of @xmath68 decay as a function of the mass of charged scalar @xmath76 .
the blue band shows the experimental measurements , while the red curve represents the predicted branching ratio including new physics contributions and the horizontal purple line denotes the sm predictions at leading order of @xmath67 . ] in fig .
[ fig:1 ] , we show the branching ratio of @xmath68 decay as a function of @xmath47 with the coupling strength @xmath45 fixed .
it indicates that the charged scalars should be heavier than about @xmath77 gev to be consistent with the charmless b decays .
noticed that the leading order sm prediction is about half less than the experimental measurements , as shown in fig .
[ fig:1 ] .
this is because next - to - leading order amplitudes are not small in qcd factorization method , especially for chirally enhanced power corrections and annihilation diagrams ( see , e.g. , @xcite ) .
but for the purpose of this paper , it should be enough to be confined at leading order .
the cdf dijet anomaly was explained in this flavor symmetry model by the process @xmath78 via s - channel @xmath79 exchange , with the cross section to be about @xmath80 pb .
now to satisfy b physics constraints , the charged scalar masses have to be raised from @xmath81 gev to around @xmath77 gev . as a result
, the corresponding cross section must be well below @xmath82 pb , which is too small to account for the cdf dijet excess .
in addition , keeping @xmath48 and @xmath49 gev unchanged while rasing the mass of charged scalars to @xmath83 gev , one might worry about its correction to the electroweak parameter @xcite @xmath84 which turns out to be @xmath85 .
notice that @xmath86 gev is taken in the above formula .
unfortunately this strongly contradicts the precision electroweak constraint @xcite @xmath87 assuming the higgs mass @xmath88 gev .
therefore the mass splitting between the charged and the neutral scalars should be quite small to satisfy the precision electroweak constraint , which means the neutral scalar masses should also be raised from @xmath29 gev to around @xmath89 gev .
for the color - singlet scalars in this mass range , one can see from fig . 2 of @xcite that the total forward - backward asymmetry of produced top pair may even be negative , which is opposite in sign to the experimental observations .
another possibility is to reduce the coupling strength @xmath45 , and at the same time raising the charged scalar mass moderately to satisfy the constraints of electroweak parameter t and charmless b decays simultaneously . noticed that taking the neutral scalar mass fixed at @xmath29 gev , its correction to @xmath90 is @xmath91 with @xmath92 gev , which corresponds to @xmath93 .
this already deviates from the precision electroweak constraint @xmath87 by more than @xmath94 sigma .
but if the standard model higgs is very heavy , it will contribute negatively to the @xmath95 parameter . keeping only leading logarithms in the higgs mass ,
the contribution can be expressed approximately as @xcite @xmath96 where @xmath97 denotes the reference value of the higgs mass .
this means , for the sm higgs as heavy as @xmath82 tev , the precision electroweak constraint on t parameter would be around @xmath98 , which is consistent with the flavor symmetric model with @xmath92 gev . in any case
, it is unlikely for the charged scalar mass in this model to be heavier than @xmath1 gev .
one may observe from fig .
[ fig:2 ] that , taking @xmath92 gev , @xmath45 should be around @xmath2 to satisfy the restriction of @xmath68 decay .
notice that it was shown in @xcite that in this model the top quark forward - backward asymmetry @xmath99 for @xmath100 gev with @xmath48 .
it is then clear that this asymmetry must be well below @xmath4 if the coupling constant @xmath45 is lowered to around @xmath2
. therefore it should be really hard , if not impossible , to explain the measured large forward - backward asymmetry of produced top pair under this circumstance . as to the wjj anomaly ,
the resonant production @xmath101 which subsequently decays to @xmath102 can enhance the wjj cross section .
it is easy to calculate first the decay width of @xmath79 , @xmath103 with the phase factor @xmath104 .
correspondingly , the wjj cross section is found to be @xmath105 pb for @xmath106 . actually , the wjj cross section from resonant @xmath79 production is not very sensitive to the value of @xmath45 , because the total width of @xmath79 also changes with @xmath45 .
for instance , the cross section is calculated to be @xmath107 pb with even smaller @xmath108 . branching ratio of @xmath68 decay as a function of the coupling strength @xmath45 , with the charged and neutral scalar masses taken at @xmath1 gev and @xmath29 gev , respectively .
the meaning of the lines is the same as in fig .
[ fig:1 ] . ] in summary , we consider the constraints of charmless b decays on a flavor symmetric scalar model proposed in @xcite .
the color - singlet weak - doublet scalars are introduced in the model which respects the flavor symmetry of @xmath109 .
it was shown in @xcite that the top quark forward - backward asymmetry and the dijet excess at cdf could be explained simultaneously with the parameters chosen as @xmath48 , @xmath49 gev and @xmath50 gev .
however the flavor symmetry of the scalars also contributes to @xmath110 decays . with the above fiducial values of the parameters , the pure penguin decay @xmath68 is predicted to have a branching ratio about one hundred times larger than the experimental results .
to avoid this constraint , the charged scalars should be heavier than around @xmath77 gev with @xmath48 fixed . as a result
, the production cross section of dijet plus a w boson would be too small to account for the cdf dijet excess .
furthermore , the precision electroweak constraints force the neutral scalar masses to be also around @xmath89 gev .
then it also becomes hard for this model to account for the forward - backward asymmetry in top quark pair production .
another possibility is to raise the charged scalar mass so that @xmath111 decay channel is allowed kinematically . in this scenario
the wjj cross section is enhanced due to the resonant production of @xmath112 so that the coupling constant @xmath45 may be lowered to evade the b physics constraint . specifically , one may take @xmath92 gev , @xmath49 gev and @xmath106 . with this parameter
set , the wjj cross section is found to be @xmath94 pb , which is in the right range to explain the cdf dijet excess .
but the scalar mass splitting will contribute to @xmath113 , which is about @xmath3 deviation from the precision electroweak constraint .
in addition , the smaller coupling strength will lead to too small @xmath9 forward - backward asymmetry to account for the experimental measurements . |
the notion of co - hopficity plays an important role in group theory . recall that a group @xmath0 is said to be _ co - hopf _
if @xmath0 is not isomorphic to a proper subgroup of itself , that is , if every injective homomorphism @xmath2 is surjective .
a group @xmath0 is _ almost co - hopf _ if for every injective homomorphism @xmath3 we have @xmath4<\infty$ ] . clearly , being co - hopf implies being almost co - hopf .
the converse is not true : for example , for any @xmath5 the free abelian group @xmath6 is almost co - hopf but not co - hopf .
it is easy to see that any freely decomposable group is not co - hopf . in particular ,
a free group of rank at least 2 is not co - hopf .
it is also well - known that finitely generated nilpotent groups are always almost co - hopf and , under some additional restrictions , also co - hopf @xcite .
an important result of sela @xcite states that a torsion - free non - elementary word - hyperbolic group @xmath0 is co - hopf if and only if @xmath0 is freely indecomposable
. partial generalizations of this result are known for certain classes of relatively hyperbolic groups , by the work of belegradek and szczepaski @xcite .
co - hopficity has also been extensively studied for 3-manifold groups and for kleinian groups .
delzant and potyagailo @xcite gave a complete characterization of co - hopfian groups among non - elementary geometrically finite kleinian groups without 2-torsion .
a counterpart agebraic notion is that of hopficity .
a group @xmath0 is said to be _ hopfian _ if every surjective endomorphism @xmath2 is necessarily injective , and hence is an automorphism of @xmath0 .
this notion is also extensively studied in geometric group theory . in particular , an important result of sela
@xcite shows that every torsion - free word - hyperbolic group is hopfian .
the notion of hopficity admits a number of interesting `` virtual '' variations .
thus a group @xmath0 is called _ cofinitely hopfian _ if every endomorphism of @xmath0 whose image is of finite index in @xmath0 , is an automorphism of @xmath0 , see , for example @xcite . a key general theme in geometric group theory is the study of `` large - scale '' geometric properties of finitely generated groups .
recall that if @xmath7 and @xmath8 are metric spaces , a map @xmath9 is called a _ coarse embedding _ if there exist monotone non - decreasing functions @xmath10 such that @xmath11 , that @xmath12 , and such that for all @xmath13 we have @xmath14 if @xmath15 is a path metric , then for any coarse embedding @xmath9 the function @xmath16 can be chosen to be affine , that is , of the form @xmath17 for some @xmath18 .
a coarse map @xmath19 is called a _ coarse equivalence _ if @xmath19 is _ coarsely surjective _
, that is , if there is @xmath20 such that for every @xmath21 there exists @xmath22 with @xmath23 .
a map @xmath9 is called a _ quasi - isometric embedding _
if @xmath19 is a coarse embedding and the functions @xmath24 in @xmath25 can be chosen to be affine , that is , of the form @xmath26 , @xmath27 where @xmath28 , @xmath29 .
finally , a map @xmath9 is a _ quasi - isometry _ if @xmath19 is a quasi - isometric embedding and @xmath19 is coarsely surjective .
the notion of co - hopficity has the following natural counterpart for metric spaces .
we say that a metric space @xmath30 is _ quasi - isometrically co - hopf _ if every quasi - isometric embedding @xmath31 is coarsely surjective , that is , if every quasi - isometric embedding @xmath31 is a quasi - isometry .
more generally , a metric space @xmath30 is called _ coarsely co - hopf _ if every coarse embedding @xmath31 is coarsely surjective . clearly , if @xmath30 is coarsely co - hopf then @xmath30 is quasi - isometrically co - hopf . if @xmath0 is a finitely generated group with a word metric @xmath32 corresponding to some finite generating set of @xmath0 , then every injective homomorphism @xmath2 is a coarse embedding .
this easily implies that if @xmath33 is coarsely co - hopf then the group @xmath0 is almost co - hopf
. the real line @xmath34 is coarsely co - hopf ( and hence quasi - isometrically co - hopf ) .
this follows from the fact that any coarse embedding must send the ends of @xmath34 to distinct ends .
since @xmath34 has two ends , a coarse embedding induces a bijection on the set of ends of @xmath34 .
it is then not hard to see that a coarse embedding from @xmath34 to @xmath34 must be coarsely surjective .
see @xcite for the formal definition of ends of a metric space .
[ ex : rooted ] the rooted regular binary tree @xmath35 is not quasi - isometrically co - hopf .
we can identify the set of vertices of @xmath35 with the set of all finite binary sequences .
the root of @xmath35 is the empty binary sequence @xmath36 and for a finite binary sequence @xmath37 its left child is the sequence @xmath38 and the right child is the sequence @xmath39 .
consider the map @xmath40 which maps @xmath35 isometrically to a copy of itself that `` hangs below '' the vertex @xmath41 .
thus @xmath42 for every finite binary sequence @xmath37 .
then @xmath19 is an isometric embedding but the image @xmath43 is not co - bounded in @xmath35 since it misses the entire infinite branch located below the vertex @xmath44 .
[ ex : free ] consider the free group @xmath45 on two generators .
then @xmath46 is not quasi - isometrically co - hopf .
the cayley graph @xmath30 of @xmath46 is a regular @xmath47-valent tree with every edge of length @xmath44 .
we may view @xmath30 in the plane so that every vertex has one edge directed upward , and three downward . picking a vertex @xmath48 of @xmath30 , denote its left branch by @xmath49 and the remainder of the tree by @xmath50 .
we have @xmath51 , and @xmath49 is a rooted ternary tree . define a quasi - isometric embedding @xmath52 by taking @xmath19 to be a shift on @xmath49 ( defined similarly to example [ ex : rooted ] ) and the identity on @xmath50 .
the map @xmath19 is not coarsely surjective , but it is a quasi - isometric embedding .
moreover , for any vertices @xmath53 of @xmath30 we have @xmath54 .
one can also see that @xmath45 is not quasi - isometrically co - hopf for algebraic reasons .
let @xmath55 with @xmath56\ne 1 $ ] .
then there is an injective homomorphism @xmath57 such that @xmath58 and @xmath59 .
this homomorphism @xmath19 is always a quasi - isometric embedding of @xmath60 into itself . if , in addition , @xmath61 and @xmath62 are chosen so that @xmath63 then @xmath64=\infty$ ] and the image @xmath65 is not co - bounded in @xmath60 .
thus , the group @xmath46 is not almost co - hopf and not quasi - isometrically co - hopf .
there do exist finitely generated groups that are algebraically co - hopf but not quasi - isometrically co - hopf .
the simplest example of this kind is the solvable baumslag - solitar group @xmath66 .
it is well - known that @xmath67 is co - hopf . to see that @xmath67 is not quasi - isometrically co - hopf we use the fact that @xmath67 admits an isometric properly discontinuous co - compact action on a proper geodesic metric space @xmath30 that is `` foliated '' by copies of the hyperbolic plane @xmath68 .
we refer the reader to the paper of farb and mosher @xcite for a detailed description of the space @xmath30 , and will only briefly recall the properties of @xmath30 here .
topologically , @xmath30 is homeomorphic to the product @xmath69 where @xmath70 is an infinite 3-regular tree ( drawn upwards ) : there is a natural projection @xmath71 whose fibers are homeomorphic to @xmath72 .
the boundary of @xmath70 is decomposed into two sets : the `` lower boundary '' consisting of a single point @xmath61 and the `` upper boundary '' @xmath73 which is homeomorphic to the cantor set ( and can be identified with the set of dyadic rationals ) . for any bi - infinite geodesic @xmath74 in @xmath70 from @xmath61 to a point of @xmath73
the full-@xmath75-preimage of @xmath74 in @xmath30 is a copy of the hyperbolic plane @xmath68 ( in the upper - half plane model ) .
the @xmath75-preimage of any vertex of @xmath70 is a horizontal horocycle in the @xmath68-``fibers '' .
any two @xmath76-fibers intersect along a complement of a horoball in @xmath68 .
similar to the above example for @xmath60 , we can take a quasi - isometric embedding @xmath77 whose image misses an infinite subtree in @xmath70 and such that @xmath78 for any vertices @xmath53 of @xmath70 .
it is not hard to see that this map @xmath19 can be extended along the @xmath75-fibers to a map @xmath79 such that @xmath80 is a quasi - isometric embedding but not coarsely surjective . since @xmath30 is quasi - isometric to @xmath67 , it follows that @xmath67 is not quasi - isometrically co - hopf .
grigorchuk s group @xmath0 of intermediate growth provides another intersting example of a group that is not quasi - isometrically co - hopf .
this group @xmath0 is finitely generated and can be realized as a group of automorphisms of the regular binary rooted tree @xmath35 .
the group @xmath0 has a number of unusual algebraic properties : it is an infinite 2-torsion group , it has intermediate growth , it is amenable but not elementary amenable and so on .
viii in @xcite for detailed background on the grigorchuk group .
it is known that there exists a subgroup @xmath81 of index 16 in @xmath0 such that @xmath82 is isomorphic to a subgroup of index 64 in @xmath0 .
the map @xmath83 , @xmath84 is clearly a quasi - isometric embedding which is not coarsely surjective .
since both @xmath81 and @xmath82 are quasi - isometric to @xmath0 , it follows that @xmath0 is not quasi - isometrically co - hopf . for gromov - hyperbolic groups and spaces quasi - isometric co - hopficity
is closely related to the properties of their hyperbolic boundaries .
we say that a compact metric space @xmath81 is _ topologically co - hopf _ if @xmath81 is not homeomorphic to a proper subset of itself .
we say that @xmath81 is _ quasi - symmetrically co - hopf _ if every quasi - symmetric map @xmath85 is surjective .
note that for a compact metric space @xmath81 being topologically co - hopf obviously implies being quasi - symmetrically co - hopf .
[ ex : merenkov ] a recent important result of merenkov @xcite shows that the converse implication does not hold .
he constructed a round sierpinski carpet @xmath86 such that @xmath86 is quasi - symmetrically co - hopf .
since @xmath86 is homeomorphic to the standard `` square '' serpinski carpet , clearly @xmath86 is not topologically co - hopf .
it is well - known ( see , for example , @xcite ) that if @xmath87 are proper gromov - hyperbolic geodesic metric spaces , then any quasi - isometric embedding @xmath88 induces a quasi - symmetric topological embedding @xmath89 between their hyperbolic boundaries .
it is then not hard to see that if @xmath0 is a word - hyperbolic group whose hyperbolic boundary @xmath90 is quasi - symmetrically co - hopf ( e.g. if it is topologically co - hopf ) , then @xmath0 is quasi - isometrically co - hopf .
this applies , for example , to any word - hyperbolic groups whose boundary @xmath90 is homeomorphic to an @xmath91-sphere ( with @xmath5 ) , such as fundamental groups of closed riemannian manifolds with all sectional curvatures @xmath92 .
the main result of this paper is the following : [ thm : a ] let @xmath0 be a non - uniform lattice in a rank - one semi - simple real lie group other than @xmath93 .
then @xmath0 is quasi - isometrically co - hopf .
thus , for example , if @xmath94 is a complete finite volume non - compact hyperbolic manifold of dimension @xmath95 then @xmath96 is quasi - isometrically co - hopf .
note that if @xmath0 is a non - uniform lattice in @xmath93 then the conclusion of theorem [ thm : a ] does not hold since @xmath0 is a virtually free group .
if @xmath0 is a uniform lattice in a rank - one semi - simple real lie group ( including possibly a lattice in @xmath93 ) then @xmath0 is gromov - hyperbolic with the boundary @xmath90 being homeomorphic to @xmath97 ( for some @xmath5 ) . in this case
it is easy to see that @xmath0 is also quasi - isometrically co - hopf since every topological embedding from @xmath97 to itself is necessarily surjective .
[ conv : x ] from now on and for the remainder of this paper let @xmath98 be a rank - one negatively curved symmetric space with metric @xmath15 ( or just @xmath99 in most cases ) .
namely , @xmath30 is isometric to a hyperbolic space @xmath100 ( with @xmath95 ) , @xmath101 ( with @xmath102 ) , @xmath103 over the reals , complexes , or quaternions , or to the octonionic plane @xmath104 .
if @xmath0 is as in theorem [ thm : a ] , then @xmath0 acts properly discontinuously ( but with a non - compact quotient ) by isometries on such a space @xmath30 and there exists a @xmath0-invariant collection @xmath105 of disjoint horoballs in @xmath30 such that @xmath106 is compact . the `` truncated '' space @xmath107 , endowed with the induced path - metric @xmath108 is quasi - isometric to the group @xmath0 by the milnor - schwartz lemma
. thus it suffices to prove that @xmath109 is quasi - isometrically co - hopf .
richard schwartz @xcite established quasi - isometric rigidity for non - uniform lattices in rank - one semi - simple lie groups and we use his proof as a starting point .
first , using coarse cohomological methods ( particularly techniques of kapovich - kleiner @xcite ) , we prove that spaces homeomorphic to @xmath110 with `` reasonably nice '' metrics are coarsely co - hopf .
this result applies to the euclidean space @xmath110 itself , to simply connected nilpotent lie groups , to the rank - one symmetric spaces @xmath30 mentioned above , as well as to the horospheres in @xmath30 .
let @xmath111 be a quasi - isometric embedding .
schwartz work implies that for every peripheral horosphere @xmath112 in @xmath113 there exists a unique peripheral horosphere @xmath114 of @xmath30 such that @xmath115 is contained in a bounded neighborhood of @xmath114 .
using coarse co - hopficity of horospheres , mentioned above , we conclude that @xmath19 gives a quasi - isometry ( with controlled constants ) between @xmath112 and @xmath114 .
then , following schwartz , we extend the map @xmath19 through each peripheral horosphere to the corresponding peripheral horoball @xmath116 in @xmath30 .
we then argue that the extended map @xmath117 is a coarse embedding . using coarse co - hopficity of @xmath30
, it follows that @xmath118 is coarsely surjective , which implies that the original map @xmath111 is coarsely surjective as well .
it seems likely that the proof of theorem [ thm : a ] generalizes to an appropriate subclass of relatively hyperbolic groups .
however , a more intriguing question is to understand what happens for higher - rank lattices : let @xmath0 be a non - uniform lattice in a semi - simple real lie group of rank @xmath119 .
is @xmath0 quasi - isometrically co - hopf ? unlike the groups considered in the present paper , higher - rank lattices are not relatively hyperbolic .
quasi - isometric rigidity for higher - rank lattices is known to hold , by the result of eskin @xcite , but the proofs there are quite different from the proof of schwartz in the rank - one case .
another natural question is : let @xmath0 be as in theorem [ thm : a ] . is @xmath0 coarsely co - hopf ?
our proof only yields quasi - isometric co - hopficity , and it is possible that coarse co - hopficity actually fails in this context .
the result of merenkov ( example [ ex : merenkov ] ) produces the first example of a compact metric space @xmath81 which is quasi - symmetrically co - hopf but not topologically co - hopf .
topologically , @xmath81 is homeomorphic to the standard sierpinkski carpet and there exists a word - hyperbolic group ( in fact a kleinian group ) with boundary homeomorphic to @xmath81 .
however , the metric structure on the sierpinski carpet in merenkov s example is not `` group - like '' and is not quasi - symmetric to the visual metric on the boundary of a word - hyperbolic group .
does there exist a word - hyperbolic group @xmath0 such that @xmath90 ( with the visual metric ) is quasi - symmetrically co - hopf ( and hence @xmath0 is quasi - isometrically co - hopf ) , but such that @xmath90 is not topologically co - hopf ?
in particuar , do there exist examples of this kind where @xmath90 is homeomorphic to the sierpinski carpet or the menger curve ?
the above question is particularly interesting for the family of hyperbolic buildings @xmath120 constructed by bourdon and pajot @xcite . in their examples @xmath121 is homeomorphic to the menger curve , and it turns out to be possible to precisely compute the conformal dimension of @xmath121 .
note that , similar to the sierpinski carpet , the menger curve is not topologically co - hopf .
are the burdon - pajot buildings @xmath120 quasi - isometrically co - hopf ?
equivalently , are their boundaries @xmath121 quasi - symmetrically co - hopf ?
it is also interesting to investigate quasi - isometric and coarse co - hopficity for other natural classes of groups and metric spaces . in an ongoing work ( in preparation ) , jason behrstock , alessandro sisto , and harold sultan study quasi - isometric co - hopficity for mapping class groups and also characterize exactly when this property holds for fundamental groups of 3-manifolds .
_ acknowledgement : _ the authors would like to thank misha kapovich for useful conversations .
recall that , by convention [ conv : x ] , @xmath30 is a rank one symmetric space different from @xmath68 .
namely , @xmath30 is isometric to a hyperbolic space @xmath100 ( with @xmath95 ) , @xmath101 ( with @xmath122 ) , @xmath103 over the reals , complexes , or quaternions , or to the octonionic plane @xmath104 .
we recall some properties of @xmath30 .
see @xcite , chapter ii.10 , for details .
let @xmath123 be a basepoint and @xmath124 a geodesic ray starting at @xmath41 .
the associated function @xmath125 given by @xmath126 is known as a busemann function on @xmath30 .
a _ horosphere _ is a level set of a busemann function .
the set @xmath127 is a _
horoball_. up to the action of the isometry group on @xmath30 , there is a unique busemann function , horosphere , and horoball .
a busemann function @xmath128 provides a decomposition of @xmath30 into _ horospherical coordinates _ , a generalization of the upper - halfspace model .
namely , let @xmath129 and decompose @xmath130 as follows : given @xmath22 , flow along the gradient of @xmath131 for time @xmath128 to reach a point @xmath132 , and write @xmath133 .
in horospherical coordinates , the @xmath112-fibers @xmath134 are geodesics , the @xmath135-fibers @xmath136 are horospheres , and the sets @xmath137 are horoballs . other horoballs appear as closed balls tangent to the boundary @xmath138 . if @xmath139 is a metric space and @xmath20 , a path @xmath140\to m$ ] , parameterized by arc - length , is called a _ @xmath141-rough geodesic _ in @xmath94 , if for any @xmath142 $ ] we have @xmath143 if @xmath144 are metric spaces , a map @xmath145 is _ coarsely lipschitz _ if there exists @xmath146 such that for any @xmath147 we have @xmath148 .
if @xmath149 is a path metric space then it is easy to see that @xmath145 is coarsely lipschitz if and only if there exist constants @xmath150 such that for any @xmath147 with @xmath151 we have @xmath152 .
the following two lemmas appear to be well known folklore facts : [ lemma : horoball1 ] there exists @xmath146 with the following property : let @xmath105 be a horoball in @xmath30 , @xmath153 and @xmath154 .
let @xmath131 be the point in @xmath105 closest to @xmath155 .
then the piecewise geodesic @xmath156 \cup [ b , x_2]$ ] is a @xmath141-rough geodesic .
acting by isometries of @xmath30 , we may assume that @xmath105 is a fixed horoball that is tangent to the boundary of @xmath30 in the horospherical model .
we may also assume that @xmath131 is the top - most point of @xmath105 , so that @xmath155 lies in the vertical geodesic passing through @xmath131 .
see figure [ fig : horoball1 ] .
consider the `` top '' of @xmath105 , i.e. the maximal subset of @xmath157 that is a graph in horospherical coordinates .
considering the riemannian metric on @xmath30 in horospherical coordinates , one sees that the geodesic @xmath158 $ ] must pass through the top of @xmath105 . setting @xmath141 to be the radius of the top of @xmath105 , centered at @xmath131 , completes the proof .
[ lemma : horoball2 ] let @xmath159 be disjoint horoballs , and @xmath160 . let @xmath161 $ ] be the minimal geodesic between @xmath162 and @xmath163 . then @xmath164 \cup [ b_1 , b_2 ] \cup [ b_2 , x_2]$ ] is a @xmath141-rough geodesic , for the value of @xmath141 in lemma [ lemma : horoball1 ] . the proof is analagous to that of lemma [ lemma : horoball1 ] .
we may normalize the horoballs @xmath159 as in figure [ fig : horoball2 ] .
the normalization depends only on the distance @xmath165 .
any geodesic @xmath158 $ ] must then pass through compact regions near @xmath166 and @xmath167 .
let @xmath168 be the radius of this region in @xmath162 .
fixing @xmath162 and varying @xmath163 , set @xmath169 .
the value @xmath168 remains bounded if the distance between the horoballs goes to infinity ( converging to the constant @xmath141 in lemma [ lemma : horoball1 ] ) .
thus , the infimum is attained and @xmath170 .
this completes the proof .
[ lemma : horoballs3 ] let @xmath159 be disjoint horoballs , @xmath171 , @xmath172 .
denote the minimal geodesic between @xmath162 and @xmath163 by @xmath161 $ ] . then @xmath173 .
fix @xmath174 and allow @xmath175 and @xmath172 to vary with the restriction @xmath176 . define a function @xmath19 on the interval @xmath177 $ ] by @xmath178 where the supremum is over all combinations of the variables with the restriction stated above , and @xmath166 denotes the closest point of @xmath162 to @xmath163 . then @xmath19 is a decreasing function , since increasing @xmath179 pushes the horoballs farther apart and forces @xmath155 closer to @xmath180 . in particular , @xmath181 since necessarily @xmath182 .
conversely , @xmath183 , taking @xmath184 .
we then have for any choice of disjoint @xmath159 and @xmath185 in the corresponding horoballs , that @xmath186 as desired .
let @xmath187 be a negatively curved rank one symmetric space .
a _ truncated space _
@xmath113 is the complement in @xmath30 of a set of disjoint open horoballs .
a truncated space is _ equivariant _ if there is a ( non - uniform ) lattice @xmath188 that leaves @xmath113 invariant , with @xmath189 compact .
we will consider @xmath113 with the induced path metric @xmath108 from @xmath30 . under this metric
, curvature remains negative in the interior of @xmath113 .
the curvature on the boundary need not be negative . for an extensive treatment of truncated spaces ,
see @xcite .
note that truncated spaces are , in general , not uniquely geodesic .
specifically , if @xmath30 is not a real hyperbolic space , then components of @xmath190 ( which come from horospheres in @xmath30 ) are isometrically embedded in @xmath109 copies of non - uniquely - geodesic riemannian metrics on certain nilpotent groups . in particular
, @xmath109 is not necessarily a @xmath191-space .
let @xmath30 be a negatively curved rank one symmetric space and @xmath188 a non - uniform lattice .
then @xmath192 is a finite - volume manifold with cusps . in @xmath30
, each cusp corresponds to a @xmath193-invariant family of horoballs . removing the horoballs produces an equivariant truncated space @xmath113 whose quotient @xmath189 is the compact core of @xmath192 .
[ prop : uniform ] let @xmath30 be a negatively curved rank one symmetric space and @xmath194 an equivariant truncated space .
then the inclusion @xmath195 is a coarse embedding .
since @xmath108 and @xmath15 are path metrics with the same line element , we have @xmath196 to get the lower bound , define an auxilliary function @xmath197 let @xmath81 be a compact fundamental region for the action of @xmath193 on @xmath113 . because @xmath193 acts on @xmath113 by isometries with respect to both metrics @xmath15 and @xmath108 , we may equivalently define @xmath198 by @xmath199 because @xmath81 is compact and the metrics @xmath15 , @xmath108 are complete , @xmath200 for @xmath201 .
furthermore , @xmath202 \rightarrow [ 0 , \infty]$ ] is continuous and increasing , with @xmath203 .
because horospheres have infinite diameter for both @xmath15 and @xmath108 ( they are isometric to appropriate nilpotent lie groups with left - invariant riemannian metrics , see @xcite ) , we also have @xmath204 .
let @xmath205 be an increasing homeomorphism of @xmath206 $ ] with @xmath207 for all @xmath208 and consider its inverse @xmath209 . for @xmath210
we then have @xmath211 this concludes the proof . a more precise quantitative version of proposition [ prop : uniform ] can be obtained by studying geodesics in @xmath113 ,
see @xcite .
for this section , let @xmath212 be a truncated space , with @xmath213 , and @xmath214 a @xmath108-quasi - isometric embedding . to ease the exposition , we refer to the target truncated space as @xmath215 .
[ lemma : schwartz1 ] there exists @xmath146 so that for every boundary horosphere @xmath112 of @xmath113 , there exists a boundary horosphere @xmath114 of @xmath216 such that @xmath115 is contained in a @xmath141-neighborhood of @xmath114 .
using nearest - point projection , we may assume @xmath217 .
[ defi : extension ] let @xmath218 be horoballs with boundaries @xmath219 .
a point in @xmath112 corresponds , in horospherical coordinates , to a geodesic ray in @xmath116 .
a map @xmath220 then extends to a map @xmath221 in the obvious fashion . in view of lemma [ lemma : schwartz1 ]
, a @xmath108-quasi - isometric embedding @xmath222 likewise extends to a map @xmath223 by filling the map on each boundary horoball .
[ lemma : schwartz2 ] a quasi - isometry @xmath224 induces a quasi - isometry @xmath221 , with uniform control on constants . one considers the metric on the horospheres of @xmath105 parallel to @xmath112 , or alternately fixes a model horosphere and varies the metric .
one then shows that if @xmath19 is a quasi - isometry with respect to one of the horospheres , it is also a quasi - isometry with respect to the horospheres at other horo - heights .
one then decomposes the metric on @xmath105 into a sum of the horosphere metric and the standard metric on @xmath34 , in horospherical coordinates .
this replacement is coarsely lipschitz , so the extended map is also coarsely lipschitz .
taking the inverse of @xmath19 completes the proof .
let @xmath30 be a simplicial complex and @xmath225 nested compacts with @xmath226 .
compactly supported cohomology @xmath227 is defined by @xmath228 for a compact space @xmath30 , @xmath229 but the two do not generally agree for unbounded spaces .
we have @xmath230 and @xmath231 for a non - trivial truncated space @xmath113 .
in fact , one has the following lemma .
[ lemma : everything ] let @xmath232 be a closed subset
. then @xmath233 if and only if @xmath234 .
it is well - known that the choice of nested compact sets does not affect @xmath235 .
choose the sequence @xmath236 , the intersection of a closed ball and @xmath237 .
with respect to the subset topology of @xmath237 , the boundary of @xmath238 is given by @xmath239 .
we have by excision @xmath240 note that @xmath241 and @xmath242 , so @xmath243 .
thus , if @xmath244 , then @xmath245 .
that is , @xmath246 is a compact set in @xmath247 , and @xmath248 .
thus , if @xmath234 , we have @xmath249 . otherwise , @xmath250 .
compactly supported cohomology is not invariant under quasi - isometries or uniform embeddings .
the remainder of this section is distilled from @xcite , where compactly supported cohomology is generalized to a theory invariant under uniform embeddings . for our purposes ,
the basic ideas of this theory , made explicit below , are sufficient .
let @xmath30 be a simplicial complex with the standard metric assigning each edge length @xmath44 . recall that a _ chain _ in @xmath30 is a formal linear combination of simplices .
the _ support _ of a chain is the union of the simplices that have non - zero coefficients in the chain .
the _ diameter _ of a chain is the diameter of its support .
an acyclic metric simplicial complex @xmath30 is @xmath251__-uniformly acyclic _ _ if there exists a function @xmath252 such that any closed chain with diameter @xmath99 is the boundary of a @xmath253-chain of diameter at most @xmath254 .
if @xmath30 is @xmath251-uniformly acyclic for all @xmath251 , we say that it is _
uniformly acyclic_. likewise , we say that a metric simplicial complex @xmath30 is @xmath251__-uniformly contractible _ _ if there exists a function @xmath252 such that every continuous map @xmath255 with image having diameter @xmath99 extends to a map @xmath256 with diameter at most @xmath254 .
if @xmath30 is @xmath251-uniformly contractible for all @xmath251 , we say it is _
uniformly contractible_.
rank one symmetric spaces and nilpotent lie groups ( with left - invariant riemannian metrics ) are uniformly contractible and uniformly acyclic .
[ lemma : approx ] let @xmath30 , @xmath149 be uniformly contractible and geometrically finite metric simplicial complexes and @xmath257 a uniform embedding .
then there exists an iterated barycentric subdivision of @xmath30 and @xmath258 depending only on the uniformity constants of @xmath259 and @xmath149 such that @xmath19 is approximated by a continuous simplicial map with additive error of at most @xmath260 .
we first approximate @xmath19 by a continuous ( but not simplicial ) map by working on the skeleta of @xmath30 . starting with the @xmath41-skeleton ,
adjust the image of each vertex by distance at most @xmath44 so that the image of each vertex of @xmath30 is a vertex of @xmath149 .
next , assuming inductively that @xmath19 is continuous on each @xmath251-simplex of @xmath30 , we now extend to the @xmath253 skeleton using the uniform contractibility of @xmath149 .
since error was bounded on the @xmath251-simplices , it remains bounded on the @xmath253-skeleton .
now that @xmath19 has been approximated by a continous map , a standard simplicial approximation theorem replaces @xmath19 by a continuous simplicial map , with bounded error depending only on the geometry of @xmath30 and @xmath149 ( see for example the proof of theorem 2c.1 of @xcite ) .
[ lemma : coho ] let @xmath30 and @xmath149 be uniformly acyclic simplicial complexes and @xmath257 a uniform embedding .
suppose furthermore that @xmath19 is a continuous simplicial map .
then if @xmath261 .
we first consturct a left inverse @xmath262 of the map @xmath263 induced by @xmath19 on the chain complex of @xmath30 , up to a chain homotopy @xmath264 .
that is , @xmath264 will be a map @xmath265 satisfying , for each @xmath266 , the homotopy condition @xmath267 and furthermore with diameter of @xmath268 controlled uniformly by the diameter of @xmath269 .
we start with the @xmath41-skeleton .
each vertex @xmath270 is the image of some vertex @xmath271 ( not necessarily unique ) .
set @xmath272 , and extend by linearity to @xmath273 . to define @xmath264 ,
let @xmath62 be an arbitrary vertex in @xmath30 and note that @xmath274 .
we have to satisfy @xmath275 .
since @xmath30 is acyclic , there exists a 1-chain @xmath276 satisfying this condition .
furthermore , note that @xmath277 is , by construction , a vertex such that @xmath278 . since @xmath19 is a uniform embedding , @xmath279
is uniformly bounded above .
thus , @xmath276 may be chosen using _ uniform
_ acyclicity so that its diameter is also uniformly bounded above .
assume next that @xmath262 and @xmath264 are defined for all @xmath280 with uniform control on diameters .
let @xmath112 be a @xmath251-simplex in @xmath30 .
then @xmath281 is a chain in @xmath30 whose diameter is bounded independently of @xmath112 .
then , by uniform acyclicity there is a chain @xmath114 with @xmath282 .
we define @xmath283 . as before
, we need to link @xmath114 back to @xmath112 .
we have @xmath284 by the homotopy condition [ fla : homotopy ] , we further have @xmath285 thus , by bounded acyclicity there is a @xmath253 chain @xmath286 such that @xmath287 as desired .
we extend both @xmath262 and @xmath264 by linearity to all of @xmath288 and @xmath289 , respectively . to conclude the argument , let @xmath81 be a compact subcomplex of @xmath30 and consider the complex @xmath290 .
the maps @xmath264 and @xmath291 on @xmath292 induce maps on @xmath293 , and the condition @xmath294 remains true for the induced maps and chains . because chain - homotopic maps on @xmath295 induce the same maps on homology
, we have , for @xmath296 , @xmath297 .
conversely , @xmath298 is the identity on cell complexes , so still the identity on homology .
thus , @xmath299 . by duality , @xmath300 .
taking @xmath238 to be an exhaustion of @xmath30 by compact subcomplexes and taking a direct limit , we conclude that @xmath301 .
[ cor : isomorphism ] let @xmath30 and @xmath149 be uniformly acyclic simplicial complexes and @xmath257 a uniform embedding .
there exists an @xmath258 depending only on the uniformity constants of @xmath302 , and @xmath149 so that @xmath303 .
lemma [ lemma : approx ] approximates @xmath19 by a continuous simplicial map , within uniform additive error .
lemma [ lemma : coho ] shows that the resulting approximation induces an isomorphism on compactly supported cohomology .
[ thm : coarse ] let @xmath304 be a manifold homeomorphic to @xmath247 , with @xmath15 a path metric that is uniformly acyclic and uniformly contractible . for each pair of non - decreasing functions @xmath305 with @xmath306 and @xmath307
, there exists a @xmath308 such that any @xmath309-coarse embedding @xmath52 is @xmath308-coarsely surjective . by corollary [ cor : isomorphism ]
, there is a uniform @xmath258 such that @xmath310 . by lemma [ lemma : everything ] , @xmath311 .
taking @xmath312 completes the proof .
let @xmath212 and @xmath215 be equivariant truncated spaces and @xmath313 a quasi - isometric embedding .
then @xmath19 is coarsely surjective with respect to the truncated metric @xmath108 . by lemma [ lemma : schwartz1 ]
, we may assume that @xmath19 maps boundary horospheres of @xmath113 to boundary horospheres of @xmath216 . by theorem [ thm : coarse ]
, @xmath19 is a surjection up to a constant independent of the boundary horosphere in question .
we then have an extension @xmath314 , as in definition [ defi : extension ] . by lemma [ lemma : schwartz2 ] , for each boundary horoball @xmath105 , the restriction @xmath315 is a quasi - isometry . by assumption , @xmath316 is a @xmath108-quasi - isometry , so @xmath316 is a @xmath99-uniform embedding by proposition [ prop : uniform ] . since @xmath30 is a path metric space , @xmath317 is then coarsely lipschitz on all of @xmath30 . we now show that @xmath317 is a uniform embedding by establishing a lower bound for distances between image points . recall that all distances are measured with respect to @xmath318 unless another metric is explicitly mentioned .
let @xmath319 so that @xmath317 is coarsely @xmath320-lipschitz and @xmath321 is coarsely @xmath320-co - lipschitz for every boundary horoball @xmath105 .
let @xmath322 be increasing proper functions so that @xmath19 is an @xmath309-uniform embedding .
let @xmath323 with @xmath324 .
we need to provide a lower bound for @xmath325 in terms of @xmath326 .
clearly , the lower bound will go to @xmath327 since @xmath317 is an isometry along vertical geodesics in horoballs .
there are four cases to consider ; in all cases we can ignore additive noise by working with sufficiently large @xmath326 and slightly increasing @xmath320 . 1 .
let @xmath328 for the same horoball @xmath105
. then @xmath329 .
2 . let @xmath330 .
this case is controlled by the uniform embeddings @xmath331 and @xmath332 ( proposition [ prop : uniform ] ) and the @xmath108-quasiisometry constants of @xmath19 .
3 . let @xmath333 for a horoball @xmath105 .
let @xmath334 be the closest point to @xmath155 .
then by lemma [ lemma : horoball1 ] , @xmath156 \cup [ b , x_2]$ ] is a @xmath141-quasi - geodesic for a universal @xmath141 depending only on @xmath30 and @xmath335 ( see also figure [ fig : horoball1 ] ) .
we consider two sub - cases : + suppose that @xmath336 .
let @xmath337 be the closest point to @xmath338 .
then by definition of @xmath131 , we have @xmath339 using lemma [ lemma : horoballs3 ] , we conclude @xmath340 + suppose , instead , that @xmath341 .
then we have the estimate @xmath342 .
we also have @xmath343 , so @xmath344 .
consider now @xmath345 , the closest point to @xmath346 . by lemma [ lemma : horoball1 ] ,
thus , @xmath348 4 .
let @xmath160 be in disjoint horoballs .
this case is identical to the previous one , except one uses lemma [ lemma : horoball2 ] rather than [ lemma : horoball1 ] .
we have then provided a lower bound for @xmath349 for any pair of points @xmath323 .
thus , the extended map @xmath317 is a coarse embedding . by theorem [ thm :
coarse ] , @xmath317 is then coarsely surjective .
namely , there exists @xmath258 so that @xmath350 ( the neighborhood is taken with respect to @xmath99 ) .
let @xmath351 be an arbitrary point .
since @xmath317 is coarsely surjective , there exists @xmath352 so that @xmath353 .
if @xmath354 , then we have shown that @xmath355 . otherwise , @xmath37 is contained in a horoball associated with @xmath113 . in appropriate horospherical coordinates ,
the horoball is given by @xmath356 and @xmath37 can be written as @xmath357 , with @xmath358 .
likewise , @xmath359 has coordinates @xmath360 , with @xmath361 .
furthermore , we have @xmath362 .
now , @xmath351 , so it has horospherical coordinates @xmath363 with @xmath364 .
it is easy to see that @xmath365 thus , for an arbitrary @xmath351 we have @xmath366 .
because @xmath332 is a uniform embedding , this implies that @xmath222 is coarsely surjective .
m. bridson and a. haefliger , _ metric spaces of non - positive curvature _ , grundlehren der mathematischen wissenschaften [ fundamental principles of mathematical sciences ] , vol . 319 , springer - verlag , berlin , 1999 . | we prove that if @xmath0 is a non - uniform lattice in a rank - one semi - simple lie group @xmath1 then @xmath0 is quasi - isometrically co - hopf .
this means that every quasi - isometric embedding @xmath2 is coarsely surjective and thus is a quasi - isometry .
[ multiblock footnote omitted ] |
microsoft kinect@xmath0 ( hereafter , simply ` kinect ' ) @xcite , a low - cost , portable motion - sensing hardware device , was developed by the microsoft corporation ( microsoft , usa ) as an accessory to the xbox @xmath2 video - game console ( 2010 ) .
the sensor is a webcamera - type , add - on peripheral device , enabling the operation of xbox via gestures and spoken commands . in 2011 , microsoft released the software - development kit ( sdk ) for kinect , thus enabling the development of applications in several standard programming languages .
the first upgrade of the sensor ( ` kinect for windows v2 ' ) , both hardware- and software - wise , tailored to the needs of xbox one , became available ( in july 2014 ) for general development and use .
the present paper is part of our research programme , aiming at investigating the possibility of involving either of the kinect sensors in the analysis of motion data of subjects walking or running ` in place ' ( e.g. , on commercially - available treadmills ) .
if successful , kinect could become an interesting alternative to marker - based systems ( mbss ) in capturing data for motion analysis , one with an incontestably high benefit - to - cost ratio .
it is rather surprising that only one study , addressing the possibility of involving kinect in the analysis of walking and running motion ( i.e. , not in a static mode or in slow motion ) , has appeared so far @xcite . using similar methodology to
the one followed herein , the authors in that study came to the conclusion that the original sensor is unsuitable for applications requiring high precision .
of course , it remains to be seen whether any improvement in the overall quality of the output can be obtained with the upgraded sensor .
the present study investigates the similarity of the output of the two sensors ; in case of significant differences , the validation of the upgraded sensor , on the basis of a marker - based system ( mbs ) , would be called for .
the detailed description of the methodology we follow herein may be found in ref .
@xcite ; we will refer to that paper whenever necessary .
the present paper has been organised in five sections . in section [
sec : method ] , we provide a short description of the hardware used in our study and give a summary of our methodology in the analysis of the human - motion data . in section [ sec : acquisition ] , we give details on the experimental part of the study .
the results , obtained from the analysis of the data of section [ sec : acquisition ] , are presented in section [ sec : results ] .
we discuss the main conclusions of the paper in the last section .
in the original sensor , the skeletal data ( ` stick figure ' ) of the output comprises @xmath3 time series of three - dimensional ( 3d ) vectors of spatial coordinates , i.e. , measurements of the ( @xmath4,@xmath5,@xmath6 ) coordinates of the @xmath3 nodes which the sensor associates with the axial and appendicular parts of the human skeleton . in coronal ( frontal ) view of the subject ( sensor view ) , the kinect coordinate system is defined with the @xmath4 axis ( medial - lateral ) pointing to the left ( i.e. , to the right part of the body of the subject being viewed ) , the @xmath5 axis ( vertical ) upwards , and the @xmath6 axis ( anterior - posterior ) away from the sensor .
the nodes @xmath7 to @xmath8 are main - body nodes , identified as hip_center , spine , shoulder_center , and head .
the nodes @xmath9 to @xmath10 relate to the left arm : shoulder_left , elbow_left , wrist_left , and hand_left ; similarly , the nodes @xmath11 to @xmath12 on the right arm are : shoulder_right , elbow_right , wrist_right , and hand_right .
the eight remaining nodes pertain to the legs , the first four to the left ( hip_left , knee_left , ankle_left , and foot_left ) , the remaining four to the subject s right ( hip_right , knee_right , ankle_right , and foot_right ) leg . in the upgraded sensor ,
some modifications have been made in the naming ( and placement ) of some nodes .
the original node hip_center has been replaced by spine_base ( and appears slightly shifted downwards ) ; the original node spine has been replaced by spine_mid ( and appears slightly shifted upwards ) ; finally , the original node shoulder_center has been replaced by neck ( and also appears slightly shifted upwards ) .
five new nodes have been appended at the end of the list , one of which is a body node ( spine_shoulder , node @xmath13 ) , whereas four nodes pertain to the subject s hands , hand_tip_left ( @xmath14 ) , thumb_left ( @xmath15 ) , hand_tip_right ( @xmath16 ) , and thumb_right ( @xmath17 ) . in both versions , parallel to the captured video image , kinect acquires an infrared image ; captured with a ccd camera , this infrared image enables the extraction of information on the depth @xmath6 .
the sampling rate in the kinect output ( for the video and the skeletal data , for both versions of the sensor ) is @xmath18 hz . the subject s motion on the treadmill is split into two components : the motion of the subject s ` centre of mass ' ( cm ) , which should be considered as one reference point , moving synchronously with the subject s physical cm , and the motion of the subject s body parts relative to the cm .
the coordinates of the cm are extracted from the main - body , the shoulder , and the hip nodes .
prior to further processing , the cm offsets ( @xmath19,@xmath20,@xmath21 ) are removed from the data at all times ; thus , the motion is defined relative to the subject s cm .
the largeness of the subject s ` stray ' motion on the treadmill is assessed on the basis of the root - mean - square ( rms ) of the distributions of @xmath19 , @xmath20 , and @xmath21 .
ideally , the period of the gait cycle @xmath22 is defined as the time lapse between successive time instants corresponding to identical postures of the human body ( position and direction of motion of the human - body parts with respect to the cm ) .
( of course , the application of ` identicalness ' in living organisms is illusional ; no two postures can ever be expected to be identical in the formal sense . )
we define the period of the gait cycle as the time lapse between successive most distal positions @xmath6 of the same lower leg , which is identified herein as the ankle ; one could also use the midpoints of the ankle and foot nodes , yet our past experience with the original sensor suggests caution with the foot nodes due to frequent artefacts , in particular in dorsal positions .
the arrays of the time instants , at which the left or right lower leg is at its most distal position with respect to the instantaneous cm of the subject , are used in timing the motion of the left or right part of the human body .
the period of the gait cycle is related to two other quantities which are used in the analysis of motion data . *
the stride length @xmath23 is the product of the velocity @xmath24 and the period of the gait cycle : @xmath25 . *
the cadence @xmath26 is defined as the number of steps per unit of time ; a commonly - used unit is the number of steps per min . using the time - instant arrays from the analysis of the left and right lower - leg signals ( as described in subsection [ sec : period ] ) , each time series ( involving one specific node and spatial direction ) was split into one - period segments , which were subsequently superimposed and averaged , to yield a representative movement for the node and spatial direction over the gait cycle .
average waveforms for all nodes and spatial directions , representing the variation of the motion of that node ( in 3d ) within the gait cycle , were extracted separately for the left and right nodes of the extremities ; waveforms were also extracted for the important angles introduced in subsection 3.1 of ref .
@xcite . in case
that left / right ( l / r ) information is not available ( as , for example , in the case of the trunk angle ) , the right lower leg was used in the timing .
all waveforms were subsequently @xmath27-centred ; the removal of the average offsets is necessary , given that the two systems yield output which can not be thought of as corresponding to the same anatomical locations .
the left and right waveforms yield two new waveforms , identified as the ` l / r average ' ( lra ) and the ` right - minus - left difference ' ( rld ) ; the l / r differences in the output must be investigated as we intend to use kinect in order to extract the asymmetrical features in the motion .
the similarity of corresponding waveforms is judged on the basis of five scoring options : pearson s correlation coefficient , the zilliacus error metric , the rms error metric , whang s score , and theil s score . assuming that a ( @xmath27-centred ) waveform , obtained from the original sensor , is denoted by @xmath28 and the corresponding ( @xmath27-centred ) waveform from the upgraded sensor by @xmath29 , the five scoring options are defined in eqs .
( 8)-(12 ) of ref .
@xcite ( where alternative ways for testing the similarity of the output of different systems are also discussed ) .
all waveforms are obtained herein using @xmath30 histogram bins in the data processing . at fixed velocity of the treadmill belt
, we will investigate the variation of these scores for the nodes and spatial directions of the extremities .
tests will be performed on the lra waveforms , as well as on those corresponding to the rld waveforms . after studying the goodness of the association of the waveforms at fixed velocity
, we will investigate velocity - dependent effects .
the data acquisition involved one male adult ( zhaw employee ) , with no known motion problems , walking and running on a commercially - available treadmill ( horizon laufband adventure 5 plus , johnson health tech .
gmbh , germany ) .
the placement of the treadmill in the laboratory of the institute of mechatronic systems ( school of engineering , zhaw ) , where the experimentation took place , was such that the subject s motion be neither hindered nor influenced in any way by close - by objects .
prior to the data - acquisition sessions , the kinect sensors were properly centred and aligned .
the sensors were then left in the same position , untouched throughout the data acquisition .
the original sensor also provides information on the elevation ( pitch ) angle at which it is set . during our extensive tests
, we discovered that this information is not reliable , at least for the particular device we used in our experimentation . to enable the accurate determination of the elevation angle of the sensor
, we set forth a simple procedure .
the subject stands ( in the upright position , not moving ) at a number of positions on the treadmill belt , and static measurements ( e.g. , @xmath9 s of kinect data ) at these positions are obtained and averaged .
the elevation angle of the sensor may be easily obtained from the slope of the average ( over a number of kinect nodes , e.g. , of those pertaining to the hips , knees , and ankles ) ( @xmath5,@xmath6 ) coordinates corresponding to these positions .
the output data , obtained from the original sensor , were corrected ( off - line ) accordingly , to yield the appropriate spatial coordinates of the kinect nodes in the ` untilted ' coordinate system . to prevent the original sensor from re - adjusting the elevation angle during the data acquisition ( which is a problematic feature ) , we attached its body unto a plastic structure mounted on a tripod . mounted on the same structure , about @xmath31 cm above the original sensor , was the upgraded sensor . a dedicated correction for the tilt effects in the output of the upgraded sensor ( which does not provide information on the elevation angle )
was performed as aforementioned .
the data from the original sensor were corrected by ( a rotation around the @xmath4 axis by ) @xmath32 , whereas those from the upgraded sensor by @xmath33 .
these corrections were confirmed by measurements taken with a laser distance meter .
interestingly , the value of the elevation angle , obtained from the sdk of the original sensor in this configuration , was only @xmath34 .
it is worth mentioning that , as we are interested in capturing the motion of the subject s lower legs ( i.e. , of the ankle and foot nodes ) , the kinect sensors must be placed at such a height that the number of lost lower - leg signals be kept reasonably small .
our past experience dictated that the kinect sensor be placed close to the minimal height recommended by the manufacturer , namely around @xmath35 ft off the ( treadmill - belt ) floor .
placing the sensor higher ( e.g. , around the midpoint of the recommended interval , namely at @xmath8 ft off the treadmill - belt floor ) leads to many lost lower - leg signals ( the ankle and foot nodes are not tracked ) , as the lower leg is not visible by the sensor during a sizeable fraction of the gait cycle , shortly after toe - off ( to ) .
the kinect sensor may lose track of the lower parts of the subject s extremities ( wrists , hands , ankles , and feet ) for two reasons : either due to the particularity of the motion of the extremity in relation to the position of the sensor ( e.g. , the identification of the elbows , wrists , and hands becomes problematic in some postures , where the viewing angle of the ulnar bone by kinect is small ) or due to the fact that these parts of the human body are obstructed ( behind the subject ) for a fraction of the gait cycle . assuming that these instances remain rare ( e.g. , below about @xmath36 of the available data in each time series , i.e. , one frame in @xmath18 ) , the missing values may be reliably obtained ( interpolated ) from the well - determined ( tracked ) data .
although , when normalised to the total number of the available values , the untracked signals usually appear ` harmless ' , particular attention was paid in order to ensure that no node be significantly affected , as in such a case the interpolation could have yielded unreliable results .
five velocities were used in the data acquisition : walking - motion data were acquired at @xmath9 km / h ; running - motion data at @xmath10 , @xmath11 , @xmath31 , and @xmath37 km / h . at each velocity setting , the subject was given @xmath7 min to adjust his movements comfortably to the velocity of the treadmill belt .
the kinect output spanned slightly more than @xmath35 min at each velocity .
the variation of the distance between the subject and the kinect sensors was monitored during the data acquisition ; it ranged between about @xmath38 and @xmath39 m , i.e. , well within the limits for the use of the sensors set by the manufacturer . the recording on the two measurement systems started simultaneously .
regarding the subject s ` stray ' motion , the rms average values ( over all velocities ) of the @xmath19 , @xmath20 , and @xmath21 distributions from the original sensor were : @xmath40 , @xmath41 , and @xmath42 mm ; the corresponding values from the upgraded sensor were : @xmath43 , @xmath44 , and @xmath45 mm .
averages of the period of the gait cycle @xmath22 , of the cadence @xmath26 , and of the stride length @xmath23 are given in table [ tab : comparison ] , separately for the two measurement systems at all velocities ; the agreement between corresponding values is very satisfactory .
we commence with the lra waveforms .
the goodness of the association of the waveforms obtained with the two measurement systems for the eight node levels of the extremities ( shoulder , elbow , wrist , hand , hip , knee , ankle , and foot ) and spatial directions was investigated as follows . separately for each of the five scoring options of section [ sec : options ] , velocity , and spatial direction , each node level was ranked according to the goodness of the association between the waveforms obtained from the two measurement systems .
the node level with the worst association was given the mark of @xmath27 , whereas the one with the best association was assigned the mark of @xmath46 .
the sum of the ranking scores over all velocities and scoring options yielded an @xmath47 ` matrix of goodness of the association ' ( @xmath10 node levels of the extremities , @xmath48 spatial directions ) ; the minimal entry in this matrix may be as small as @xmath27 ( which , in fact , was the overall score of the hips in the @xmath6 direction ! ) , whereas the maximal value may be as large as @xmath49 ( the maximal value was obtained for the foot nodes in the @xmath6 direction , a score of @xmath50 ) .
it was found that the nodes with the worst association ( henceforth , nwa ) between the two measurement systems correspond to the shoulders ( in all three spatial directions ) and the hips ( in the @xmath4 and @xmath6 directions ) ; the knees in the @xmath5 direction complete the first quartile of the similarity - index distribution .
the average value of pearson s correlation coefficient for the nwa was found equal to @xmath51 and the rms of the distribution to @xmath52 .
we subsequently pursued the investigation of systematic differences in the performance of the sensors regarding : a ) the upper and lower parts of the human body ( i.e. , upper- versus lower - extremity nodes ) and b ) the three spatial directions . * to assess the similarity of the waveforms , obtained from the two measurement systems for the nodes of the upper and lower extremities , one - factor anova tests were performed , separately for each of the five scoring options of subsection [ sec : options ] , on the scores extracted at each velocity , for all upper - extremity nodes and spatial directions , and all lower - extremity nodes and spatial directions . assuming a significance level @xmath53 ( the value which most statisticians adopt as the outset of the statistical significance ) , none of these tests resulted in significant effects , at any velocity ; the minimal p - value from these tests was about @xmath54 .
the exclusion of the scores of the nwa does not affect the results significantly , except at @xmath37 km / h where the resulting p - value dropped to @xmath55 ( which , nonetheless , exceeds the adopted @xmath56 level ) .
we thus conclude that the analysis can not support any significant differences in the performance of the two sensors for the upper and lower extremities . *
we next addressed the goodness of the association of the waveforms with regard to the three spatial directions @xmath4 , @xmath5 , and @xmath6 .
although the corresponding anova tests did not reveal significant effects ( except at @xmath10 km / h , where the p - value was @xmath57 ) , the p - values were generally small , i.e. , close to our @xmath56 threshold .
after the exclusion of the scores of the nwa , all running - motion data yielded p - values below @xmath56 .
the best - matching waveforms correspond to the depth @xmath6 ; the average values of pearson s correlation coefficient in the @xmath4 , @xmath5 , and @xmath6 directions were found to be : @xmath58 , @xmath59 , and @xmath60 , respectively .
it must be mentioned that pronounced velocity - dependent effects were observed in the overall similarity of the output of the two sensors .
the association of the waveforms , obtained from the two kinect sensors , appears to deteriorate with decreasing velocity ; for instance , the average value ( over all the nodes and spatial directions ) of pearson s correlation coefficient dropped ( almost linearly ) from @xmath61 ( at @xmath37 km / h ) to @xmath62 ( at @xmath9 km / h ) .
the exclusion of the scores of the nwa increases the overall similarity , leaving its velocity dependence almost intact ; for instance , the average value of pearson s correlation coefficient dropped ( almost linearly ) from @xmath63 ( at @xmath37 km / h ) to @xmath64 ( at @xmath9 km / h ) .
similar conclusions were drawn after examining the results obtained with the other four scoring options . we have not found a plausible explanation of such a velocity dependence in the similarity of the output of the two sensors .
we also compared the @xmath5 waveforms of the lower legs .
this comparison is interesting for two reasons .
first , the lower - leg signals are used in timing the motion ; second , we intend to use these signals in order to obtain the times ( expressed as fractions of the gait cycle ) of the initial contact ( ic ) and the to @xcite ; the difference of these two values is the stance fraction . shown in figs .
[ fig : lly_l ] and [ fig : lly_r ] are the waveforms obtained from the outputs of the two measurement systems , separately at each velocity .
the waveforms represent the variation of the raw signals , i.e. , the @xmath5 offsets of the subject s cm have not been removed .
the salient feature in the waveforms obtained with the original sensor is a pronounced peak appearing around the ic ; this peak is reduced in the waveforms obtained with the upgraded sensor .
we also investigated the goodness of the association of the rld waveforms . to this end
, two - sided t - tests were performed on the score distributions between corresponding lra and rld waveforms , a total of @xmath65 tests : five scoring options , three tests per scoring option ( paired , homoscedastic , and unequal - variance ) , and five velocities .
the p - values , obtained from the majority of these tests for the running - motion data , were found to be small , below @xmath56 ; the median p - value was equal to @xmath66 , whereas the minimal one was @xmath67 .
( after the exclusion of the scores of the nwa , the median p - value dropped to @xmath68 , whereas the minimal one to @xmath69 . )
the analysis showed that pearson s correlation coefficients of the tests on the rld waveforms were systematically below those of the lra waveforms , whereas all other scores were larger for the rld waveforms , thus indicating a poorer association of the waveforms obtained from the two measurement systems for the rlds compared to the lras . as a result , the similarity between the rld waveforms , obtained with the two kinect sensors , is weaker than between the lra waveforms .
we also examined the similarity of the waveforms for the important angles , introduced in subsection 3.1 of @xcite .
the only waveforms which match well are those for the hip and knee angles in the sagittal plane , and ( to a lesser degree ) those for the hip angles in the coronal plane . in the sagittal plane ,
the average values of pearson s correlation coefficients for the lras were equal to @xmath70 and @xmath71 for the hip and knee angles , respectively . in the coronal plane ,
the average value of pearson s correlation coefficient for the lras was equal to @xmath72 for the hip angles .
the corresponding rld waveforms do not much well ; in the sagittal plane , the average values of pearson s correlation coefficients were equal to @xmath73 and @xmath74 ; the agreement for the hip angles in the coronal plane was found to be even poorer ( pearson s correlation coefficient came out around @xmath75 ) .
the average value of pearson s correlation coefficients for the lras in the case of the knee angle in 3d was @xmath76 ; for the corresponding rlds , @xmath77 ; the left and right knee angles in 3d are shown in figs .
[ fig : ka3d_l ] and [ fig : ka3d_r ] .
we finally address the comparison of the roms obtained from the waveforms of two measurement systems waveforms .
it might be argued that one could simply use in a study the roms , rather than the waveforms , as representative of the motion of each node .
of course , given that each waveform is essentially replaced by one number , the information content in the roms is drastically reduced compared to that contained in the waveforms .
shown in fig .
[ fig : romlras ] is a scatter plot of the roms of the lra waveforms .
the ideal relation between these two quantities is a straight line with slope equal to @xmath7 ; pearson s correlation coefficient between the two rom arrays came out equal to @xmath78 . shown in fig .
[ fig : romrlds ] is a scatter plot of the roms of the rld waveforms . a straight line with slope equal to @xmath7 is the expected relation here too ; however , the scattering of the values is larger and pearson s correlation coefficient dropped to @xmath79 .
again , we draw the conclusion that the similarity of the rld waveforms , obtained with the two kinect sensors , is smaller to that of the lra waveforms . an unweighted quadratic fit , constrained to pass through the origin of the plot , to the roms of the lra waveforms resulted in curvature and slope values equal to @xmath80 mm@xmath81 and @xmath82 , respectively , indicating that ( in comparison to the upgraded sensor ) the original sensor overestimates ( on average ) the roms ( in the domain of the values of fig . [
fig : romlras ] ) by about @xmath83 .
these results do not change significantly after the exclusion of the scores of the nwa .
the present study is part of a broader programme , aiming at investigating the involvement of the microsoft kinect@xmath0 ( hereafter , ` kinect ' ) sensor in the analysis of human motion . in this work ,
we set on assessing the similarity in the performance of the two ( original and upgraded ) versions of the sensor for data acquired from one subject walking and running on a commercially - available treadmill .
an estimate of the degree of the association of the output of these systems is obtained after following the methodology of ref .
@xcite and comparing the waveforms extracted for the nodes of the extremities from the two measurement systems , as well as of those for the important angles ( defined in subsection 3.1 therein ) .
our comparative study of the two kinect sensors identifies the similarities and the differences in their performance , but ( of course ) can not enable conclusions regarding which of the two sensors performs better . *
the worst association is obtained for the waveforms of the shoulders ( in all three spatial directions ) , of the hips ( in the @xmath4 and @xmath6 directions ) , and of the knees ( in the @xmath5 direction ) . *
the association of the waveforms in the two lateral directions ( @xmath4 and @xmath5 ) is poorer than that in the @xmath6 direction ( depth ) . *
the association of the waveforms deteriorates with decreasing velocity of the walker / runner . *
the @xmath5 waveforms of the lower legs come out different . a characteristic peak , which is present in the data of the original sensor shortly before the initial contact , appears reduced in the data of the upgraded sensor . * exempting the hip and knee angles in the sagittal plane , and the lateral hip angles in the coronal plane , the association of the waveforms of the important angles is poor . *
the association of the ranges of motion ( roms ) is poor .
it appears that the original sensor overestimates the roms ( in the domain of the values of the present paper ) by about @xmath83 . *
the association between the waveforms , entering the investigation of the asymmetry of the motion , is poorer than the results obtained for the average waveforms for the left and right parts of the human body .
as the original sensor has been proven unsuitable for applications requiring high precision @xcite , it must be investigated whether the differences detailed herein constitute an improvement in the performance of the upgraded sensor . as a result , a dedicated validation study for the upgraded sensor , on the basis of a marker - based system , is called for . 99 http://www.xbox.com/en-gb/kinect/ , http://www.microsoft.com/en-us/kinectforwindows/ a. pfister , a.m. west , s. bronner , j.a .
noah , comparative abilities of microsoft kinect and vicon 3d motion capture for gait analysis , j. med .
technol . 38 ( 2014 ) 274 - 280 .
malinowski , e. matsinos , s. roth , on using the microsoft kinect@xmath0 sensors in the analysis of human motion , arxiv:1412.2032 [ physics.med-ph ] .
s. unpuu , the biomechanics of walking and running , clin .
sport med .
13 ( 1994 ) 843 - 863 .
novacheck , the biomechanics of running , gait posture 7 ( 1998 ) 77 - 95 .
the average values of the period of the gait cycle @xmath22 , of the cadence @xmath26 , and of the stride length @xmath23 at the five velocity settings used in the data acquisition ( see section [ sec : acquisition ] ) , separately for the two kinect sensors .
@xmath11 km / h & @xmath31 km / h & @xmath37 km / h + + @xmath22 ( s ) & @xmath84 & @xmath85 & @xmath86 & @xmath87 & @xmath88 + @xmath26 ( steps / min ) & @xmath89 & @xmath90 & @xmath91 & @xmath92 & @xmath93 + @xmath23 ( m ) & @xmath94 & @xmath95 & @xmath96 & @xmath97 & @xmath98 + + @xmath22 ( s ) & @xmath99 & @xmath100 & @xmath101 & @xmath102 & @xmath103 + @xmath26 ( steps / min ) & @xmath104 & @xmath105 & @xmath106 & @xmath107 & @xmath108 + @xmath23 ( m ) & @xmath109 & @xmath110 & @xmath111 & @xmath112 & @xmath113 + waveforms for the raw @xmath5 coordinate of the left lower leg ( ankle ) obtained , separately for each velocity , from the two kinect sensors .
the quantity @xmath114 is the fraction of the gait cycle .
the difference in the values simply reflects the higher position of the upgraded sensor on the mount , see section [ sec : acquisition].,width=585 ] the ranges of motion ( roms ) of the left / right average ( lra ) waveforms , obtained with the original kinect sensor , plotted versus those obtained with the upgraded kinect sensor , at all velocity settings used in the data acquisition ( see section [ sec : acquisition]).,width=585 ] | the present paper is part of a broader programme , exploring the possibility of involving the microsoft kinect@xmath0 sensor in the analysis of human motion . in this study ,
the output obtained from the two available versions of this sensor is critically examined .
we demonstrate that the two outputs differ in regard to the variation of the physical quantities involved in the modelling of the human motion .
as the original sensor has been found unsuitable for applications requiring high precision , the observed differences in the output of the two sensors call for the validation of the upgraded sensor on the basis of a marker - based system .
+ _ pacs : _ 87.85.gj ; 07.07.df , biomechanics , motion analysis ,
treadmill , kinect @xmath1e - mail : evangelos[dot]matsinos[at]zhaw[dot]ch , evangelos[dot]matsinos[at]sunrise[dot]ch |
the collective magnetic dipole mode ( m1 ) was one of the central subjects in the last two decades .
the interest has grown with time due to the wealth of experimental data provided by several groups performing either @xmath2 experiments at backward angles with high resolution or @xmath3 measurements using brehmstralung radiation .
the theoretical work in this field preceded the experiment .
indeed lo iudice and palumbo @xcite predicted the existence of this state from an anti - phase angular oscillation of proton and neutron axially symmetric and rigid systems .
the model is known under the name of two rotor model ( trm ) , and the predicted mode as scissors mode .
later on iachello predicted the existence of such a mode at lower energy within the interacting boson approximation ( iba ) @xcite .
the experimental discovery of this state in @xmath4gd , by achim richter and his collaborators @xcite , in an @xmath2 experiment , was a stimulus for both experimental @xcite and theoretical @xcite intensive activities .
thus , many descriptions have been proposed and some authors reactivated some old own papers devoted in the past to closely related topics .
both phenomenological@xcite and microscopic @xcite descriptions have been proposed .
the phenomenological formalisms provide an intuitive basis for the description of the new mode , although many details such as the fragmentation and the magnitude of the total strength can not be accounted for .
that is the reason why several improvements have been proposed along the time .
the rigid motion of the proton - neutron system was replaced by a share motion@xcite or the motion of two liquid drops@xcite .
the interacting boson approximation was extended by distinguishing between the proton and neutron bosons @xcite and pointed out that the f spin symmetry plays an important role for the structure of the magnetic state .
also the coherent state model@xcite was generalized to gcsm@xcite , which provides a new boson description of m1 states .
the microscopic formalisms use the qrpa approach within a deformed single particle basis and predict indeed that the m1 strength is distributed among many qrpa states . the nice feature brought by the microscopic theories
is that the low lying states with relatively large m1 transition probability are excited mostly by the orbital part of the transition operator .
such type of states are conventionally called scissors like states since their structure confirm the hypothesis of the trm model that the mode is due to the convection current created by the protons motion . early microscopic description of the state as a doorway state orthogonal to the spurious rotational modes predicted fragmentation @xcite the fragmentation effect was confirmed experimentally and as a consequence a new question was opened , namely , whether there are orbital states in the region beyond 4 mev where the spin - flip excitations dominate .
a long standing debate was about the mixture in the predicted mode , of spurious components of isoscalar nature@xcite .
several projection procedures were proposed@xcite to purify the theoretical state from spurious components .
many review articles discussing the results on this topic appeared in journals or proceedings of conferences and summer schools .
since we focus our attention on a specific description , we might unavoidably omit some important contributions to the field from our list of references and therefore , we advise the reader to consult the most recent reviews from refs.@xcite .
the present paper is devoted to the study of the orbital and spin excitations in the even - even isotopes of sm , using a projected single particle basis @xcite .
some ingredients like the projected states and the model hamiltonian are similar to those used in ref.@xcite .
the new point of the present investigation consists of the definition of the mean field . here
the mean field of the single particle motion is defined by averaging a particle core hamiltonian on a coherent state describing a phenomenological deformed core .
the coupling term comprises a monopole - monopole and a quadrupole - quadrupole interaction .
the monopole term is calculated by exploiting the restriction for the volume conservation . in this way
the single particle energies , approximated by the average of the mean field hamiltonian on the projected states , have a complex dependence on the deformation parameter .
we recall that in ref .
@xcite the single particle energies depend linearly on deformation .
therefore our aim is to investigate whether by considering more realistic single particle energies one obtains an improvement of the quantitative description given in ref . @xcite .
as we shall see at the end of this paper many details shown by the experimental data are nicely described by our approach which in fact confirm our expectation that in order to describe the non - global features of the system one should define properly the mean field of the single particle motion .
the results are described according to the following plan . in section 2 ,
we present briefly the projected single particle basis .
also , here we describe the mean field and derive analytical expressions for the single particle energies .
the model hamiltonian and the qrpa states are introduced in section 3 . the reduced probability for exciting these states from the ground state is also given .
applications to the even - even isotopes of sm are described in section 4 .
the main results are summarized in section 5
technical details about the construction of the angular momentum projected single particle basis appropriate for the description of the single particle motion in a deformed mean field generated by the particle core interaction , are given in ref .
aiming at a self - consistent presentation , we give here the main ingredients of the method used in the above quoted reference . the single particle mean field is determined by a particle - core hamiltonian :
@xmath5 here @xmath6 denotes the spherical shell model hamiltonian while @xmath7 is a harmonic quadrupole boson ( @xmath8 ) hamiltonian associated to a phenomenological core .
the interaction of the two subsystems is accounted for by the third term of the above equation , written in terms of the shape coordinates @xmath9 .
the quadrupole shape coordinates are related to the quadrupole boson operators by the canonical transformation : @xmath10 where @xmath11 is an arbitrary c number .
the monopole shape coordinate is to be determined from the volume conservation condition . in the quantized form ,
the result is : @xmath12.\ ] ] averaging @xmath13 on the eigenstates of @xmath6 , hereafter denoted by @xmath14 , one obtains a deformed boson hamiltonian whose ground state is described by a coherent state @xmath15|0\rangle_b,\ ] ] with @xmath16 standing for the vacuum state of the boson operators and @xmath17 a real parameter which simulates the nuclear deformation . on the other hand ,
the average of @xmath13 on @xmath18 is similar to the nilsson hamiltonian @xcite . due to these properties
, it is expected that the best trial functions to be used to generate , through projection , a spherical basis are : @xmath19 the upper index appearing in the l.h .
side of the above equation suggests that the product function is associated to the particle core system the projected states are obtained in the usual manner by acting on these deformed states with the projection operator @xmath20 a certain subset of projected states is orthogonal : @xmath21.\ ] ] the main properties of these projected spherical states are : a ) they are orthogonal with respect to i and m quantum numbers .
b ) although the projected states are associated to the particle - core system , they can be used as a single particle basis .
indeed when a matrix element of a particle like operator is calculated , the integration on the core collective coordinates is performed first , which results in obtaining a final factorized expression : one factor carries the dependence on deformation and one is a spherical shell model matrix element .
c ) the connection between the nuclear deformation and the parameter @xmath17 entering the definition of the coherent state ( 2.4 ) is readily obtained by requiring that the strength of the particle - core quadrupole - quadrupole interaction be identical to the nilsson deformed term of the mean field : @xmath22 here @xmath23 and @xmath24 denote the frequencies of nilsson s mean field related to the deformation @xmath25 by : @xmath26 the average of the particle - core hamiltonian @xmath27 on the projected spherical states defined by eq.(2.7 ) has the expression @xmath28 \frac{(\omega^2_{\perp}-\omega^2_z)}{90}.\end{aligned}\ ] ] here we used the standard notation for the clebsch gordan coefficients @xmath29 .
@xmath30 stands for the following integral @xmath31^kexp[d^2p_2(x)]dx , ~k=0,1\ ] ] it is worth mentioning that the norms for the core s projected states as well as the matrix elements of any boson operator on these projected states can be fully determined once the overlap integrals defined in ( 2.11 ) , are known @xcite . since the core contribution does not depend on the quantum numbers of the single particle energy level , it produces a shift for all energies and therefore is omitted in eq.(2.10 ) .
however when the ground state energy variation against deformation is studied , this term should be considered as well .
the first term from ( 2.10 ) is , of course , the single particle energy for the spherical shell model state @xmath14 . according to our remark concerning the use of the projected spherical states for describing the single particle motion ,
the average values @xmath32 may be viewed as approximate expressions for the single particle energies in deformed nilsson orbits .
we may account for the deviations from the exact eigenvalues by considering , later on , the exact matrix elements of the two body interaction when a specific treatment of the many body system is applied .
it is worth mentioning that the dependence of the new single particle energies on deformation is similar to that shown by the nilsson model .
this is clearly seen in fig . 1 and fig .
2 described in section 4 .
we remark that the energies shown in the above mentioned plot depend on deformation in a different manner than those obtained by one of the present authors in ref .
@xcite . indeed therein , they depend linearly on deformation while here non - linear effects are present .
the difference between the two sets of energies is caused by the fact that here the volume conservation condition was used for the monopole shape coordinate while in ref.@xcite this term was ignored and moreover there is no account of the volume conservation condition . in this respect ,
one expects that this difference in the single particle energies will induce some modifications of the results of ref.@xcite concerning the fragmentation process of the m1 strength .
although the energy levels are similar to those of the nilsson model , the quantum numbers in the two schemes are different . indeed here we generate from each j a multiplet of @xmath33 states as i , which plays the role of the nilsson quantum number @xmath34 , runs from 1/2 to j. on the other hand , for a given i there are @xmath35 degenerate sub - states while the nilsson states are only double degenerate . as explained in ref.@xcite , the redundancy problem can be solved by changing the normalization of the model functions : @xmath36 due to this normalization , the states @xmath37 used to calculate the matrix elements of a given operator should be multiplied with the weighting factor @xmath38 .
concluding , the projected single particle basis is defined by eq.(2.7 ) .
the projected states are thought of as eigenstates of an effective hamiltonian , with the corresponding energies given by eq.(2.10 ) .
in the states introduced in the previous section , we consider a system of z protons and n neutrons interacting among themselves in the following manner . the nucleons of similar isotopic charge interact through pairing interaction , dipole - dipole and quadrupole quadrupole interaction . protons and neutrons interact only through dipole and quadrupole two body interaction .
the many body hamiltonian describing such a system is written in second quantization as : @xmath39 note that although the model hamiltonian is formally identical to that used in ref.@xcite , the mean field parts of the two schemes are different from each other
. therefore even if the procedures used are similar , the final results are different . before applying the qrpa treatment , it is convenient to rewrite the qq interaction by re - coupling the initial and final nucleons in a different way and keep in the final result only the dipole - dipole term . by
doing this , the new expression is no longer separable .
the many body hamiltonian is written first in quasiparticle representation , by using the bogoliubov - valatin transformation , defined by the coefficients @xmath40 and @xmath41 . at this stage
one can define the qrpa dipole phonon operator which creates , by acting on the vacuum state @xmath42 , the qrpa states : @xmath43|0\rangle\equiv\gamma_{1\mu}^+(\lambda)|0\rangle,\ ] ] where @xmath44 and @xmath45 denote the dipole operator constructed out of two quasiparticle creation operators @xmath46 and @xmath47 and two annihilation operators @xmath48 and @xmath49 , respectively @xmath50 the phonon amplitudes are obtained by solving the qrpa equations with the normalization condition : @xmath51 the solutions of the qrpa equations are labeled by the index @xmath52 defined by the ordering relation for the corresponding energies ( @xmath53 ) : @xmath54 the qrpa states can be excited from the ground state , the vacuum state of the phonon operator , by the transition operator : @xmath55.\ ] ] where the effective gyromagnetic factors are obtained by accounting for the renormalization induced by the angular momentum carried by the core system : @xmath56 the reduced transition probability is defined by : @xmath57 the involved matrix element has the expression : @xmath58,~~ { \hat i}=\sqrt{2i+1}.\end{aligned}\ ] ] with obvious notations for the reduced matrix elements for the orbital angular momentum ( l ) and spin operator ( @xmath59 ) .
the qrpa equations were solved for @xmath60sm isotopes . the single particle energies are those given by eq .
( 2.10 ) , with the spherical shell model energies @xmath61 determined by the frequency @xmath62 and the strengths c and d for the spin orbit and @xmath63 terms , respectively @xmath64 where @xmath65 for protons and @xmath66 for neutrons @xcite .
the single particle energies can be considered either as functions of the nuclear deformation @xmath67 by means of the frequencies @xmath23 , @xmath24 , or as functions of the deformation parameter @xmath17 due to eq .
( 2.8 ) . in the later situation ,
an additional parameter is introduced,@xmath11 , which remained undetermined when the shape coordinates were transformed in boson variables . in order to compare the present results with those reported in ref.@xcite
it is useful to define the strength of the particle core qq interaction so that the corresponding matrix elements have similar expressions in the two descriptions .
it results @xmath68 for illustration , proton and neutron single particle energies are presented as functions of the deformation parameter @xmath17 in the figures 1 and 2 , respectively , for @xmath69sm .
the particle - core coupling constant @xmath70 was taken equal to that given in ref.@xcite . in principle
the deformation parameter should be determined by requiring that the system of z protons and n neutrons filling the energy levels in figs . 1 and 2 respectively , in accordance to the pauli exclusion principle , has a minimum energy with respect to changing the deformation . by this procedure ,
one obtains the equilibrium value of @xmath17 specified by the dashed vertical line .
however in the generalized coherent state model ( gcsm)@xcite , which is also based on coherent states description , the deformation parameter was fixed by a fit of the experimental e2 strength to the lowest @xmath71 states .
it turns out that this procedure works well for deformed nuclei but not unambiguously for near spherical nuclei .
the value obtained for the case of @xmath69sm , by using this method , is mentioned by vertical full lines in figs . 1 , 2 .
the deviation of this value from the equilibrium deformation might be corrected in two ways ; either by modifying the strength @xmath70 ( or , equivalently , the parameter @xmath11 ) so that the two values are the same , or by considering the normalization of the single particle energies due to the motion of the core .
indeed , we recall that the term of the model hamiltonian which corresponds to the core motion was set to zero when the average on the coherent state describing the phenomenological core was performed .
the average of the core hamiltonian is a function of @xmath17 which influences , to a large extent , the equations for energy minimum .
the fermi energies for protons and neutrons are pointed by arrows in the corresponding graphs .
the strength parameters of the model hamiltonian which were used in our calculations are collected in table 1 .
as we have already mentioned , we want to outline the effect on the m1 strength coming from the fact that we use a different set of single particle energies than in ref.@xcite . such an effect could be best determined if the strength parameters used in the two descriptions were close to each other .
indeed the parameters from table 1 are , with few exceptions , practically the same as in ref .
we modified the pairing strengths so that the gap parameters produced with the new set of single particle energies were the same as in the previous treatment .
moreover these are the same as those used by hamamoto and magnusson in ref@xcite . only for @xmath72sm and @xmath69sm one notices a deviation of the proton gap of about 100kev .
also the strength of the qq interaction for @xmath73sm and @xmath74sm is smaller in the present case by a factor of about 2 .
the strengths of the pairing interaction are consistent with the results obtained by nilsson @xcite while the strengths of dipole and quadrupole interactions are close to the parameters used by de coster and heyde @xcite and by hamamoto and magnusson@xcite .
the results of the bcs treatment are shown in terms of quasiparticle energies for @xmath69sm , in figs . 3 and 4 .
from these figures , one sees that despite the fact that the considered nucleus is a deformed one , the quasiparticle energies exhibit a shell like structure manifested by some energy gaps .
the quasiparticle energies were used to solve the qrpa equations .
further , the transition operator given by eq .
( 3.5 ) is used to excite the resulting states . the reduced transition probability for such excitation to take place
is calculated by means of relation ( 3.8 ) which contains in a factorized form the contribution coming from the bv and qrpa transformation amplitudes as well as from the matrix elements of the orbital and spin operators | a many body hamiltonian comprising pairing , quadrupole - quadrupole and spin - spin interaction is treated within a projected spherical basis with the aim of describing the detailed structure of the magnetic states of orbital and spin - flip nature in @xmath0sm . the mean field for the single particle motion takes care of the volume conservation of a phenomenological core which results in having a complex dependence on deformation for single particle energies .
several experimental features like the total orbital and total spin strengths , the shape of the orbital strength distribution , the hump structure of the spin strength distribution , the quadratic dependence of the orbital strength on nuclear deformation , the saturation effect of the orbital strength with respect to the casten s factor p are very well described within a quasiparticle random phase approximation formalism .
comparison of present results with those obtained in some previous treatments are also given .
# 1@xmath1#1 |
entanglement @xcite , i.e. , correlations between parts of a quantum system that defy any classical description , lies at the heart of quantum mechanics .
it is the basis for many applications of quantum information theory , such as quantum teleportation @xcite , quantum computing @xcite , quantum cryptography @xcite , and quantum metrology @xcite .
the first experimental demonstration of entanglement has been achieved by violating bell s inequality @xcite with polarization - entangled optical photon pairs generated during spontaneous parametric down - conversion in a nonlinear crystal @xcite . in many applications , it is desirable to have a source of entangled pairs of spatially separated particles .
such pairs are called epr pairs in reference to the seminal work of einstein , podolsky and rosen on the completeness of quantum mechanics @xcite .
compared to quantum optical scenarios , the generation of _ electronic _ epr pairs is rather challenging .
epr pairs of electrons are nonetheless highly desirable because an on - demand generation of such pairs would facilitate certain quantum communication tasks in solid - state devices @xcite .
theoretically , a conventional @xmath0-wave superconductor provides a natural source for electronic epr pairs @xcite : the electrons in a bcs superconductor form spin singlet cooper pairs in the ground state .
following theoretical proposals @xcite , the coherent splitting of cooper pairs , originating from a superconducting electrode , into two spatially separated electrons on neighboring quantum dots ( qds ) has recently been demonstrated experimentally @xcite .
while measurements of the current flowing out of the qds have indeed demonstrated the splitting of cooper pairs , the detection of the spin entanglement of the expected electronic singlet state has so far remained elusive .
similar devices can be used as a tool to detect unconventional pairing in superconductors @xcite or to entangle mechanical resonators @xcite .
detecting the entanglement of electronic epr pairs is not as straightforward as it is with their counterparts in quantum optics .
several works @xcite propose to violate a bell - type inequality with current noise measurements .
however , this will require accurate measurements of the cross - correlations between the currents from the two qds .
a measurable signal only emerges if these currents are large enough , i.e. , for a strong coupling of the qds to the measurement device ( `` open quantum dots '' ) .
this conflicts with the requirement of isolating the qds from the environment ( `` closed quantum dots '' ) , which is necessary for splitting cooper pairs coherently in the first place .
moreover , most existing proposals involve the use of strong ferromagnets and complex sample geometries , and neglect ( possibly long - range ) electron - electron interactions when computing the current - current correlations . as shown in ref .
@xcite , such interactions can reduce the measured entanglement signal .
another approach , which is closest in spirit to our work , is taken in @xcite .
these authors investigate the possibility to transfer electron spin entanglement to photon polarization states .
these particular schemes however , suffer from a low detection efficiency and require the use of additional quantum resources to generate a pure two - photon state @xcite .
very recently , the possibility of generating polarization entangled photons in a superconducting p - n junction has been proposed @xcite . ]
our proposal avoids the above difficulties by converting the spin entanglement of a _ single _ cooper pair into polarization entanglement of a _ single _ pair of optical photons and requires only classical resources such as laser drives and tunable gate voltages .
because photons do not interact with each other , photonic entanglement is more robust to perturbations than electronic entanglement , and can be detected using standard bell - type measurements .
provided sufficiently independent cavities are used ( see appendix . [
sec : effect - cavity - cross ] and ref .
@xcite ) , there can be no doubt that the entanglement ultimately measured stems from the split cooper pair because our entanglement transfer scheme involves only local operations .
since our scheme does not involve the measurement of electronic currents , it works even in closed qds .
finally , we show that the entanglement transfer can be carried out on time scales small compared to @xmath1 , the intrinsic coherence time of the qds ( see appendix [ sec : typic - valu - param ] ) .
let us first present our proposed experimental setup in more detail .
a schematic drawing of a possible realization is shown in fig .
[ fig : setup ] .
our starting point is the typical setup for cooper pair splitters , i.e. , a superconductor which is tunnel coupled to two nearby qds @xcite .
the spacing between the qds should be smaller than the superconducting coherence length and the qds are assumed to be in the coulomb blockade regime such that adding an electron to the qds requires a large charging energy @xmath2 .
the onsite energies of the qds can be tuned via gate voltages . splitting a cooper pair into a singlet state shared between the two
qds becomes energetically possible if the total energy of the singlet state coincides with the chemical potential of the superconductor .
both qds are embedded into optical cavities that serve as frequency filters allowing only certain desired optical transitions .
the small distance between the qds rules out conventional optical cavities , but photonic crystal cavities are nowadays easy to manufacture at the required length scales and can have optical linewidths and frequencies compatible with our proposal @xcite .
moreover , cavities with high quality factors and directional out - coupling of photons into a narrow solid angle for high - efficiency collection have been fabricated @xcite and self - assembled qds have been successfully embedded into photonic crystals in several experiments @xcite .
we will now present our scheme for transferring the spin entanglement of a cooper pair onto the polarization state of a photon pair . for simplicity
, we discuss a left - right symmetric setup . to be specific , we assume that the qds are realized as self - assembled gaas qds . the relevant electronic levels are thus generated from the light - hole ( @xmath3 ) and heavy - hole ( @xmath4 ) bands forming the valence band , as well as the conduction band ( @xmath5 ) .
the energy difference between the hole bands @xmath6 is of the same order as the superconducting gap @xmath7 , whereas the transition frequency between the valence band and the conduction band is in the optical frequency range @xcite .
moreover , we assume that a weak magnetic field is applied which causes a zeeman splitting @xmath8 ( with @xmath9 ) of all electronic levels .
are occupied and transitions into them are forbidden because of the pauli principle ( hash marks ) .
[ fig : setupschematics ] ] the entanglement transfer can be split into initialization and three phases , which we discuss next . a schematic level diagram along with the essential steps of our scheme is shown in fig .
[ fig : setupschematics ] .
_ initialization : _ initially , the gate voltages of the qds are tuned in such a way that the lowest light - hole states on each qd , @xmath10 and @xmath11 , are occupied .
furthermore the heavy - hole level resides in the superconducting gap but is detuned with respect to the chemical potential of the superconductor . _ phase 1 : _ the splitting of a cooper pair is achieved by tuning the gate voltages to bring the heavy - hole level into resonance with the chemical potential of the superconductor .
single - particle tunneling is suppressed due to the large superconducting gap .
furthermore , the large onsite coulomb interaction in the qd suppresses the tunneling of both electrons of a cooper pair onto the same qd .
the cooper pair splitting process , where one electron tunnels to each qd , is thus the dominant process @xcite .
when the separation between the qds is much smaller than the superconducting coherence length , the cooper pair splitting rate is ( see appendix [ supp : sw ] ) @xmath12^{-1/2}$ ] . here , @xmath13 ( @xmath14 ) denotes the electronic tunnel amplitude between the superconductor and the left ( right ) qd and @xmath15 denotes the normal - state density of states of the superconductor . if @xmath16 , where @xmath1 is the intrinsic coherence time of the qds , this process is coherent and leads to rabi oscillations between the superconductor and the heavy - hole states on the qds .
ideally , after half a period , the double - qd is occupied by a singlet state and the oscillation is stopped by detuning the heavy - hole level away from resonance . _ phase 2 : _ next , the electrons in the qds are excited from the heavy - hole to the conduction band .
this is achieved by switching on a strong drive laser on each of the two qds , with frequency @xmath17 .
a linearly polarized drive can be used , which induces spin conserving transitions .
after half a rabi period the laser is switched off , having lifted the singlet state into the conduction band levels .
if the zeeman splittings of the heavy hole and conduction bands differ , one may use two drive lasers per qd to satisfy the resonance conditions for the two different transition frequencies simultaneously .
the duration of this step is inversely proportional to the drive strength and can thus be made fast compared with @xmath1 .
_ phase 3 : _ the resonance frequencies of the two optical cavities are chosen to be close to the transition frequency between the conduction band and the light - hole band , @xmath18 . furthermore , the cavity linewidth @xmath19 is assumed to be much smaller than the frequency separation between the light- and heavy - hole bands , i.e. , @xmath20 .
therefore , the decay of the conduction band electrons into the light - hole band due to the dipole coupling of strength @xmath21 between electrons and photons , will be strongly enhanced , whereas the decay into heavy - hole states is suppressed .
since the lowest light - hole state @xmath22 is always occupied , a conduction band electron in the states @xmath23 or @xmath24 can only transition to the empty @xmath25 state via the emission of a linearly or circularly polarized photon , respectively . to investigate these three phases ,
we have numerically solved the schrdinger equation for the full system in the coherent limit @xmath26 ( see appendix [ sec : numerical - simulation ] ) .
figure [ fig : combinedplot ] shows the evolution during each phase of the occupation of the electronic levels and the cavity mode for an optimal choice of the drive strengths and drive durations . in the limit @xmath27 , the emitted photons undergo coherent oscillations between the qd and the cavity .
ideally , after half a rabi period @xmath28 , the electrons in both qds occupy the @xmath25 states while the electronic entanglement has been transferred to the photons .
in a real experiment , the photons need to be extracted from the cavities for measurement .
this is achieved by coupling each cavity to a continuum of modes , e.g. , as provided by a waveguide .
hence , the cavity acquires a finite loss rate @xmath29 .
as discussed further below , we will focus on the weak coupling limit , where @xmath30 . in this limit
, the coherent oscillations are suppressed and the photons are emitted into the continuum on a time scale @xmath31 .
, of the occupation probability of various electronic and photonic modes .
at the end of phase one , marked by @xmath32 , the electronic occupation per spin in the heavy - hole band reaches its maximum ( @xmath33 ) . during phase two , between @xmath34 and @xmath35 ,
the electronic population is transferred from the heavy - hole band to the conduction band .
once the conduction band is occupied the electrons can transition from the conduction band to the light - hole band by emitting photons into the cavity . at the end of phase three , marked by @xmath36 ,
the electronic system is in the product state @xmath37 , and the entanglement has been transferred to the photonic degree of freedom ( see text and fig . [
fig : lognegplot ] ) .
[ fig : combinedplot ] ] once both photons have been emitted , the electronic singlet has been transferred onto a two - photon state , ideally given by @xmath38 here @xmath39 , with @xmath40 , represents the photon states emitted into the continuum modes with either linear ( @xmath41 ) or circular ( @xmath42 ) polarization , and @xmath43 denotes the corresponding transition frequency .
importantly , because of the finite linewidth of the electronic levels , the emitted photons are spread out in frequency .
let us characterize the frequency overlap by @xmath44 . the normalization of the above photonic state @xmath45 is then given by @xmath46 . as we show next , the entanglement of the state ( [ eq:25 ] )
can be detected by standard polarization measurements , as long as @xmath47 is below a certain threshold value .
the density matrix of the polarization degree of freedom is obtained by tracing @xmath48 in eq .
( [ eq:25 ] ) over the frequency degree of freedom , @xmath49,\end{aligned}\ ] ] where we have introduced the shorthand notation @xmath50 and similar for the other two - photon polarization states . in the limit @xmath51
, corresponding to distinguishable frequencies , the state ( [ eq:8 ] ) is separable : @xmath52 with @xmath53 . in the other limit @xmath54 , corresponding to indistinguishable frequencies , the state ( [ eq:8 ] ) is maximally entangled : @xmath55 with @xmath56 . to obtain the latter expression , we have decomposed the circularly polarized state as a superposition of two orthogonal linearly polarized states @xmath57 . thus , depending on the value of @xmath47 , the polarization degree of freedom may or may not be entangled .
states because transitions to the latter are blocked ( see text).[fig : purcell_levels ] ] an experimentally accessible way of demonstrating entanglement in polarization is provided by the violation of the clauser - horne - shimony - holt ( chsh ) variant of bell s inequality @xcite ; by now a standard technique of quantum optics @xcite . in our case , we find that @xmath58 violates the chsh inequality if ( see appendix [ sec : chsh - ineq - entangl ] ) @xmath59 to relate @xmath47 with the parameters of our model , we use the weisskopf - wigner ( ww ) theory @xcite of the purcell effect , which allows us to derive analytically the state of the photons emitted into the continuum by the electronic system via the cavity . the relevant part of the level scheme is depicted in fig .
[ fig : purcell_levels ] .
we consider the zero - temperature limit where the cavity is initially empty . in each qd
, the problem then separates into two independent purcell emission processes , corresponding to transitions from the conduction band levels @xmath23 and @xmath24 into the unoccupied light - hole state @xmath25 ( blue and red arrows in fig .
[ fig : purcell_levels ] ) . for each of these transitions ,
the photon emission process can be described by the jaynes - cummings hamiltonian @xcite , where the cavity mode is coupled to a bosonic quasi - continuum . within the ww theory ,
the associated schrdinger equation can be solved analytically ( see appendix [ supp : ww ] ) and the solution is given by @xmath60 here @xmath61 denotes a state with @xmath62 electrons in the conduction band level , @xmath63 photons in the cavity mode and @xmath64 photons with momentum @xmath65 in the continuum ( @xmath66 denotes the vacuum state of the continuum ) . since we want to extract the photons quickly and avoid coherent oscillations between the cavity and the electrons , we focus on the weak - coupling , near - resonant regime where @xmath67 . here
@xmath68 is the detuning of the cavity mode @xmath69 from the spin - conserving and spin - flipping electronic transitions with frequencies @xmath70 and @xmath71 , respectively . at long time
@xmath72 , @xmath73 and @xmath74 vanish ( see fig . [
fig : purcell_snap ] , panel ( c ) ) , while the amplitude of the emitted photon @xmath75 asymptotically goes towards ( see fig .
[ fig : purcell_snap ] panels ( a ) and ( b ) and appendix [ supp : ww ] ) @xmath76 where @xmath77 , @xmath78 , @xmath79 is the photon frequency in the continuum , and @xmath80 is the coupling constant between the cavity mode and the continuum .
the state of the emitted photon can be written as @xmath81 in the long time limit , the distribution of the emitted photons is centered on the frequency @xmath82 and its width is determined by the purcell rate @xmath83 , because this is the smaller of the two rates @xmath84 and @xmath83 .
equation ( [ eq:11 ] ) allows us to evaluate the overlap in frequency of two photons emitted during the spin - conserving and spin - flipping transitions . for @xmath85
, we can expand to leading order in @xmath86 and find @xmath87 hence , from eqs .
( [ eq:13 ] ) and ( [ eq:14 ] ) , we find that the state of the emitted photons is entangled in polarization if @xmath88 thus , if the linewidth of the conduction band levels induced by the purcell effect is larger than the zeeman splitting of the conduction band doublet , it is possible to violate bell s inequality , thereby demonstrating the entanglement of the split cooper pair .
so far we have discussed the ideal case without any imperfections . to quantify the sensitivity of the proposed scheme to realistic parameter variations , we use the numerical simulations for the coherent system ( @xmath26 ) . in this case , the irreversible purcell emission in phase three is replaced by coherent rabi oscillations between the electronic system and the cavity ( see fig . [
fig : combinedplot ] ) .
after half - integer multiples of the rabi period , the photonic state in the cavities is ideally given by eq .
( [ eq:25 ] ) .
the fidelity of the actually generated photonic state computed numerically with this ideal state is shown in the top panel of fig .
[ fig : lognegplot ] . to quantify the entanglement of the photonic state generated in the cavity , we compute its logarithmic negativity @xcite .
this is shown in the bottom panel of fig .
[ fig : lognegplot ] . both the fidelity and the logarithmic negativity are shown as a function of the asymmetry of the electron - photon coupling strengths for the two polarizations @xmath89 and as a function of the detuning @xmath90 between the cavity mode and the electronic transition frequency between conduction band and light - hole band .
we stress that since only local unitary operations are applied to each qd , the positivity of the logarithmic negativity of the photonic state bears witness to the entanglement of the split cooper pair .
as expected , the optimal entanglement transfer takes place closest to resonance and for equal coupling strengths .
however , sizeable and detectable photon entanglement remains even away from the optimal point . in appendix
[ sec : entangl - pres - electr ] , we further show that photonic entanglement persists even in the presence of finite electronic decoherence .
finite temperature effects are not included in the present work but could be an interesting topic for future investigation . of the numerically computed photonic state @xmath91 with the ideal state @xmath45 of eq .
( [ eq:25 ] ) . _
lower panel : _ logarithmic negativity of @xmath92 as a function of photon coupling asymmetry @xmath93 and detuning @xmath94 between the cavity resonance and the electronic transition frequency . the larger the value of the logarithmic negativity , the more entanglement is present .
[ fig : lognegplot ] ]
in conclusion , we have proposed a hybrid electro - optical scheme to detect entanglement of split cooper pairs . by mapping the spin entanglement of electrons to the polarization entanglement of optical photons
, we avoid several difficulties of previous proposals .
provided cavities with little cross - talk at a distance smaller than the superconducting coherence length can be fabricated @xcite ( see also appendix [ sec : effect - cavity - cross ] ) , our scheme could be implemented by combining state - of - the - art technologies : a photonic crystal cavity with coupling strength @xmath95 @xcite and linewidth @xmath96 leads to @xmath97 .
this exceeds the zeeman splitting @xmath98 corresponding to a magnetic field of @xmath99 . with a typical cooper pair splitting rate of @xmath100 @xcite , the entire entanglement transfer can thus be performed fast compared to typical decoherence rates , @xmath101 @xcite .
our scheme can thus be used to verify the entanglement - preserving nature of the cooper pair splitting process @xcite ; a crucial step towards realizing a reliable source of electronic epr pairs in the solid state .
we would like to acknowledge stimulating discussions with christoph bruder .
this work was financially supported by the swiss snf and the nccr quantum science and technology .
sw acknowledges support from the itn cqom .
tls is supported by the national research fund , luxembourg ( attract 7556175 ) .
in this appendix , we present a systematic derivation of the cooper pair tunneling rate using the schrieffer - wolff ( sw ) transformation .
the system under investigation is described by the hamiltonian @xmath102 , where @xmath103 describes a conventional bcs superconductor . here , @xmath104 is the quasiparticle annihilation operator ( with momentum @xmath105 and spin @xmath106 ) , which is defined by @xmath107 , where @xmath108 denotes the bcs ground state .
the quasiparticle energies are given by @xmath109 , where @xmath7 is the superconducting gap and @xmath110 is the normal - state single - electron energy as measured from the chemical potential of the superconductor ( henceforth set to zero ) .
only the electronic states in the heavy - hole ( @xmath4 ) band of the quantum dots are included in the derivation of the cooper pair tunneling rate .
states in the light - hole and conduction bands can be safely ignored due to their larger detuning from the superconductor s chemical potential .
the hamiltonian describing the quantum dots is @xmath111 where @xmath112 and @xmath113 denotes the left and right quantum dots , respectively .
moreover , @xmath114 is the corresponding number operator given in terms of the electronic creation ( @xmath115 ) and annihilation ( @xmath116 ) operators for the electrons in the heavy - hole bands . for simplicity
, we have assumed that the left and right quantum dots have the same orbital and zeeman energies .
furthermore , the quantum dots are assumed to be in the coulomb blockade regime , so double occupancy of the heavy - hole bands is forbidden .
the tunnel coupling between the superconductor and the quantum dots is described by @xmath117 where @xmath118 is the corresponding electron tunneling amplitude and @xmath119 creates an electron ( with spin @xmath120 ) at position @xmath121 in the superconductor . going to momentum space , we can express the tunnel coupling in terms of the electron creation ( @xmath122 ) and annihilation ( @xmath123 ) operators for the superconductor , which are related to the quasiparticle operators via the bogoliubov transformation @xmath124 here , @xmath125 and @xmath126 are the usual bcs coefficients .
we express the original hamiltonian as @xmath127 .
we wish to determine a unitary transformation @xmath128 that eliminates @xmath129 to linear order in @xmath118 .
choosing the anti - hermitian operator @xmath130 such that @xmath131=-k_{{\rm sc}-{\rm qd}},\label{eq : s7}\ ] ] the transformed hamiltonian becomes , to second order in @xmath132 @xmath133 + \mathcal{o}(w_{\alpha}^3 ) .
\label{eq : formalsw}\ ] ] the solution of ( [ eq : s7 ] ) is given by @xmath134 where @xmath135 the effective hamiltonian at low temperatures and for large coulomb repulsion is then obtained by projecting @xmath136 onto the subspace where all quasiparticle states are empty and the two heavy - hole states of a given quantum dot contain at most one electron .
to second order in @xmath118 , we obtain @xmath137,\end{aligned}\ ] ] where @xmath138 .
the first term in the brackets describes the coherent cooper pair splitting while the second and third terms describe an effective spin - conserving inter - dot coupling .
we note that the latter two terms are suppressed by a small factor @xmath139 compared to the first one , and therefore can be safely ignored .
the sum over @xmath140 can be performed by linearizing the spectrum around the fermi energy and using @xmath141 .
the effective hamiltonian can then be written as @xmath142 where @xmath143 here , @xmath144 is the fermi momentum , @xmath145 is the superconducting coherence length , @xmath15 is the normal - state density of states at the chemical potential of the superconductor , and @xmath146 on resonance , i.e. , for @xmath147 and in the limit @xmath148 , eq .
( [ eq : gc ] ) reduces to the expression given in the main text .
to describe the dynamics of our entanglement transfer scheme , we use a real - time simulation of the system from the initial emission of the cooper pair into the quantum dots to the final emission of the polarization entangled photons into the cavities .
we will distinguish three phases : in phase one , we use the gates to load a singlet into the heavy - hole state of the quantum dots .
this phase is described by the hamiltonian , @xmath149 , where ( for @xmath150 , @xmath151 , and @xmath152 ) , @xmath153 the left and right dots are described by the hamiltonians @xmath154 , which contain the different zeeman - split orbital energies , @xmath155 , and the charging energy @xmath2 .
the electronic creation and annihilation operators for the individual orbitals are denoted by @xmath156 and @xmath157 , respectively .
the corresponding number operators are @xmath158 and the total number of particles on a given dot is denoted by @xmath159 .
the amplitude of the proximity coupling @xmath160 can be found using a schrieffer - wolff transformation , see eq .
( [ eq : gc ] ) .
it is the dominant coupling mechanism near resonance , i.e. , for @xmath161 , because all other possible tunneling terms between the superconductor and the quantum dots are strongly suppressed for large @xmath2 or @xmath7 .
the hamiltonian @xmath162 describes a time - dependent shift of the onsite energies , and will be used to establish the resonance condition for half a rabi period , @xmath163 , where @xmath164 denotes the heaviside function . at time @xmath165 , there is a high probability that a singlet occupies the quantum dots . in phase two ,
the singlet state is pumped from the heavy - hole band into the conduction band , and in phase three , the conduction band electrons transition to the light - hole band emitting photons .
these phases are governed by the hamiltonian @xmath166 , where @xmath167 .
\notag\end{aligned}\ ] ] for the numerical simulation , we use two optical cavity modes with linear and circular polarizations and frequencies @xmath168 and @xmath169 , respectively , to simulate the effect of a single cavity mode with a nonzero linewidth .
the cavity modes are described by @xmath170 .
the drive hamiltonians @xmath171 model the effect of a drive laser with frequency @xmath172 and causes spin - conserving rabi oscillations between the heavy hole and conduction band .
we assume that its amplitude has the form @xmath173 , where @xmath174 is about half a rabi period .
note that in order for the drive to efficiently transfer both spin states of the heavy - hole doublet , the width of its frequency spectrum @xmath175 should be larger than the detuning due to different zeeman splittings in the heavy hole and conduction bands .
alternatively one may use two narrow bandwidth lasers tuned on resonance with each transition . at the end of phase two ( at @xmath176 ) , the singlet state will then reside in the conduction band .
once the conduction band is occupied , the electrons can transition from the conduction band to the light - hole band by emitting photons into the cavity .
this is described by the coupling hamiltonian @xmath177 , which leads to rabi oscillations between the electrons and the cavity photons . in this process , the electron may ( or may not ) flip its spin , thereby emitting a circularly ( linearly ) polarized photon .
importantly , we assume that the gate voltages ensure that the lowest heavy - hole state at energy @xmath178 is always occupied , so that transitions into this state are blocked due to pauli exclusion principle .
for the numerical simulation , we assume that the photon frequencies are close to resonance with the respective transitions , i.e. , @xmath179 and @xmath180 . again , after half a rabi period @xmath181 ( at time @xmath182 ) , ideally the electronic system is in the product state @xmath183 , whereas the photon degree of freedom should now be entangled . a plot of the numerical result is shown in fig .
[ fig : lognegplot ] of the main text .
it shows the transfer of electron population between the heavy - hole band at the beginning ( @xmath184 ) and the light - hole band at the end ( @xmath185 ) for an optimal choice of drive durations and strengths .
moreover , it shows an increase in the photon occupation of the cavities , which are assumed to be empty before the beginning ( @xmath184 ) , towards the end of phase three ( @xmath185 ) . using these numerical results
, it is convenient to quantify the entanglement in the final state by calculating the logarithmic negativity of the photon state .
the logarithmic negativity is given by @xmath186 , where @xmath187 $ ] denotes the density matrix of photons , and @xmath188 means partial transposition with respect to either subsystem @xmath189 or @xmath190 .
we investigated the logarithmic negativity in the final photon state as a function of the ratios @xmath89 and @xmath90 ( see fig . [
fig : lognegplot ] of the main text ) .
let us stress that the entanglement witnessed by the positive values of the logarithmic negativity can only stem from the entanglement of the split cooper pair , since only local unitary operations are performed on the two subsystems .
the chsh variant of bell s inequality used in this work to demonstrate entanglement is expressed in terms of the photon polarization correlation function @xmath191.\end{aligned}\ ] ] an appropriate choice for the operators @xmath189 , @xmath192 , @xmath190 and @xmath193 is conveniently given by @xmath194 in terms of the pseudo - pauli operators @xmath195 here @xmath196 is the relative angle between the polarizing beam splitter settings of the left and right observers .
a state is entangled in polarization if @xmath197 ) of the main text that @xmath198 over @xmath199 yields @xmath200 and the condition for entanglement of @xmath58 given by eq .
( [ eq:13 ] ) of the main text .
in this appendix we derive analytically the amplitudes @xmath73 , @xmath74 and @xmath75 of the weisskopf - wigner ( ww ) ansatz of eq .
( [ eq:21 ] ) of the main text .
since the left and right subsystems evolve independently at this stage we suppress the @xmath201 index , and focus only on one side of the system . the photon pair state is then immediately obtained by linearity .
our starting point is the hamiltonian ( we set @xmath202 ) @xmath203 for the spin - conserving transition with transition frequency @xmath204 , @xmath205 represents the pauli matrix for the effective two - level system consisting of spin-@xmath206 conduction band level and the spin-@xmath206 light - hole state , i.e. , @xmath207^\dag.\end{aligned}\ ] ] for the spin - flipping transition with frequency @xmath208 , the pauli matrices are defined analogously .
next , @xmath209 ( @xmath210 ) represents the annihilation ( creation ) operator for a photon in the cavity mode with frequency @xmath211 .
the coupling strengths between the electronic transition and the cavity mode is denoted with @xmath212 , and @xmath213 ( @xmath214 ) denotes the annihilation ( creation ) operator for a photon with frequency @xmath215 in the continuum . in the wide - band limit , the coupling strength @xmath216 between the one dimensional quasi - continuum and the cavity mode determines the cavity linewidth as @xmath217 where @xmath218 is the quasi - continuum mode volume and @xmath219 the velocity of light . substituting the ansatz of eq .
( [ eq:21 ] ) in the main text into the associated schrdinger equation yields the differential equations @xmath220 the solution is most easily obtained by laplace transform using the initial conditions @xmath221 , @xmath222 . in laplace space , we then find the following algebraic equations ( @xmath0 denotes the laplace variable ) @xmath223 solving for the laplace amplitudes we find @xmath224 here , we have applied the ww approximation and introduced the cavity damping rate @xmath19 according to @xmath225 the imaginary part @xmath226 yields a frequency renormalization similar to the lamb - shift , which we shall ignore in the following , as this shift is typically small in the optical frequency regime .
the ww approximation is valid for weak enough damping , such that @xmath227 and is essentially equivalent to a born - markov approximation as we have established by comparing the analytic results below for the intra - cavity and electronic states with a numerical lindblad master equation calculation ( not shown ) .
note that in the optical regime , the above condition is easily satisfied and the ww approximation is expected to be adequate .
the poles of @xmath228 and @xmath229 are found to be given by @xmath230 where we have defined the detuning @xmath231 .
@xmath232 has an additional imaginary pole at @xmath233 . in the regime of interest
@xmath234 , the poles are well approximated to order @xmath235 and @xmath236 by @xmath237 in the optical regime , the frequency shifts may further be safely neglected since @xmath238 .
hence , @xmath239 and @xmath240 .
the inverse laplace transform of the amplitudes amounts to a summation over residues and yields , to second order in @xmath241 and @xmath242 @xmath243\label{eq:2s},\\ c_a(t)&=\frac{-ig}{\delta_k^c-\delta_k^0}\left(e^{(-i\omega_0^{}-\kappa_0^{})t}-e^{(-i\omega_c^{}-\kappa_c^{})t}\right)\label{eq:3s},\\ c_{k}(t)&=-\nu_0g\bigg[\frac{e^{-i\omega_k^{}t}}{\delta_k^c\delta_k^0 } + \frac{1}{\delta_k^c-\delta_k^0}\left(\frac{e^{(-i\omega_c^{}-\kappa_c^{})t}}{\delta_k^c}-\frac{e^{(-i\omega_0^{}-\kappa_0^{})t}}{\delta_k^0}\right)\bigg]\label{eq:4s},\end{aligned}\ ] ] with @xmath244 in the long time limit @xmath72 we obtain from eq .
( [ eq:4s ] ) the results of eq .
( [ eq:11 ] ) in the main text .
[ fig : purcell_snap ] of the main text illustrates eqs .
( [ eq:2s ] ) to ( [ eq:4s ] ) for the resonant case @xmath245 . ]
in the main text we have assumed that the photon emission process takes place on a time scale short compared to the coherence time @xmath1 of the qds . here
, we investigate numerically the effect of electronic decoherence on the chsh inequality violation of the intra - cavity photon state for a perfect cavity with @xmath246 . to this end
we consider the reduced qd model consisting of two three - level `` atoms '' , each resonantly coupled to two cavity modes as depicted in fig .
[ fig : reduced_model ] . we shall consider electronic relaxation determined by the non - radiative relaxation rates @xmath247 and @xmath248 for transitions from states @xmath249 and @xmath250 to the state @xmath251 as well as electronic dephasing with rates @xmath252 and @xmath253 due to fluctuations of the corresponding transition energies .
this corresponds to an effective coherence time for the double dot of @xmath254 .
as before , the hamiltonian of both qd - cavity systems decouples into a sum as @xmath255 with ( we set @xmath256 ) @xmath257 note that , in order to most clearly distinguish the effect of decoherence from other effects , we choose equal coupling strength @xmath212 for both transitions .
the evolution of the state of the system @xmath258 , is described within the born - markov approximation by the zero - temperature lindblad master equation @xmath259+\sum_{\alpha = l , r}\left(\gamma_{01}^{}\mathcal{d}[\sigma_{01}^{\alpha,- } ] + \gamma_{02}^{}\mathcal{d}[\sigma_{02}^{\alpha,- } ] \right)\rho\nonumber\\ & + \sum_{\alpha = l , r}\left(\gamma_{01}^{\varphi}\mathcal{d}[\sigma_{01}^{\alpha , z } ] + \gamma_{02}^{\varphi}\mathcal{d}[\sigma_{02}^{\alpha , z } ] \right)\rho,\end{aligned}\ ] ] with @xmath260\rho=\left(2o\rho o^\dagger - o^\dagger o\rho - \rho o^\dagger o\right)/2 $ ] , @xmath261 and @xmath262 .
the photonic state @xmath263 $ ] , is obtained by tracing @xmath258 over the electronic degrees of freedom .
[ fig : bell_relax ] shows the chsh correlation of eq .
( [ eq:1s ] ) as a function of the relative angle @xmath199 between left and right observers , for @xmath264 .
while decoherence clearly weakens the chsh correlation , this simulation essentially demonstrates that as long as the coupling strength @xmath212 is large compared with the decoherence rate @xmath265 of the qd , the chsh inequality can be violated . ]
in this appendix we investigate quantitatively to what extent cavity cross - talk affects our entanglement transfer scheme . to this end
we perform a numerical simulation of the coherent system including a polarization conserving cavity - cavity coupling term of the form @xmath266 .
we use the notation of appendix [ sec : entangl - pres - electr ] . .
when the inter - cavity coupling is weaker than roughly the electron - photon coupling in each cavity , the chsh inequality is violated ( shaded area).[fig : inter_cav ] ] fig .
[ fig : inter_cav ] shows the maximal bell inequality violation achievable after half a ( bare ) rabi period @xmath267 .
we see that while the correlation signal is clearly reduced by a finite inter - cavity coupling , it still surpasses the threshold of @xmath268 demonstrating entanglement , roughly as long as @xmath269 .
furthermore , in the experimentally relevant case where the photons are coupled out of the cavities , the population of the cavity modes remains small at all times and is of order @xmath270 ( see appendix .
[ supp : ww ] and panel ( c ) of fig .
[ fig : purcell_snap ] ) .
hence the effective cavity - cavity coupling rate is reduced compared with the coherent case .
in this appendix , we estimate the typical range of values for various parameters used in our theory . a typical value of the coupling strength between the quantum dot and the cavity field is @xmath271
ghz [ ] . choosing a cavity decay rate of @xmath272 ghz , leads to an induced bandwidth of @xmath273 ghz for the conduction band levels .
the zeeman splitting for a magnetic field strength of 10 mt is @xmath274 ghz .
thus for these parameters by using a magnetic field which is weaker than 10 mt , bell s inequality can be violated .
a typical intrinsic coherence time of the self - assembled quantum dots is @xmath275 @xmath276s ( @xmath277 ghz ) [ ] .
a typical value of cooper pair splitting rate is @xmath278 ghz [ ] .
thus the various conditions ( @xmath279 , @xmath280 , @xmath281 , and @xmath16 ) essential for demonstrating entanglement in our scheme , can be satisfied with current technology . | based on the bardeen cooper schrieffer ( bcs ) theory of superconductivity , the coherent splitting of cooper pairs from a superconductor to two spatially separated quantum dots has been predicted to generate nonlocal pairs of entangled electrons . in order to test this hypothesis
, we propose a scheme to transfer the spin state of a split cooper pair onto the polarization state of a pair of optical photons .
we show that the produced photon pairs can be used to violate a bell inequality , unambiguously demonstrating the entanglement of the split cooper pairs . |
advances in experimental setups along the past two decades have allowed to grow in a controlled way , and characterize , tiny structures and devices , paving the way for the slow development of those fields of materials science now covered under the umbrella of nanoscience .
ever more powerful computers and canny algorithms have also allowed to simulate larger and larger clusters of atoms or molecules , filling partially the bridge between theory and experiments .
molecular dynamics packages based on density functional theory ( dft)@xcite represent a specially useful set of theoretical tools in the analysis of such materials and devices . among them , those which use basis sets ( bs ) of localized wave functions are particularly attractive since , one the one hand , can be easily adapted to work with order - n algorithms , and on the other , they can be written in a tight - binding language , which allows for easier analytical approaches .
siesta is a simulation package which literally implements the tight - binding philosophy .
@xcite indeed , it uses norm - conserving pseudopotentials @xcite to integrate away core energy levels and very flexible bs made up of numerical atomic - like wave - functions to handle valence electrons .
the use of minimal bases allow for fast calculations which already provide a qualitative understanding of the simulated material .
accurate calculations can be performed at a higher numerical cost , using more complete bs .
assessment of the degree of reliability of a bs might be essential , since competing would - be ground states may in some instances have small energy differences .
such analyses have already been performed for selected molecules and solids @xcite , but not for magnetic materials .
those studies show how both the number of wave functions used as well as their extent are variational parameters , providing therefore a path for systematic improvements of the accuracy of a simulation .
a similar study for magnetic elements or materials seems to be highly desirable , since they have their own peculiarities and , in particular , are usually tougher to simulate .
we have performed an exhaustive study of the degree of accuracy of the basis for iron in most of its bulk phases as well as for small clusters .
we find that siesta provides a highly accurate description of the systems we have scrutinized , provided that a large numbers of extended orbitals is used .
we will show below how bs regarded as fairly complete for simpler elements can provide disastrous results for this transition metal .
the siesta package has built into it the possibility to cope with non - collinear commensurate spin structures but only in the local density approximation ( lda ) .
we have therefore coded the ability to compute non - collinear arrangements of the spin moments in the generalized gradient approximation ( gga ) , since lda fails to provide adequate ground states and lattice constants of a number of magnetic transition metals .
moreover , we have included the possibility to simulate non commensurate spiral structures @xcite . while bcc ( @xmath0)iron is firmly established to be a ferromagnet , the scenario for fcc ( @xmath1)iron is more complex , since it stands at a crossing point between high spin ferromagnetic and antiferromagnetic states , and the actual realization depends sensitively on its actual atomic volume and , possibly , strains @xcite .
tsunoda discovered a decade ago that @xmath1iron could be stabilized as pellets of radii up to 100 nm , with a lattice constant of 3.577 @xcite .
he also found that the magnetic structure of the pellets was helicoidal , with pitch vector @xmath2 .
a number of authors have subsequently looked for theoretical low - energy collinear and non - collinear states appearing in such @xmath1phase . @xcite .
knpfle and co - workers @xcite , who used gga or lda and a full - potential implementation of density functional theory found that the ground state was indeed a spiral with the correct pitch vector .
but since the augmented spherical wave ( asw ) method tends to overestimate the atomic volume , it is difficult to extract what their equilibrium lattice constant @xmath3 might be .
we have found a spiral ground state with @xmath3 of 3.56 , in excellent agreement with the experimental data of tsunoda .
we find two local minima when we plot the energy of the spiral state as a function of the pitch vector , at @xmath4 and @xmath5 , as other authors did @xcite .
we label these two states as s1 and s2 .
the state s1 is the global minimum for lattice constants down to 3.47 , and only below it is s2 the ground state .
we believe that our results represent a significant methodological advance for the simulations of magnetic systems using the siesta package since , on the one hand , we lay down a firm ground for the reliability of atomic bases of different sizes and , on the other , we allow for the description of interesting non - collinear and spiral structures of the spin .
the layout of this article is simple .
section ii provides the theoretical backbone of the article ; subsection iia is a brief reminder of the non - collinear formalism as is applied to dft ; we present subsequently details of our implementation of non - collinearity and non commensurate spiral arrangements of spin for such structures .
section iii is devoted to show and discuss our results for the stability of the different states of iron , both within lda and gga , using different bs , up to an optimal choice .
we finish the article with a short summary .
we devote this section to provide details of our implementation of non - collinear gga and of unconmensurate spiral arrangements .
we believe it is useful to supply first a backbone of non - collinear dft @xcite , which will help us discuss similarities and differences with the latest approaches @xcite . 1 . in a non - collinear material , the direction of the magnetization vector @xmath6 changes at each place in the sample , according to the angles @xmath7 , @xmath8 , + @xmath9 2 .
the density matrix can be decomposed in terms of the electronic density @xmath10 and @xmath11 as + @xmath12 + where @xmath13 denote the three pauli matrices .
there is a single rotation matrix , @xmath14 which brings @xmath15 to collinear form , + @xmath16 3 .
the total energy is a functional of the density matrix , @xmath17=t[\nn]+e_h[\nn]+e_{xc}[\nn]$ ] which , upon variation provides with the effective single - particle hamiltonian + @xmath18\right)\,\tilde{i}+v_{xc}[\nn].\end{aligned}\ ] ] + @xmath19 and @xmath20 are the kinetic and hartree energy functionals , while + @xmath21 + takes account of exchange and correlation . @xmath22 and @xmath23 , in eq .
( 4 ) , are the corresponding potentials . 4 .
the spinor eigenfunctions of @xmath24 , @xmath25 can be used to compute the density matrix , since @xmath26 each eigenfunction can individually be rotated to bring it back to collinear form + @xmath27 5 . the exchange and correlation potential matrix , which is obtained by functional differentiation of the exchange and correlation energy , can be uniquely decomposed in terms of pauli matrices , + @xmath28 + where @xmath29 and @xmath30 . 6 . in lda approximation
, @xmath31 is a function of only one vector , @xmath32 so that it must be proportional to it .
it is then easily shown that @xmath33 is diagonalized by the same rotation matrix as @xmath34 , + @xmath35 + where @xmath36 $ ] .
this can be interpreted physically as rotating the whole system into a collinear reference frame , where @xmath37 can be computed as in conventional lda .
the gga expression for the exchange and correlation energy contains also the vector @xmath38 .
therefore the gga potential includes both spin stiffness and antisymmetric exchange terms ( dzyaloshinskii - moriya @xcite ) + @xmath39 + this implies that @xmath40 can not be fully diagonalized by the @xmath41 rotation matrices , + @xmath42 total free energy as a function of ( a ) number of k points in half the brillouin zone ( with a grid cutoff of 400 ry ) and ( b ) grid cutoff ( with 4.000 k points in half the brillouin zone ) .
an optimized double - zeta ( dz ) basis was used . ] an accurate description of non - collinearity must include a spin stiffness term @xcite , as is the case for classical localized spins .
for instance , the heisenberg hamiltonian is @xmath43 where @xmath44 is the spin stiffness .
our recipe to partly take account of gradient corrections is to neglect the stiffness contribution and evaluate only @xmath45 .
we rotate the density matrix to bring it into collinear form , @xmath46 .
we then compute its gradients @xmath47 in this collinear reference frame , and the collinear potential , @xmath48 .
we finally rotate back .
knpfle and coworkers chose an apparently different approach @xcite .
they rotated both @xmath15 and its gradient , @xmath49 , and then evaluated @xmath23 with the diagonal terms of both matrices , discarding the non - diagonal terms of @xmath49 .
but since @xmath50 where @xmath51 , both approaches are analytically equivalent . kleinman and bylander @xcite added a spin stiffness term to their lda exchange and correlation potential from which they obtained a spiral ground state with @xmath52 at the atomic volume of bulk copper .
the effective lda hamiltonian does not commute with the pauli matrices @xmath53 unless the system be paramagnetic or collinear ( @xmath7 be equal to @xmath54 ) .
therefore the expectation values of @xmath55 are not conserved in the iterative selfconsistency process of dft .
states with @xmath56 can be shown to be metastable , and therefore @xmath7 is conserved in this case .
a convenient variational wave function for either molecules ( or solids ) with non - collinear ( commensurate ) magnetic moments is @xmath57 for solids , the above wave function can be easily rewritten so that it explicitly satisfies block theorem . for helicoidal arrangement of spins of pitch vector @xmath58 ,
the dft hamiltonian commutes with the operator @xmath59 , which translates by a lattice vector @xmath60 and then rotates about the z - axis .
since a wave function of the kind @xmath61 is an eigenfunction of @xmath62 , a generalized bloch theorem holds . @xcite it must be stressed that such wave function is not an eigenstate of @xmath63 since , as noted above , a rotation by a constant angle @xmath64 about the y - axis , @xmath65 , does not commute with the hamiltonian , unless @xmath66 .
we have checked numerically that for any wave - function of the form above , the angle @xmath64 is indeed not conserved by the application of @xmath63 , unless @xmath56 , which corresponds to a metastable situation . such wave function
must therefore be regarded as purely variational .
energy difference ( a ) and number of iterations
( b ) as a function of the electronic temperature .
the calculation was performed with 4.000 k points , a grid cutoff of 400 ry ( 27.000 points ) and an optimized dz basis . ]
siesta uses norm - conserving pseudopotentials ( ncps ) @xcite optimized so that their local part be smooth @xcite .
izquierdo and coworkers @xcite proposed to generate the ncps for iron from the atomic configuration [ ar]3d@xmath674s@xmath68 , with core radii for 4s , 4p , 3d and 4f orbitals set equal to 2.00 a.u .
they found that the optimal radius for partial - core corrections was 0.7 a.u .
we have looked carefully for a pseudopotential which could produce a better fit to bulk bcc iron .
our first criterium to determine the cutoff radii was to compare the eigenvalues of the valence shells of atomic iron obtained from the ncps and the all - electrons hamiltonians and try to minimize their differences .
we found that the radii obtained using such procedure were very different from each other .
moreover , they produced fits to the bulk bcc phase of poor quality as compared with the proposal of izquierdo et al .
inclusion of the partial - core 3p levels into the valence did not help , mostly due to the fact that 3s electrons are still taken as part of the core and therefore there remains a strong overlap between valence and pseudo - core charge .
siesta performs brillouin zone integrations on a grid of monkhorst - pack special points typically extended to cover half of it @xcite .
the integrand is also smeared by a fermi function .
hamiltonian matrix elements are partly computed on a real space grid , whose fineness @xmath69 is controlled by a grid cutoff , @xmath70 .
since the energy of the different states need not shift rigidly when increasing accuracy and moreover competing ground states for ( fcc ) iron have energy differences as tiny as 5 mev , we decided to set the number of k points , the electronic temperature and the grid cutoff to match an accuracy of about 1 mev .
[ kpoints ] and [ temp ] show typical results for the convergence of the energy of bcc ferromagnetic iron as a function of those parameters .
all the calculations shown there were performed using the gga functional as parametrized by perdew and coworkers @xcite and optimized either double - zeta or triple - zeta bases ( see below ) .
we find then that we need 4.000 k points and up to 700 ry ( which corresponds to 50.000 points in the real space grid ) to meet the desired accuracy .
[ temp ] shows that increasing the temperature to an optimal value of about 200 or 300 k speeds the convergence of the selfconsistent process significantly while not damaging the accuracy required for the energy .
( a ) evolution of the free energy of the three most stable states of iron as a function of the size of the bs .
af = antiferromagnetic , hs = high - spin , fm = ferromagnetic .
( b ) cohesive energy curves of those same three states , for the two minima found using the bsd .
the discontinuous curves correspond to radii of 6 a. u. and the continuous to 10 a. u.the calculations have been performed using the gga approximation . ]
siesta allows for a great flexibility in the use of bs of wave functions which describe valence electrons . for each species of atom
, one may specify one shell of s , p , d and f orbitals . within each shell
, one may choose how many wave functions having the required angular symmetry are needed .
a single - zeta basis ( sz ) is equivalent to choosing just one .
completion of the basis leads to double - zeta and triple - zeta bases ( dz , tz ) .
in addition , one may polarize an orbital ( p ) , which means adding wave functions which correspond to one higher angular momentum unit @xcite . the minimum basis required to accommodate the eight valence electrons of iron
would be sz for both s and d orbitals , which provides a total of 6 wave functions per spin .
siesta is set up to the default maximum bs tztp , which corresponds to 72 wave functions ( wf ) per spin .
using more wave is equivalent to filling up the hilbert space and provides a better variational estimate of the ground state .
junquera and coworkers @xcite pointed out that the confinement radius of each orbital are also variational parameters .
very fast calculations or simulations of a large number of atoms may therefore be performed by the use of a single - zeta basis of rather confined orbitals .
such calculations usually provide much of the features of a material or device .
but they are usually regarded are pretty inaccurate , and dz bases with polarized s orbitals are rather used .
we have minimized a few bs ranging from sz - sz - sz ( 9 wf ) to tztp - tztp - tztp ( 72 wf ) .
[ hilbert ] ( a ) shows how the convergence of the energy for the three most stable states of bulk iron , e. g. : bcc ferromagnetic , fcc ferromagnetic high spin and fcc antiferromagnetic as a function of the number of orbitals used in the bs .
we find that a tz - tz - tz bs ( 27 wf ) is essentially converged for p , d and f orbitals , since the free energy of the three states changes only a little if we polarize this bs .
while we have not checked explicitly that a fourth zeta for the s - orbital may still change somewhat the energy , a inspection of the curve induces us to believe that our results are completely converged .
it is also apparent that the states do not shift rigidly upon improving the accuracy .
we shall see below that such an effect is specially damaging in the lda approximation .
free energy and magnetic moment of the ground and lowest excited states of bulk iron as predicted by lda approximation , using ( a ) bsd and ( b ) bst , as a function of the atomic volume . ]
we have paid special attention to the minimization of the bs dzp - sz - dz ( bsd ) and tz - tz - tz ( bst ) , where we have used a grid software program to look for local minima of the energy as a function of the radii of the first zeta of s , p and d orbitals .
we find a first local minimum for somewhat confined radii of about 6 a.u . and a deeper one for radii of about 10 a.u .
we found that the energy still decreased upon looking further away , but thought it worthless to attempt to look for such next minimum .
[ hilbert ] ( b ) shows that extended radii improve both the energy and the lattice constant substantially .
for instance , the lattice constant of the bcc ferromagnetic state obtained using bsd in gga approximation , as predicted by the first minimum is 2.90 , while the second one gives @xmath71 .
( a ) free energy and ( b ) magnetic moment as a function of pitch vector @xmath58 of the spiral state for different lattice constants ranging from 3.48 to 3.61 . ]
we find that dzp bs predict erroneous orderings of the ground and first excited states .
such effect is particularly apparent in the lda approximation .
[ stateslda ] ( a ) shows that bsd erroneously predicts that the lda ferromagnetic bcc state is more stable that the paramagnetic fcc one .
[ stateslda ] ( b ) shows that usage of more complete bs correct such a mistake .
under such proviso , siesta provides pretty accurate results for the lda predictions of the different physical magnitudes .
for instance , the lattice constant , magnetic moment and bulk modulus of ferromagnetic bcc iron are found to be 2.76 , @xmath72 and 2.68 mbar , which compare extremely well with the best all - electrons plane - wave calculations @xcite .
moreover , the lattice constant for paramagnetic fcc , 3.38 , is also very similar to all - electrons estimate of 3.375 , while the energy difference between both states is somewhat underestimated ( 55 versus about 70 - 80 mev ) .
@xcite .
dzp basis sets also provide awkward results in the gga approximation , even though the relative stability of the lowest energy states is correct now , see fig .
[ statesgga ] .
nevertheless , we find that the shape of the energy curves of the fcc states change significantly when we increase the size of the basis from bsd to bst .
now , since the spiral state smoothly interpolates between the ferromagnetic hs and the antiferromagnetic ones , we have increased further the size of the bs .
we have included more polarization orbitals of p , d and f symmetry , and have found that the energy of the three curves is essentially converged ( see fig .
[ hilbert ] ) .
we find equilibrium lattice constant , magnetic moment and bulk modulus of 2.85 , @xmath73 and 1.83 mbar for the ground state , which compare reasonably well with former all - electrons or ultrasoft - pseudopotentials - based plane waves calculations @xcite .
we have computed the properties of the spiral state for lattice constants well below 3.54 , so that the binding energy curve is a clear parabola , with a minimum at @xmath74 , very close to the experimental value ( 2.577 ) .
we have also simulated clusters with a number of atoms ranging from 2 to 5 , using a bst and non collinear gga , as shown in table [ clusters ] .
our calculations compare very well with previous theoretical simulations @xcite and even improve them when comparisons are made with the experimental values found for the @xmath75 cluster @xcite .
llll & @xmath76 ( ) & @xmath77 ( ev / ) & @xmath78 ( @xmath79 + fe@xmath80 & 2.02 & 1.51 & 6.00 + fe@xmath81 d@xmath82 & 2.28 & 1.72 & 5.62 + fe@xmath81 c@xmath83 & 2.27 & 1.88 & 10.00 + fe@xmath84 c@xmath85 & 2.30 & 2.21 & 14.00 + fe@xmath84 t@xmath86 & 1,2@xmath873,4 2.27 & 2.31 & 14.00 + & 1@xmath872 , 3@xmath874 2.65 & & + fe@xmath88 d@xmath89 & 1@xmath872,3 2@xmath873 2.43 & 2.58 & 17.07 + & 1,2,3@xmath874,5 2.37 & & + free energy and magnetic moment of the ground and lowest excited states of bulk iron as predicted by lda approximation , using ( a ) bsd and ( b ) bst , as a function of the atomic volume . ]
we turn now to the predictions for the spiral state in lda .
we scan the energy as function of pitch vector along the @xmath91x and xw directions , where we find the two minima @xmath92 and @xmath93 we talked about in the introduction . on closer inspection of fig .
[ spirallda ] , we see that the energy curves for lattice constants equal or larger than 3.58 @xmath94 only have the @xmath92 minimum .
the second minimum appears when we decrease @xmath76 at or below 3.54 , becoming lowest in energy at @xmath95 .
the curves corresponding to smallest lattice constants are very shallow .
their two minima have almost the same energy up to 1 mev , and are separated by energy barriers as small as 4 mev . if lda were accurate enough for iron , one would expect both phases not only to coexist but also to change dynamically from one to the other .
we finally discuss our results for the spiral structures in gga , where we find that the s2 state has already clearly developed when @xmath96 , but that the ground state is s1 down to lattice constants of 3.47 .
it can be seen again that the energy curves change rather much when we increase the size of the basis .
one of the reasons is that the shape of the binding energy parabola also changes significantly ( see fig .
[ statesgga ] ) .
but for bst , the ferromagnetic state has always considerably higher energy and there is a clearer asymmetric double - well structure with activation barriers of about 5 - 7 mev .
marsman and hafner have also performed simulations of @xmath90-iron under tetrahedric , orthorombic and monoclinic distortions .
however they obtained for the undistorted case a spiral state s2 with equilibrium lattice constant @xmath97 , much smaller than the experimental one .
they also found that the equilibrium lattice constant for s1 was @xmath98 . on the contrary
, we obtain a state s1 with @xmath74 , much closer to experiments .
( a ) free energy and ( b ) magnetic moment as a function of pitch vector @xmath58 of the spiral state for different lattice constants ranging from 3.48 to 3.61 . ]
we have performed a thorough study of the different iron phases in order to provide a good basis for future calculations .
we have paid special attention to the minimization of the different parameters used in the lcao - dft siesta code .
we found that a grid cutoff of 700 ry , 4.000 k points and temperature smaller than 300 k are needed to meet an accuracy of about 1 mev .
we showed that a double - zeta basis is not accurate enough to supply the correct ground state ordering of the iron phases , since for lda it predicts that the bcc phase is less stable than the fcc , which is contrary to all previous calculations .
however , when we use a triple - zeta basis the results change dramatically for both lda and gga .
we obtain for the ferromagnetic bcc and equilibrium lattice constant of 2.85 , a magnetic moment of 2.31 @xmath99 and a bulk modulus of 1.83 mbar , in excellent agreement with experiments .
we have also simulated iron clusters and we found a better estimate of the properties of these materials than previous works . finally , we have also made a profound study of the @xmath90 phase and we found an equilibrium lattice constant of about 3.56 , closer to the experimental value of 3.577 , but with a spiral state s1 instead of the experimental s2 .
however these simulations agree with previous works and even improve them .
j. f. wishes to thank to the authors and developers of the siesta code for a number of helpful discussions and e - mail exchanges : p. ordejn , j. soler , e. artacho , a. garca - arribas and j. junquera .
he also wishes to acknowledge discussions with a. vega and s. bouarab .
optimization of the bs was carried out with the use of the package innergrid which was generously supplied by grid systems .
the work presented here has been funded by the spanish mcyt and the european union under project no .
bfm2000 - 0526 and contract no . hprn - ct-2000 - 00144 , respectively .
v. m. g .- s .
thanks the ministerio espaol de educacin , cultura y deporte for a fellowship . | we have implemented non - collinear gga and a generalized bloch s theorem to simulate unconmensurate spiral arrangements of spins in a density functional theory code based on localized wave functions .
we have subsequently performed a thorough study of the different states of bulk iron .
we determine the minimal basis set required to obtain reliable orderings of ground and excited states .
we find that the most stable fcc phase is a spiral with an equilibrium lattice constant 3.56 . |
this work was supported by epsrc under the oxford condensed matter theory grant gr / m04426 .
the ideas developed initially at an itp ( santa barbara ) workshop in 1997 , and rbs would like to thank the participants , and particularly the directors s. nagel , a. liu and s. edwards , for sharing their insights .
see for example : _ jamming and rheology _ , edited by a. j. liu and s. r. nagel ( taylor & francis , london , 2001 ) , and overviews therein by d. levine ; d. j. durian and a. j. liu ; and j. kurchan . h. m. jaeger , c. h. liu and s. r. nagel , , 40 ( 1989 ) j. b. knight , h. m. jaeger and s. r. nagel , , 3728 ( 1993 ) j. b. knight , c. g. fandrich , c. n. lau , h. m. jaeger and s. r. nagel , , 3957 ( 1995 ) e. r. nowak , j. b. knight , e. ben - naim , h. m. jaeger and s. r. nagel , , 1971 ( 1998 ) e. ben - naim , j. b. knight and e. r. nowak , h. m. jaeger and s. r. nagel , physica d * 123 * , 380 ( 1998 ) c. a. angell and w. sichina , ann .
* 279 * , 53 ( 1976 ) h. m. jaeger and s. r. nagel and r. p. behringer , , 1259 ( 1996 ) h. j. herrmann , in _ disorder and granular media _ , edited by d. bideau and a. hansen ( north - holland , amsterdam , 1993 ) , p.305 - 320 d. c. hong , s. yue , j. k. rudra , m. y. choi and y. w. kim , phys . rev .
e * 50 * , 4123 ( 1994 ) t. boutreux and p. g. de gennes , physica a * 224 * , 59 ( 1997 ) m. nicodemi , a. coniglio and h. j. herrmann , phys .
e * 55 * , 3962 ( 1997 ) e. caglioti , v. loreto , h. j. herrmann and m. nicodemi , phys .
79 * , 1575 ( 1997 ) d. a. head and g. j. rodgers , j. phys .
a * 31 * , 107 ( 1998 ) d. a. head and g. j. rodgers , j. phys .
a * 32 * 1387 ( 1999 ) p. l. krapivsky and e. ben - naim , , 6778 ( 1994 ) l. f. cugliandolo , j. kurchan and l. peliti , , 3898 ( 1997 ) l. bellon , s. ciliberto and c. laroche , europhys
* 53 * , 511 ( 2001 ) j. jckle and s. eisinger , z. phys .
b * 84 * , 115 ( 1991 ) p. sollich and m. r. evans , , 3238 ( 1999 ) k. g. wilson and j. kogut , phys . rep .
* 12*c , 75 ( 1974 ) | a diffusion - deposition model for glassy dynamics in compacting granular systems is treated by time scaling and by a method that provides the exact asymptotic ( long time ) behavior .
the results include vogel - fulcher dependence of rates on density , inverse logarithmic time decay of densities , exponential distribution of decay times and broadening of noise spectrum .
these are all in broad agreement with experiments .
the main characteristics result from a marginal rescaling in time of the control parameter ( density ) ; this is argued to be generic for glassy systems .
glassy dynamics occurs with similar characteristics in a remarkably diverse range of systems @xcite .
this letter attempts to give reasons for this similarity .
+ it begins by considering an idealized model of granular materials , which is perhaps the simplest class of real systems showing glassy behavior .
the model is treated by approximate scaling and then by asymptotically exact methods .
the behavior is similar to that found in simulations and in the chicago experiments @xcite on real granular systems . because we expect diverging characteristic times ( while no diverging length has yet been seen in glassy systems ) the scaling procedure rescales time , @xmath0 . the consequent scaling equation for the control parameter , in this case the density @xmath1 , has a particular ( marginal ) form .
we show at the end of the letter that this leads directly to the best known characteristic features of glassy dynamics .
we also argue at the end that this marginal form arises when are excessively sensitive to changes of the control parameter , as is the case in glassy dynamics in general , so arriving at a universal scaling scenario .
+ all the following characteristics of glassy behavior @xcite are shared by super cooled molecular glass - forming liquids , structural glasses , foams , colloids and even ( shaken ) granular systems : @xmath2 extreme slowing of rates with typically a vogel - fulcher dependence on control parameters ( temperature , or density , etc ) ; @xmath3 slow decay of correlation functions and of the control parameter itself , often fitted to stretched exponential or inverse logarithmic time dependence ; @xmath4 associated aging phenomena ; @xmath5 characteristic broadening of noise spectra .
+ this has raised questions of universality , and it suggests that , despite obvious differences in fundamental mechanisms , something generic might already be present and discernable in simple models .
granular materials show all the standard characteristics when compacting under shaking , even though thermal aspects play no role @xcite .
the simplicity of models ( e.g. exclusion models ) appropriate for shaken granular systems @xcite makes them ideal candidates for such investigation , and we consider one such model here .
+ previous investigations have shown @xcite how a simple one - dimensional deposition - evaporation model ( continuum car parking ) can yield characteristic behavior .
even the noise spectrum ( obtained through simulations on the model @xcite ) agrees well with experimental observations @xcite .
+ the model considered here is closely related to the continuum car parking model and consists of unit sized blocks performing a random walk on a ring of size @xmath6 .
they only interact through hard core potentials and as soon as a gap of unit size opens up between two adjacent blocks , an additional block will be deposited in the gap and the diffusion will continue ( see fig .
[ mo ] ) . diffusion - deposition process on a ring , scaledwidth=40.0%,scaledwidth=9.0% ] this model allows rearrangement ( diffusion ) of the grains to enable another grain to move ( deposit ) .
the model mimics the actual mechanism of rattling within and escape from / entry into rearranging cages formed by surrounding grains . under compaction ,
typical free volume elements become smaller , drastically reducing the probability of finding one big enough to accommodate a block , and hence lengthening time scales .
a qualitative scaling description of this effect is as follows .
consider elementary walkers each carrying a gap of size @xmath7 , which is the average free volume per block .
for the formation of a gap of unit size , @xmath8 such elementary walkers have to coincide .
a typical initial state has on average one elementary walker on each lattice site .
since walkers can move a distance of order @xmath9 in time @xmath0 , by @xmath10 walkers from @xmath11 sites around a particular site can have accumulated there , and do so with approximate probability per unit time @xmath12 it is straight forward to show that this result applies in any dimension .
this rate of unit gap formation is equivalent to a characteristic inverse time @xmath13 for deposition / motion from or into a cage .
equation ( [ probrate ] ) implies that under rescaling of @xmath14 by a dilatation factor @xmath15 , @xmath11 goes to @xmath16 where @xmath17 .
this equality can be written in terms of the free volume density parameter @xmath18 as the scaling equation @xmath19 in this treatment @xmath20 .
however , the exact discussion of the asymptotic behavior obtained below , implies that the proper form of ( [ scaleq ] ) has @xmath21 a constant . in either form
, the scaling equation has a marginal form , in which to leading order the parameter does not rescale .
this marginality can also be shown to occur in the continuous car parking model and we argue below that this is a key feature of all adequate models of glassy behavior , having its physical origin in the exceptional slowing in such systems .
+ first we turn to the full analytic treatment of the model . at some instant
there are @xmath11 blocks on the ring of size @xmath6 .
we choose to consider the gaps between adjacent blocks , @xmath22 , as the dynamical variables .
the vector @xmath23 then performs a random walk on the hyper - surface ( see fig . [ pi ] )
@xmath24 @xmath25 tracing out a random walk trajectory on @xmath26 , being reflected once at @xmath27 and then passing through the portal at @xmath28,scaledwidth=35.0% ] to incorporate the hard - core interactions and the deposition process we impose the boundary conditions : @xmath29 reflecting ( corresponding to two blocks bouncing off each other ) and @xmath30 a _ portal _ that will transfer the system to @xmath31 ( corresponding to a gap of unit size forming and a block being deposited ) .
+ in what follows , unless stated otherwise , @xmath32 means `` asymptotically correct in the large system , high density limit '' , and the quantity @xmath1 ( or equivalently @xmath7 ) will be used as our control parameter .
the time evolution of @xmath33 is governed by @xmath34 here @xmath35 is the time between deposition events at time @xmath0 ( i.e. the total time spent on @xmath36 ) , for the specific realization considered . in order to estimate @xmath37
we define the cross - section @xmath38 as the ratio between the portal area and the total boundary area ( each @xmath39-dimensional ) on each hyper - surface @xmath40
. this will allow us to find the probability of hitting a portal , given that we hit a boundary . through direct volume consideration one gets , after some lengthy analysis , the asymptotically exact result @xmath41 the same analysis tells us that the distances between boundaries of @xmath42 scale as @xmath18 . therefore if we let @xmath43 be the average time it takes for @xmath44 to go between boundaries , we have @xmath45 , where @xmath46 is a constant inversely proportional to the diffusion constant of the blocks .
define , with a slight abuse of notation , @xmath47 to be the time between deposition events at a given density .
since the portal cross section is very small for high densities we expect the random walk to be essentially ergodic between the deposition events .
therefore we can get the distribution for @xmath47 through considering repeated bouncing off the reflecting boundaries ( blocks colliding ) until finally hitting a portal ( deposition event ) .
the probability of hitting a portal on the @xmath46:th contact with the boundary is then @xmath48 , so the probability distribution for @xmath47 is @xmath49 where @xmath50 is a multiple of @xmath43 . in the limit @xmath51 ( [ dist2 ] ) can be considered to be a continuous exponential distribution . by increasing @xmath6 we can , for any fixed density , have an arbitrary number of essentially independent deposition events in any finite time interval . from the law of large numbers
we thus know that the @xmath33-distribution can be made arbitrarily sharp in the large system limit .
therefore @xmath52 . to solve ( [ drhodt ] ) for @xmath53 we need to calculate the average deposition rate at time @xmath0 , @xmath54 .
this is easily done using ( [ dist2 ] ) , @xmath55 which inserted into ( [ drhodt ] ) gives the asymptotic solution @xmath56 for large @xmath0 .
the asymptotic solution ( [ asymp ] ) is not valid at densities low enough to be suitable for comparison with practical simulations . since the cross section ( [ cross ] ) becomes small ( i.e. ergodicity holds between deposition events ) already at quite moderate densities we instead compare a numerical integration of ( [ drhodt ] ) to simulation data ( see fig .
[ rho ] ) .
+ least square fit , with respect to the constant @xmath46 in ( [ rate ] ) , of the numerical integration of ( [ drhodt ] ) , compared to data from a simulation of the evolving density using parallel updates on a ring of size @xmath57.,scaledwidth=45.0%,scaledwidth=15.0% ] to examine the power spectrum we need to include the effects of fluctuations .
since ergodicity holds on @xmath58 , in the densely packed limit , we can write the master equation for the probability distribution of @xmath1 at time @xmath0 , @xmath59 , as , @xmath60 this is easily solved by implementing laplace transform techniques on the time variable . for the initial condition @xmath61 and in the continuum limit the solution is ( to the highest order of @xmath6 in the exponential ) @xmath62 where @xmath63 is the continuum probability density and @xmath64 . in order to keep the density evolution rate finite in the continuum limit @xmath14 has been rescaled as @xmath65 . since @xmath66 diverges strongly as @xmath67 ,
the approximations used to arrive at ( [ appro ] ) break down , at any finite @xmath6 , in the dense limit @xmath68 . however , for large @xmath6 the average of @xmath1 is independent of @xmath6 , while the variance can be made as small as we please .
so for large @xmath6 , matching of ( [ appro ] ) to a gaussian in @xmath1 gives the correct distribution for large @xmath6 .
using the two variable gaussian approximation the two - time density correlation function can be calculated for a system which starts at density @xmath69 and is aged for a time @xmath70 from whereon measurements are performed : @xmath71 in the last step we have used our particular form of @xmath66 and assumed that we are in a region where @xmath72 , but where @xmath73 .
the above result shows that our model exhibits `` simple aging '' in the sense that the two - time correlation function is only a function of @xmath74 . the density - density response coefficient , corresponding to perturbing the density at time @xmath70 and then measuring the response in the density at time @xmath75 , is proportional to ( [ rr ] ) ( ignoring an additive constant ) ( [ rr ] ) .
this gives the simplest form of the fluctuation dissipation violation investigated in @xcite .
we now define the ( complex ) power - spectrum for the aged system as @xmath76 letting @xmath77 denote the exponential integral , this can be written in the scaling form @xmath78 since the rates in our system are of the order of the inverse aging time ( i.e. @xmath79 ) , a fairly large @xmath80 must be considered in order to sample the effects of aging .
therefore there are three asymptotic regimes ( see fig .
[ pow ] ) ; @xmath81 the reason for the lower cut - off in @xmath50 is that the values of @xmath82 are unphysical .
further we know that on short time - scales the system is time translation invariant and therefor the real part of @xmath83 corresponds to the `` ordinary '' power - spectra in this limit .
we naively stretch this analogy to the whole range of possible @xmath84 ( and note that the the imaginary part of @xmath83 stays close to the real part of @xmath83 for the longer time scales ( see fig .
[ pow ] ) ) .
log - log - plot of the real and imaginary parts of the power spectrum scaling function @xmath85 versus @xmath86 for @xmath87 .
the asymptotic regions and the asymptotic behavior ( see equations ( [ regions ] ) ) are indicated.,scaledwidth=45.0% ] since our model only allows deposition of blocks ( grains ) it corresponds to the limit of very weak tapping in @xcite . taking this into account a qualitative agreement can be seen between our analytical results ( see fig . [ pow ] ) and the experimental results presented in @xcite .
further , assuming that the density increases with the depth we can interpret this as an increase of the effective aging time as measurements are done closer to the bottom of the container . doing this
we see that the bulk of the power spectra is moved towards lower frequencies while the maximum height seems to increase with aging .
this is in full accord with the scaling form of @xmath83 .
we now discuss further the nature of time scaling in such models and the question of universality . under rescaling of time ( @xmath88 ) ,
the general renormalization group transformation @xcite of a control parameter @xmath1 , such as the density ( or temperature ) can be written as @xmath89 . if there is a divergence of the characteristic time scale @xmath14 at @xmath90 , then @xmath91 is a fixed point of the transformation . and
linearizing about the fixed point using @xmath92 gives in general @xmath93 where ( from the semi - group property @xmath94 ) the eigenvalue @xmath95 ( if non - zero ) has to have a standard power - law dependence on the time dilation factor @xmath15 , of the form @xmath96 , with @xmath97 .
this leads to @xmath98 , which is ordinary critical slowing .
but the essential characteristic of glassy behavior in general is that the slowing of characteristic rates with the control parameter is excessive , more extreme than with ordinary critical slowing .
then the control parameter has to change only marginally to stretch the time scales significantly .
that is the case when , in ( [ scaleps ] ) , @xmath99
. then higher order terms in the expansion in @xmath18 have to be retained , and ( [ scaleps ] ) becomes of the marginal form @xmath100 where the @xmath15-dependence is again set by the semi - group property @xcite .
as was seen earlier , this form , with @xmath101 , applies to the diffusion - deposition model ( and also the continuum car parking model , etc ) .
eliminating @xmath15 between the scaling equations for @xmath18 and for time gives a generalized vogel - fulcher form : @xmath102 this specific equation is a direct result of the qualitative `` stronger than ordinary critical slowing '' essential characteristic of glassy dynamics .
it suggests that the common vogel - fulcher form ( @xmath101 ) is one of various generic universal forms distinguished by the value of @xmath80 .
the second commonest form would then be the exponential - inverse - squared form ( @xmath103 ) , which has appeared in a glassy model with constrained dynamics @xcite .
+ a second characteristic of glassy systems , including the models discussed here , and responsible for meta stability , is that the control parameter is one whose mean and local fluctuations evolve with the slowed internal dynamics .
its evolution towards the fixed point value then becomes of form @xmath104 when the generalized vogel - fulcher form applies .
this scenario applies , with @xmath101 , to the analytic treatment of the density evolution of the diffusion - deposition model , equation ( [ asymp ] ) .
the slow evolution of the control parameter is in general related to the broadening @xmath105 frequency scale of the noise spectrum .
this time - scaling scenario seems equally applicable for thermal glasses . it does not require a diverging length in glasses , nor a real static equilibrium transition
it does not so far explain possible / universal power laws in the noise spectrum , nor the non universal differences which distinguishes various types of `` glass '' . |
the principle of equal a priori probabilities , which is at the foundation of statistical mechanics , postulates subjective ignorance at the core of our understanding of macroscopic systems .
it assumes that an isolated system is described by the microcanonical ensemble : an equal mixture over all possible microscopic configurations . on the other hand ,
quantum mechanics provides us with a powerful theory for understanding microscopic systems .
it states that an isolated system is described by a pure state , which leaves no room for subjective ignorance . in the attempt to provide a consistent foundation of quantum statistical mechanics
a purely quantum mechanical explanation of the effectiveness of statistical ensembles has been debated for quite a long time . recently
an attempt to provide an alternative foundation to statistical mechanics was suggested , in which entanglement is viewed at the origin of generalized canonical ensembles . in this approach , there is no need to assume the principle of equal a priori probability , and the isolated system can be in a pure state , consistent with quantum mechanics @xcite .
the mathematical justification for the effectiveness of this approach is based on the concentration - of - measure phenomenon , which has also appeared in the literature under the name of typicality @xcite . simplifying the results of @xcite
, one can prove that the vast majority of pure states picked uniformly at random in a sufficiently large hilbert space will be almost indistinguishable at the level of sufficiently small subsystems , stemming from the appropriate statistical mechanical ensemble .
therefore it is not necessary to assume subjective ignorance over the state of the system , describing it with a mixed state , since any pure state will typically give very similar results . despite the elegance and simplicity of this approach
there are aspects which need to be addressed in order to understand its concrete effectiveness .
for example , it has already being pointed out in other contexts @xcite that it is extremely inefficient to reach or create a random pure state in a large system .
this is due to the fact that the uniformly distributed random unitary generating the state requires an exponential number of parameters to be specified , and any dynamical process corresponding to it will take an exponentially long time .
this is in contrast with the relatively fast process of equilibration , which naturally leads the subsystem to its generalized canonical ensemble .
therefore , if typicality has to provide an explanation for the statistical mechanical ensembles corresponding to equilibrium states , then it should also be able to account for fast equilibration processes .
one way of solving this problem is to restrict the class of allowed states in the isolated system .
this kind of approach has also been considered in ref.@xcite . in previous work , we reported first results on the occurrence of typicality in random matrix product states @xcite . constraining the states of the system to be mps - like
is one way to avoid the efficiency problem , mentioned above , in the realization of the concentration - of - measure phenomenon .
mpss are an example of physically meaningful states
. they can be ground states of gapped hamiltonians @xcite , and they are at the core of some computational techniques , based on quantum information theory , which have been recently introduced @xcite .
the principal reason for the effectiveness of mps is that the ground states of many relevant systems can be well approximated by them ( see @xcite for a recent review and original references ) . here
we present more details about typicality in mpss giving further evidence of its relevance . in sec .
ii we provide some background on the literature concerning typicality in random quantum states . in sec .
iii we define the particular set of random mps we are interested in and review previous results on the concentration - of - measure phenomenon in this context @xcite . in sec .
iv we provide a detailed numerical analysis of the initial results discussed in ref . @xcite and present an analytical derivation of the average non - homogeneous rmps .
we also study specific statistical properties of rmps and show how they can be used to efficiently approximate some characteristics of general random pure states .
v is devoted to conclusions .
arguments related to typicality have been presented throughout the quantum and statistical mechanics literature since the very beginning of this field ( e.g. @xcite and chapter 6 in @xcite ) .
more recently , lebowitz has emphasized the importance of typicality in classical statistical mechanics in the context of the second law of thermodynamics @xcite . in refs .
@xcite the importance of typicality was recognized also in the context of quantum statistical mechanics .
a different and more heuristic approach was used in @xcite .
a nice review with original contributions can be found in @xcite . in the following ,
we provide a brief overview of some of these works .
the work in @xcite makes direct use of the concentration - of - measure phenomenon and focuses on the properties of the subsystem s states .
the concentration - of - measure phenomenon is a well known topic in the mathematical literature and appears in many different contexts @xcite . roughly speaking ,
suppose we have a function @xmath0 , whose domain is an n - dimensional vector space equipped with a probability measure .
if @xmath1 does not oscillate too much then , for @xmath2 sufficiently large , @xmath1 is almost constant , with very high probability .
more formally : @xmath3 where @xmath4 is the average value of @xmath1 , @xmath5 are some universal positive constants , and @xmath6 is the lipschitz constant of @xmath1 . in this sense @xmath1 is concentrated around its average value .
the lipschitz constant can be defined as the supremum of the gradient . within this formalism
one can prove , for example , that the area of a hypersphere concentrates around the equator @xcite . when the domain of @xmath1 is a hypersphere the concentration - of - measure phenomenon is also referred to as levi s lemma @xcite .
@xmath7 , the tensor product of a subsystem @xmath8 and a bath @xmath9 .
we denote the dimensions of the corresponding hilbert spaces with @xmath10 and @xmath11 respectively .
the states in the isolated system @xmath12 may satisfy some constraints or restrictions , for example belonging to a small energy - shell with respect to a hamiltonian .
we now consider a uniformly random state @xmath13 .
this can be achieved by generating @xmath14 with a random unitary matrix , distributed according to the haar measure , acting on some reference state in @xmath15 .
consider now the state of the subsystem @xmath16 .
the results in @xcite show that , for almost all @xmath14 , @xmath17 will be very close in trace norm to the generalized canonical distribution .
this is proved by defining the function @xmath18 , for which the lipschitz constant is 2 , and using levi s lemma to obtain a concentration - of - measure result on the function @xmath19 .
then one has to show that also the average value @xmath20 is small : most states are almost at the same distance from the average state , and this distance is small .
this last result comes from the following inequalities @xcite @xmath21}\leq\sqrt{d_{s}^{2}/d_{r}}.\ ] ] from this , using chebyshev s inequality and the fact that @xmath22 , one obtains a first concentration result , i.e. a bound on the probability of fluctuations .
however , levi s lemma gives a much stronger bound @xcite
@xmath23}\right]\leq2e^{-cd_{r}\epsilon^{2}},\ ] ] with @xmath24 .
note that the probability is exponentially suppressed in the dimension of the hilbert space and not just in the size of @xmath12 , which would be the result obtained using chebyshev s inequality . from the bound on the state distance one can easily derive a similar result for any observable , in this way establishing typicality for the subsystem .
it is worth emphasizing that this derivation of a generalized canonical ensemble is fully consistent with the quantum formalism and makes no use of the principle of equal a priori probability . on the other hand ,
the entanglement between the subsystem and the bath is seen as the main mechanism through which incomplete knowledge over the state of the subsystem emerges .
@xmath25 this kind of probability distribution gives an average state which is diagonal and coincides with the dephased state @xcite .
defining @xmath26 , and assuming that the purity of the average state is low ( @xmath27\ll1 $ ] ) , one can show that the variance in the expectation value of @xmath28 is small @xmath29\ ] ] with @xmath30 the normalized variance of the random variables @xmath31 .
note that the operator norm @xmath32 quantifies the range of possible values for the observable @xmath33 . from the above inequality
it follows that a random pure state is likely to yield expectation values for @xmath33 which are very close to the ensemble average @xmath34 $ ] .
this behavior occurs for probability distributions such that @xmath27\ll1 $ ] , and for observables , where @xmath32 does not increase linearly with the dimension of the hilbert space ( which can be argued to be the physically most interesting , see also the comments in @xcite ) .
a trivial way to guarantee that @xmath32 does not increase is to restrict to local observables , supported on finitely many subsystems . from the bound on the variance
one can use chebyshev s inequality to bound the probability that @xmath33 will be far way from its average value @xmath35 .
in fact , in refs . @xcite it has been shown in great generality that most of the time the state is close to some fixed state , the time averaged one , and in this sense there is equilibration of the system .
an interest the typicality arguments reviewed above provide a consistent foundation for quantum statistical mechanics , although some questions can still be raised .
in particular , one can argue about the exponential amount of resources necessary to generate uniformly distributed random states and the fact that apparently dynamical process can generate such a state efficiently . in this sense , one might also wonder whether most of the states in the hilbert space are physically realizable .
a matrix product state is a pure quantum state whose coefficients are specified by a product of matrices . for the case of periodic boundary conditions ( pbc )
a mps can be written as @xmath36\cdots a^{i_{n}}[n]\right)|i_{1}\cdots i_{n}\rangle,\ ] ] whereas with open boundary conditions ( obc ) one has @xmath37\cdots a^{i_{n}}[n]|\phi_{f}\rangle|i_{1}\cdots i_{n}\rangle,\ ] ] with @xmath38 specifying the states at the boundaries , and @xmath39 is a local basis at site @xmath40 .
the matrices @xmath41,a^{2}[k],\dots , a^{d}[k]\}$ ] , with @xmath42 , are @xmath43-dimensional complex matrices , where @xmath44 is the local hilbert space dimension .
we denote as homogeneous mpss the states for which the set @xmath41,a^{2}[k],\dots , a^{d}[k]\}$ ] is the same for all sites @xmath40 . by definition ,
a mps is specified by the set @xmath41,a^{2}[k],\dots , a^{d}[k]\}$ ] .
however there may be different sets of matrices that originate the same mps .
in @xcite it was shown that this gauge degree of freedom can be fixed using a canonical form . in the present work , using the sequential generation of mpss introduced in @xcite , we consider random mpss with a clear operational definition .
consider a quantum spin chain initially in a product state @xmath45 ( with @xmath46 ) and an ancillary system in the state @xmath47 .
let @xmath48 $ ] be a unitary operation on @xmath49 , acting on the ancillary system and the @xmath40th site of the chain ( see fig .
[ fig : fig1 ] ) .
the @xmath50 $ ] matrices are then defined by @xmath51\equiv\langle i,\alpha|u[k]|0,\beta\rangle,\label{eq : amat}\ ] ] where the greek indices refer to the ancilla space and the latin indices to the physical space ( see @xcite for related works on truncated random unitaries ) . for homogeneous mpss
the index @xmath40 is removed , implying that the unitary interaction is the same for all sites in the spin chain . due to unitarity
we have @xmath52^{\dagger}a^{i}[k]=\mathbb{i}_{\chi}$ ] for all @xmath40 in the bulk ( see @xcite for more details ) .
this property , together with a proper normalization of the boundaries , corresponds to a mps of unit - norm . letting the ancilla interact sequentially with the @xmath53 sites of the chain and assuming that the ancilla decouples in the last step ( this can be done without loss of generality , as shown in @xcite ) ,
the state on @xmath54 is described by @xmath55 which is a homogeneous mps with open boundary conditions .
it can be proved @xcite that the set of states generated in this way is equal to the set of obc - mpss .
we choose the interaction characterizing the rmps ensemble to be represented by random unitary matrices @xmath48 $ ] distributed according to the haar measure . ]
since any state can be described by a mps when @xmath56 , typicality naturally holds true for mpss with this scaling of @xmath43 .
what is not obvious is whether one can have typicality in the set of rmpss where @xmath43 increases at most polynomially with the number of particles @xmath57 , with some constant @xmath58 . in @xcite , using a concentration - of - measure result for the group of unitary matrices @xcite , we were able to prove typicality for the expectation values of local observables over @xmath59 sites @xmath60\right)\left(\bigotimes_{k = l+1}^{n}\mathbb{i}[k]\right),\ ] ] with respect to normalized rmps @xmath14 generated according to the sequential construction .
let us define @xmath61}\prod_{k = l+1}^{n}e_{\mathbb{i}[k]}\right),\label{eq : f_def}\ ] ] where the transfer operator @xmath62 is defined as @xmath63}\equiv\sum_{i_{k},j_{k}=1}^{d}\langle i_{k}|o[k]|j_{k}\rangle a^{i_{k}}[k]\otimes a^{j_{k}}[k]^{*},\ ] ] @xmath64}\equiv\sum_{i_{k}=1}^{d}a^{i_{k}}[k]\otimes a^{i_{k}}[k]^{*}.\ ] ] since the @xmath65matrices are sub - blocks or random unitaries @xmath66 , they define a random variable @xmath67 satisfying @xcite ( @xmath68 are positive constants ) @xmath69\leq c_{1}\exp\left(-c_{2}\epsilon^{2}d\frac{\chi}{n^{2}}\right).\label{eq : com_f}\ ] ] this means that if @xmath70 with @xmath71 increasing the size of the system renders the expectation values more concentrated around their averages , i.e. statistical fluctuations will be suppressed .
is @xmath72dimensional , implying a concentration result for @xmath57and @xmath73 from this result one can also derive a weaker concentration result for the probability of fluctuation of the trace distance at the sub - system level . defining @xmath74 , and analogously for the average state ,
we have ( see appendix c in @xcite for a derivation ) @xmath75\leq4^{l}c_{1}exp(-c_{2}\epsilon^{2}d\frac{\chi}{n^{2}}).\label{eq : com_state}\ ] ] this shows that although the number of mps is much smaller than the total number of states , picking random matrix product states @xmath14 according to the above construction still provides a concentration - of - measure result for suitable random variables defined through @xmath76 this fact could be exploited numerically for the efficient simulation of quantum systems .
@xcite provides a possible indication of the role that the concentration of measure may play in numerical simulations . in this work a different class of random mps
is constructed and used for the efficient sampling of local observables .
we suspect that the efficiency of the sampling is due to a concentration of measure phenomenon .
we point out that the above result concerning the subsystem state is weaker than the one obtained in @xcite , where the probability of fluctuations is exponentially suppressed in the dimension of the total hilbert space , i.e. doubly - exponentially in @xmath53 .
similar weaker bounds have been obtained in the context of k - designs @xcite .
in this section , we discuss numerical signatures of typicality for the case of random pure states distributed according to the haar measure . these simulations are useful in illustrating typicality and the tightness of the theoretical bounds , and they also serve as a reference point for understanding the properties of rmps . we first want to understand the effect of the number @xmath77 of sampled states used in the averages.in fig .
[ fig : d1-micro ] we evaluate the trace distance between the empirical average random state and its exact value for the circular unitary ensemble ( cue ) which is the ensemble of unitary matrices corresponding to the haar measure , as a function of @xmath77:d when increasing the number of sampled states . for this purpose
we consider the average hilbert - schmidt distance for a sub - system of one qubit , @xmath78 , which is easier to evaluate numerically than the trace distance.from fig .
[ fig : conv - d2 ] we see that 500 states seem already sufficient to suppress the statistical fluctuations in the subsystem . ) , as the number of states sampled is increased .
the size of the system is 3 , 6 and 8 qubits , going from the bottom to the top curve . ] , as the number of states used is increased , and for different bath sizes . ]
now , using 500 random pure states , we compare the bound given in @xcite for @xmath79 and its empirical value obtained from the simulations . fig .
[ fig:4eps ] shows how the analytical value and the numerical results are pretty close and scale in the same way with the bath size @xmath80 . in the figure
we also plot the actual value of the average distance using the trace norm and hilbert - schmidt norm .
( with @xmath81 and @xmath11 the dimension of the system and the bath ) , and the numerical average value of @xmath79 and @xmath82 for general random states distributed with the haar measure . ]
again we need to estimate the size @xmath77 of the sample , in order to avoid statistical fluctuations that would be too strong .
we look at how the average trace distance from the average rmps for the subsystem , @xmath83 , converges when the sample size is increased .
[ fig : conv - mps - obc20 ] shows this quantity for homogeneous mpss with obc , @xmath84 and @xmath85 .
the different curves in the figure are for different bath sizes : in the right we have the curves for baths of 20 , 30 and 40 qubits ( from top to bottom with 30 and 40 almost indistinguishable ) .
it may appear that for 500 states the average value has not totally converged .
however , if one compares more bath sizes , as done in the upper part of fig .
[ fig : conv - mps - obc20 ] with 8 , 12 , 20 and 40 qubits , it can be seen that the fluctuations are small in the scale of interest . , for homogeneous mpss and obc as the number of states used in the average increases for @xmath84 and different bath sizes .
] , for homogeneous mpss and obc as the number of states used in the average increases for @xmath86 ( from top to bottom ) . ]
@xmath79 as a function of the difference between the size of the system and the subsystem , for fixed but different values of @xmath43 and for @xmath87 .
the symbols denote homogeneous mps , whereas the symbols joined by lines denotes non - homogeneous mps with the same value of @xmath43 .
periodic boundary conditions are imposed . ] as a function of the difference between the size of the system and the subsystem for different values of @xmath43 and for @xmath88 .
the symbols denote open boundary conditions , whereas the symbols joined by lines denote periodic boundary conditions , both for homogeneous mps . ]
we now consider the dependence of @xmath79 in @xmath43 , for fixed system and bath sizes . in fig .
[ fig : logd1-d ] it can be seen that @xmath79 also decreases with @xmath43 , until a value that depends on the bath size . for small system ( up
to 8 qubits ) we checked that this limiting value is the same as the one for the cue . since @xmath43 can be viewed as a sort of correlation length , this behavior is easily explained considering that when the correlation length is of the order of the size of chain , @xmath79 can not decrease anymore . as a function of @xmath43 for different values of the bath s size @xmath89 .
we consider homogeneous mpss with obc . ] as a function of the difference between the size of the system and the subsystem for @xmath90 .
we consider homogeneous mpss with obc . ]
an important aspect in the study of the concentration - of - measure for rmps is the characterization of the average state .
below we provide an analytical expression for the averaged mps state and numerical verifications of the results obtained .
we use bold faced letters to denote vectors : @xmath91 .
let @xmath92a^{i_{2}}[2]\cdots a^{i_{n}}[n]\right)|i_{1}i_{2}\cdots i_{n}\rangle\ ] ] be an instance of rmps , and the @xmath33-matrices are defined as @xmath93\equiv{\rm tr}_{\mathcal{f}}\left[\left(|i_{k}\rangle\langle0|_{\mathcal{f}}\otimes\mathbb{i}_{\mathcal{a}}\right)\cdot u\left[k\right]\right],$ ] where @xmath94 is the ancillary hilbert space and @xmath95 is the physical hilbert space . for brevity of notation
we will not always specify the site index of the @xmath33-matrices and @xmath66-matrices .
the density matrix @xmath96 is given by @xmath97 @xmath98 @xmath99 the factor in the coefficient of the density matrix can also be written as @xmath100 @xmath101=\ ] ] @xmath102,\ ] ] where @xmath8 is the operator which cyclically permutes the states in @xmath103 @xmath104 the coefficient of the density matrix ( [ eq : avedm ] ) can then be rewritten as @xmath105 \right ) \otimes \left . \left
( \bigotimes_{k=1}^{n}u[k]^ { * } \right ) \right].\end{aligned}\ ] ] in a more compact form we can define a new averaged density matrix in @xmath106 @xmath107\right)\otimes\left(\bigotimes_{k=1}^{n}u[k]^{*}\right)}],\label{eq : average_state}\ ] ] which is related to the previous @xmath108 in the following way @xmath109 in the previous expression the quantity @xmath110\right)\otimes\left(\bigotimes_{k=1}^{n}u[k]^{*}\right)}\ ] ] carries information about the average over the haar measure . since , for non - homogeneous rmps , permutations of the factors inside the integral are allowed , one sees that @xmath110\right)\otimes\left(\bigotimes_{k=1}^{n}u[k]^{*}\right)}=\otimes_{k=1}^{n}\overline{u[k]\otimes u[k]^{*}}=\pi_{\chi d}^{\otimes n},\ ] ] where @xmath111 is the density matrix of the maximally entangled state , ( see the appendix in @xcite ) @xmath112 now let us go back to the expression for @xmath113 .
since the operator @xmath8 that cyclically permutes the ancilla spaces acts on @xmath53-tensor copies of the same state , it does not change the state @xmath114\right)\otimes\left(\bigotimes_{k=1}^{n}u[k]^{*}\right)}\right]\ ] ] @xmath115,\ ] ] and the trace will just restrict the projector to the maximally entangled state over the physical subspace @xmath116=\pi_{d}^{\otimes n}.\ ] ] now it is easy to see that @xmath117 @xmath118 this proves that the average non - homogeneous rmps is the completely mixed state for any value of @xmath43 @xmath119 since the average state is the same for the two ensembles , all the functional depending only on the first moment of the distribution will be identical in the non - homogeneous rmps ensemble and cue ensemble .
this is the case for the expectation value of observables @xmath120 the case of homogeneous rmps is more complicated , although a close expression can be obtained starting from eq .
( [ eq : average_state ] ) .
the problem comes from the fact that now it is not anymore possible to factorize the average in the tensor product of unitaries , but nevertheless the integral can be expressed as @xcite @xmath121 where the vector @xmath122 is obtained superimposing the column of the matrix @xmath123 , an orthonormal representation of the permutation @xmath124 acting on @xmath53 elements . in general
it is hard to find such a representation for large @xmath53 and @xmath44 . in this section
we numerically check the previous results . as the number of states used in the average is increased , for @xmath125 ( bottom ) and @xmath126 ( top ) .
for each system size we considered the case of mpss with @xmath127 and @xmath128 and general states ( cue ) , however the different plots can barely be distinguished in the plot scale . ] varying @xmath43 for different system sizes @xmath53 .
the numbers of states used is @xmath129 .
it can be seen that there is no dependence on @xmath43 as expected from our analytical result in eq .
[ eq : result ] .
the small fluctuations are due to the finite sampling . ]
now we want to obtain more insight into the average mps , studying its behavior for larger system .
we use the hilbert - schmidt norm between the average mps and the completely mixed state , since this can be calculated efficiently , in contrast to the trace norm .
numerically , when we originate a number @xmath77 of rmps @xmath130 the ensemble average is given by @xmath131 we then have @xmath132=\text{tr}\left[(\overline{\rho}^{r})^{2}\right]-\frac{1}{d^{n}}.\ ] ] is is clear from this expression that a necessary condition for a small distance is a low purity of the average states .
one may expect that this limit is not reachable for mixed states constructed with mpss , since the entanglement of each element in the ensemble has a bound depending on @xmath43 .
the purity of the average mps is given by @xmath133=\frac{1}{r^{2}}\sum_{i , j=1}^{r}\frac{|\langle\psi_{i}|\psi_{j}\rangle|^{2}}{\langle\psi_{i}|\psi_{i}\rangle\langle\psi_{j}|\psi_{j}\rangle},\ ] ] and the overlap between different mpss can be efficiently evaluated . before showing the results , let us rewrite the previous formula , remembering that we want the purity to attain its minimum value @xmath134 @xmath133=\frac{1}{r}+\frac{1}{r^{2}}\sum_{i\neq j}\frac{|\langle\psi_{i}|\psi_{j}\rangle|^{2}}{\langle\psi_{i}|\psi_{i}\rangle\langle\psi_{j}|\psi_{j}\rangle}.\ ] ] in the limit of a large sample , the first term vanishes , and the second term should converge to @xmath134 .
it is clear that for large systems , @xmath135 , the principal limitation comes from the finite size of the sample .
note that these expressions are also valid for general states .
let us study how much the second term on the left hand side of the previous equation differs from @xmath136 this is shown in fig .
[ fig : purity ] as a function of the system size , and for different values of @xmath43 .
an exponential decrease of the purity is observed for systems of up to 50 qubits . in the plot we considered values of @xmath137 and @xmath138 .
however , they all collapse to the same point .
in fact , fig . [
fig : purity - chi ] shows that there is no relevant dependence on @xmath43 with a very high accuracy .
we also study the relative error of the purity in relation to the analytical value .
this is an upper bound for the trace distance ( again the term @xmath139 is neglected):@xmath140-\frac{1}{r})-\frac{1}{d^{n}}}{\frac{1}{d^{n}}}=d^{n}||\overline{\rho}^{r}-\frac{\mathbb{i}}{d^{n}}||_{2}^{2}-\frac{d^{n}}{r}.\ ] ] fig .
[ fig : errorpurity ] shows this expression as a function of the number of qubits and for different values of @xmath141 and @xmath138 ( from top to bottom ) . note that already for @xmath127 the error is around 10% and does not increase with the system size .
some fluctuations are observed , and we believe they are due to the finite size of the sample . as a function of the number of qubits .
the continous line corresponds to @xmath142 .
the points are for values of @xmath137 and @xmath138 , but the difference can hardly be seen . ] , neglecting the first term @xmath139 as a function of @xmath43 for different number of qubits @xmath53 .
there is no dependence on @xmath43 as expected . ] , as a function of the number of qubits @xmath53 and for different values of @xmath43 .
note that already for @xmath127 the error is around 10% and does not increase with the system size .
the fluctuations are due to the finite size of the sample . ]
the scheme for the sequential generation of rmps can be viewed as an algorithm for the efficient generation of random states .
the computational efficiency is given by the number of random unitary matrices needed , which equals the size of the system , and by the size of these matrices , which depends on @xmath143 in this section we study how well the ensemble of non - homogeneous rmps mimics a subset of statistical features associated with the haar measure .
in particular we consider the average bipartite entanglement , the minimum eigenvalue distribution and higher moments of the reduced density matrix .
the distribution of entanglement produced by the present scheme for the generation of random states can be a relevant quantity for quantum information tasks .
the average bipartite entanglement ( abe ) or global entanglement @xcite , here denoted with @xmath144 is a measure of multipartite entanglement .
it is defined as @xmath145,\ ] ] where @xmath146 is the reduced density matrix of the i - th qubit .
@xmath147 can have values between 0 and 1 , ranging from a product state to a maximally entangled state . in fig .
[ fig : abe ] we show histograms of q obtained over @xmath148 realizations of rmps for different system sizes .
the distribution shows the same features as the ones obtained with other random circuit schemes @xcite . increasing the dimension of the hilbert space the values of @xmath147 concentrate around the average .
this is a manifestation of the concentration of measure phenomenon . in fig .
[ fig : avebe_4_6](a ) we plot the difference between the average value of @xmath147 for rmps and the exact value known for the cue ensemble : @xmath149 .
the simulation shows that the average @xmath150 , for @xmath43 sufficiently large , depends only on the size of the system and the larger the system the closer the average value is to the cue exact result : @xmath151 , with @xmath2 the size of the system .
although from the numerical simulations the difference between the two is always finite .
this behavior is similar to other random circuit constructions @xcite , but in our case the size of the system plays the role of the depth of the random circuit and , for a fixed @xmath152 determines the computational cost of the simulation .
this is also an indication of the fact that the second moment of the rmps @xmath153 is different from @xmath154 .
this can be seen from the fact that the average purity is a functional of two copies of the random state : @xmath155 , where @xmath8 is the swap operator .
the discrepancy between the rmps and cue ensembles can then be detected from the second moment of the distributions .
as a function of the rmps rank @xmath143 ( b ) standard deviation of @xmath147 for a system of @xmath2 qubits . ( c )
@xmath156 with @xmath157 and @xmath158 .
( d ) @xmath159 as functions of @xmath143 ] in fig .
[ fig : avebe_4_6](b ) we plot the decrease of the standard deviation of @xmath147 as a function of the system size @xmath2 , with @xmath160 .
the exponential decrease is again a signature of the concentration - of - measure phenomenon for rmps : for large system size almost all rmps have @xmath147 exponentially close to the average value @xmath150 .
the other quantities we want to compare with the ones obtained from the cue are the higher moments of the reduced density matrix evaluated for a partition of the system in two parts , @xmath33 and @xmath9 , of dimension @xmath161 and @xmath11 respectively .
calculations in the cue provide the following exact results @xcite @xmath162 @xmath163 @xmath164 fig .
[ fig : avebe_4_6](c ) shows how the distance of the averaged rmps value from the cue results depends on @xmath43 .
as can be seen a finite and relatively small value of @xmath43 is sufficient to guarantee a very good approximation of the cue value .
again , after some point there is no improvement in increasing the value of @xmath152 as already observed for the average bipartite entanglement in fig .
[ fig : avebe_4_6](a ) .
we also consider the statistical properties of the minimum eigenvalue of the reduced density matrix of a subsystem @xmath33 of dimension @xmath161 .
this is a quantity related to the entanglement of the subsystem , and its values can range from @xmath165 , for product states , to @xmath166 , for maximally entangled states @xcite .
[ fig : avebe_4_6](d ) shows the dependence on @xmath43 of @xmath159 , where @xmath167 and @xmath157 ( in a system of @xmath168 qubits ) .
the figure indicates an exponential convergence to the cue value .
the fluctuations seen in the plot are due to the finite size of the sampling set and to the small value of the quantity that we want to estimate ( of the order of @xmath169 ) .
this result is again an indication of the good accuracy that can be obtained in approximating some properties of the cue with the rmps states .
in conclusion , in this work we have studied in detail a set of random matrix product states ( rmps ) introduced in ref . @xcite . as already pointed out in that reference
, rmpss can be a useful tool to address foundational problems of quantum statistical mechanics .
in particular here we have proved that the set of non - homogeneous rmps and the set of uniformly distributed general random states have the same average state .
this property , together with the validity of the concentration - of - measure phenomenon , implies that any generalized canonical state can be approximated by the reduced density matrix of a random matrix product state , as long as the average random mps coincide with the associated averaged microcanonical ensemble .
let us call @xmath170 the totally mixed state of a global hilbert space @xmath15 satisfying some set of restrictions denoted with @xmath12 ( e.g. having a fixed energy ) , and whose mps states are denoted with @xmath171 . assuming that the average random mps satisfies @xmath172 and substituting this in eq.[eq : com_state ] it follows with @xmath174 since @xmath175 is nothing but the generalized canonical state this prove our statement .
notice that this is true only assuming measure concentration in the restricted space and the equality @xmath172 .
the present work focused on the case when the hilbert space has no restrictions .
an interesting future direction of research would be to check the identity between @xmath176 and @xmath177 when some kind of constraints are imposed .
another interesting application of rmpss is in the field of pseudo - random quantum circuits .
we show that statistical properties of general quantum random states , which are computational resources for some quantum information tasks @xcite , are very well approximated by rmpss .
since this states can be generated efficiently , as long as @xmath43 scales polynomially in the size of the system , they constitute an efficient tool for the approximate simulation of random quantum states . | we study the set of random matrix product states ( rmps ) introduced in pra * 81 * , 032336 as a tool to explore foundational aspects of quantum statistical mechanics . in the present work ,
we provide an accurate numerical and analytical investigation of the properties of rmps .
we calculate the average state of the ensemble in the non - homogeneous case , and numerically check the validity of this result .
we also suggest using rmps as a tool to approximate properties of general quantum random states .
the numerical simulations presented here support the accuracy and efficiency of this approximation .
these results suggest that any generalized canonical state can be approximated with high probability by the reduced density matrix of a random mps , if the average mps coincide with the associated microcanonical ensemble . |
small oscillations about the stable equilibrium of a many - body ground state are quantized as bosonic quasi - particles or bosons . in various physical contexts
the linearized equations of motion for such excitations are known as the random phase approximation , or rpa equations for short @xcite .
concrete examples are furnished by the vibrational excitations of a solid , the spin waves of a magnet , the electromagnetic modes of an optical medium , or the density oscillations of a bose - einstein condensate . constrained by the requirement of dynamical stability ,
the hamiltonian @xmath2 of any vibrational or quasi - boson system of the mentioned kind must lie in a positive cone , @xmath3 , of so - called elliptic symplectic generators .
it should be stressed that although @xmath2 is hermitian as an operator in fock space , the quantum - to - classical mapping sends @xmath2 to an rpa generator @xmath4 which is in general neither hermitian nor anti - hermitian as a linear operator on the classical phase space . in view of this , a distinctive feature of the set of elliptic generators @xmath5 is that they can be brought to diagonal form ( with real frequencies , corresponding to stable oscillatory motion ) by real bogoliubov transformations , i.e. , by conjugating with elements of the real symplectic group @xmath6 . in this general
setting , our goal is to investigate what happens with observables such as the spectral statistics and the transport properties when the bosonic system is strongly disordered . in particular , we wish to understand whether there exists some low - energy universality , possibly of an unusual type , due to the interplay between disorder and the geometry of the cone @xmath3 .
( for example , a high degree of low - temperature universality is known to be observed @xcite in strongly disordered solids as well as amorphous or glassy systems . ) motivated by this question , in the present paper we initiate the study of a class of semi - realistic random matrix models for disordered bosons . by construction ,
the probability measures of the models we propose are supported on @xmath3 .
thus , unlike @xcite , the unphysical behavior of runaway motion associated with complex frequencies is excluded .
our rpa generators @xmath7 have the particular feature of being sums of a deterministic and a random part .
for simplicity we focus here on models without time - reversal symmetry , although tr - invariant models of a similar kind can be treated with little extra effort . by using a variant of the efetov - wegner supersymmetry method
, we will derive an approximation for the density of states of mean - field or self - consistent type , reminiscent of the so - called ` coherent potential approximation ' ( cpa ) @xcite . in the longer term ,
the goal is to develop a description of our disordered boson models by field theories of the non - linear sigma model type .
( as is well known @xcite , such a description has proven very useful for the case of disordered fermions ) . in that formulation , universality ( if any ) is expected to emerge whenever the renormalization group flow gets attracted to a few - parameter manifold of renormalizable field theories .
the plan of the paper is this . in section [ sect:2 ]
we outline the basic setting and in particular , we review the notion of positive cone of elliptic symplectic generators . we also introduce the random models to be considered and give a summary of the analytical results obtained .
section [ sect:3 ] is concerned with the derivation of the coherent potential approximation for our models . for pedagogical reasons ,
we first discuss the zero - dimensional case in some detail .
the extension to @xmath1 dimensions is given in section [ sect:4 ] .
there we also show some results for the numerical solution of the cpa equation .
in this section , we start with some background on mathematical formulation and describe a class of random models which are tractable by the superbosonization variant of the efetov - wegner supersymmetry method .
we then give a summary of our analytical results , relegating the presentation of numerical results to the end of the paper .
let @xmath8 be a set of creation and annihilation operators for bosonic quasi - particles with quantum numbers @xmath9 .
( for example , @xmath10 might be a discrete set of momenta selected by periodic boundary conditions in a finite box . )
such operators span a hermitian symplectic vector space @xmath11 with symplectic form @xmath12 : \ ; w \times w \to \mathbb{c}$ ] defined by the canonical commutation relations @xmath13 = \delta_{k k^\prime } \ ; , \qquad [ a_k , a_{k^\prime } ] = 0 \ , , \qquad [ a_k^\dagger , a_{k^\prime}^\dagger ] = 0 \,.\ ] ] we now assume that we are given a linear hamiltonian dynamics on @xmath14 .
this may be interpreted either as a linear hamiltonian flow on @xmath14 viewed as a classical phase space , or as a quantum time evolution on @xmath14 viewed as a subspace of the associative algebra of polynomials in @xmath15 , @xmath16 ( the so - called weyl algebra ) . in either case ,
the dynamical equations are @xmath17 ( the bar means complex conjugation . ) in order for the canonical commutation relations to be invariant under the dynamics , we require that @xmath18 and @xmath19 .
thus @xmath20 are the matrix elements of an anti - hermitian matrix @xmath21 , while @xmath22 are those of a complex symmetric matrix @xmath23 .
altogether , these conditions mean that @xmath24 is the generator of a symplectic transformation .
more precisely , defining the lie algebra , @xmath25 , of the complex symplectic group by the linear condition @xmath26 @xmath4 lies in a non - compact real form @xmath27 determined by @xmath28 it should be mentioned that this description is appropriate in the absence of time - reversal invariance . if time reversal is a symmetry of the physical system , then the time - evolution generator @xmath4 is subject to additional complex anti - linear conditions .
equations ( [ eq : rpa ] ) arise as the equations of motion for a system of non - interacting bosons with hamiltonian @xmath29 and dynamics @xmath30 $ ] .
alternatively , one may imagine that they arise as an approximation to the collective motion of an interacting many - particle system ; as a particular example we mention density oscillations of a fluid .
in the latter case , equations ( [ eq : rpa ] ) go under the name of random phase approximation ( rpa ) .
the characteristic frequencies of the dynamical system ( [ eq : rpa ] ) or equivalently , the single - boson energies of the hamiltonian @xmath2 , can be computed as the eigenvalues of @xmath4 . owing to the symplectic condition @xmath31 the characteristic polynomial satisfies @xmath32 .
the eigenvalues of @xmath4 therefore come as pairs @xmath33 .
if @xmath4 lies at some random position in the real symplectic lie algebra @xmath34 , then its eigenvalues will typically be complex , since @xmath35 is neither hermitian nor anti - hermitian . in the present context ,
however , complex eigenvalues are forbidden , as they correspond to the unphysical situation of unstable motion .
in fact , the physical requirement of stability of the rpa dynamics dictates that the spectrum of @xmath4 must lie on the imaginary axis , so that the normal modes of the bosonic system are vectors in @xmath14 with periodic time dependence ( @xmath36 ) .
moreover , the second - quantized hamiltonian @xmath2 in ( [ eq:2ndq ] ) must have a ground state in fock space . by standard considerations of linear algebra ,
all these stability conditions are met if and only if @xmath4 lies in the set @xmath37 we refer to @xmath3 as the _ positive cone of elliptic generators _ in @xmath34 .
it is a fact that every @xmath5 can be brought to diagonal form by a real bogoliubov transformation , i.e. an element @xmath38 of the real symplectic group @xmath6 , which is defined by the condition @xmath39 in the present paper we consider rpa generators @xmath4 , or equivalently hamiltonians @xmath2 , which are a sum of two parts : @xmath40 the term @xmath41 is the deterministic ( i.e. , non - random ) part of @xmath4 . while the formalism developed below can in principle handle any choice of @xmath41 , the explicit calculations presented in section [ sect : apply ] will be carried out for a simple concrete model of @xmath41 with unit mass matrix and elastic constants given by a discrete laplacian .
a precise description of the concrete model for @xmath41 is as follows .
let @xmath42 be a cubic lattice in @xmath1 space dimensions and associate with each site @xmath43 the operators @xmath44 and @xmath45 for boson creation and annihilation , respectively .
we then take the second - quantized hamiltonian to be @xmath46 where the sum for the second term on the right - hand side is over nearest neighbor pairs of sites @xmath47 of @xmath48 .
such a hamiltonian is easily diagonalized by fourier transforming to momentum space .
the spectrum of single - boson energies @xmath49 as a function of the wave vector @xmath50 comes out to be @xmath51 note that @xmath52 for small @xmath53 , which tells us that the speed of sound in units of the lattice spacing is @xmath54 . by computing the rpa generator from the dynamical equation
@xmath55 $ ] for @xmath56 and @xmath57 we obtain the expression @xmath58 where @xmath59 is the scaled lattice laplacian ( with diagonal part removed ) which has eigenvalue spectrum @xmath60 .
next , we tensor up the model by introducing @xmath0 identical bands . mathematically speaking , we pass from the symplectic vector space ( for each @xmath43 ) @xmath61 to the tensor product @xmath62 and take the generator @xmath41 to be @xmath63 this means that creation operators @xmath64 and annihilation operators @xmath65 get an extra band index @xmath66 .
note that in the physical setting of lattice vibrations a reasonable choice of @xmath0 in @xmath1 dimensions is @xmath67 due to the vector nature of lattice displacements .
we turn to @xmath68 , the second term in ( [ eq : mz3 ] ) , which is random .
a particular feature of our disordered model is that we take @xmath68 to be diagonal in the sites @xmath43 of the lattice .
for simplicity we begin the discussion with the very special case of @xmath48 consisting of just a single site . the full model to be discussed later
is obtained by repeating the single - site discussion at every site of @xmath42 . with the single site of the lattice we associate a hermitian vector space @xmath69 with symplectic structure @xmath70 .
in order to implement the positivity condition [ see eq .
( [ eq : cone ] ) ] for @xmath4 to be in the cone @xmath3 , we let @xmath71 where @xmath72 is a rectangular linear operator @xmath73 mapping @xmath14 into an auxiliary vector space @xmath74 .
the dimension @xmath75 is a parameter of our model .
it may be bigger or smaller than @xmath76 .
a special role is played by the choice @xmath77 , as this is the minimal dimension for the operator @xmath68 to have full rank .
it is easy to see that for @xmath78 the symplectic condition @xmath79 holds if and only if @xmath72 satisfies the reality condition @xmath80 this condition fixes a real form , say @xmath81 , of the complex vector space @xmath82 .
note that if @xmath83 has full rank then @xmath84 and @xmath85 .
disorder is introduced by declaring the matrix elements of @xmath72 to be gaussian random variables .
more precisely , we define the probability measure for @xmath86 as @xmath87 where @xmath88 is lebesgue measure on the normed vector space @xmath81 and @xmath89 is a normalization constant .
the parameter @xmath90 is a measure of the disorder strength .
we mention in passing that the model for @xmath91 with probability measure ( [ eq : meas ] ) ( and @xmath92 ) is equivalent to the random matrix model studied in @xcite by different methods .
finally , we describe the generalization to an arbitrary lattice or graph @xmath48 . with each lattice site
@xmath43 we associate one copy @xmath93 of the hermitian symplectic vector space @xmath94 .
the total physical space then is the orthogonal sum @xmath95 .
note that @xmath96 where @xmath97 denotes the number of sites of @xmath48 .
the full generator of the dynamics is @xmath7 where the deterministic part @xmath41 may in principle be any element of the positive cone @xmath98 . for concreteness
we let @xmath42 and take @xmath41 to be the generator described in section [ sect:2.2.1 ] .
the random part @xmath68 is a sum @xmath99 of @xmath100 made from independent and identically distributed random operators @xmath101 . in other words ,
the distribution for @xmath68 is given by the product distribution @xmath102 while our interest will ultimately be in correlation functions and transport properties , we here take a first step by studying the average resolvent of the time - evolution generator @xmath103 @xmath104 where the symbol @xmath105 means the expectation value with respect to the probability measure ( [ eq : meas - lambda ] ) . notice that by the symplectic property @xmath106 the resolvent satisfies @xmath107 , so @xmath108 is an odd function of the frequency parameter @xmath109 .
because the support of our probability measure is contained in the positive cone of elliptic elements , @xmath3 , the eigenvalue spectrum of the random operator @xmath4 is always imaginary and @xmath110 is analytic in the right and left halves of the complex @xmath111-plane . in the following we assume @xmath112 .
it is a standard fact that the local density function @xmath113 of the characteristic boson frequencies @xmath114 can be computed from @xmath115 we now come to our main result .
fixing the ratio @xmath116 we take the large-@xmath0 limit of the model with dynamical generator @xmath7 on @xmath42 as described above .
we then claim that in this limit @xmath110 is expressed by @xmath117^d } \frac{d^d k } { z^2 + p^2 + p\nu(2-\delta_k ) + \nu^2 ( 1-\delta_k ) } \,,\ ] ] where the complex and energy - dependent quantity @xmath118 is a solution of the self - consistency equation @xmath119^d } \frac{d^d k}{(2\pi)^d}\ ; \frac{p + \nu ( 1 - \frac{1}{2}\delta_k ) } { z^2 + p^2 + p\nu(2-\delta_k ) + \nu^2 ( 1-\delta_k ) } \,.\ ] ] @xmath118 plays the role of a ` self energy ' or ` coherent potential ' .
we briefly discuss some features of the solution in two extreme cases .
there is only one relevant parameter , @xmath120 . in the limit of weak disorder ( @xmath121 ) one
infers that @xmath122 and @xmath123^d } \frac{d^d k } { z^2 + \nu^2 ( 1-\delta_k)}\ ] ] is simply the cauchy transform of the local density of frequencies of the deterministic generator @xmath41 . on the other hand , for strong disorder ( @xmath124 )
the coherent potential @xmath125 becomes large and the system ( [ eq : g(z ) ] , [ eq : cpa ] ) simplifies to @xmath126 a special situation arises for @xmath127 . in this case
it follows by a short computation from ( [ eq : rmt - limit ] ) that the scaled function @xmath128 satisfies an equation , @xmath129 which was derived and solved by lueck , sommers , and one of the authors @xcite .
the analysis of @xcite shows that the density of states behaves as @xmath130 near @xmath131 in this case .
numerical results for @xmath132 and for the more general situation of intermediate disorder strengths will be presented in sections [ sect : rmt - limit ] and [ sect : apply ] .
in the sequel we explain how to arrive at our main equations ( [ eq : g(z ) ] ) and ( [ eq : cpa ] ) . for pedagogical reasons
we describe the method first for the simple situation of a single site ( the zero - dimensional case ) . throughout this section
we let @xmath69 and @xmath133 .
our plan is to compute the average trace of resolvent ( [ eq : resolvent ] ) by a variant of the efetov - wegner supersymmetry method .
the first step of this method is to express @xmath134 as a gaussian berezin ( super-)integral . to get started , we use the elementary identity @xmath135 and then write each of the two determinants as a gaussian integral using ordinary integration variables for the determinant in the denominator and anti - commuting variables for that in the numerator . in the case of the ordinary gaussian
integral , there exists a convergence issue because the elements @xmath35 have indefinite real and imaginary parts in general .
it is therefore crucial that all our generators @xmath4 , constrained to lie in the positive cone @xmath3 , satisfy the inequality @xmath136 . to take advantage of this positivity property
, we express the determinant as follows : @xmath137 note that owing to @xmath112 and @xmath138 , the @xmath139 matrix of operators in ( [ eq:3.3 ] ) has positive real part .
next , we introduce symmetric complex bilinear inner products @xmath140 for each of the two vector spaces @xmath74 and @xmath14 .
these inner products are consistent with the hermitian structures of @xmath74 and @xmath14 in the sense that , e.g. for @xmath74 , the sesqui - bilinear form @xmath141 agrees with the hermitian scalar product of @xmath74 .
we then express the reciprocal determinant @xmath142 as an integral over two complex vectors @xmath143 and @xmath144 @xmath145 where it is understood that we are integrating with the product of lebesgue measures for @xmath74 and @xmath14 .
the normalization is chosen in such a way that @xmath146 .
we emphasize that the integral ( [ eq : boson - det ] ) converges absolutely due to @xmath147 and @xmath148 . in the case of the determinant itself
we integrate in the sense of berezin ( i.e. , we actually differentiate ) with respect to two independent vectors @xmath149 and @xmath150 whose components are grassmann variables : @xmath151 again , it is understood that we are integrating with the flat berezin form , i.e. , the product of all partial derivatives w.r.t . the grassmann variables .
the bar in the present instance means nothing but independence , e.g. , of @xmath152 from @xmath149 .
we now multiply the two gaussian integral formulas ( [ eq : boson - det ] ) and ( [ eq : fermion - det ] ) and take the disorder average inside the absolutely convergent integral to obtain @xmath153 this integral representation is a suitable starting point for further analysis . next , we compute the disorder expectation value in ( [ eq : mz-3.6 ] ) . for that we introduce the quadratic quantities @xmath154 where @xmath155 is meant as a linear transformation from @xmath14 to @xmath74 with coefficients in the even part of a grassmann algebra , and similar for @xmath156 with the roles of @xmath14 and @xmath74 reversed .
we then have @xmath157 by completing the square and shifting variables . at this point
we make the observation that @xmath158 depends on @xmath159 , @xmath160 , @xmath149 , @xmath152 only through scalar products such as @xmath161 , @xmath162 , @xmath163 .
these share the feature of invariance under the group @xmath164 of real orthogonal transformations of @xmath133
. it will be useful to organize all these @xmath164-scalars into a supermatrix : @xmath165 two of the matrix entries vanish since @xmath166 and , similarly , @xmath167 . we also have @xmath168 .
we further note the expression @xmath169 for the supertrace of @xmath170 .
there exist certain linear dependencies amongst the matrix elements of @xmath170 . to describe them we need the operation @xmath171 of taking the supertranspose : @xmath172 with its help we can formulate the symmetries of @xmath170 as follows : @xmath173 we now arrange the remaining integration variables @xmath174 , @xmath175 , @xmath150 , and @xmath176 in the form of rectangular supermatrices : @xmath177 more
precisely , @xmath178 is to be viewed as a linear mapping from @xmath14 into the superspace @xmath179 with grassmann - even resp .
grassmann - odd matrix coefficients on the even resp .
odd positions of @xmath180 .
it is easy to check the identity @xmath181 which lets us re - express our disorder average as @xmath182 the next step is to carry out the integral over @xmath174 , @xmath175 , @xmath150 , and @xmath176 , thereby eliminating @xmath178 and @xmath184 from the calculation
. this will be straightforward to do because the dependence on these variables is gaussian . as a preparatory step ,
we verify from @xmath185 the relation @xmath186 where @xmath187 by using the results ( [ eq : mz-3.10],[eq : mz-3.11 ] ) in equation ( [ eq : mz-3.6 ] ) we then arrive at our next formula : @xmath188 the integral on the right - hand side is still over the original variables @xmath159 , @xmath160 , @xmath149 , @xmath152 in @xmath170 and @xmath174 , @xmath175 , @xmath150 , @xmath176 in @xmath178 , @xmath184 .
finally , by using a standard formula for gaussian berezin superintegrals we perform the integral over @xmath178 and @xmath184 .
this results in @xmath189 the superdeterminant here is over the tensor product space @xmath190 .
we recall that the superdeterminant of a supermatrix is defined by @xmath191 the integral above is still over the variables @xmath159 , @xmath160 , @xmath149 , and @xmath152 entering via their scalar products into the supermatrix @xmath170 . by scaling these integration variables so that @xmath192 , we obtain the following expression for the generating function of our problem : @xmath193 the symbol @xmath194 stands for the identity in superspace @xmath179 . superbosonization is a change of variables @xcite which lets us switch from integrating over a large number of vector - type variables , to integrating over a smaller number of matrix - type variables . in the present context
these are the components of the vector variables @xmath159 , @xmath160 , @xmath149 , @xmath152 and the matrix elements of the supermatrix @xmath170 , respectively .
such a reduction of the number of integration variables does not come for free but requires the integrand to be invariant under one of the lie groups @xmath195 , @xmath196 , or @xmath197 .
there exists a version of superbosonization for each of these classical lie symmetries .
as we have seen , our integrand is expressed in terms of quadratic invariants of the orthogonal group @xmath198 .
therefore we now recall from @xcite the superbosonization identity for the case of @xmath164-symmetry .
the @xmath164-superbosonization identity reads @xmath199 where on the left - hand side we integrate with the flat berezin form @xmath200 and on the right - hand side the berezin integration form is @xmath201 where @xmath202 is still the flat berezin form ( i.e. , the product of differentials for the even variables and partial derivatives for the odd variables ) .
the domain of integration for the so - called boson - boson block @xmath203 [ see eq .
( [ eq : def - p ] ) ] is the space of positive hermitian @xmath204 matrices @xmath205 subject to @xmath206 . in the fermion - fermion sector , the integration domain is the space of unitary @xmath204 matrices @xmath207 subject to the symmetry relation @xmath208 .
these matrix spaces are diffeomorphic to the symmetric spaces @xmath209 and @xmath210 respectively
. by applying the superbosonization identity ( [ bosonize1 ] ) to the integral representation ( [ eq : om(z ) ] ) , we obtain our final result for the generating function : @xmath211 notice that a substitution @xmath212 was also made . by the relation @xmath213 this removes the multiplicative constant @xmath214 from ( [ eq : om(z ) ] ) . the result ( [ eq : final ] ) is exact and mathematically rigorous for @xmath215 .
( in the present case of @xmath164-symmetry the superbosonization identity fails for @xmath216 ; see @xcite . ) from it we get the average trace of resolvent by differentiating at coinciding points @xmath217 : @xmath218 to conclude this section we consider the special limit of vanishing deterministic generator @xmath219 . in that case our expression simplifies to @xmath220 where all superdeterminants and supertraces are over @xmath179 .
recalling the parameter @xmath221 we see that our integral is of the form @xmath222 with @xmath223 we now investigate the random - matrix limit @xmath224 with @xmath225 held fixed . in this limit the integral for @xmath226 can be computed by the saddle - point or laplace method . by the principles of supersymmetry , the leading contributions to the integral at @xmath217 can be shown @xcite to come from saddle points which are multiples @xmath227 of the identity .
we here omit the details of the calculation and present only the outcome . by execution of the saddle - point method
we find that @xmath228 where @xmath118 is a solution of the saddle - point equation @xmath229 in section [ sect:2.3 ] [ see eq .
( [ eq : rmt - limit ] ) ] we already remarked that for @xmath127 this is equivalent to an equation analyzed and solved in @xcite .
hence in what follows we focus on @xmath230 .
we begin with the case @xmath231 .
a plot of the density of states for @xmath232 is shown in figure [ fig : rmt ] .
we see that there is a gap at low frequencies .
this feature can be understood in the same way as the marcenko - pastur law @xcite for rectangular wishart matrices .
indeed , recall that our random generator is @xmath78 where @xmath233 is rectangular of size @xmath234 .
the non - zero eigenvalues of @xmath68 coincide with those of the operator @xmath235 but the latter has @xmath236 additional eigenvalues at zero by rank - nullity . in the large-@xmath0 limit the level repulsion due to this macroscopic number of zero modes produces a spectral gap of size proportional to @xmath237 .
the gap closes as @xmath238 approaches unity , leading at @xmath127 to the situation investigated in @xcite .
for @xmath239 it is the operator @xmath83 which by rank - nullity has @xmath240 zero modes , and the same goes for @xmath241 .
therefore the density of states contains a dirac mass @xmath242 at zero in this case .
a plot of the density of states for @xmath239 is shown in figure [ fig : rmt ] , where see that the dos approaches a finite value at zero frequency . (
the argument of macroscopic level repulsion does not apply here , as the operator @xmath68 is neither hermitian nor anti - hermitian . )
the discussion above is concerned with the so - called bulk scaling limit .
another limit of interest is the edge - scaling limit at @xmath243 where one sends @xmath224 while keeping @xmath244 fixed . for @xmath127
this limit was thoroughly investigated in @xcite , while for @xmath231 the situation is trivial because of the absence of states at @xmath243 . for @xmath239 the edge - scaling limit
was studied in @xcite .
the hessian of the function @xmath245 at the saddle point @xmath246 has eigenvalues of order @xmath109 . therefore ,
in edge scaling @xmath247 this saddle point is not isolated and one has to work with a whole supermanifold of saddle points .
( technically speaking , the saddle - point supermanifold is a riemannian symmetric superspace @xmath248 of type @xmath249 . )
the law for the density of states in the limit @xmath250 turns out @xcite to be the universal law for systems of class @xmath251 in the symmetry classification of @xcite .
we now turn to the @xmath1-dimensional model described in section [ sect:2.2 ] .
the procedure of deriving the coherent potential approximation ( [ eq : g(z ) ] , [ eq : cpa ] ) for this model remains essentially the same as before .
again , our first step is to express the determinants in ( [ eq : mz-3.1 ] ) as gaussian integrals over vector variables @xmath159 , @xmath160 , @xmath149 , @xmath152 for @xmath252 and @xmath174 , @xmath175 , @xmath150 , @xmath176 for @xmath253 .
the gaussian integral representation has the effect of factorizing the independent random variables associated with different sites of the lattice .
the disorder average can therefore be carried out for each site separately . by the local @xmath164 gauge symmetry of the model ,
the integrand after disorder averaging depends only on @xmath164 gauge invariant combinations of the fundamental variables @xmath159 , @xmath160 , @xmath149 , @xmath152 . these organize into supermatrices @xmath170 as before .
thus we introduce such a supermatrix @xmath254 for each site @xmath43 and switch to integrating over @xmath254 by the superbosonization formula ( [ bosonize1 ] ) . because the dependence on the variables @xmath174 , @xmath175 , @xmath150 , @xmath176 is still gaussian , they can again be integrated out to produce a superdeterminant . in this way
we obtain @xmath255 where @xmath256 denotes the orthogonal projector from @xmath257 onto @xmath258
. the only difference of any essence from our earlier result ( [ eq : final ] ) is that the integral now is over a field of supermatrices @xmath259 instead of a single supermatrix @xmath170 .
the operator @xmath260 is diagonal on @xmath14 but ( for generic @xmath254 ) non - diagonal on @xmath179 . on the other hand ,
the operator @xmath261 is trivial on superspace but couples the sites of the graph @xmath48 .
the inverse square root of @xmath262 is raised to the @xmath263 power because each of the @xmath0 bands of the deterministic limit contribute the same factor .
we now face the task of analyzing the model ( [ eq : final ] ) by the field - theoretic methods of gradient expansion and renormalization .
( note that a closely related problem has already been tackled in @xcite . ) hoping to make progress with this in a future publication , we here take a first step by computing the local density of states .
let us finally work out the mean - field solution of the model ( [ eq : final ] ) with deterministic generator @xmath264 as defined in ( [ eq : mz6 ] ) .
writing the integrand as @xmath265 ( for @xmath266 ) we take the general variation of @xmath267 : @xmath268 for large @xmath0 we expect the field integral to be essentially given by a spatially homogeneous saddle point @xmath269 ( independent of @xmath43 ) and small fluctuations around it .
therefore , after setting @xmath270 we look for solutions of @xmath271 of this very form .
the variational equation @xmath271 then reduces to an equation of self - consistent mean - field type : @xmath272 where we have used the property that the laplacian @xmath59 is diagonal with eigenvalues @xmath273 in momentum space . by evaluating the trace of the matrix inverse
we immediately arrive at equation ( [ eq : cpa ] ) . within this mean - field ( or coherent potential ) approximation scheme , we obtain the expression ( [ eq : g(z ) ] ) for the average resolvent trace @xmath110 .
let us finish by showing some numerical results for the model in dimension @xmath274 . in this case
the density of states from ( [ eq : mz-2.21 ] ) for the pure system ( @xmath275 ) is @xmath276 as is seen in figure [ fig : cpa ] , the van hove singularity at @xmath277 is still visible for @xmath278 ( and @xmath279 . as the disorder strength @xmath90
is increased , the bulk of the spectrum is pushed to higher frequencies and a peak begins to develop at small frequencies ( see the plot for @xmath280 ) . at values of @xmath90
much larger than the sound velocity @xmath281 we recover the random - matrix limit shown in figure [ fig : rmt ] .
99 p. ring and p. schuck , _ the nuclear many - body problem _
( springer , new york , 1980 ) r.o .
pohl , x. liu , and e. thompson , rev .
* 74 * ( 2002 ) 991 x. barillier - peruisel , o. bohigas , and h.a .
weidenmller , ann . phys .
* 324 * ( 2009 ) 1855 j. korringa , j. phys .
solids * 7 * ( 1958 ) 252 j.l .
beeby , phys .
* 135 * ( 1963 ) a130 p. soven , phys
* 156 * ( 1967 ) 809 f. evers and a.d .
mirlin , rev .
* 80 * ( 2008 ) 1355 t. lueck , h .- j .
sommers , m.r .
zirnbauer , j. math .
* 47 * ( 2006 ) 103304 p. littelmann , h .- j .
sommers , m.r .
zirnbauer , commun .
* 283 * ( 2008 ) 343 s. schmittner , diploma thesis , universitt zu kln ( october 2010 ) v.a .
marcenko and l.a .
pastur , math .
ussr - sb * 1 * ( 1967 ) 457 a. altland and m.r .
zirnbauer , phys .
b * 55 * ( 1997 ) 1142 | a family of random models for bosonic quasi - particle excitations , e.g. the vibrations of a disordered solid , is introduced .
the generator of the linearized phase space dynamics of these models is the sum of a deterministic and a random part .
the former may describe any model of @xmath0 identical phonon bands , while the latter is a @xmath1-dimensional generalization of the random matrix model of lueck , sommers , and zirnbauer ( lsz ) .
the models are constructed so as to exclude the unphysical occurrence of runaway solutions . by using the efetov - wegner supersymmetry method in combination with the new technique of superbosonization
, the disordered boson model is cast in the form of a supermatrix field theory .
a self - consistent approximation of mean - field type arises from treating the field theory as a variational problem .
the resulting scheme , referred to as a coherent potential approximation , becomes exact for large values of @xmath0 . in the random - matrix limit , agreement with the results of lsz is found .
the self - consistency equation for the full @xmath1-dimensional problem is solved numerically . |
the universe is presently undergoing an accelerated phase of expansion .
several candidates have been proposed in the literature to explain this cosmic accelerated expansion , such as dark energy models , the generalized chaplygin gas , modified gravity and scalar - tensor theories , tachyon scalar fields and specific braneworld models , such as the dvali - gabadadze - porrati ( dgp ) model .
however , it was pointed out that assuming that dark energy is governed by a perfect fluid equation of state may systematically induce wrong results and be misleading in inferring the nature of dark energy @xcite . in this spirit
, an alternative model was recently proposed without the presence of exotic fluids and modifications of the friedmann equations , using a more complicated equation of state , namely , the van der waals ( vdw ) equation of state @xcite , given by @xmath0 where @xmath1 is the energy density and @xmath2 the pressure of the vdw fluid .
the accelerated and decelerated periods depend on the parameters , @xmath3 , @xmath4 and @xmath5 of the equation of state , and in the limiting case @xmath6 , one recovers the dark energy equation of state , with @xmath7 .
it was also stressed that the perfect fluid equation of state @xmath8 reflects an approximation of cosmic epochs describing stationary situations , where phase transitions are not considered @xcite .
thus , one of the advantages of the vdw model is that it describes the transition from a matter field dominated era to a scalar field dominated epoch , without introducing scalar fields .
furthermore , it is useful to explain the universe with a minimal number of ingredients , and the vdw gas actually treats dark matter and dark energy as a single fluid . the vdw quintessence scenario has also been confronted successfully with a wide variety of observational tests , by constraining the free parameters @xcite .
the success of this model has stimulated several different approaches .
recently , a model with a binary mixture whose constituents are described by a vdw fluid and by a dark energy density was also proposed @xcite .
it is interesting to note that this model can simulate several aspects , namely , an inflationary period where the acceleration grows exponentially ; and a present accelerated period where the dark energy density dominates over the energy density of the vdw fluid .
the construction of a general scheme where the vdw ideal fluid has a mathematically equivalent representation as a scalar - tensor theory with a specific potential , as well as explicit examples were also explored in ref .
another approach has been analyzed in the context of brane cosmology @xcite , where the cosmological fluid on the brane is modeled by the vdw equation of state .
it was shown that this model reproduces several features , namely , an initial accelerated epoch where the vdw fluid behaves like a scalar field with a negative pressure , and a present accelerated phase due to a cosmological constant on the brane . despite of the fact that the vdw equation of state represents a spatially homogeneous cosmic fluid and
is assumed not to cluster , due to gravitational instabilities inhomogeneities may arise .
thus , the solutions outlined in this paper may possibly originate from density fluctuations in the cosmological background , resulting in the nucleation through the respective density perturbations .
therefore , the pressure in the vdw equation of state may be regarded a radial pressure , and the transverse pressure is determined through the einstein field equations .
the justification of this approach may be considered as an extension of the construction of inhomogeneous wormhole spacetimes supported by phantom energy @xcite .
the latter construction was motivated by the analysis carried out in ref .
@xcite , where time - dependent solutions were constructed with ghost scalar fields . in this model , it was shown that the radial pressure is negative everywhere and equals the transverse pressure far from the throat , showing that ghost scalar fields behave essentially as dark energy .
now , in the vdw context , one may also consider , in principle , that for large radial distances in these inhomogeneous solutions the radial pressure equals the transverse pressure , and thus behaving essentially as a vdw cosmological fluid , consequently justifying the above approach .
the transverse pressure obtained from the field equations should not be considered typical of the cosmic fluid driving the accelerated expansion , as it is only a feature of these inhomogeneous vdw configurations .
it is also important to emphasize that the equation of state leading to the acceleration of the universe on large scales , is an average equation of state corresponding to a background fluid .
so in the inhomogeneities giving rise to the configurations discussed in this work , it is also important to justify if one may really use an equation of state that in fact corresponds to a background fluid . for this , it is important to understand the evolution of structure formation in the vdw scheme .
it was shown that for large redshifts the vdw model is similar to the @xmath9cdm model @xcite , where the vdw dark matter plays the role of a cosmological constant , showing that structure formation could evolve in a similar manner as in the @xmath9cdm model , although these points need to be further investigated @xcite .
now , it was also shown that due to a negative adiabatic sound speed , perturbations are unstable in the vdw quintessence scenario , and inevitably the role of anisotropic stresses and entropy production has to be taken into account .
thus , as a first approximation and as a working hypothesis , one may argue that this also justifies the above approach of using an equation of state that corresponds to a background fluid , with a radial pressure .
we shall denote these inhomogeneous spherically symmetric solutions as _ van der waals quintessence stars_. it is also interesting to note that the thermodynamic properties of a general relativistic isotropic pressure self - gravitating ball of fluid , undergoing a generic first order phase transition and governed by the vdw equation of state , was studied in ref .
@xcite . in this work
, we shall generalize the latter to an anisotropic pressure general relativistic model , and we shall consider two specific class of solutions of vdw quintessence stars , namely , gravastars and traversable wormholes .
the `` gravastar '' ( _ grav_itational _ va_cuum _ star _ ) picture developed by mazur and mottola @xcite , initially arouse from a considerable amount of scepticism related to spacetime singularities and event horizons in the schwarzschild black hole solution ( see ref .
@xcite for a review on specific misconceptions and ambiguities related to these issues ) . in the gravastar model ,
the interior schwarzschild solution is replaced with a de sitter condensate , and thus , does away with the problem of the singularity at the origin and the event horizon . in this model ,
the quantum vacuum undergoes a phase transition at or near the location where the event horizon is expected to form . in this spirit , it is interesting to note that several models have been proposed to some extent in the literature @xcite .
recently , motivated by dark energy quintessence , which is a possible candidate responsible for the late - time cosmic accelerated expansion , a generalization of the gravastar model has been proposed @xcite , by considering an interior solution governed by the dark energy equation of state , @xmath10 .
stable dark energy stellar models were analyzed in ref .
@xcite , and it was found that large stability regions exist that are sufficiently close to where the event horizon is expected to form .
it was also emphasized that it would be difficult to distinguish the exterior geometry of the dark energy stars from an astrophysical black hole . in this context , it is interesting to note that stars supported by a generalized chaplygin gas equation of state have been explored @xcite . in this work
, we shall also be interested in investigating the construction of traversable wormholes @xcite supported by the vdw equation of state .
an analogous approach was recently carried out in the presence of phantom energy @xcite and the generalized chaplygin gas @xcite .
we will show that traversable wormhole solutions may be constructed using the vdw equation of state , which are either asymptotically flat or possess finite dimensions , where the exotic matter is confined to the throat neighborhood .
the latter solutions are constructed by matching an interior wormhole geometry to an exterior schwarzschild vacuum .
analogously to their phantom and chaplygin counterparts , these vdw quintessence wormholes have far - reaching astrophysical and cosmological consequences , such as the production of closed timelike curves and the consequent violation of causality .
this paper is outlined in the following manner : in sec .
[ sec : eq - structure ] , the equations of structure for van der waals quintessence stellar models are presented . in sec .
[ sec : gravastar ] , an exact solution of a gravastar is found , and the respective properties and characteristics are analysed . in sec .
[ sec : wormhole ] , an exact solution of a traversable wormhole geometry is also found . in sec .
[ conclusion ] , we conclude .
consider a static and spherically symmetric spacetime , given by the following metric , in curvature coordinates @xmath11 where @xmath12 and @xmath13 are arbitrary functions of the radial coordinate , @xmath14 .
the function @xmath13 is the quasi - local mass , and is denoted as the mass function . the einstein field equation , @xmath15 , where @xmath16 is the einstein tensor and @xmath17 the stress - energy tensor , provides the following relationships @xmath18 where the prime denotes a derivative with respect to the radial coordinate , @xmath14 .
@xmath19 is the energy density , @xmath20 is the radial pressure , and @xmath21 is the lateral pressure measured in the orthogonal direction to the radial direction . equation ( [ anisottov ] ) corresponds to the anisotropic pressure tolman - oppenheimer - volkoff ( tov ) equation , and may be obtained using the conservation of the stress - energy tensor , @xmath22 .
the four - velocity of a static observer , at rest at constant @xmath23 , is @xmath24 .
the observer s four - acceleration is @xmath25 , so that taking into account metric ( [ metric ] ) , we have @xmath26 and @xmath27 note that from the geodesic equation , a radially moving test particle , which initially starts at rest , obeys the following equation of motion @xmath28 @xmath29 is the radial component of proper acceleration that an observer must maintain in order to remain at rest at constant @xmath30 .
one may consider that the geometry is attractive if @xmath31 , i.e. , observers must maintain an outward - directed radial acceleration to keep from being pulled into the star ; and repulsive if @xmath32 , i.e. , observers must maintain an inward - directed radial acceleration to avoid being pushed away from the star .
this distinction depends on the sign of @xmath33 , as is transparent from eq .
( [ radial - acc ] ) . in particular , for a constant redshift function , @xmath34 ,
static observers are also geodesic .
thus , the convention used is that @xmath35 is positive for an inwardly gravitational attraction , and negative for an outward gravitational repulsion @xcite . despite of the fact that , in a cosmological context , the van der waals fluid is considered homogeneous
, inhomogeneities may arise through gravitational instabilities , resulting in a nucleation of the cosmic fluid due to the respective density perturbations .
thus , these solutions may possibly originate from density fluctuations in the cosmological background .
consider the vdw equation of state for an inhomogeneous spherically symmetric spacetime , given by @xmath36 in the present context of relativistic stellar models , we shall consider the pressure in eq .
( [ vdwstar ] ) as the radial pressure , @xmath37 , and the tangential pressure may be obtained via eq .
( [ anisottov ] ) .
note that the limiting case @xmath38 , reduces to the dark energy equation of state , with @xmath39 .
using eqs .
( [ mass])-([pressure ] ) , then eq . ( [ vdwstar ] ) provides the following relationship @xmath40 note that now we have four equations , eqs .
( [ mass])-([anisottov ] ) and eq .
( [ vdwstar ] ) , with five unknown functions of @xmath14 , i.e. , @xmath12 , @xmath13 , @xmath19 , @xmath20 and @xmath21 . to solve the system ,
one may adopt different strategies , namely , one may model an appropriate spacetime geometry by imposing @xmath13 and/or @xmath12 by hand and consequently determine the stress - energy tensor components . in counterpart
, one may consider an adequate source of the spacetime geometry by imposing the stress - energy components , and consequently determine the metric fields .
another useful and interesting relationship may be obtained from eqs .
( [ mass ] ) , ( [ anisottov ] ) and eq . ( [ vdwstar ] ) , which provide @xmath41 + m'r\left[1+\frac{\gamma}{1-\frac{\beta m'}{4\pi r^2}}-\frac{\alpha m'}{4\pi r^2 } \right]\phi ' \bigg\ } \label{tovdark2 } \,.\end{aligned}\ ] ] @xmath42
is denoted the anisotropy factor , as it is a measure of the pressure anisotropy of the fluid comprising the vdw quintessence star .
the factor @xmath43 represents a force due to the anisotropic nature of the stellar model , which is repulsive , i.e. , being outward directed if @xmath44 , and attractive if @xmath45 .
@xmath46 corresponds to the particular case of an isotropic pressure vdw quintessence star .
one may , in principle , construct asymptotically flat spacetimes , where @xmath47 , and @xmath48 , as @xmath49 .
an alternative approach is to consider a cut - off of the stress - energy tensor at a junction radius @xmath50 .
for instance , consider for simplicity that the exterior solution is the schwarzschild spacetime , given by @xmath51 @xmath52 may be interpreted as the vdw quintessence star s total mass . in this case
the spacetimes given by the metrics eq .
( [ metric ] ) and ( [ metricvacuumlambda ] ) are matched at @xmath50 , and one has a thin shell surrounding the star . using the darmois - israel formalism ,
the surface stresses are given by @xmath53 \ , , \label{surfpressure}\end{aligned}\ ] ] where the overdot denotes a derivative with respect to the proper time , @xmath54 ; @xmath55 is the surface energy density and @xmath56 the surface pressure ( see refs .
@xcite for details ) .
the static case is given by taking into account @xmath57 .
the total mass of the vdw quintessence star , for the static case , is given by @xmath58 \,,\ ] ] where @xmath59 is the surface mass of the thin shell , and is defined as @xmath60 .
the mazur - mottola model is constituted by an onion - like structure with five layers , including two thin - shells , with surface stresses @xmath61 and @xmath62 , where @xmath55 is the surface energy density and @xmath63 is the surface tangential pressure .
the interior of the solution is replaced by a segment of de sitter space , which is then matched to a finite thickness shell of stiff matter with the equation of state @xmath64 .
the latter is further matched to an external schwarzschild vacuum with @xmath65 .
a simplification of the mazur - mottola configuration , and its dynamic stability , was considered in ref .
@xcite using a three - layer solution @xcite , i.e. , a de sitter interior solution was matched to a schwarzschild exterior solution at a junction surface , comprising of a thin shell with surface stresses @xmath55 and @xmath63 .
consider the specific case of a constant energy density , @xmath66 , so that eq .
( [ mass ] ) provides the following mass function @xmath67 where for simplicity , the definition @xmath68 is used . using eq .
( [ vdweos ] ) , one finds the following expression @xmath69 where , for notational convenience , the constant @xmath70 is defined by @xmath71 now , in order to obtain an accelerated behaviour of the universe , the condition @xmath72 , should be obeyed .
we shall use this condition , in the present inhomogeneous spacetime , which yields the following inequality @xmath73 note that in order to provide a positive energy density , one may also impose the following simultaneous conditions , in terms of the vdw equation of state parameters @xmath74 see ref .
@xcite for further details .
taking into account the mass function given by eq .
( [ const - rho ] ) , and using eq .
( [ mass ] ) , we have @xmath75 , so that inequality ( [ acc - univ ] ) takes the following form @xmath76 from which one readily verifies , using eq .
( [ grav - phi ] ) , that @xmath77 , providing the necessary repulsive character , which is a fundamental property of gravastar models .
note that as we are considering positive energy densities , so that @xmath78 , in the gravastar models , we have @xmath79 . equation ( [ grav - phi ] ) , can be integrated to provide the following spacetime metric @xmath80 the stress - energy scenario is given by @xmath75 and @xmath81\,.\end{aligned}\ ] ] the anisotropy factor is provided by the following relationship @xmath82 which is plotted in fig . [ fig : delta ] .
note that @xmath46 , i.e. , @xmath83 , at the center @xmath84 as was to be expected . and @xmath85.,width=268 ]
the spacetime metric representing a spherically symmetric and static wormhole is given by eq .
( [ metric ] ) , where @xmath12 and @xmath86 are arbitrary functions of the radial coordinate , @xmath14 , denoted as the redshift function , and the form function , respectively @xcite .
the radial coordinate has a range that increases from a minimum value at @xmath87 , corresponding to the wormhole throat , to infinity . to be a wormhole solution a flaring out condition of the throat
is imposed , i.e. , @xmath88 @xcite . from this
we verify that at the throat @xmath89 , the condition @xmath90 is imposed to have wormhole solutions . for the wormhole to be traversable
, one must demand that there are no horizons present , which are identified as the surfaces with @xmath91 , so that @xmath12 must be finite everywhere .
note that the condition @xmath92 is also imposed .
consider , for instance , the specific case of of @xmath93 , which obeys the conditions of a wormhole , so that eq .
( [ vdweos ] ) provides the following solution @xmath94\,\ln(r^2-r_0 ^ 2)+\frac{\pi r_0 ^ 2 \gamma}{8\pi r_0 ^ 2 + \beta}\ln(8\pi r^4+\beta r_0 ^ 2 ) \nonumber \\ & & + \left(\frac{\alpha}{16\pi r_0 ^ 2}-\frac{1}{2}\right)\ln(r ) -\frac{\alpha}{32\pi r^2 } -\frac{\alpha r_0 ^ 2}{64\pi r^4 } -\frac{\sqrt{2\pi \beta}\,r_0\gamma}{2(8\pi r_0 ^ 2 + \beta)}\ ; \arctan\left(\frac{2r^2}{r_0}\sqrt{\frac{2\pi}{\beta}}\right)+c\ , , \label{phi}\end{aligned}\ ] ] where @xmath95 is a constant of integration .
we verify the existence of an event horizon , @xmath96 , from the first term in the right - hand - side .
thus , fine - tuning the @xmath5 parameter as @xmath97 eq . ( [ phi ] ) reduces to @xmath98 this now corresponds to a traversable wormhole , as one verifies the absence of an event horizon . note that this solution is not asymptotically flat , so that one needs to match the latter to an exterior vacuum spacetime .
if we consider a further imposition , namely , @xmath99 , then eq .
( [ fine - tune ] ) becomes zero , @xmath100 , so that the redshift function assumes a particularly simple form , given by @xmath101 the vdw equation of state , eq .
( [ vdweos ] ) , for this particular choice of parameters is @xmath102 , implying a negative radial pressure .
note that this specific choice of a wormhole is asymptotically flat .
the stress - energy components take the following form @xmath103 this solution corresponds to a rather exotic form of a vdw fluid , due to the presence of a negative energy density . the null energy condition , defined as @xmath104 , where @xmath105 any null vector , is also violated , i.e , @xmath106 , which is a necessary condition of wormholes .
it is also of interest to note that the vdw cosmic fluid with an isotropic pressure may be completely determined by only one parameter in the equation of state , so that one may consider @xmath5 as the only independent parameter needed to describe the vdw fluid @xcite .
therefore , using a wide variety of values of @xmath5 constrained by observations considered in ref .
@xcite , one may obtain restrictions on the parameters @xmath3 and @xmath4 , through eq .
( [ fine - tune ] ) , in the inhomogeneous solutions considered here . on the other hand ,
although the particular case of @xmath100 is not included in the best fits parameter range @xcite , it is not excluded from the observational constraints .
this should not be considered a serious shortcoming , as the parameters @xmath3 , @xmath4 and @xmath5 are defined using the critical values of the energy density , the isotropic pressure and the critical volume @xcite , and in the presence of anisotropic pressures the respective definitions should be extended to include the transverse pressure . despite of this fact , and due to the analytical complexity of eq .
( [ phi2 ] ) , we shall , for simplicity , consider the case @xmath100 and further analyze the physical properties and characteristics of this specific wormhole solution below , which prove to be extremely interesting .
one may also quantify the `` total amount '' of energy condition violating matter , which amounts to calculating the `` volume integral quantifier '' defined as @xmath107 @xcite .
the amount of violation is defined as the extent to which this integral becomes negative . using the `` volume integral quantifier '' , which provides information about the `` total amount '' of averaged null energy condition ( anec ) violating matter in the spacetime ( see ref .
@xcite for details ) , given by @xmath108dv$ ] , with a cut - off of the stress - energy at @xmath50 , we have @xmath109'\;dr \ , .
\label{vol - int}\end{aligned}\ ] ] now , using the form function given by @xmath93 , and the redshift function provided by eq . ( [ phi3 ] ) , and evaluating the integral , one finally ends up with the following simplified expression for the `` volume integral quantifier '' @xmath110 by taking the limit @xmath111 , one readily verifies that @xmath112 .
it is also interesting to note that in the limit @xmath113 , the volume integral tends to a finite value , i.e. , @xmath114 .
this proves that as in the specific case of phantom @xcite and chaplygin @xcite wormholes , one may theoretically construct a wormhole with arbitrarily small amounts of a vdw quintessence fluid . as emphasized in ref .
@xcite , this result is not unexpected , however , it is interesting to note the relative ease with which one may theoretically construct wormholes supported by infinitesimal amounts of exotic fluids used in cosmology to explain the present accelerated cosmic expansion .
it is also of a particular interest to analyze the traversability conditions
. we will be interested in specific solutions for traversable wormholes and assume that a traveller of an absurdly advanced civilization , with human traits , begins the trip in a space station in the lower universe , at proper distance @xmath115 , and ends up in the upper universe , at @xmath116 .
the proper distance is given by @xmath117 .
assume that the traveller has a radial velocity @xmath118 , as measured by a static observer positioned at @xmath14 .
one may relate the proper distance travelled @xmath119 , radius travelled @xmath120 , coordinate time lapse @xmath121 , and proper time lapse as measured by the observer @xmath122 , by the following relationships @xmath123 see @xcite for details . for a convenient trip through the wormhole , certain conditions
should also be imposed @xcite .
firstly , the entire journey should be done in a relatively short time as measured both by the traveller , @xmath124 , and by observers who remain at rest at the stations , @xmath125 . for simplicity
, we shall take into account non - relativistic and constant traversal velocities , @xmath126 .
thus , @xmath124 and @xmath125 are given by @xmath127 considering the redshift function provided by eq .
( [ phi3 ] ) , and @xmath128 , so that @xmath129 , we verify that both @xmath124 and @xmath130 are bounded from below by @xmath131 .
an important traversability condition required is that the acceleration felt by the traveller should not exceed earth s gravity @xcite .
thus , the traveller s four - acceleration expressed in his proper reference frame , considering once again non - relativistic and constant traversal velocities , @xmath132 , yields the following restriction ( see ref .
@xcite for details ) @xmath133 which is readily verified at the throat , where the conditions are most severe .
another important condition is that an observer traversing through the wormhole should not be ripped apart by enormous tidal forces .
thus , it is required that the tidal accelerations felt by the traveller should not exceed , for instance , the earth s gravitational acceleration @xcite .
the constraint @xmath134 provides the tidal acceleration restrictions as measured by a traveller moving radially through the wormhole , given by the following inequalities @xmath135 \right \label{radialtidalconstraint } \\
\left | \frac{\gamma ^2}{2r^2 } \left [ v^2\left ( b'-\frac{b}{r } \right ) + 2(r - b)\phi ' \right ] \right | \,\big|\eta^{\hat{2}'}\big| \leq g_\oplus \ , . \label{lateraltidalconstraint}\end{aligned}\ ] ] we refer the reader to ref .
@xcite for details related to the deduction of these conditions . the factors
@xmath136 and @xmath137 appearing in inequalities ( [ radialtidalconstraint])-([lateraltidalconstraint ] ) are the spatial separations between radial and lateral parts of the traveller s body , as measured in his proper reference frame . for computational purposes we may assume @xmath138 . the radial tidal constraint , eq .
( [ radialtidalconstraint ] ) , constrains the redshift function , and the lateral tidal constraint , eq .
( [ lateraltidalconstraint ] ) , constrains the velocity with which observers traverse the wormhole .
these inequalities are particularly simple at the throat , @xmath87 , and reduce to @xmath139 taking into account the redshift function given by eq .
( [ phi3 ] ) , and the form function considered above , i.e. , @xmath93 , we verify that evaluated at the throat we have @xmath140 and @xmath141 . thus , inequality ( [ 1st ] ) provides the restriction @xmath142 .
considering the minimum value , i.e. , @xmath143 , and that @xmath144 , we verify that the minimum value of the throat is given by @xmath145 .
using the latter approximation , and taking into account non - relativistic and constant traversal velocities , @xmath126 , then inequality ( [ 2nd ] ) provides the following restriction for the velocity , @xmath146 .
for instance , considering that the traversal velocity be of the order @xmath147 , and assuming that the traversal times take their minimum value , we have @xmath148 .
this provides an extremely reasonable traversable wormhole , useful for travellers with human traits .
in conclusion , noting that the vdw quintessence equation of state is an interesting scenario for describing the late universe , we have explored the construction of relativistic stellar models , in particular gravastars and traversable wormholes , supported by a vdw equation of state .
these solutions may possibly originate from density fluctuations in the cosmological background , resulting in the nucleation through the respective density perturbations .
we have found an exact solution of a gravastar model , by considering the specific case of a constant energy density .
relatively to traversable wormhole geometries , we also found an exact solution of an asymptotically flat wormhole , as well as solutions , where the exotic matter is constrained to the throat neighborhood , by considering a matching of an interior wormhole geometry to an exterior schwarzschild vacuum solution .
it was also found that one may theoretically construct a wormhole with arbitrarily small amounts of a vdw quintessence fluid .
several characteristics and properties of these traversable wormholes using the traversability conditions were also explored .
this work was partially funded by fundao para a cincia e tecnologia ( fct )
portugal through the grant sfrh / bpd/26269/2006 .
s. capozziello , v. f. cardone , s. carloni , d. de martino , m. falanga , a. troisi and m. bruni , `` constraining van der waals quintessence by observations , '' jcap 0504 , 005 ( 2005 ) , [ arxiv : astro - ph/0410503 ] .
s. sushkov , `` wormholes supported by a phantom energy , '' phys
d * 71 * , 043520 ( 2005 ) [ arxiv : gr - qc/0502084 ] ; f. s. n. lobo , `` phantom energy traversable wormholes , '' phys .
d * 71 * , 084011 ( 2005 ) [ arxiv : gr - qc/0502099 ] ; f. s. n. lobo , `` stability of phantom wormholes , '' phys .
d * 71 * , 124022 ( 2005 ) [ arxiv : gr - qc/0506001 ] .
p. o. mazur and e. mottola , `` gravitational condensate stars : an alternative to black holes , '' [ arxiv : gr - qc/0109035 ] ; p. o. mazur and e. mottola , `` dark energy and condensate stars : casimir energy in the large , '' [ arxiv : gr - qc/0405111 ] ; r. doran , f. s. n. lobo and p. crawford , `` interior of a schwarzschild black hole revisited , '' [ arxiv : gr - qc/0609042 ] .
i. dymnikova , `` vacuum nonsingular black hole , '' gen .
* 24 * , 235 ( 1992 ) ; i. dymnikova , `` spherically symmetric space - time with the regular de sitter center , '' int . j. mod .
d 12 , 1015 - 1034 ( 2003 ) [ arxiv : gr - qc/0304110 ] ; i. dymnikova and e. galaktionov , `` stability of a vacuum nonsingular black hole , '' class .
* 22 * , 2331 - 2358 ( 2005 ) [ arxiv : gr - qc/0409049 ] .
g. chapline , `` dark energy stars , '' [ arxiv : astro - ph/0503200 ] .
f. s. n. lobo , `` stable dark energy stars , '' class .
quant . grav .
* 23 * , 1525 ( 2006 ) [ arxiv : astro - ph/0508115 ] .
o. bertolami and j. paramos , `` the chaplygin dark star , '' phys .
d * 72 * , 123512 ( 2005 ) [ arxiv : astro - ph/0509547 ] ; n. bilic , g. b. tupper and r. d. viollier , `` born - infeld phantom gravastars , '' jcap * 0602 * , 013 ( 2006 ) [ arxiv : astro - ph/0503427 ] ; f. s. n. lobo , `` chaplygin traversable wormholes , '' phys .
d * 73 * 064028 ( 2006 ) , [ arxiv : gr - qc/0511003 ] . t. a. roman , `` inflating lorentzian wormholes , '' phys .
d * 47 * , 1370 ( 1993 ) [ arxiv : gr - qc/9211012 ] . | the van der waals quintessence equation of state is an interesting scenario for describing the late universe , and seems to provide a solution to the puzzle of dark energy , without the presence of exotic fluids or modifications of the friedmann equations . in this work
, the construction of inhomogeneous compact spheres supported by a van der waals equation of state is explored .
these relativistic stellar configurations shall be denoted as _ van der waals quintessence stars_. despite of the fact that , in a cosmological context , the van der waals fluid is considered homogeneous , inhomogeneities may arise through gravitational instabilities . thus , these solutions may possibly originate from density fluctuations in the cosmological background .
two specific classes of solutions , namely , gravastars and traversable wormholes are analyzed .
exact solutions are found , and their respective characteristics and physical properties are further explored . |
let @xmath2 be a separable hilbert space and let @xmath3 be a densely defined linear operator with domain @xmath4 .
[ def - n - ran ] _ the set of complex numbers : @xmath5 is called the _ numerical range _ of the operator @xmath3 .
_ [ rem-1 - 1 ] ( a ) it is known that the set @xmath6 is convex ( the toeplitz - hausdorff theorem ) , and in general is neither open nor closed , even for a closed operator @xmath3 .
+ ( b ) let @xmath7 be complement of the numerical range closure in the complex plane .
then @xmath8 is a connected open set except the special case , when @xmath9 is a strip bounded by two parallel straight lines .
below we use some important properties of this set , see e.g. ( * ? ? ?
* ch.v ) , or ( * ? ? ?
* ch.1.6 ) . recall that @xmath10 is called a _
deficiency _ ( or _ defect _ ) of a closed operator @xmath3 in @xmath2 .
[ prop - n - ran ] ( i ) let @xmath3 be a closed operator in @xmath2 . then for any complex number @xmath11
, the operator @xmath12 is injective .
moreover , it has a closed range @xmath13 and a constant deficiency @xmath14 in each of connected component of @xmath15 .
+ ( ii ) if @xmath16 for @xmath11 , then @xmath8 is a subset of the resolvent set @xmath17 of the operator @xmath3 and @xmath18 ( iii ) if @xmath19 is dense and @xmath20 , then @xmath3 is closable , hence the adjoint operator @xmath21 is also densely defined .
[ spect - n - ran - bounded ] for a bounded operator @xmath22 the spectrum @xmath23 is a subset of @xmath24 . for unbounded operator @xmath3
the relation between spectrum and numerical range is more complicated .
for example , it may very well happen that @xmath23 is not contained in @xmath9 , but for a closed operator @xmath3 the essential spectrum @xmath25 is always a subset of @xmath9 .
the condition @xmath26 in proposition [ prop - n - ran ] ( ii ) serves to ensure that for those unbounded operators one gets @xmath27 i.e. , the same conclusion as in corollary [ spect - n - ran - bounded ] for bounded operators .
[ sect - oper ] _
operator @xmath3 is called _ sectorial _ with semi - angle @xmath28 and a vertex at @xmath29 if @xmath30 if , in addition , @xmath3 is closed and there is @xmath31 such that it belongs to the resolvent set @xmath32 , then operator @xmath3 is called @xmath1-sectorial . _
[ rem-1 - 2 ] let @xmath3 be @xmath1-sectorial with the semi - angle @xmath28 and the vertex at @xmath29 .
then it is obvious that the operators @xmath33 and @xmath34 belong to the same sector @xmath35 for any non - negative parameters @xmath36 .
in fact @xmath37 , i.e. the operator @xmath38 has the vertex at @xmath39 . some of important properties of the @xmath1-sectorial operators are summarized by the following [ hol - contr ] if @xmath3 is @xmath1-sectorial in @xmath2 , then the semigroup @xmath40 generated by the operator @xmath3 : + ( i ) is _ holomorphic _ in the open sector @xmath41 ; + ( ii ) is a contraction , i.e. @xmath42 is a subset of the unit disc @xmath43 for @xmath41 .
the notion of the _ quasi - sectorial _ contractions was introduced in @xcite to study the operator - norm approximations of semigroups . in paper @xcite this class of contractions appeared in analysis of the operator - norm error bound estimate of the exponential trotter product formula for the case of accretive perturbations .
further applications of these contractions which , in particular , improve the rate of convergence estimate of @xcite for the euler formula , one can find in @xcite , @xcite and @xcite .
[ alpha - domain ] _ for @xmath44 we define in the complex plane @xmath45 a _ closed domain _ :
@xmath46 this is a convex subset of the unit disc @xmath47 , with _ `` angle '' _ ( in contrast to _ tangent _ ) touching of its boundary @xmath48 at only one point @xmath49 , see figure 1 .
it is evident that @xmath50 . _
[ q - s - cont](quasi - sectorial contractions @xcite ) _ a contraction @xmath51 on the hilbert space @xmath2 is called _ quasi - sectorial _ with semi - angle @xmath52 with respect to the vertex at @xmath49 , if @xmath53 .
_ notice that if operator @xmath51 is a _ quasi - sectorial _ contraction , then @xmath54 is an @xmath1-_sectorial _ operator with vertex @xmath29 and semi - angle @xmath55 .
the limits @xmath56 and @xmath57 correspond , respectively , to non - negative ( i.e. _ self - adjoint _ ) and to _ general _ contraction .
the _ resolvent _ of an @xmath1-sectorial operator @xmath58 , with semi - angle @xmath59 $ ] and vertex at @xmath29 , gives the first non - trivial ( and for us a _ key _ ) example of a quasi - sectorial contraction .
[ th - res ] let @xmath58 be @xmath1-sectorial operator with semi - angle @xmath60 $ ] and vertex at @xmath29 .
then @xmath61 is a family of quasi - sectorial contractions which numerical ranges @xmath62 for all @xmath63 . :
first , by virtue of proposition [ prop - n - ran ] ( ii ) we obtain the estimate : @xmath64 which implies that operators @xmath65 are contractions with numerical ranges @xmath66 .
next , by remark [ rem-1 - 2 ] for all @xmath67 one gets @xmath68 , where @xmath69 , i.e. for any @xmath63 the numerical range @xmath70 .
similarly , one finds that @xmath71 , i.e. , @xmath72 .
therefore , for all @xmath73 we obtain : @xmath74 moreover , since @xmath75 , by definition [ alpha - domain ] we get @xmath76 , i.e. for these values of @xmath55 the operators @xmath65 are quasi - sectorial contractions with numerical ranges in @xmath77 .
now we are in position to prove the _ main _ theorem establishing a relation between quasi - sectorial contraction semigroups and a certain class of @xmath1-sectorial generators .
[ th - main ] let @xmath58 be an @xmath1-sectorial operator with semi - angle @xmath78 $ ] and with vertex at @xmath29 .
then @xmath79 is a quasi - sectorial contraction semigroup with numerical ranges @xmath80 for all @xmath63 .
the proof of the theorem is based on a series of lemmata and on the numerical range _ mapping _
theorem by kato @xcite ( see also an important comment about this theorem in @xcite ) .
[ kato - map]_@xcite _ let @xmath81 be a rational function on the complex plane @xmath45 , with @xmath82 .
let for some compact and convex set @xmath83 the inverse function @xmath84 , where @xmath85 is a convex kernel of @xmath86 , i.e. , a subset of @xmath86 such that @xmath86 is _ star - shaped _ relative to any @xmath87 . if @xmath51 is an operator with numerical range @xmath88 , then @xmath89 . notice that for a convex set @xmath86 the corresponding _ convex _ kernel @xmath90 .
[ z - n ] let @xmath91 be complex functions , for @xmath92 and @xmath93 . then the sets @xmath94 are convex and domains @xmath95 for any @xmath93 , if @xmath96
. [ euler]_(euler formula ) _ let @xmath58 be an @xmath1-sectorial operator .
then for @xmath73 one gets the strong limit @xmath97 the next section is reserved for the proofs .
they refine and modify some lines of reasonings of the paper @xcite .
this concerns , in particular , a corrected proofs of proposition [ th - res ] and theorem [ th - main ] ( cf .
theorem 2.1 of @xcite ) , as well as reformulations and proofs of propositions [ kato - map ] and lemma [ z - n ] .
( lemma [ z - n ] ) : + let @xmath98 , then one gets @xmath99 . therefore , for the mappings @xmath100 one obtains @xmath101 for any @xmath102 .
thus , it rests to check the same property only for images @xmath103 of the sub - domain : @xmath104 see definition [ alpha - domain ] and figure 1 . for @xmath105 , two segments of tangent straight intervals : @xmath106 are correspondingly _ upper _ @xmath107 and _ lower _ @xmath108 non - arc parts of the total boundary @xmath109 ; they also coincide with a part of the boundary @xmath110 connected to the vertex @xmath49 .
now we proceed by induction .
let @xmath111 .
then one obviously obtain : @xmath112 .
for @xmath113 the boundary @xmath114 of domain @xmath115 is a union @xmath116 of the contour @xmath117 and its conjugate @xmath118 . since @xmath119 for all @xmath105 ,
the contour @xmath120 see ( [ sub - dom ] ) .
the same is obviously true for the image of the lower branch @xmath121 .
if @xmath75 , one gets : @xmath122 where @xmath123 , and @xmath124 therefore , @xmath125 . since the same is also true for the image of the lower branch @xmath121 , we obtain @xmath126 and by consequence @xmath127 for @xmath75 .
now let @xmath128 and suppose that @xmath129 .
then the image of the @xmath130 order mapping of domain @xmath131 is : @xmath132 and since @xmath129 , we obtain @xmath133 by the same reasoning as for @xmath113 .
[ max - im ] let @xmath134 .
then @xmath135 and @xmath136 for @xmath137 . in the limit @xmath138
this implies that @xmath139 , @xmath140 and @xmath141 by the same reasoning one gets the estimates similar to ( [ estim - im ] ) and ( [ im - lim ] ) for @xmath142 .
hence , @xmath143 , cf .
( [ im+ ] ) .
notice that in spite of the arc - part of the contour @xmath144 shrinks in the limit @xmath138 to zero , we obtain @xmath145 for the left extreme point of the projection on the real axe ( @xmath137 ) of the image @xmath146 . since @xmath147 , for @xmath75 , the arguments ( [ im - lim ] ) and ( [ re - lim ] ) bolster the conclusion of the lemma [ z - n ] .
( lemma [ euler ] ) : + by ( [ f - contr ] ) we have for @xmath148 @xmath149 and since @xmath58 is @xmath1-sectorial , we also get that @xmath150 .
then the _ hille - yosida _ theory ensures the existence of the contraction semigroup @xmath79 , and the standards arguments ( see e.g. ( * ? ? ?
* ch.v ) , or ( * ? ? ?
* ch.1.1 ) ) yield the convergence of the euler formula ( [ eul - s - lim ] ) in the strong topology .
( theorem [ th - main ] ) : + take @xmath151 and the compact convex set @xmath152 , see lemma [ z - n ] . since the set @xmath153 is _ convex _ , its convex kernel @xmath85 exists and @xmath90 . then by proposition
[ kato - map ] we obtain that @xmath154 , if the numerical range @xmath155 .
let contraction @xmath156 .
since by proposition [ th - res ] for any @xmath157 we have @xmath158 and since @xmath159 , we can choose @xmath90 .
then by the kato numerical range mapping theorem ( proposition [ kato - map ] ) we get : @xmath160 similarly , take the contraction @xmath161 . since ( [ 1st - step ] ) is valid for any @xmath73 , it is true for @xmath162 . then by definition of @xmath85 one has @xmath163 .
now again the proposition [ kato - map ] implies : @xmath164 therefore , we obtain @xmath165 , for any @xmath166 . by lemma [ euler ]
this yields @xmath167 for any unit vector @xmath168 .
therefore , the numerical ranges of the contraction semigroup @xmath169 for all @xmath73 , if it is generated by @xmath1-sectorial operator with the semi - angle @xmath78 $ ] and with the vertex at @xmath29 .
\1 . notice that definition [ q - s - cont ] of _ quasi - sectorial _
contractions @xmath51 is quite restrictive comparing to the notion of _ general _ contractions , which demands _ only _ @xmath170 . for the latter case one
has a well - known _
chernoff lemma _
@xcite : @xmath171 which is _ not _ even a convergent bound . for quasi - sectorial contractions we can obtain a much stronger estimate @xcite : @xmath172 convergent to zero in the uniform topology when @xmath173 .
notice that the rate of convergence @xmath174 obtained in @xcite with help of the _ poisson representation _ and the _ tchebychev inequality _ is not optimal . in @xcite , @xcite and @xcite this estimate was improved up to the _ optimal _ rate @xmath175 , which one can easily verify for a particular case of self - adjoint contractions ( i.e. @xmath56 ) with help of the spectral representation .
the inequality ( [ n-1/3 ] ) and its further improvements are based on the following important result about the upper bound estimate for the case of _ quasi - sectorial _ contractions : [ c - n - estim ] if @xmath51 is a quasi - sectorial contraction on a hilbert space @xmath2 with semi - angle @xmath176 , i.e. the numerical range @xmath177 is a subset of the domain @xmath131 , then @xmath178 for the proof see lemma 3.1 of @xcite .
another application of quasi - sectorial contractions generalizes the chernoff semigroup approximation theory @xcite , @xcite to the operator - norm approximations @xcite .
[ norm - appr ] let @xmath179 be a family of uniformly quasi - sectorial contractions on a hilbert space @xmath2 , i.e. such that there exists @xmath180 and @xmath181 , for all @xmath182 .
let @xmath183 and let @xmath184 be a closed operator with non - empty resolvent set , defined in a closed subspace @xmath185 .
then the family @xmath186 converges , when @xmath187 , in the uniform resolvent sense to the operator @xmath184 if and only if @xmath188 here @xmath189 denotes the orthogonal projection onto the subspace @xmath190 .
we conclude by application of theorem [ th - main ] and proposition [ c - n - estim ] to the euler formula @xcite , @xcite , @xcite .
[ norm - appr - euler ] if @xmath58 is an @xmath1-sectorial operator in a hilbert space @xmath191 , with semi - angle @xmath192 $ ] and with vertex at @xmath29 , then @xmath193 moreover , uniformly in @xmath194 one has the error estimate : @xmath195
i would like to thank professor mitsuru uchiyama for a useful remark indicating a flaw in our arguments in section 2 of @xcite , revision of this part of the paper @xcite is done in the present manuscript .
i also thankful to vincent cachia and hagen neidhardt for a pleasant collaboration .
chernoff , _ product formulas , nonlinear semigroups , and addition of unbounded operators . _
memoirs of the american mathematical society , no . 140 .
american mathematical society , providence , r. i. , 1974 . | we revise the notion of the _ quasi - sectorial _ contractions .
our main theorem establishes a relation between semigroups of _ quasi - sectorial _ contractions and a class of @xmath0sectorial generators .
we discuss a relevance of this kind of contractions to the theory of operator - norm approximations of strongly continuous semigroups .
operator numerical range ; @xmath1-sectorial generators ; contraction semigroups ; quasi - sectorial contractions ; holomorphic semigroups ; semigroup operator - norm approximations .
47a55 , 47d03 , 81q10 |
in photoionized nebulae both h ii regions and planetary nebulae ( pne) optical recombination lines ( orls ) provide abundance values that are systematically larger than those obtained using collisionally excited lines ( cels ) ( see e. g. * ? ? ?
* ; * ? ? ?
* , and references therein ) .
this is known as the _ abundance discrepancy problem_. it is one of the major unresolved problems in nebular astrophysics , being known of for more than seventy years @xcite , and has far - reaching consequences on the measurement of abundances throughout the universe , most often measured using cels from ionized gas .
the abundance discrepancy factor ( _ adf _ ) is defined as the ratio between the abundances derived from orls and cels and is usually between 1.5 and 3 ( see e. g. * ? ? ?
* ; * ? ? ?
* ) , but in pne it has a significant tail extending to much larger values . it has recently been shown that the largest abundance discrepancy factors ( up to @xmath2300 ) , are reached in pne with close binary central stars that have undergone a common envelope ( ce ) phase @xcite .
the data presented in these papers support the hypothesis that two different gas phases may coexist @xcite : hot , standard metallicity gas at @xmath210000 k where the cels can be efficiently excited , and much cooler ( @xmath21000 k ) gas clumps with a highly enhanced content of heavy elements ( which is the cause of the cooling ) where only orls form . as this dual nature is not predicted by mass loss theories , these results have added a new , unexpected ingredient to the abundance discrepancy problem : very high discrepancy factors should be explained in the framework of ( interacting ) binary evolution . some of the proposed explanations are naturally linked to binarity , such as for instance high - metallicity nova - like ejecta @xcite .
other appealing explanations involve the presence of planetary debris that survived the whole evolution of the central star , or the tidal destruction , accretion and ejection of jupiter - like planets ( see the discussion in * ? ? ?
* ) , but it is difficult at this stage to favor any scenario because observational constraints are still very scarce .
the existing long - slit studies point out that the plasma emitting in o ii orls ( the brightest metal orls in the optical spectrum ) is generally more concentrated in the central parts of the nebulae than the cel emitting zones of the same ion ( e. g. * ? ? ?
* ; * ? ? ?
however , no direct imaging of the gas component where the bulk of the metal orls is emitted has ever been attempted , mainly because o ii orls are faint and suitable filters are generally not available .
ngc 6778 ( @xmath1 = 19@xmath3 18@xmath4 24.94@xmath5 , @xmath6 = @xmath701@xmath8
35@xmath9 47.641@xmath10 , png034.506.7 ) was found to show unusually strong orls reflecting a high _ adf _ value ( average value @xmath220 , see * ? ? ?
it hosts a close binary central star with an orbital period of 3.7 hours , making it one of the shortest period binary central stars known @xcite . in the present work , we take advantage of the large collecting area of the 10.4 m gran telescopio canarias ( gtc ) telescope and its blue tunable filter ( btf ) to study the 2d emission profiles of the o ii orl and [ o iii ] cels for the first time .
we obtained narrow band imagery of ngc6778 using the optical system for imaging and low resolution integrated spectroscopy ( osiris , @xcite ) on the 10.4 m gran telescopio canarias ( gtc ) , at the observatorio del roque de los muchachos on la palma .
the osiris detector consists of a mosaic of two marconi ccds , each with 2048 @xmath11 4096 pixels and a total unvignetted field of view of @xmath12 arcmin@xmath13 , giving a plate scale of 0.127 arcsec / pix .
we selected the 2@xmath112 binning mode and the 200 khz readout mode .
we used the osiris blue tunable filter ( btf ) for our observations .
it consists of a fabry - perot etalon which allows narrow band imaging at a selected wavelength .
however , fabry - perot etalons cause a radial wavelength dependence in the form of `` rings '' of constant wavelength centred on the optical axis .
therefore , in order to sample the pn o ii emission at 4649 + 50 , the btf tuning was displaced to [email protected] nm , with an effective bandpass of 0.85 nm .
this ensures that the wavelength selection is the desired one to isolate o ii emission when the target is located 1 arcmin from the tf optical center .
there was no need to scan the tf as the nebula is sufficiently small ( @xmath1525 arcsec diameter excluding the faintest collimated outflows ) that the wavelength variation from one side to the other of the nebula is @xmath153 at the selected location
. data for ngc6778 were taken in service mode during the night of 2015 august 15 under director s discretionary time .
a total integration time of 3 @xmath11 600 s was obtained , with small offsets to remove the sky background contribution .
the osiris btf images were bias- and flat - field corrected in the usual way using iraf .
sky emission was removed by producing an image containing only the sky emission from the series of dithered images .
this sky image was then subtracted from the original images of the source , resulting in images cleaned from any background and sky emission . to save telescope time , owing to the large exposure times needed to obtain the desired s / n ratio in the o ii lines , no attempt was made to subtract the negligible nebular continuum .
additionally , we show [ o iii ] @xmath05007 , h@xmath1 and [ n ii ] @xmath06548 narrow - band images taken from @xcite ( see sect .
[ results ] ) .
all the images were registered to the [ o iii ] image pixel scale ( 0.189 arcsec pix@xmath16 ) and astrometrized using the 2mass catalog and the script designed by @xcite .
ngc6778 was observed with vimos @xcite , on the eso - vlt ut3 melipal on the night of 2007 september 14 .
vimos is equipped with an integral field unit containing 6400 fibers and has a changeable scale on the sky that was set to 0.33 arcsec per fiber .
this provided a sky coverage of [email protected] arcsec@xmath13 with an image formed by an array of 40@xmath1140 fibers .
we obtained observations in the vimos high - resolution mode , providing a reciprocal dispersion of 0.6 pix@xmath16 and a usable range from 3900 to 7400 adding the blue and red grisms of the spectrograph .
the object was observed with an exposure time of 300 s in both the red and blue configurations .
the reduction was performed with the vimos pipelines available at the instrument website . for sky subtraction
we observed fields offset by 40 arcsec .
the sky level was estimated as an average for each pixel along the dispersion axis and then subtracted from the object data cube .
we also applied a correction for the differential atmospheric extinction using the central star of ngc6778 in the data cube and performing the necessary shifts along the dispersion axis . to extract the final vimos emission line maps we used a set of idl routines which fit gaussian profiles to the lines of interest in each spaxel in the data cube . in fig .
[ fig1 ] we show the area covered by the vimos ifu by overlaying the vimos map taken in the light of h@xmath1 on an alfosc - not narrow - band image centered on the same line from @xcite .
the ifu was positioned to cover the central [email protected] arcsec@xmath13 of the nebula .
map ( color image ) overlaid on alfosc - not narrow band image showing the area covered by the vimos ifu ( [email protected] arcsec@xmath13 ) .
the alfosc - not image was taken from @xcite . ]
in the upper - left panel of fig . [ fig2 ]
we show the osiris - gtc btf image centered on the o ii @xmath174649 + 51 orls showing the spatial distribution of o@xmath18 from these lines . in the upper - right panel
, we show the narrow - band [ o iii ] @xmath05007 cel image from @xcite .
the lower panels present narrow - band images in h@xmath1 ( left ) and [ n ii ] @xmath06548 cel ( right ) from the same authors .
it is known that in several pne the o ii orls are contaminated by fluorescent n iii lines and/or c iii orls . in their deep vlt - fors2 spectrum of ngc6778 , @xcite did not report any contamination in the wavelength range covered by our tunable filter . in fig .
[ fig3 ] , we show a section of the observed spectrum of ngc6778 around the multiplet 1 of o ii .
it is clear that there is no emission from the c iii @xmath04647.42 line , which is the brightest one of the multiplet 1 of c iii .
therefore , all the emission covered by the tunable filter comes purely from o ii recombination lines . from inspection of fig .
[ fig2 ] , it is clear that the [ o iii ] @xmath05007 and h@xmath1 emissions share a very similar spatial distribution in the central parts of ngc6778 ( see also the discussion in * ? ? ?
on the other hand , the [ n ii ] @xmath06548 emission has a very different spatial distribution .
@xcite showed that the innermost region of ngc6778 can be described as a distorted and fragmented [ n ii]-bright equatorial ring aligned along a line close to the east - west direction .
this [ n ii ] cel innermost region is characterized by the presence of filamentary low - ionization structures ( lis ) .
their origin , however , remains unknown but has been linked to post - ce nebulae through photoionization of neutral material deposited in the orbital plane during the ce phase ( see * ? ? ?
* ; * ? ? ?
* and references therein ) .
the most important finding from inspection of fig .
[ fig2 ] is that in ngc6778 the location of the o@xmath18 ions producing the o ii orls at 4649 + 50 does not match the spatial distribution of the same ions emitting in the [ o iii ] @xmath05007 cel .
the o ii orl emission is concentrated inside the [ o iii ] cel emission or hi emission .
this is the first time that the different distribution of o ii orl and nebular [ o iii ] cel emission has been shown using direct imaging .
this finding clearly supports the hypothesis of the existence of two separate plasmas , with the additional indication that they are not well mixed , perhaps because they were produced in distinct ejection events .
previous works based on long - slit and 2d spectrophotometry in partial regions of pne found similar results .
hence , this behaviour seems to be a standard characteristic of the orl emission in pne with high _ adf_s ( see e. g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we have analyzed the morphology of the emission coming from various other lines on monochromatic emission line maps taken with vimos - vlt . in the left column of fig .
[ fig4 ] , we show the c ii @xmath04267 and o ii @xmath174649 + 50 maps which characterize the spatial distribution of pure c@xmath18 and o@xmath18 recombination emission . although the c ii emission has a rather low contrast relative to the background , its global morphology resembles the o ii @xmath174649 + 50 emission , showing local peaks at very similar spaxels . the right column of fig .
[ fig4 ] shows the resemblance of the auroral [ o iii ] @xmath04363 cel and the nebular [ o iii ] @xmath05007 cel emission maps .
the morphologies of o ii @xmath04649 + 50 and [ o iii ] @xmath05007 emissions are remarkably different , and are consistent with what is found from the gtc and not imagery ( see fig .
[ fig1 ] ) .
on the other hand , it is clear that the [ o iii ] @xmath04363 and [ o iii ] @xmath05007 cel emissions have a very different spatial distribution , but [ o iii ] @xmath04363 resembles the o ii orl emission .
this causes a radially increasing [ o iii]5007/4363 flux ratio , which may be interpreted as being due to a markedly negative electron temperature gradient ( * ? ? ?
* estimated a negative gradient of more than 1000 k , see their fig .
this effect has been found in other pne ( see * ? ? ?
* ; * ? ? ?
* for the pne ngc6153 and ngc6720 , respectively ) .
however , the presence of extremely high - density clumps ( @xmath19@xmath2010@xmath21 @xmath22 ) in the inner regions would cause the quenching of nebular [ o iii ] lines in the central part of the pn owing to collisional de - excitation effects .
the effect of high - density ionized gas in the determination of chemical abundances from orls and cels has been studied in detail in i. e. the orion nebula , by @xcite and @xcite .
this would affect the determination of physical conditions ( @xmath19 and @xmath23 ) and abundances from cels , creating a spurious temperature gradient .
moreover , if these clumps are h - poor and only contribute a few percent of the total h mass of the pn , as it has been proposed for pn , they would not be seen in hi images .
therefore , our observations support the presence of a metal rich , dense component in the inner regions of ngc 6778 , where orl emission mainly form , in addition to the `` standard '' , low density and h - rich gas component where cels are emitted .
other than its location in the inner nebular regions , which may indicate a younger origin , the morphology of the orl emitting gas does not provide any other clues to its origin .
detailed 3d photoionization models are therefore needed to fully constrain the nebula and shed further light on its formation .
all these results are in very good agreement with what was found by @xcite , although they were limited to a 1-d portion of the nebulae as selected in their long - slit fors2 observations .
4267 and o
ii @xmath174649 + 50 orls ; right column : auroral [ o iii ] @xmath04363 and nebular @xmath05007 cels .
the maps area is [email protected] arcsec@xmath13 and the orientation is north up and east left .
position of the central star is marked with a white cross . ] finally , some kinematical studies point to the presence of a colder plasma where orls are preferentially emitted .
from very high spectral resolution observations ( r=150,000 ) of two moderate to high _ adf _ pne ( ngc7009 with _ adf_@xmath25 and ngc6153 with _
adf_@xmath210 ) @xcite found that the full width at half - maximum ( fwhm ) of o ii orls were significantly smaller than that shown by [ o iii ] cels .
they also found that auroral and nebular [ o iii ] lines in ngc6153 showed different velocity profiles , which implies large temperature fluctuations .
such large temperature fluctuations can also contribute to the observed _ adf _ ( see * ? ? ? * and references therein ) .
@xcite made a detailed kinematical study of orls and cels in a limited region of pn ngc7009 .
they found that orls exhibit different kinematics to cels but no clear conclusion could be drawn because of the limited spatial coverage .
@xcite analyzed high resolution ( r@xmath233,000 ) spectra of the halo pn bobn 1 .
they found a relatively low _ adf _
( @xmath22.5 ) and o ii and ne ii orls expansion velocities lower than that of [ o iii ] and [ ne iii ] cels , which they attributed to different thermal motions .
it is worth mentioning that in the case of ngc6778 , where the o ii orl emission comes from the innermost zones of the pn , the kinematics can be affected by the expansion velocity field combined with the ionization stratification of the gas in the pn .
this would result in narrower o ii orls compared to nebular [ o iii ] cels .
we present direct imaging in the light of the o ii @xmath174649 + 50 orls for the first time in a pn . we have found that the distribution of the o@xmath18 emission coming from these lines in ngc6778 is more centrally concentrated than the emission coming from either the strongest [ o iii ] @xmath05007 cel or the h@xmath1 rl . also , from 2d reconstructed vimos monochromatic emission line maps , we find that the auroral [ o iii ] @xmath04363 cel emission resembles the o ii orls emission but not the nebular [ o iii ] emission . this may be due to presence of a high - density , h - poor gas component in the inner regions of the nebula , which could be responsible for the bulk of the o ii emission and cause the abundance discrepancy .
however , we do not have enough information to generalize the scenario of metal - rich inclusions to explain the bulk of low - to - moderate adf pne or h ii regions .
moreover , the effect of high - velocity flows , such as in hh objects , and/or high - density inclusions , such as proplyds , has definitely something to say in the abundance discrepancy problem in h ii regions .
the physical origin of the enhancement of the o ii emission in the central parts of this pn is still elusive , but should almost certainly be linked to the binary evolution of its central star .
additional observations are needed to put this important result on solid ground and further progress in our understanding of the origin of the phenomenon .
determining the spatial distribution of the orls and cels , and confronting their dynamics in several pne with high _ adf _ values would help constrain the geometry and ejection mechanism of the metal rich ( h - poor ) component .
this work is based on observations obtained with the 10.4 m gtc , operating on the island of la palma at the spanish observatories of the roque de los muchachos of the instituto de astrofsica de canarias .
we thank the anonymous referee for his / her comments .
this research has been supported by the spanish ministry of economy and competitiveness ( mineco ) under grants aya2012 - 35330 , aya2011 - 22614 , and aya2012 - 38700 .
jgr acknowledges support from severo ochoa excellence program ( sev-2011 - 0187 ) postdoctoral fellowship .
prg was supported by a ramn y cajal fellowship ( ryc2010 - 05762 ) .
jgr acknowledges fruitful discussions with c. esteban . | we present the first direct image of the high - metallicity gas component in a planetary nebula ( ngc6778 ) , taken with the osiris blue tunable filter centered on the o ii @xmath04649 + 50 optical recombination lines ( orls ) at the 10.4 m gran telescopio canarias .
we show that the emission of these faint o ii orls is concentrated in the central parts of the planetary nebula and is not spatially coincident either with emission coming from the bright [ o iii ] @xmath05007 collisionally excited line ( cel ) or the bright h@xmath1 recombination line . from monochromatic emission line maps taken with vimos at the 8.2 m very large telescope ,
we find that the spatial distribution of the emission from the auroral [ o iii ] @xmath04363 line resembles that of the o ii orls but differs from nebular [ o iii ] @xmath05007 cel distribution , implying a temperature gradient inside the planetary nebula . the centrally peaked distribution of the o ii emission and the differences with the [ o iii ] and hi emission profiles are consistent with the presence of an h - poor gas whose origin may be linked to the binarity of the central star .
however , determination of the spatial distribution of the orls and cels in other pne , and a comparison of their dynamics is needed to further constrain the geometry and ejection mechanism of the metal - rich ( h - poor ) component and hence , understand the origin of the abundance discrepancy problem in pne . |
in recent years fluctuation diamagnetism and the nernst effect in the pseudogap regime of cuprate superconductors have generated great interest both theoretically @xcite and experimentally @xcite .
experiments have found a very large diamagnetic and nernst response in the enigmatic pseudogap phase of the cuprates .
a large diamagnetic signal naturally points towards fluctuating superconductivity as one of the possible origins .
also , the fact that the nernst response is usually very small in typical nonmagnetic metals and a much stronger response is observed in the vortex - liquid regime as is expected in a fluctuating superconductor supports this point of view .
compared to conventional superconductors , the nernst and diamagnetic response have been found to exist over an anomalously large region @xcite in the pseudogap phase , extending to temperatures far above the superconducting transition temperature @xmath0 ( see however ref . ) .
this , and other mysterious features of the pseudogap phase@xcite , have lead to an intense debate over whether these responses originate purely from superconducting ( sc ) fluctuations @xcite or have a significant contribution from quasiparticles and other possible competing orders @xcite .
another important piece of the debate is related to the fact that the putative boundary @xcite of the large nernst and diamagnetic response regime , the so - called ` onset ' temperature @xmath1 , tracks @xmath2 and follows a dome - shaped curve as a function of doping @xmath3 instead of tracking the pseudogap temperature scale @xmath4 , which monotonically decreases with @xmath3 .
this has been argued as evidence against a pairing origin of the pseudogap , mainly due to the expectation that if the pseudogap arises from pairing then sc fluctuations and associated nernst and diamagnetic responses should persist all the way up to pseudogap temperature @xcite .
several theoretical works in the past have studied the nernst effect and diamagnetism in the models of sc fluctuations in various parameter regimes .
one of the pertinent issues in this context is the relative importance of amplitude and phase fluctuations of the sc order parameter @xmath5 and the role of vortices in the observed signal .
microscopically , the effect of gaussian fluctuations around the bcs state has been investigated@xcite near @xmath6 and , more recently , over a broad range of temperature and magnetic field @xcite . also , fluctuations beyond that of the bcs paradigm have been studied using a more phenomenological description @xcite .
the thermo - electric response has also been calculated in numerical simulations @xcite capturing fluctuations beyond the gaussian level through a ginzburg - landau functional , modeling overdoped cuprates .
other complementary theoretical works , more relevant for the underdoped region , have utilized a ` phase - only ' description , by studying a two - dimensional ( 2d ) xy model and its variants via various numerical @xcite and analytical methods @xcite .
such a description for underdoped cuprates is based on the phase - fluctuation scenario @xcite , where , unlike in bcs theory , @xmath6 is controlled by the superfluid density @xmath7 rather than the pairing gap scale @xmath8 . as a result , superconductivity gets destroyed at @xmath6 by strong phase fluctuations in the underdoped regime , whereas local pairing survives up to a much higher temperature scale @xmath9 .
the importance of phase fluctuations has also been emphasized in the analysis of diamagnetism @xcite and the nernst effect @xcite by ong and co - workers , but this interpretation of the nernst data has been challenged recently @xcite .
however , there is considerable independent evidence in underdoped cuprates that the sc order is destroyed by phase - disordering @xcite rather than a gap collapse .
this is also what one expects in a doped mott insulator @xcite .
studies of fluctuation diamagnetism @xcite based on the phase - fluctuation scenario have mostly ignored the effect of amplitude fluctuations and thus are constrained to describe only the extreme underdoped part of the cuprate phase diagram .
a complete theoretical calculation of either diamagnetism or the nernst effect based on a single model of sc fluctuations over the entire range of experimentally realized doping has so far not been performed . in this work
, we aim to address the above issue and calculate the fluctuation diamagnetism ( an equilibrium property , unlike the nernst effect which is a consequence of nonequilibrium transport ) based on a recently proposed phenomenological model of high-@xmath0 superconductors @xcite .
this model is of the ginzburg - landau ( gl ) type with doping and temperature dependent coefficients and is motivated by a large amount of spectroscopic data obtained from the cuprates such as arpes@xcite , stm @xcite as well as thermodynamic and transport measurements@xcite .
the starting point of our description is the premise that to understand several aspects of their phenomenology , the free energy of cuprate superconductors can be expressed solely as a functional of the complex pair amplitude ( e.g. for investigating the role of pairing fluctuations in diamagnetic response ) , other degrees of freedom e.g. electrons , cdw being not explicit .
there are two main inputs to our phenomenological theory@xcite
( i ) a pairing temperature scale @xmath4 below which local pairing amplitude becomes substantial and ( ii ) the superfluid density @xmath10 , that linearly increases with doping for small x , as implied by well - known uemura correlation @xcite . as shown in the schematic phase diagram for our model in fig .
[ fig.phasediagram ] , the temperature scale @xmath4 mimics the doping dependence of the pseudogap temperature scale @xcite and is much larger than @xmath7 in the underdoped regime .
the model effectively interpolates between a phase - only description in the extreme underdoped side to a conventional ginzburg - landau model in the overdoped region and also accesses the intermediate regime around optimal doping .
the model is also able to produce several other experimentally observed properties of the cuprates , such as the doping dependence of the phase stiffness or @xmath11 and the fluctuation specific heat @xcite and when coupled to nodal quasiparticles can produce fermi arcs with details that agree well with experiments @xcite .
a summary of the model and the parameters occurring in it is provided in appendix a. we note that there is considerable evidence for other ordering tendencies in the cuprates , e.g. , nematic@xcite , stripes@xcite , checkerboard@xcite , circulating current @xcite and charge density wave ( cdw ) @xcite . the strength and significance of each varies with the material , doping and temperature .
recently detection of short - range cdw order by x - ray @xcite and nmr measurements @xcite in the pseudogap and superconducting states of underdoped cuprates has attracted a lot of attention . in zero magnetic field , short - range cdw order competes with superconductivity and seems to become long - ranged only at high fields , presumably giving rise to fermi surface reconstruction , as suggested by quantum oscillation experiments @xcite .
motivated by these findings a phenomenological nonlinear sigma model ( nlsm ) , in terms of a coupled sc and cdw order parameter has been proposed recently @xcite .
subsequent work @xcite has pointed out that the observed magnetization and nernst effect could arise primarily from the vortex physics that results after the charge ordering degrees of freedom have been integrated out .
further , it also attributes the rise of the x - ray structure factor , seen in experiments , to a proliferation of vortices .
our theory is similar in spirit to these ideas and the parameters in our gl - like free energy functional can be thought to be renormalized values of those that occur in a theory with a larger order parameter space after the other orders have been integrated out . in this work
we obtain the magnetization by performing classical monte - carlo ( mc ) simulations on the gl - like model described above .
calculations of the nernst effect will be reported elsewhere @xcite .
as our model effectively reduces to a phase - only description on the underdoped side for the temperature range @xmath12 , vortices play a major role in determining the nernst and diamagnetic responses in this regime .
the vortex core energy is one of the important quantities in this regard ; we calculate its doping dependence from our model and show that it is consistent with the available estimates @xcite for underdoped cuprates .
our main results could be summarized as follows
we calculate the diamagnetic response over the entire doping - temperature phase diagram for a range of magnetic fields and show that our results match well with the available experiments .
we obtain a region of enhanced fluctuation diamagnetism in the pseudogap phase extending far above the transition temperature @xmath6 and show that the boundary of the region , namely the onset temperature @xmath13 , follows a dome shaped curve tracking @xmath6 as a function of doping ( fig .
[ fig.phasediagram ] ) . in the hole doping @xmath3 and temperature @xmath14 plane .
the local pairing temperature scale @xmath4 , ( solid black line ) is an input to our phenomenological model [ eq . ] and it mimics the experimental pseudogap temperature scale .
the model reproduces a dome - shaped superconducting region ( shaded in pink ) .
the region of enhanced fluctuation diamagnetism ( shaded in blue ) in the pseudogap phase and corresponding onset temperature @xmath1 are shown .
the two arcs shown by dotted lines denote regions where quantum fluctuation effects , as well as other low - energy degrees of freedom , such as electronic and spin plus their coupling with pair degrees of freedom , need to be explicitly included in the free energy functional.,height=226 ] the paper is organized as follows . in sec .
[ sec.glfunctional ] , we provide a brief introduction to the model employed and describe the details of the calculation of the magnetization in sec .
[ sec.mag ] with sec .
[ sec.method ] outlining the methodology of the numerical simulation .
we then present our results for magnetization and vortex core - energy in sec .
[ sec.results ] with a discussion in sec .
[ sec.disc ] and a brief conclusion in sec .
[ sec.conc ] . in appendix [ app.parameters ] , we also describe in detail the various features of the model and parameters in it along with a summary of how the cuprate dome can be obtained from it . a discussion of the effects of quantum phase fluctuations and other competing orders on the superconducting dome is given in appendix [ app.tcquantum ] .
in our model , the highly anisotropic cuprate superconductor is modeled as a weakly coupled stack of @xmath15 planes . as a first approximation
, we ignore the interplane coupling .
the free energy functional @xcite @xmath16 is defined on the @xmath17 planes of the superconductor .
the model describes the free energy as a functional in which the effects of the fluctuations of the order parameter phase and magnitude are coupled , and the relative importance of the latter increases with hole doping @xmath3 .
the pairing field @xmath18 , with amplitude @xmath19 and phase @xmath20 , is defined on the sites @xmath21 of a square lattice .
microscopically , the field @xmath22 is expected to be related to the complex spin - singlet pairing amplitude defined on the cu - o - cu bonds @xcite . the functional @xmath23 is defined as [ eq.functional ] @xmath24 where @xmath25 represents pairs of nearest neighbour sites .
the details of the choice of parameters for a specific cuprate material , e.g. bi2212 , that we study here , are discussed in appendix [ app.parameters ] .
we note that the two main inputs from cuprate phenomenology are the doping and temperature dependence of the parameters @xmath26 and @xmath27 .
@xmath28 is assumed a doping independent positive number .
the effect of a magnetic field is incorporated in our model through a bond flux @xmath29 , which modifies @xmath30 to @xmath31 in @xmath32 [ eq . ] . assuming extreme type - ii ( infinite penetration depth ) limit as appropriate for the cuprate superconductors , the magnetic field @xmath33 is given by the condition @xmath34 , where @xmath35 is a sum over a plaquette of the lattice and @xmath36 is the magnetic flux per plaquette in units of the universal flux quantum @xmath37 . as mentioned earlier
the lattice constant @xmath38 introduces a field scale @xmath39 , which , in principle , can be deduced by comparing our results with experimental data .
here we study the phase diagram as a function of @xmath40 .
introducing a magnetic field gives rise to a diamagnetic moment @xmath41 in the system .
the principal goal of our calculation is to find @xmath41 as a function of @xmath14 , @xmath33 and @xmath3 . to achieve this
, we first define the diamagnetic current along a bond between nearest neighbour sites @xmath21 and @xmath42 as @xmath43 @xmath41 is then obtained from @xmath44 by an appropriate integration . to this end
we perform our calculations in a cylindrical geometry with periodic boundary conditions along the @xmath3 and zero current conditions along the @xmath45 direction .
the magnetic field is radially outwards which yields a magnetization @xmath41 along the same direction .
we work in the landau gauge where the @xmath29 are non - zero only for @xmath25 along the @xmath45 direction and @xmath46 is only along the @xmath3 direction .
we integrate the current along the @xmath45 direction from the edge of the sample to the middle to obtain @xmath47 .
we use metropolis sampling for the mc simulations .
we have verified that our results are independent of gauge choice and also boundary conditions .
we would like to emphasize that there is no meissner effect in our system since it is strictly two dimensional .
thus , the expression for the current we use is different from the standard london expression @xmath48 , which holds only for the so - called coulomb gauge ( @xmath49 ) and where the complex superconducting order parameter is spatially uniform . here , we are concerned with diamagnetism at field scales for real materials much higher than @xmath50 , where the meissner effect sets in . in our numerical grid , the direction of the current is always in the periodic direction for all choice of gauges as can be seen from the gauge invariance of the expression for the current eq .. gauge invariance here is invariance under the simulatenous transformations @xmath51 and @xmath52 for an arbitrary function @xmath53 of @xmath54 .
note that such a transformation is not allowed if the @xmath55 is spatially uniform ( a necessary condition for the london expression for the current ) .
as mentioned earlier , we have verified the gauge invariance of our results by choosing different gauges for @xmath56 in our simulation .
we perform our simulations using the standard metropolis monte - carlo scheme . for obtaining the superfluid density @xmath7
, we perform the simulations on a square lattice of size 100@xmath57100 with periodic boundary conditions along both directions . at each value of doping @xmath3 we let the system equilibrate for about @xmath58 mc steps per site and then average over @xmath59 mc steps . as mentioned in section [ sec.superconductingtc ] we determine the actual transition temperature @xmath6 accurately by employing the standard finite - size scaling analysis @xcite of kt transition for small system sizes . to calculate the magnetization we perform our simulations on a cylindrical grid with periodic boundary conditions in one direction and zero current conditions along the other .
the magnetic flux is in the the radial direction of the cylinder with uniform flux per plaquette .
the allowed values of flux are in the radial direction and determined by the condition of zero flux in the axial direction .
the current is in the azimuthal direction ( in the direction in which we have periodic boundary conditions ) , is maximum at the edges and falls to zero and changes direction at the centre . as a consistency check ,
we calculate the magnetization at zero , @xmath60 and @xmath61 flux / plaquette and find it to be zero as it should be . for each @xmath62 , we perform @xmath63 mc steps per site for equilibration and a further @xmath64 steps for thermal averaging for our largest system size .
even though we perform the simulation for a single 2d layer we assume that the actual 3d system is a collection of 2d layers with a negligible josephson coupling between them .
thus , the only possible interaction between the pancake vortices is electromagnetic in nature .
this type of interaction has been shown to not change the bkt universality class of the superfluid transition and gives a very small non - universal correction to the superfluid density jump @xcite .
we thus also ignore the electromagnetic interaction among the vortices in different layers .
the conversion from 2d magnetization to 3d magnetization involves division by an appropriate length along the @xmath65 axis . for bi2212 ,
the lattice spacing along the @xmath65 axis is 3.07 nm and the appropriate length is half this value @xmath66 nm @xcite .
the dimensionless temperature is converted into kelvin by multiplying with @xmath67 which is suitably chosen for bi2212 ( see appendix [ app.parameters ] ) .
in this section we report our results for fluctuation diamagnetism . based on these ,
we analyse the superconducting fluctuation regime in the pseudogap phase .
we also discuss the role of vortex core - energy in our model .
we have obtained the values of magnetization as a function of temperature , doping and field in our model .
the overall features of diamagnetism over the phase diagram are summarized in fig.[fig : magcolor ] through color map plots of the strength of diamagnetic signal as a function of @xmath14 and @xmath33 for three different values of doping from underdoped to overdoped .
having already fixed the parameters of our model , we convert the magnetization to physical units as mentioned above .
the magnetic field also could be converted to tesla by a suitable choice of the field scale @xmath68 or equivalently the coarse - graining length @xmath38 in our model
. we can achieve good qualitative and reasonable quantitative agreement with the experimental data of li _ et al . _
@xcite with magnetization values at worst being within a factor of 2 with the measured values by choosing @xmath69 t , i.e. @xmath70 . fig .
[ fig : magvsf ] shows the magnetization as a function of magnetic field for @xmath71 and @xmath72 at different temperatures .
the main qualitative observation that one can immediately make is that as the field decreases the magnetization appears to go to zero for temperatures @xmath73 and to diverge for @xmath74 .
this is consistent with the predictions of a renormalization group calculation in the vicinity of the bkt transition @xcite .
as mentioned earlier , we do not have a meissner effect in our calculations .
the full meissner effect diamagnetic signal for real materials is much larger than the values of magnetization that we obtain here . for the range of magnetic fields
we apply , it is of the order of @xmath63 a / m per tesla . since this is several orders of magnitude larger than the fluctuation diamagnetic response we calculate ( about 1500 a / m at its largest ) , we do not show it in the plots with the experimental data .
our assumption of working with a strictly two dimensional system is valid only for fields substantially larger than @xmath50 , which holds for the specific range of fields for which we perform our calculations . in fig .
[ fig : magvst ] , we show plots of the magnetization as a function of temperature for two different values of doping .
it is evident that there is a significant diamagnetic signal persisting much above @xmath6 as seen in experiments @xcite .
to clearly demonstrate this point , we have plotted the magnetization above in the @xmath3-@xmath14 plane for @xmath76 for two values of magnetic field as in fig .
[ fig : magxt ] .
the diamagnetic signal is found to extend till a temperature which is weakly dependent on the field and approximately scales as @xmath77 for the particular choice of parameters here .
experimentally , both the nernst effect @xcite and diamagnetism @xcite have been seen to track the superconducting dome .
the persistence of the nernst and diamagnetism signal over a dome - shaped region above @xmath6 , instead of the entire pseudogap state till @xmath4 , has been argued as evidence against the pairing origin of the pseudogap line .
this is due to the expectation that if the pseudogap line is related to pairing then superconducting fluctuations should continue till @xmath78 .
however , on the basis of our results , we can argue that this expectation is not justified since pairing fluctuations identifiable as superconducting fluctuations , e.g. those detected through nernst effect or diamagnetism , are mainly controlled by @xmath7 or the @xmath6 scale .
the gl - like model we study has the pseudogap temperature @xmath4 explicitly set as the local pairing scale by construction , but , even then , the diamagnetic signal tracks the superconducting @xmath6 that is governed by the superfluid density rather than the pairing scale @xmath78 on the underdoped side .
plane for two different values of @xmath79 .
it can be clearly seen that diamagnetic signal follows the superconducting dome .
the magnetization at these fields is obtained by interpolating between our numerical data points for fixed values of fields.,width=11,height=302 ] the putative boundary of the region of substantial fluctuation diamagnetism can be defined as the onset temperature @xmath80 . experimentally the onset temperature is inferred from both nernst @xcite effect and diamagnetism @xcite measurements . in the former case @xmath75
is defined as the temperature at which the measured nernst signal starts deviating from the high - temperature background quasiparticle contribution and , in the latter case , as the temperature where the magnetization starts decreasing rapidly away from a weakly @xmath14-dependent paramagnetic van vleck signal . in our model @xmath75
can be deduced by defining , albeit in an ad hoc manner , a threshold value of the magnetization .
however , more concretely , a good qualitative measure of the onset temperature can be obtained from our model by estimating the transition temperature @xmath81 via a single - site mean - field approximation @xcite .
@xmath82 gives a measure of the temperature scale corresponding to the local superfluid density and is found to be approximately @xmath83 .
fluctuation effects destroy the global phase coherence and reduce the mean - field transition temperature from @xmath81 to the actual transition temperature @xmath6 .
however , one would expect a manifestation of substantial sc fluctuations , such as fluctuation diamagnetism , to persist over a temperature range @xmath84 .
these considerations lead to the schematic phase diagram of fig .
[ fig.phasediagram ] .
the suppression of @xmath2 relative to @xmath4 is an indicator of the strength of superconducting fluctuations .
the fact that @xmath2 is dome shaped while @xmath4 decreases monotonically with increasing @xmath3 shows that fluctuations get weaker with increasing doping .
the two extremes of the strength of fluctuations are represented by the phase - only @xmath85 model ( ` strong fluctuations ' ) and the gaussian model ( ` weak fluctuations ' ) .
the magnetization in the former model has been calculated and found to be in reasonable agreement with experimental data @xcite .
our data for extreme underdoped samples are in agreement with the results of these calculations as shown in fig [ fig.mag_2dud ] .
the strong fluctuation limit corresponds to the fluctuations essentially in the phase @xmath86 of the superconducting order parameter with the amplitude @xmath55 being frozen .
as @xmath3 increases , amplitude fluctuations start becoming more significant even as the overall strength of fluctuations decreases till one arrives at regime where the fluctuations are gaussian and can not be divided into contributions from amplitude and phase in any meaningful sense . as can be seen from fig .
[ fig : magvsf ] , our calculations show that the qualitative behavior obtained from the @xmath85 model persists even when amplitude fluctuations develop changing only the overall magnitude of the magnetization
. 20 d. podolsky , s. raghu and a. vishwanath , phys .
99 * , 117004 ( 2007 ) .
v. oganesyan , d. a. huse and s. l. sondhi , phys .
b * 73 * , 094503 ( 2006 ) .
s. mukerjee and d. a. huse , phys .
b * 70 * , 014506 ( 2004 ) .
i. ussishkin , s. l. sondhi and d. a. huse , phys .
lett . * 89 * , 287001 ( 2002 ) .
m. n. serbyn , m. a. skvortsov , a. a. varlamov and v. galitski , phys .
lett . * 102 * , 067001 ( 2009 ) .
s. s. chung , p. kakshvili and c. j. bolech , phys .
b * 86 * , 134525 ( 2012 ) .
g. wachtel and d. orgad , phys .
b * 90 * , 184505 ( 2014 ) .
x. jiang , d. li and b. rosenstein , phys .
b * 89 * , 064507 ( 2014 ) .
y. wang , s. ono , y. onose , g. gu , y. ando , y. tokura , s. uchida , and n. p. ong , science * 299 * , 86 ( 2003 ) . a. lascialfari , a. rigamonti , l. romano , p. tedesco , a. varlamov and d. embriaco , phys .
b * 65 * , 144523 ( 2002 ) .
a. lascialfari , a. rigamonti , l. romano , a. a. varlamov and i. zucca , phys .
b * 68 * , 100505(r ) ( 2003 ) .
y. wang , l. li , m. j. naughton , g. d. gu , s. uchida and n. p. ong , phys .
95 * , 247002 ( 2005 ) .
y. wang , l. li and n. p. ong , phys .
b * 73 * , 024510 ( 2006 ) .
l. li , y. wang , s. komiya , s. ono , y. ando , g. d. gu and n. p. ong , phys .
b * 81 * , 054510 ( 2010 ) l. li , y. wang and n. p. ong , phys .
b * 87 * , 056502 ( 2013 ) .
j. chang _ et al .
_ , nat .
phys . * 8 * , 751(2012 ) . o. cyr - choiniere _ et al .
_ , nature ( london ) * 458 * 743 ( 2009 ) . g. yu _
_ , arxiv:1210.6942 [ cond-mat.supr-con ] .
kokanovic et al .
b * 88 * 060505(r ) ( 2013 ) t. timusk and b. statt , rep
* 62 * , 61 ( 1999 ) .
r. daou _ et al .
_ , nature ( london ) * 463 * , 519 ( 2010 ) .
a. hackl , m. vojta and s. sachdev , phys .
b * 81 * , 045102 ( 2010 ) .
a. lee , n. nagaosa and x. g. wen , rev .
phys . * 78 * , 17 ( 2006 ) .
k. michaeli and a. m. finkelstein , europhys .
lett . * 86 * , 27007 ( 2009 ) .
a. levchenko , m. r. norman and a. a. varlamov , phys .
b * 83 * , 020506(r ) ( 2011 ) .
s. raghu , d. podolsky , a. vishwanath and d. a. huse , phys . rev .
b * 78 * , 184520 ( 2008 ) .
l. benfatto , c. castellani and t. giamarchi , phys .
lett . * 99 * , 207002 ( 2007 ) .
l. benfatto , c. castellani and t. giamarchi , phys .
lett . * 98 * , 117008 ( 2007 ) .
l. benfatto , c. castellani and t. giamarchi , phys .
b * 77 * , 100506 ( r ) ( 2008 ) .
v. j. emery and s. a. kivelson , nature ( london ) * 374 * , 434 ( 1995 ) .
y. j. uemura _ et al .
_ , phys .
lett . * 62 * , 2317 ( 1989 ) .
d. m. broun _ et al .
_ , phys .
lett . * 99 * , 237003 ( 2007 ) .
i. hetel , t. r. lemberger and m. randeria , nat .
phys . * 3 * , 700 ( 2007 ) .
a. paramekanti , m. randeria and n. trivedi , phys .
lett . * 87 * , 217002 ( 2001 ) .
s. banerjee , t. v. ramakrishnan , c. dasgupta , phys .
b * 83 * , 024510 ( 2011 ) .
a. damascelli , z. hussain and z .- x .
shen , rev .
phys . * 75 * , 473 ( 2003 ) .
o. fischer _ et al .
_ , rev .
79 * , 353 ( 2007 ) .
k. sarkar , s. banerjee , s. mukerjee and t. v. ramakrishnan ( unpublished ) .
s. a. kivelson , e. fradkin and v. j. emery , nature * 393 * , 550 ( 1998 ) .
s. a. kivelson _ et al .
_ , rev .
phys . * 75 * , 1201 ( 2003 ) .
j. e. hoffman _ et al .
_ , science * 295 * , 4650 ( 1995 ) .
b. fauque , y. sidis , v. hinkov , s. pailhes , c. t. lin , x. chaud and p. bourges , phys .
lett . * 96 * , 197001 ( 2006 ) .
e. fradkin , s. a. kivelson and j. m. tranquada , arxiv:1407.4480 [ cond-mat.supr-con ] .
a. j. achkar _ et al .
_ , phys .
lett . * 109 * , 167001 ( 2012 ) .
j. chang _ et al .
. phys .
* 8 * , 871 ( 2012 ) .
g. ghiringhelli _ et al .
_ , science * 337 * , 821 ( 2012 ) .
t. wu _ et al .
_ , nature * 477 * , 191 ( 2011 )
. n. doiron - leyraud _ et al .
_ , nature * 447 * , 565 ( 2007 ) .
l. e. hayward , d. g. hawthorn , r. g. melko and s. sachdev , science * 343 * , 1336 ( 2014 ) .
l. e. hayward , a. j. achkar , d. g. hawthorn , r. g. melko and s. sachdev , phys .
b * 90 * , 094515 ( 2014 ) .
g. wachtel and d. orgad , phys .
b * 91 * , 014503 ( 2015 ) .
g. baskaran and p. w. anderson , phys .
b * 37 * , 580 ( 1988 ) .
w. e. lawrence and s. doniach , proc .
12@xmath87 int .
conf . low temp .
, e. kanada , ed .
( kyoto 1970 , keigaku publ .
co. 1971 ) , p-361 .
p. minnhagen and p. olsson , phys .
lett . * 67 * , 1039 ( 1991 ) . c. ebner and d. stroud , phys .
b * 23 * , 6164(r ) ( 1981 ) .
e. w. carlson , s. a. kivelson , v. j. emery and e. manousakis , phys .
b * 83 * , 612 ( 1999 ) .
a. paramekanti , m. randeria , t. v. ramakrishnan and s. s. mandal , phys .
b * 62 * , 6786 ( 2000 ) .
m. franz and a. p. iyengar , phys .
lett . * 96 * , 047007 ( 2006 ) .
m. dzrazga _ et al .
_ , phys .
a * 143 * , 267 ( 1990 ) ; z. phys .
b : condens .
matter * 74 * , 67 ( 1989 ) .
s. banerjee , t. v. ramakrishnan , c. dasgupta , phys .
b * 84 * 144525 ( 2011 ) .
d. r. nelson and j. m. kosterlitz , phys .
lett . * 39 * , 1201 ( 1977 ) . h. weber and p. minnhagen , phys .
b * 37 * 10 ( 1988 ) .
k. s. raman , v. oganesyan , and s. l. sondhi , phys .
b * 79 * , 174528 ( 2009 ) .
v. g. kogan , m. ledvij , a. y. simonov , j. h. cho and d. c. johnston , phys .
lett . * 70 * , 1870 ( 1993 ) .
y. ando and k. segawa , phys .
lett . * 88 * , 167005 ( 2002 ) .
f. rullier - albenque , h. alloul and g. rikken , phys .
b * 84 * , 014522 ( 2011 ) .
a. allais , d. chowdhury , and s. sachdev , nature communications * 5 * , 5771 ( 2014 ) .
a. r. moodenbaugh _ et al .
_ , phys .
b * 38 * , 4596 ( 1988 ) ; j. m. tranquada _
et al . _ ,
nature * 375 * , 561 ( 1995 ) . | using a recently proposed ginzburg - landau - like lattice free energy functional due to banerjee et al .
phys . rev .
b 83 , 024510 ( 2011 ) we calculate the fluctuation diamagnetism of high-@xmath0 superconductors as a function of doping , magnetic field and temperature .
we analyse the pairing fluctuations above the superconducting transition temperature in the cuprates , ranging from the strong phase fluctuation dominated underdoped limit to the more conventional amplitude fluctuation dominated overdoped regime . we show that a model where the pairing scale increases and the superfluid density decreases with underdoping produces features of the observed magnetization in the pseudogap region , in good qualitative and reasonable quantitative agreement with the experimental data . in particular , we explicitly show that even when the pseudogap has a pairing origin the magnetization actually tracks the superconducting dome instead of the pseudogap temperature , as seen in experiment .
we discuss the doping dependence of the ` onset ' temperature for fluctuation diamagnetism and comment on the role of vortex core - energy in our model . |
it is well - known that the faddeev - popov construction , which is the standard analytic method for fixing the gauge , is not sufficient in order to treat the infrared regime of qcd .
the problem originates in the existence of gribov copies @xcite which are ignored in the faddeev - popov construction .
gribov copies do not play any role in the ultraviolet regime .
for that reason , the yang - mills action complemented by the faddeev - popov action is a good starting point to analyze both gauge - invariant and non - invariant quantities at momentum scales much bigger than @xmath0 gev .
however , at low momenta the question of how to include the effect of gribov copies in a lagrangian still remains unsolved even though some ideas have been already developed . among those ,
the most popular is probably the gribov
zwanziger proposal @xcite .
it consists in restricting the functional integral to the first gribov region by adding several new fields .
unfortunately it was shown in @xcite that the first gribov region also includes many gribov copies so that the gribov ambiguity is not completely removed . moreover ,
the gribov - zwanzinger procedure relies on some formal manipulations that are not fully justified from first principles . in the past decades ,
lattice simulations improved considerably and became the most reliable technique to describe the infrared region of qcd @xcite .
in particular , quenched lattice simulations in landau gauge @xcite and unquenched simulations @xcite have demonstrated that the gluon propagator acquires a finite value at low momentum , contrary to the original belief .
numerical simulations succeeded in convincing the community that , in this gauge , the gluon propagator behaves in the ir as if it was massive . even before that , the infrared regime was studied through several semi - analytical methods , among which the more popular were based on dyson - schwinger ( ds ) equations . depending on the precise implementation , two solutions were observed , called scaling and decoupling ( or massive ) solutions .
the scaling solution is characterized by a gluon propagator that vanishes as a power law at low momentum @xcite . on the other hand
, the decoupling solution corresponds to a finite gluon propagator @xcite , in agreement with lattice results .
acknowledging the fact that no analytic method can deal with the gribov ambiguity in a fully consistent way , two of the authors have proposed an alternative strategy to study analytically the ir regime of the theory .
the idea relies on working with the simplest lagrangian which reproduces lattice results . to account for the decoupling solution observed on the lattice in a minimal way ,
the faddeev - popov lagrangian is complemented by a mass term for the gluons . in euclidean space ,
the lagrangian density reads : @xmath1 where @xmath2 is the coupling constant , @xmath3 are euclidean dirac matrices satisfying @xmath4 , the flavor index @xmath5 runs over the @xmath6 quark flavors and @xmath7 the latin indices correspond to the @xmath8 gauge group , @xmath9 are the generators of the algebra in the fundamental representation and @xmath10 are the structure constants .
the model described by eq .
( [ eq_cf ] ) is a particular case of the curci - ferrari model @xcite .
it is well known that the mass term violates the brst symmetry of the faddeev - popov lagrangian .
however , it is still symmetric under a modified brst symmetry responsible for its renormalizability .
this lagrangian can also be motivated from first principles by taking into account gribov copies @xcite . a great advantage of the phenomenological model described by eq .
( [ eq_cf ] ) is that it is very simple and allows to perform perturbative calculations very easily .
moreover , we found @xcite that there exist renormalization schemes in which the coupling constant remains finite at all momentum scales .
the absence of landau pole allows us to implement perturbation theory even in the infrared regime .
the quenched one - loop calculations for the two - point @xcite and three - point @xcite correlation functions compare very well with lattice simulations .
the previous model was also used for studying the two - point correlation functions at finite temperature @xcite where it reproduces at a qualitative level the properties of the gluon and ghost propagators . in this article
, we pursue our systematic comparison of the correlation functions obtained within the model ( [ eq_cf ] ) with those extracted from lattice simulations .
we include here dynamical quarks and compute , at one loop , the two - point correlation function for the gluon , ghost and quark , for arbitrary dimension , number of colors ( @xmath11 ) and flavors ( @xmath6 ) .
the rest of the article is organized as follows . in sect .
[ glgh ] , we discuss the 1-loop calculation for the gluon and ghost propagators which are expressed in terms of passarino - veltman integrals @xcite
. we present our results in arbitrary dimension and give explicit expressions in the physically relevant case @xmath12 .
[ quark ] is devoted to the quark propagator . in sect .
[ sec_renorm ] we introduce the renormalization schemes and discuss the implementation of the renormalization group . in sect .
[ sec_res ] we perform a comparison of lattice correlation functions with our unquenched perturbative results in the gluon and ghost sector .
we focus on @xmath13 with two light degenerates quarks ( @xmath14 ) and with two light quarks and two heavier quarks ( @xmath15 ) . in sect .
[ sectionnf21 ] we work with @xmath16 .
the quark propagator was extracted from lattice simulations in this case , which enables us to make a comparison with our analytical results . finally , in sect .
[ sec_error ] we estimate the two - loop contributions which gives an indication of the error bars on our results .
the gluon and ghost propagators have been extensively studied in lattice simulations , both in the quenched and unquenched case .
it is found that the addition of the sea quarks does not change qualitatively the behavior of both propagators .
it however tends to lower the plateau observed at small momenta for the gluon propagator .
we parametrize the gluon and ghost 2-point vertex functions in the following way @xmath17,\\ & \gamma^{(2)}_{c^a { \overline c}^b}(p)=\delta^{ab } \frac{p^2}{j(p ) } , \end{aligned}\ ] ] where @xmath18 and @xmath19 are the transverse and parallel projector respectively , defined as : @xmath20 in linear covariant gauges and in particular in the linear version of the landau gauge without the inclusion of the mass term , the gluon self - energy is transverse . here , the longitudinal part is non - zero but it is controlled by a non - renormalization theorem ( see @xcite ) .
it is important to stress , however , that this longitudinal part has no impact on the gluon propagator , which is transverse in landau gauge even in presence of the mass term .
similar remarks apply to higher vertex functions , that contains longitudinal parts , but which do not contribute to the corresponding correlation functions .
the function @xmath21 ( usually called the dressing function of the ghost ) and the @xmath22 ( the transverse part of the two - point gluon vertex , related to the gluon dressing function @xmath23 ) explicitly appear in the ghost and gluon propagators and are therefore of special interest in order to compare our calculations with lattice results .
as the longitudinal part of the two - point gluon vertex is not directly accessible in lattice simulations , we do not describe it here . at one - loop ,
the gluon and ghost two - point vertex functions are given by the diagrams shown in fig .
[ diagrams ] .
first line : four diagrams contributing to the gluon self - energy .
second line : diagram contributing to the ghost self - energy , title="fig:",width=75 ] first line : four diagrams contributing to the gluon self - energy .
second line : diagram contributing to the ghost self - energy , title="fig:",width=75 ] first line : four diagrams contributing to the gluon self - energy .
second line : diagram contributing to the ghost self - energy , title="fig:",width=75 ] first line : four diagrams contributing to the gluon self - energy .
second line : diagram contributing to the ghost self - energy , title="fig:",width=79 ] first line : four diagrams contributing to the gluon self - energy .
second line : diagram contributing to the ghost self - energy , width=94 ] most of these diagrams were already computed in @xcite . in what concerns the gluon propagator
, we just need here to compute the diagram with a quark loop ( fourth diagram of fig .
[ diagrams ] ) , which can be expressed in terms of passarino - veltman integrals as @xmath24 b_0(p , m , m)\right .
\\ & \left .
+ 2 ( d-2 ) a(m)\right\ } , \end{split}\ ] ] where the index @xmath25 represents the fourth diagram in fig .
[ diagrams ] and @xmath26 is defined by @xmath27 ( in the fundamental representation , @xmath28 ) .
the @xmath29 and @xmath30 functions are the analogue of passarino - veltman integrals @xcite in euclidean space : @xmath31 @xmath32 each quark flavor contributes to the gluon vertex and it is therefore necessary to sum this diagram over the flavors , with @xmath33 replaced by the corresponding quark mass .
we have checked that ( [ eq_diag_quark_loop ] ) coincides with the expression of @xcite when the quark mass is set to zero . in @xmath34 , the diagram can be expressed in a completely analytical form : @xmath35 where @xmath36 . at one loop ,
the ghost propagator is the same as in the quenched situation due to the non - existence of a ghost - quark vertex .
however , the ghost dressing function will be indirectly influenced by the quarks through the renormalization - group flow of the coupling constant and gluon mass .
the one - loop contribution to the quark two - point vertex involves only one diagram which is shown in fig .
[ quarkdiagram ] .
it has two independent structures in dirac indices and is therefore parametrized by two scalar functions : @xmath37 both @xmath38 and @xmath39 have been determined in lattice simulations .
diagram contributing to the quark two point vertex , width=151 ] this diagram is expressed in terms of passarino - veltman integrals as follows : @xmath40b_0(p , m , m)\\&+(m^2+p^2)^2 b_0(p,0,m ) + a(m)[(2-d)m^2\\ & -m^2-p^2]+(d-2 ) m^2 a(m)\big\rbrace , \end{split}\ ] ] where @xmath41 is defined by @xmath42 .
( in the fundamental representation , @xmath43 . )
the previous expression coincides with that of @xcite when the gluon mass @xmath44 is set to zero . in @xmath34 ,
the diagram takes the analytical form : @xmath45 q -2 m^2 p^2(-2 m^2+m^2 + p^2 ) \\ & -2[2 m^6 + 3 m^4(p^2-m^2)+(m^2+p^2)^3 ] \log \left(\frac{m}{m}\right)\\ & -2 ( m^2+p^2)^3 \log \left(\frac{m^2+p^2}{m^2}\right)\big\rbrace\\ & + \frac{3 g^2 c_f m}{8 \pi ^2}\big\lbrace -\frac{1 } { \epsilon}+\log\left(\frac{m e^{\gamma/2}}{\sqrt{4 \pi } } \right)-\frac{2}{3}\\ & -\frac{k^2}{4 p^2 } q + \frac{1}{2 p^2 } ( m^2-m^2+p^2 ) \log
\left(\frac{m}{m}\right)\big\rbrace\\ & \\ & + \mathcal o(\epsilon)\end{aligned}\ ] ] with : @xmath46 . \end{split}\ ] ] it is worth to be mentioned that , at one loop , the field renormalization ( the part proportional to @xmath47 in the previous expression ) is finite .
in fact , it even vanishes in the limit of vanishing gluon mass .
in four dimensions most of the expressions presented above are divergent . in order to absorb these divergences ,
we redefine the coupling constant , masses and fields by introducing renormalization factors : @xmath48 the index @xmath49 represents the bare quantities and for now on , when not specified , all quantities are the renormalized ones .
the renormalization constants are redefined in the infrared - safe ( is ) scheme : @xmath50 the is scheme is convenient because it does not present a landau pole @xcite .
it combines the taylor scheme @xcite and a non - renormalization theorem for the gluon mass , conjectured in @xcite , and proved in @xcite .
the explicit expressions for the renormalization factors are given in @xcite for @xmath51 and for arbitrary @xmath6 their divergent parts match in landau gauge with the results presented in @xcite .
the renormalization procedure leads to finite expressions for the vertex functions .
however , as is well known , these expressions are hampered by large logarithms and we have to implement the renormalization - group procedure to control perturbation theory .
the @xmath52 functions and anomalous dimensions of the fields are : @xmath53 for completeness , we give here the contribution of the quarks to the various @xmath52 and @xmath54 functions : @xmath55 @xmath56\\ & \times \log \left(\frac{m}{m}\right)+\left(-2 m^6 - 3 m^4 \mu^2+\mu^6\right ) \log
\left(\frac{m^2+\mu^2}{m^2}\right)\\ & + \frac{1}{2 k^2}\left[2(m^2-m^2)^3 ( 2 m^2+m^2)\right.\\ & + \mu^2(m^2-m^2)(7m^4 + 6 m^2m^2 + 5 m^4)\\ & \left.+\mu^4 \left(3 m^4 - 2 m^2 m^2 - 3 m^4\right)+ \mu^6 \left(m^2+m^2\right)+\mu^8\right]q\big\rbrace\end{aligned}\ ] ] @xmath57 where @xmath58 is the expression appearing in eq .
( [ eq_kq ] ) with @xmath59 replaced by the renormalization - group scale @xmath60 . as discussed above , the ghost 2-point function is not affected by the quarks at one loop so that @xmath61 is not modified and we recall that , in the is scheme , @xmath62 and @xmath63 .
we can then use the rg equation for the vertex function with @xmath64 gluon legs , @xmath65 ghost legs and @xmath66 quark legs : @xmath67 to relate these functions at different scales , giving : @xmath68 here @xmath69 , @xmath70 and @xmath71 are obtained by integration of the beta functions with initial conditions given at some scale @xmath72 and : @xmath73 note that each quark mass @xmath74 has its own @xmath75 function that must be integrated . in our 1-loop calculation , the flow of @xmath74 does not depend on all the quark masses but only on @xmath74 itself .
as the infrared safe scheme does not present a landau pole the rg scale @xmath60 will be chosen as @xmath76 for a correlation function with typical momentum @xmath59 in order to avoid large logarithms .
in this section we compare the gluon and ghost propagator obtained in our one - loop analytical expressions and in lattice simulations , for @xmath77 with @xmath14 and @xmath15 .
when comparing our findings with the lattice data , we have to fix the initial conditions of the renormalization - group flow .
these quantities were considered here as fitting parameters and were chosen to minimize simultaneously the relative error for the gluon and ghost propagators , respectively defined as : @xmath78 there are only few parameters to choose in the fitting procedure .
they correspond to the initial conditions at some scale @xmath72 of the coupling constant and masses in the renormalization - group flow . in the @xmath15 case
there are _ a priori _ three quark masses to fit .
however we fixed , in the initial condition , the middle and heavy quarks to be twice and 20 times heavier than the lightest one ( to obtain the same mass ratios as obtained from the lattice at two gev ) .
therefore we have to fit only three parameters : the coupling constant , the gluon mass and the light quark mass , all at some renormalization scale ( we choose @xmath79 gev here ) . ) . in consequence , by using multiplicative renormalizability , we multiplied each lattice propagator by a constant in order to impose the renormalization condition ( [ rencond ] ) . in the case of unrenormalized propagators ,
this also implements the required field renormalization . ] for @xmath14 , the best fits were obtained for @xmath80 , @xmath81 gev and @xmath82 gev .
the corresponding gluon and ghost propagators are depicted in fig .
[ fig_propags_4dnf2 ] . for @xmath15 ,
the best fits were obtained for @xmath83 , @xmath84 gev and @xmath85 gev .
the corresponding gluon and ghost propagators are shown in fig .
[ fig_propags_4dnf211 ] .
the comparison is very satisfactory , with an error of at most three percent .
similar precisions were already obtained in the quenched situation , see @xcite .
it is important to stress that _ a priori _ we have three parameters to fix but , in fact , in qcd , once the scale is fixed , the coupling constant should not be arbitrary . at a given scale ,
its value should be compatible to the rg evolution from the known value ( for example , at the @xmath86 mass scale ) .
we verified that the rg evolution from the coupling @xmath87 to the @xmath86 mass scale gives a coupling at that scale of @xmath88 to be compared to @xmath89 ( see , for example @xcite ) . in order to arrive to this scale we had to use the bottom mass scale 4.2 gev above where we took into account a fifth quark in the running of the coupling .
this gives a 17% error on the coupling that is more or less the overall 1-loop calculation .
this is similar to the corresponding calculation done in the quenched case @xcite . in this sense ,
the coupling constant is not a free adjustable parameter .
however , in order to study the infrared , it is more convenient to fit the coupling at scales of the order of 1 gev . for @xmath14 .
the points are lattice data of @xcite.,title="fig : " ] for @xmath14 .
the points are lattice data of @xcite.,title="fig : " ] for @xmath15 .
the points are lattice data of @xcite.,title="fig : " ] for @xmath15 .
the points are lattice data of @xcite.,title="fig : " ] the contribution of the new diagram with an internal loop of quark is not strong enough to modify considerably the infrared behavior of the propagators due to the ir - safe structure of the quark propagator .
moreover the optimal values of the coupling constant and the gluon mass do not depend on slight changes in the quark mass .
that means that the error for the gluon and ghost propagators do not change significantly if we change the value of the light quark mass slightly .
[ errorcontourlevels ] shows the contour regions for the errors .
the contour regions are vertical , showing that the quality of the fit is almost insensitive to the quark mass ( vertical axes ) .
( left ) and @xmath90 ( right ) for @xmath12 , both for @xmath14 ( above ) and @xmath15 ( below ) .
the contourlines correspond to 8% , 10% and 12 % ( top left ) , 1.5% , 2% and 2.2 % ( top right ) , 3% , 6% and 10 % ( bottom left ) , 1% , 2% and 4% ( bottom right).the quality of the fits is almost insensitive to the value of the quark mass.,title="fig:",width=158 ] ( left ) and @xmath90 ( right ) for @xmath12 , both for @xmath14 ( above ) and @xmath15 ( below ) .
the contourlines correspond to 8% , 10% and 12 % ( top left ) , 1.5% , 2% and 2.2 % ( top right ) , 3% , 6% and 10 % ( bottom left ) , 1% , 2% and 4% ( bottom right).the quality of the fits is almost insensitive to the value of the quark mass.,title="fig:",width=158 ] + ( left ) and @xmath90 ( right ) for @xmath12 , both for @xmath14 ( above ) and @xmath15 ( below ) .
the contourlines correspond to 8% , 10% and 12 % ( top left ) , 1.5% , 2% and 2.2 % ( top right ) , 3% ,
6% and 10 % ( bottom left ) , 1% , 2% and 4% ( bottom right).the quality of the fits is almost insensitive to the value of the quark mass.,title="fig:",width=158 ] ( left ) and @xmath90 ( right ) for @xmath12 , both for @xmath14 ( above ) and @xmath15 ( below ) .
the contourlines correspond to 8% , 10% and 12 % ( top left ) , 1.5% , 2% and 2.2 % ( top right ) , 3% , 6% and 10 % ( bottom left ) , 1% , 2% and 4% ( bottom right).the quality of the fits is almost insensitive to the value of the quark mass.,title="fig:",width=158 ]
+ as expected , if the quark mass considerably increases at values of the order of the gev , the propagators tend to those obtained in the quenched approximation .
it is interesting to compare the results obtained for the gluon propagator and the ghost dressing function at one loop for different number of quarks . in fig .
[ nf ] we present the gluon propagator and the ghost dressing function for @xmath14 and @xmath15 . to compare their infrared behaviors , we normalized the curves such that they coincide at @xmath25 gev .
as was observed in lattice simulation @xcite , the addition of heavy quarks leads to a suppression of the ir saturation point .
_ and _ changing the renormalization - group trajectory . if we change only @xmath6 but use the same initial conditions ( at , say , @xmath91 gev to obtain similar correlations in the uv ) , the quarks tend to suppress the gluon propagator around 1 gev but enhance it at very low momenta . ]
we also see that the ghost dressing function is enhanced .
, title="fig : " ] , title="fig : " ]
we computed the quark propagator for arbitrary number of flavors but as the only available data for the quark sector corresponds to two degenerates light quarks plus a heavy one ( @xmath16 ) , we show the results only for this case .
as in the previous section , we fixed the ratio of the light and heavy quarks to the ratio of the bare masses used in lattice simulations .
accordingly , in the present case , we fix the mass of the heavy quark to be five times heavier than the light one at @xmath79 gev . as discussed in the previous section , the fits of the gluon and ghost propagators are rather insensitive to the choice of the quark mass ( see fig .
[ errorcontourlevels ] ) . however our results for @xmath92 depend strongly on this parameter .
therefore we fixed the fitting parameters by minimizing simultaneously the error on the gluon propagator and on the function @xmath92 , @xmath93 .
( note that the ghost propagator was not extracted from lattice simulations for @xmath16 and could not be used for fixing the parameters , see below . )
it is defined as @xmath94 note that we used a slightly different definition of the error .
the reason is that the mass function becomes rapidly close to zero when the momentum grows . accordingly a relative error point by point would be badly estimated because the signal to noise ratio of lattice data becomes of order one . in order to avoid this problem we normalized the data with a typical value of the mass in the infrared ( at the scale @xmath72 ) .
the best fit parameters are @xmath95 , @xmath81 gev and @xmath96 gev at @xmath79 gev .
we insist on that the coupling and gluon mass are fixed from the gluon propagator and once this is done , we fixed the mass parameter from the scalar part of the quark propagator .
the results obtained are shown in fig .
[ nf21m08 ] .
the gluon propagator is again reproduced with an accuracy similar to that obtained for @xmath14 and @xmath15 .
the one - loop results for @xmath92 compare correctly with lattice simulations .
in particular we retrieve the dramatic increase of the mass at a momentum scale of the order of 2 gev .
we want to stress that we could have a perfect fit of @xmath92 , but in doing so we would spoil the quality of the gluon propagator . instead , we choose to maintain the quality of the gluon propagator at the prize of a not - so - good quark mass curve . in the fig .
[ nf21m14 ] it is shown that there is a region of parameters where the fit is excellent ( see fig .
[ nf21m13 ] ) , but we insist in using a value of the coupling compatible with a good fit for the gluon dressing function . the ghost dressing function was not extracted from lattice data and we can not test our findings in this case .
however , we expect that this function is rather insensitive to the inclusion of a heavy quarks . under this hypothesis ,
it is interesting to compare our findings with the ghost dressing function for @xmath15 .
we find fits ( not shown ) of the same quality as in the previous section .
( bottom ) in @xmath12 for @xmath16 .
the parameters are @xmath95 , @xmath97 gev and @xmath98 .
the points are lattice data of @xcite.,title="fig : " ] ( bottom ) in @xmath12 for @xmath16 .
the parameters are @xmath95 , @xmath97 gev and @xmath98 .
the points are lattice data of @xcite.,title="fig : " ] .
the contourlines correspond to 10% , 20% and 30 % .,width=181 ] in @xmath12 for @xmath16 .
the parameters are @xmath99 , @xmath100 gev and @xmath101 gev .
the points are lattice data of @xcite . ] for completeness , we also compare the mass functions for the light and heavy quarks in fig .
[ fig_mass_lh ] .
if the one - loop correlation functions compare very well for the ghosts , gluons propagators and correctly for the function @xmath92 , the results for the quark renormalization function @xmath102 is not of the same quality for choosing the best fitting parameters , but concentrated on the function @xmath92 . ] , as can be seen in fig .
[ oneloopznf21 ] .
we attribute this mismatch to the fact that the one - loop contribution to this function is unusually small .
this is a consequence of the fact that , when the mass of the gluon is zero , @xmath102 has no contribution at one loop but it does have contributions at two - loops . in this situation ,
the two - loop contributions are not negligible and we do not expect a one - loop calculation to be sufficient for describing the characteristics of the field renormalization of the quarks . in the next section , we estimate the contribution of the two - loop diagrams and show that they have a small impact on our findings for the ghost and gluon propagators and on @xmath92 but that they give a large contribution to @xmath102 . in @xmath12 .
the points are lattice data of @xcite . ]
the results presented in fig . [ oneloopznf21 ] show that first order perturbation theory is not enough to reproduce the behavior of the @xmath39 in the landau gauge .
for this reason we want to estimate the importance of two loop corrections . ideally , we should compute all the two - loop diagrams and repeat the analysis performed above .
obviously this work exceeds the purpose of the present article .
instead we use a cartoon expression of the two - loop @xmath52 functions ( let us call it hybrid expression ) obtained by complementing the @xmath52 functions and anomalous dimensions derived here at one loop with the ultraviolet two - loop contribution computed in @xcite with a standard ( massless ) faddeev - popov lagrangian . to take into account the suppression of massive contribution at low momenta ,
the ultraviolet two - loop contributions have to be appropriately regularized in that regime . here
we choose to modify the anomalous dimension presented in @xcite in a minimal way by multiplying them by : @xmath103 this factor goes to one in the uv limit where the whole expression matches with @xcite .
on the other hand , in the infrared regime it goes to zero as @xmath104 , as expected . in the following , we present our results for @xmath105 gev
but we checked that we obtain the same conclusions by changing the parameter in the range 0.52 gev .
we have performed a fit of the lattice data with this hybrid flow equations .
the best fits were obtained for @xmath106 , @xmath107 gev and @xmath108 gev , again defined at @xmath79 gev .
these fits are depicted in fig .
[ zonas ] , in the center of the shaded areas .
we estimate the error coming from neglecting the 2-loop contribution as the difference between the best fit in the hybrid model and in the purely one - loop calculation . in the fig .
[ zonas ] the hybrid calculation is taken as the central value ( that includes a rough estimate of the two - loop contributions ) and the error band estimated as explained just before is represented by the shaded area .
we clearly see that higher corrections for the gluon propagator are small and compatible with the discrepancy between our one - loop results and the lattice data .
the two - loop contribution to the function @xmath38 is small but does not seem sufficient to make the one - loop results compatible with lattice data .
this could be related with the spontaneous breaking of chiral symmetry which was not treated here .
finally , the corrections to @xmath39 are large and can explain the discrepancy between the lattice data and the 1-loop results .
as mentioned in the previous section , this last function is therefore more difficult to describe and it is necessary to go to next to leading order to find a satisfactory matching . ( middle ) and
@xmath39 ( bottom ) in @xmath12 using the hybrid approximation .
the points are lattice data of @xcite.,title="fig : " ] ( middle ) and @xmath39 ( bottom ) in @xmath12 using the hybrid approximation .
the points are lattice data of @xcite.,title="fig : " ] ( middle ) and @xmath39 ( bottom ) in @xmath12 using the hybrid approximation .
the points are lattice data of @xcite.,title="fig : " ] ( middle ) and @xmath102 ( bottom ) as a function of momentum in @xmath12 . in red the hybrid calculation results ( see text ) and in black the one loop calculation and the symmetrized one loop results with respect to hybrid calculation .
the points are lattice data of @xcite and @xcite.,title="fig : " ] ( middle ) and @xmath102 ( bottom ) as a function of momentum in @xmath12 . in red the hybrid calculation results ( see text ) and in black the one loop calculation and the symmetrized one loop results with respect to hybrid calculation .
the points are lattice data of @xcite and @xcite.,title="fig : " ] ( middle ) and @xmath102 ( bottom ) as a function of momentum in @xmath12 . in red the hybrid calculation results ( see text ) and in black the one loop calculation and the symmetrized one loop results with respect to hybrid calculation .
the points are lattice data of @xcite and @xcite.,title="fig : " ]
in this article , we presented a one - loop calculation of the two - point correlation functions of qcd in the landau gauge .
the effect of the gribov copies is minimally encoded in a mass term for the gluons .
we find that the gluon and ghost propagators are reproduced with high precision .
we estimate the higher order corrections to be small which seems to indicate that the curci - ferrari model reproduces well the lattice data . on the other hand , one loop calculations are not enough to describe properly all the properties of the quark sector .
the one loop contribution to @xmath102 function is very small and it vanishes if the gluon mass goes to zero in landau gauge .
that is why the two loops contributions are important and we have to take them into account if we want to quantitatively reproduce the @xmath102 function behavior .
the mass of the quark is correctly reproduced even though the accuracy is not as good as it is for the gluon sector .
it is even unclear if a perfect matching for this quantity by such a simple one - loop calculation is possible given that we have not studied the influence of the chiral symmetry breaking .
this analysis remains for a future work .
as we discuss in the article , computing the two loop contribution can help to estimate the validity of perturbation theory in this model .
that is why we are considering in doing the two loop calculation completely .
the authors want to thank a. sternbeck , j. rodriguez - quintero and b. el - bennich for kindly making available the lattice data .
we also want to thank j. a. gracey for providing the two loop massless results for each diagram independently .
the authors want to acknowledge partial support from pedeciba and ecos programs .
m. pelez wants to thank the anii for financial support .
a. sternbeck , l. von smekal , d. b. leinweber and a. g. williams , pos lat * 2007 * , 340 ( 2007 ) [ arxiv:0710.1982 [ hep - lat ] ] .
a. cucchieri and t. mendes , phys .
lett . * 100 * ( 2008 ) 241601 and also arxiv:1001.2584 [ hep - lat ] .
a. cucchieri and t. mendes , phys
. rev .
d * 78 * ( 2008 ) 094503 .
i. l. bogolubsky _ et al .
. lett .
b * 676 * , 69 ( 2009 ) .
d. dudal , o. oliveira and n. vandersickel , phys .
d * 81 * ( 2010 ) 074505 .
p. o. bowman , u. heller , d. leinweber , m. parappilly , a. williams , and others , phys .
d * 70 * , 034509 ( 2004 ) . | we investigate the gluon , ghost and quark propagators in the landau gauge with dynamic quarks .
we perform a one - loop calculation in a model where the standard faddeev - popov lagrangian is complemented by a mass term for the gluons which is seen as a minimal way of taking into account the effect of the gribov copies .
the analytic results are compared with lattice data obtained in four dimension and for two , three and four quark flavors .
the gluon and ghost propagators are reproduced with a few percent accuracy in the whole range of accessible momenta .
the scalar part of the quark propagator is found to be in good agreement with the lattice data .
however , the quark renormalization is poorly described .
we attribute this discrepancy to the fact that the one - loop corrections to this quantity are unusually small so that the two loop contribution can not be discarded .
the results are expressed in terms of the coupling , the gluon mass and the light quark mass at 1 gev . |
diluted magnetic semiconductors @xcite represent an important class of materials and structures where ferromagnetism can be tuned by voltage @xcite .
this ability comes from the carrier - mediated character of the ferromagnetic interaction @xcite .
the ferromagnetic ordered state in these systems appears due to mobile carriers interacting with stationary spins of magnetic impurities . to date
, curie temperatures as high as @xmath0 have been observed in a technologically important class of the mn - doped iii - v semiconductor structures @xcite . when the magnetic semiconductors become combined with the conventional field - effect layered structures , the number of mobile carriers and the ferromagnetic interaction become tunable by the voltage @xcite .
this ability to externally control the properties of magnetic crystals with means other than the external magnetic field may have important device applications .
a further step from magnetic semiconductor layers would be zero dimensional systems , quantum dots ( qds ) .
magnetic quantum dots can be viewed as nano - scale memory elements where information is stored in the form of magnetic polarization .
such a system may have important advantages compared to the conventional metal spin - valve memories : ( 1 ) small sizes and relatively small number of carriers and ( 2 ) voltage control of the number of electrons which was already demonstrated in many experiments for non - magnetic qds @xcite .
therefore it is important to develop the understanding of magnetic qds with interacting carriers .
here we develop a theory of magnetic qds with many carriers where coulomb , ferromagnetic , and single - particle energies contribute to the formation of the equilibrium state . using the quasi - classical description ,
we show that a qd may be split into three phases with different physical properties .
the geometrical sizes of these phases are determined by the coulomb , ferromagnetic , and single - particle contributions to the chemical potential of a qd . for calculations ,
we employ the mean - field theory and the boltzmann kinetic equation .
this approach becomes reduced to the thomas - fermi model at low temperatures .
we should note that our approach ignores the discrete structure of single - particle spectrum of qd and is valid when electrons occupy at least several quantum levels .
at the same time , this approach has an important advantage : it allows us to describe the coulomb effects in relatively large qds where , as it is shown below , the coulomb interaction is very strong and significantly exceeds the ferromagnetic interaction and the kinetic energy of carriers .
the hole - mediated ferromagnetism in quasi - two - dimensional ( 2d ) systems is strongly anisotropic due to the heavy hole - light hole splitting in the valence band .
therefore , the magnetic polarization occurs predominantly in the growth direction .
then , two magnetic states of a qd with spin polarizations `` up '' and `` down '' can represent a single bit .
presently qds and other nano - structures doped with magnetic ( mn ) impurities attract a lot of attention . among other studies ,
several recent theoretical papers investigate qds and their electron and excitonic states in the presence of a single mn ion @xcite .
in particular , it was suggested in ref .
@xcite that a qd with a single mn ion can act as a multi - qubit which can be controlled optically .
another direction of research describes the magnetic states and polarons in qds with many mn ions and one or several carriers @xcite . among the above publications
, the paper @xcite demonstrates that the coulomb - interaction effects in few - electron qds can determine a collective magnetic state of holes and mn spins .
ferromagnetism and spin separation in digital layered structures and quantum wells were also studied in several experimental @xcite and theoretical publications @xcite .
as a model system , we are going to use a qd `` made out of '' a 2d quantum well .
such a zero - dimensional system can be fabricated by etching and lithographical methods . within the lithographical method ,
a qd can be defined , for example , by using the top metal gates ( fig .
[ fig1]a ) .
the number of carriers in such a qd is a voltage tunable parameter . .
( b ) calculated 2d spin density as function of the total 2d density of holes at various temperatures ; @xmath1 , @xmath2 , @xmath3 , @xmath4 , @xmath5 .
curie temperature for the above parameters is about @xmath6 . ] to describe a state of many carriers confined in a qd , we start from the local properties of the coupled hole - mn system in a 2d quantum well . in our system ,
a mobile hole and mn spins experience an exchange interaction : @xmath7 , where @xmath8 is the exchange interaction constant , and @xmath9 and @xmath10 are the z - components of the spin operators for a mn spin and hole , respectively ; @xmath11 and @xmath12 are the coordinates of hole and @xmath13-impurity , respectively .
the above operator describes the interaction between mn - spins and heavy holes and assumes a sufficiently large energy separation between the heavy - hole and light - hole quantized states in the valence band .
the corresponding effective spin - dependent potential of a single hole has a form : @xmath14 where @xmath15}.\end{aligned}\ ] ] here @xmath16 is the net spin 2d density , @xmath17 are the 2d densities of the spin components , @xmath18 is the brillouin function , @xmath19 is the number of cations per unit cell , and @xmath20 .
@xmath21 is the total 2d density in a quantum well . for the ground - state wave function in a quantum well
, we will use @xmath22 , where @xmath23 is the quantum well width .
the chemical potential for a 2d gas depends on @xmath24 and @xmath25 : @xmath26,\end{aligned}\ ] ] where @xmath27 .
now we calculate the spin polarization : @xmath28 , \label{nspin}\end{aligned}\ ] ] where @xmath29 .
the zener ferromagnetic phase transition occurs when eq .
[ nspin ] has a nonzero solution .
[ fig1]b shows the data for spin density @xmath30 for a gaas / algaas quantum well with the following parameters : @xmath1 , @xmath31 , @xmath3 , @xmath4 , @xmath5 .
the above exchange parameter @xmath8 is comparable to that used in other publications on magnetic semiconductors ( see e.g. @xcite ) .
since the exchange interaction is antiferromagnetic ( @xmath32 ) , the case @xmath33 corresponds to the negative average polarization of mn ions , @xmath34 .
curie temperature can be analytically calculated in the high - density limit : @xmath35 ( @xmath36 ) .
the in - plane potential in a lateral qd near its center can be approximated by the parabolic function : @xmath37 where @xmath38 is the electron charge and @xmath39 is a characteristic frequency of a confining potential .
the potential @xmath40 determines a depth of a lateral potential well . in a qd defined by metal gates ( fig .
[ fig1]a ) , the parameters @xmath40 and @xmath39 are functions of the gate voltages . in equilibrium , the carrier distribution function , which satisfies the boltzmann equation , has a form : @xmath41 where @xmath42 is the lateral position vector and @xmath43 is the in - plane momentum .
the self - consistent scalar potential of a hole is composed of two terms : @xmath44 where @xmath45 is the electrostatic potential induced by a non - uniform spatial distribution of carriers , @xmath46 . in addition , the distribution function ( [ ff ] ) depends on the spin of hole through the exchange interaction which is a function of the local spin density , @xmath47 ( see eqs .
[ uspin],[uspin0 ] ) . at the same time , the function @xmath48 itself is determined by the total local density of holes , @xmath49 , and is given by the numerical solution of eq .
[ nspin ] ( see the data in fig . [
therefore , it is convenient to regard @xmath50 as a function of @xmath51 , i.e. @xmath52=u_{spin}[j_z , n_{spin}(n_{2d}))]=u_{spin}[j_z , n_{2d}({\bf r})]$ ] . by integrating the function ( [ ff ] ) over momenta
we come to two non - local non - linear equations for the densities @xmath53 .
then , these equations can be solved for the chemical potential and rewritten in the form resembling the central equation of the thomas - fermi model : @xmath54+k_b t \ln[e^{\frac{2\pi\hbar^2 n_{j_z}}{m_{hh } k_b t}}-1 ] , \hskip 0.5 cm j_z=\pm3/2 , \label{tf}\end{aligned}\ ] ] where @xmath55 and @xmath56 , where @xmath57 is an effective non - local dielectric constant of a system with metal gates @xcite .
we should also note that @xmath45 was written as a 2d integral and this it valid if the lateral size of a qd is greater than the quantum - well width @xmath23 . in the system with the top gates closely located to a quantum well ,
the 2d integral in the equation for @xmath45 is reduced to a linear function of @xmath51 @xcite and is given by a local flat - capacitor formula : @xmath58 where @xmath59 is the distance between the qd plane and the top gate ; the distance to the back metal contact is assumed to be larger , i.e. @xmath60 ; @xmath61 is a dielectric constant of the semiconductor ( @xmath62 ) .
the approximation ( [ fc ] ) has been successfully used in the past for description of several experiments on optical and electronic properties of modulated lateral structures @xcite . by using the local approximation for the self - consistent potential ( [ fc ] )
, we reduce eqs .
[ tf ] to coupled non - linear local equations which should be solved numerically .
the total number of holes in a qd is determined by the chemical potential @xmath63 and the lateral - potential depth @xmath40 .
electrostatics of the structure under study ( fig .
[ fig1]a ) is similar to that studied in refs .
@xcite and we can use here the results of the above publications . if the barrier between the qd and back contact permits efficient tunnelling , the chemical potential in the qd coincides with the potential of the back contact ( i.e. @xmath64 ) .
simultaneously , the front barrier ( usually made of algaas ) blocks tunnelling between the qd and the top gate . also , if @xmath65 , the potential @xmath40 becomes close to @xmath66 .
[ fig2],[fig3 ] , and [ fig4 ] show numerical calculations for the local spin densities in a circular qd with @xmath67 , @xmath68 , and @xmath64 ; for the qd depth , we take @xmath69 , and @xmath70 . a qd with the minimum free energy
is circularly symmetric and can be split into different phases .
the total number of holes in a qd is given by an integral @xmath71 .
the corresponding @xmath72 for the above values of @xmath40 are estimated as @xmath73 , and @xmath74 . in qds with
relatively small @xmath72 ( fig.[fig2 ] ) , the system is split into ferromagnetic ( f ) and paramagnetic ( p ) phases . in fig .
[ fig2 ] , the carriers with spin @xmath75 are pushed away from the center of qd , the total spin of holes is negative , and the mn - subsystem has a positive magnetization .
this situation corresponds to the antiferromagnetic hole - mn coupling ( @xmath32 ) . with increasing the total number of carriers ( fig .
[ fig3 ] ) , one can see the formation of another stripe within the ferromagnetic phase .
this stripe is located between the center region of a qd and the paramagnetic phase and the holes in this stripe are almost fully spin - polarized . with further increasing @xmath72 ( fig .
[ fig4 ] ) and for relatively low temperatures , the formation of the ferromagnetic stripe ( f2 ) with filly - polarized holes becomes evident .
simultaneously , the paramagnetic stripe becomes very narrow .
such a magnetic stricture of a qd can be understood by looking at the data in fig .
[ fig1]b . at low temperatures ,
the function @xmath76 becomes very close to the linear function @xmath25 in an extended interval of @xmath25 .
for example , at @xmath77 , @xmath78 for @xmath79 . in the above interval of @xmath25 at @xmath77 ,
the hole subsystem is almost completely spin polarized .
it is interesting to estimate different types of energies contributing to the formation of stripes .
it is easy to see that the coulomb energy in eqs .
[ tf ] dominates the ferromagnetic and single - particle ( kinetic ) energies .
the coulomb energy @xmath80 for @xmath81 while the spin energy @xmath82 for @xmath83 at @xmath77 .
the single - particle kinetic energy under the similar conditions @xmath84 . for qds with more carriers and higher @xmath51 ,
the above energies become increased but the condition @xmath85 remains .
we can also obtain analytic solutions of eqs .
[ tf ] under certain conditions .
if both spin subsystems of holes ( @xmath55 ) form a degenerate fermi gas , the last term in eqs .
[ tf ] becomes proportional to the fermi energy , @xmath86 .
then , we can sum up the equations for @xmath55 .
the resulting equation does not contain the spin energy @xmath24 and can be solved analytically : @xmath87.\end{aligned}\ ] ] this formula can be applied , for example , to large qds in the spatial region of the ferromagnetic phase f1 , @xmath88 , where @xmath89 is the radius of the f1 phase ( see fig . [ fig4 ] ) .
for this phase , spin densities can also be found analytically , by using the condition @xmath90 , where @xmath91 is a positive constant equal to @xmath30 at high @xmath25 in fig .
[ fig1]b ; for @xmath77 , @xmath92 ( see fig . [
the formula ( [ na ] ) also describes the density distribution in the paramagnetic phase in the regions where the hole gas is degenerate ( @xmath93 ) . for many other regimes ,
the spin densities should be found numerically .
since the coulomb energy dominates the magnetic and kinetic terms , the total radius of a qd can be well estimated from eq .
[ na ] by putting @xmath94 .
the resulting estimate @xmath95}$ ] is valid at low temperatures .
experimentally , the stripe structure of a qd can be observed , for example , by spatially - resolved optical spectroscopy @xcite . in optical spectroscopy
, a spatial resolution can be as small as @xmath96 @xcite .
simultaneously optical emission is sensitive to the spin - polarization of carriers . in an optical experiment , a ferromagnetic qd would be excited with weak nonpolarized illumination ; the resulting local photoluminescence will be circularly polarized and reveal the formation of stripes with different magnetic structures .
optical methods can also be used to write a magnetic state of qd ( bit : `` up''-``down '' ) .
this may be done with circularly - polarized light .
polarized optical pumping can bring a qd into a required collective magnetic state with spins `` up '' or `` down '' . in order to prepare a quantum dot in a required magnetic state
, one can also use a magnetic field induced by an electric current driven through a metallic wire on the surface of a sample . another method to observe the magnetic phases in nano - structures is the electrical - capacitance spectroscopy @xcite which was successfully applied to observe , for example , compressible and incompressible stripes in electron quantum wires in the regime of the quantum hall effect @xcite .
the capacitance spectroscopy has been also applied to lateral and self - assembled quantum dots ( see e.g. @xcite ) . for the nano - structures with relatively large sizes considered in this paper ,
the signature of the ferromagnetic phase in the capacitance spectra is expected to be relatively weak because of the inequality @xmath97 .
however , the formation of the ferromagnetic phase can be recognized from a critical behavior of the capacitance spectrum as a function of temperature and voltage .
as an example , we consider here quantum wires in a structure with the interdigitated metal gate ( see inset in fig .
[ fig5 ] ) . in such a system ,
alternating voltages are applied to the metal strips located on the surface of a quantum well .
similar structures were studied experimentally in ref.@xcite .
we can calculate the capacitance of a wire as a derivative @xmath98 , where @xmath99 is the linear density of carriers in a quantum wire and @xmath100 is the length of a wire in the in - plane @xmath101-direction . in the local approximation for the coulomb potential ( eq . [ fc ] ) ,
the problems of quantum dot and wire become similar . fig .
[ fig5 ] shows the capacitance of a quantum wire as a function of voltage for two temperatures , just below and above the curie temperature @xmath102 .
one can see in fig .
[ fig5 ] that at low temperatures ( @xmath103 ) the capacitance becomes increased starting from a critical voltage @xmath104 .
this voltage corresponds to the minimum 2d density ( @xmath105 at @xmath77 ; see fig .
[ fig1]b ) which is necessary to obtain the ferromagnetic phase stripe in the center of nanowire . starting from this voltage , the central region of a nanowire contains a ferromagnetic stripe .
the capacitance of a partially ferromagnetic wire becomes increased since the spin interaction makes a lateral potential well a little deeper and a wire can accommodate more carriers at a given voltage . if temperature increases just by @xmath106 , the ferromagnetic stripe vanishes and capacitance becomes reduced .
this peculiar temperature behavior for @xmath107 can be taken as an evidence for the ferromagnetic phase . and @xmath108
. the inset on the right hand side shows a top view of a nanowire with two phases . ]
in this paper , we studied quantum dots and wires with many interacting carriers within the quasi - classical approach .
the strongest interaction in quantum dots with a relatively weak confinement and a large number of carriers comes from the coulomb forces .
however , a weaker ferromagnetic interaction determines the spin structure of a large quantum dot .
depending on the parameters , a quantum dot can be split into three phases .
the author would like to thank bruce mccombe for motivating discussions .
this work was supported by ohio university and the a.v .
humboldt foundation .
semiconductor spintronics and quantum computation , edited by d. d. awschalom , d. loss , and n. samarth ( springer - verlag , berlin , 2002 ) ; i. zutic , j. fabian , and s. das sarma , rev . mod .
76 , 323 ( 2004 ) .
a. d. giddings , m. n. khalid , t. jungwirth , j. wunderlich , s. yasin , r. p. campion , k. w. edmonds , j. sinova , k. ito , k .- y .
wang , d. williams , b. l. gallagher , and c. t. foxon , phys .
94 , 127202 ( 2005 ) .
y. d. park , a. t. hanbicki , s. c. erwin , c. s. hellberg , j. m. sullivan , j. e. mattson , t. f. ambrose , a. wilson , g. spanos , and b. t. jonker , science 295 , 651 ( 2002 ) ; h. ohno , d. chiba , f. matsukura , t. omiya , e. abe , t. dietl , y. ohno , k. ohtani , nature 408 , 944 ( 2000 ) . for the case of self - assembled quantum dots , see i.e. : h. drexler , d. leonard , w. hansen , j. p. kotthaus , and p. m. petroff , phys .
73 , 2252 ( 1994 ) ; p. schittenhelm , c. engel , f. findeis , g. abstreiter , a. a. darhuber , g. bauer , a. o. kosogov , and p. werner , j. vac .
b 16 , 1575 ( 1998 ) .
x. chen , m. na , m. cheon , s. wang , h. luo , b. d. mccombe , x. liu , y. sasaki , t. wojtowicz , j. k. furdyna , s. j. potashnik , and p. schiffer , appl .
81 , 511 ( 2002 ) .
j. fernandez - rossier and l. j. sham , phys . rev .
b 64 , 235323 ( 2001 ) ; | many - particle electron states in semiconductor quantum dots with carrier - mediated ferromagnetism are studied theoretically within the self - consistent boltzmann equation formalism .
depending on the conditions , a quantum dot may contain there phases : partially spin - polarized ferromagnetic , fully spin - polarized ferromagnetic , and paramagnetic phases .
the physical properties of many - body ferromagnetic confined systems come from the competing carrier - mediated ferromagnetic and coulomb interactions .
the magnetic phases in gated quantum dots with holes can be controlled by the voltage or via optical methods . |
we are interested in the dynamics of an asset price in a limit order market .
we aim at constructing a tractable model which captures endogenously such phenomena observed in actual markets as nonlinearity in liquidation , permanent market impact and a flash crash due to illiquidity .
the liquidity issue has been pointed out as a major risk which standard models of financial engineering have not taken into consideration . the failures of several financial institutions are often attributed to the uncovered liquidity risk .
the liquidity crisis is a rare event ; an exogenous statistical modeling of liquidity costs is therefore not sufficient for preparing ourselves for a future crisis .
an economic consideration is required for a deeper understanding of the liquidity risk .
this paper provides an utility - based asset pricing model with analytically tractable structure .
+ in a limit order market , the roles of supplier and demander of liquidity are not symmetric .
a liquidity supplier submits a _
limit order _ that quotes a price for a specified volume of an asset .
they can trade with each other by using limit orders to maximize their own utilities .
once an equilibrium is achieved , no more trade occurs among them .
however , a liquidity supplier still has an incentive to submit a limit order as long as the corresponding transaction improves her utility .
the remained limit orders form a price curve which is a nonlinear function of volume .
a liquidity demander submits a _ market order _ to buy or sell any amount of the asset according to the price curve .
taking a bertrand - type competition among liquidity suppliers into account , it is then reasonable to begin with modeling the price curve as the utility indifference price for a representative liquidity supplier ( rls hereafter ) .
+ if the rls is risk - neutral , then the utility indifference price of an asset coincides with the expected value of the future cash - flow associated with the asset .
the price curve then becomes linear in volume .
this simplest framework was adopted by many studies such as glosten and milgrom @xcite . in this study , we assume the rls to be risk - averse in managing her inventory , which results in nonlinear pricing with market impact . +
this approach differs from the classical works including garman @xcite , amihud and mendelson @xcite , ho and stoll @xcite , ohara and oldfield @xcite , where a price quote is a solution of an utility maximization problem for a market maker with exogenously given order - flow . here , we solve an utility maximization problem for a representative liquidity demander ( rld hereafter ) .
therefore , an order - flow is endogenously determined .
our model is closely related to the one considered in bank and kramkov @xcite , where they analyzed the market impact of a large trade and formulated a nonlinear stochastic integral as the profit and loss associated with a given strategy of a large trader . here , we aim at deriving the dynamics of the _ efficient price _ , that is the asset price when the rld follows an optimal strategy . unlike standard optimal investment or hedging problems ,
the asset price is nonlinear in volume and depends on the trading strategy .
if the set of the liquidity demanders consists of price takers and a single large investor , the optimal strategy can be interpreted as the solution of an optimal investment or hedging problem for the large investor under a simplified version of the model introduced by bank and kramkov @xcite .
the model represents permanent market impact , while instantaneous or temporary market impact models have been extensively considered in the literature .
see e.g. cetin et al .
@xcite , fukasawa @xcite , guant @xcite and the references therein . in the most of the preceding studies ,
the structure of market impact was exogenously modeled . here , as in bank and kramkov @xcite
, the market impact is endogenous .
+ we consider a hypothetical market where all traded securities expire at a maturity and yield cash - flow , of which the joint probability distribution is given exogenously .
this can be seen as a model of a commodity futures market .
a standard theory shows that a commodity future price is determined by an arbitrage relation to the spot price of the commodity .
this arbitrage pricing however does not work for a commodity of which the inventory costs are so expensive , such as oil or electricity .
a major factor which determines the future price of such a commodity would be the probability distribution of the spot price at the maturity , which is the idea of our model .
+ in section 2 , we give a rigorous formulation of the problem . in section 3
, we consider a lvy - driven market as an example which admits explicit results .
in particular , we observe that an optimal risk allocation between the rls and the rld is achieved in a specific situation .
we see also that an endowment shock can invoke a price crash with severe illiquidity . in section 4
, we solve the optimal hedging problem under a condition of completeness , where the optimal risk allocation is achieved . in particular , we construct a perfect hedging strategy under the market impact .
we observe that the volatility of the efficient price process depends on the convexity of an initial endowment to be hedged .
we see also that a price crash is invoked by a propagated endowment shock . in section 5 , under an incomplete situation , we establish a dynamic programming principle in a discrete time framework and give its applications .
here we rigorously formulate the problem .
let @xmath0})$ ] be a filtered probability space satisfying the usual assumptions with @xmath1 composed of the null sets .
the time @xmath2 stands for the maturity of all securities and we set @xmath3 .
assume there exists a regular conditional probability measure given @xmath4 for each @xmath5 .
by @xmath6 $ ] , we always mean the expectation of a random variable @xmath7 with respect to this regular conditional probability measure .
let @xmath8 be an @xmath9-measurable random vector which stands for the cash - flow of @xmath10 securities at time @xmath2 .
we assume @xmath11 dividend rate and @xmath11 risk - free rate .
+ the rls evaluates future cash - flow @xmath7 at time @xmath2 by an exponential utility @xmath12\ ] ] at @xmath13 $ ] , where @xmath14 is a parameter of risk - aversion .
the exponential utility with @xmath15 is interpreted as the limit of ( [ pi ] ) as @xmath16 .
more specifically , under a suitable condition on @xmath7 , @xmath17 $ ] when @xmath15 , which is the case that the rls is risk - neutral . an important property of the general exponential utility is the cash - invariance : @xmath18 for any @xmath4-measurable random variable @xmath19 .
another important property is the quasi - concavity : if @xmath20 and @xmath21 , then @xmath22 for any @xmath23 $ ] .
together with the cash - invariance property , the quasi - concavity implies the concavity : @xmath24 the rls is initially endowed with a risky asset which yields cash - flow at time @xmath2 , denoted by @xmath25 .
if the rls is holding @xmath26 units of the @xmath10 securities as her inventory at time @xmath27 $ ] , then for a volume @xmath28 , she quotes a price according to the utility indifference principle : @xmath29 note that the utility indifference price does not depend on the amount of cash held by the rls due to that her utility is exponential . if the rls is risk - neutral , then @xmath30 $ ] .
otherwise , the price depends on the inventory @xmath31 of the securities , which describes permanent market impact . for all @xmath32 and @xmath31
, @xmath33 is a convex function of @xmath34 with @xmath35 .
this implies in particular that @xmath36 for any @xmath34 and @xmath31 , which means that the selling price for an amount is higher than or equal to the buying price for the same amount .
this represents bid - ask spread that is a measure of market liquidity .
+ the rld also evaluates future cash - flow @xmath7 by an exponential utility @xmath37\ ] ] at time @xmath13 $ ] , where @xmath38 is a parameter of risk - aversion .
we allow @xmath39 by taking the limit @xmath40 .
the rld is initially endowed with a risky asset which yields cash - flow at time @xmath2 , denoted by @xmath41 .
let @xmath42 be a subset of @xmath43 with @xmath44 .
we suppose @xmath45 is finite for all @xmath46 and @xmath27 $ ] .
let @xmath47 be the set of the @xmath10-dimensional simple predictable processes @xmath48 with @xmath49 and @xmath50 for all @xmath13 $ ] .
the rld is allowed to take any element @xmath51 as her trading strategy , where @xmath52 stands for the number of shares for the @xmath53-th security held by the rld at time @xmath32 .
the price for a volume @xmath28 of the @xmath10 securities at time @xmath32 is @xmath54 .
this is because the rls holds @xmath55 units of the securities due to the preceding trades with the rld .
then the profit and loss at time @xmath2 associated with @xmath56 is given by @xmath57 the problem of the rld is to maximize the utility of her terminal wealth @xmath58 in @xmath48 on a suitably extended set of @xmath47 . + naturally arise the following questions : 1 .
does a dynamic programming principle hold for this utility maximization problem ?
2 . how is the optimal strategy @xmath59 characterized ?
how does the _ efficient price _
@xmath60 behave ?
how does the _ efficient infinitesimal price per unit _ ( eipu hereafter ) @xmath61 behave ?
how does the _ efficient price convexity _
@xmath62 behave ?
the eipu is interpreted as the asset price vector for small investors who are price takers .
the efficient price convexity is a measure of infinitesimal illiquidity .
here we give an example which admits explicit computation .
we will see an optimal risk allocation between the rls and the rld is achieved in some cases , while not in other cases .
let @xmath63 , @xmath64 and @xmath65 for a lvy process @xmath66 with @xmath67 .
the characteristic function of @xmath68 admits the lvy - khintchine representation @xmath69= iub-\frac{1}{2}\sigma^2 u^2 + \int ( e^{iux}-1-iuh(x))\nu(\mathrm{d}x),\ ] ] where @xmath70 , @xmath71 and @xmath72 is a measure on @xmath73 with @xmath74 assume that there exists an interval @xmath75 such that @xmath76 for all @xmath77 .
then , @xmath78 for all @xmath77 and by theorem 25.3 of sato @xcite , @xmath79 for all @xmath80 , where @xmath81 since @xmath82 is a concave functional , @xmath83 is a concave function with @xmath84 .
for example , @xmath85 if @xmath86 , where @xmath87 is a standard brownian motion
. we have @xmath88 if @xmath66 is a gamma process with @xmath89 following the gamma distribution with rate @xmath90 and shape @xmath91 .
another example is @xmath92 when @xmath66 is a one - sided stable process with rate @xmath93 and exponent @xmath94 .
+ suppose @xmath95 for @xmath96 and @xmath97 .
then , by ( [ pilevy ] ) , @xmath98 for @xmath99 and @xmath100 and so , @xmath101 for @xmath56 . by integration by parts ,
note that if @xmath11 is an interior point of @xmath103 , then @xmath83 is differentiable at @xmath11 and by letting @xmath16 , @xmath104.\ ] ] the process @xmath105 is a martingale and it was the asset price process when the rls was risk - neutral .
the right hand side of ( [ pl ] ) can be written as @xmath106 the first term is of the familiar form of profit and loss associated with the trading strategy @xmath48 when @xmath105 was the price process .
the second term can be both positive and negative and converges to @xmath11 as @xmath16 .
+ now , consider @xmath41 of the form @xmath107 where @xmath108 and @xmath109 is a simple predictable bounded process independent of @xmath66 .
this specification is just to obtain an explicit result .
if one interprets @xmath110 as an economic signal at time @xmath32 , then @xmath111 determines its contribution to the initial endowment @xmath41 .
+ if @xmath112 takes values in @xmath103 , then the terminal wealth of the rld is @xmath113 the last inequality follows from the convexity of @xmath83 and the equality is attained when @xmath114 since @xmath109 is independent of @xmath66 , we have @xmath115 = 1\ ] ] if , say , @xmath48 is bounded . see kallsen and shiryaev @xcite . if @xmath42 is bounded but enough large for @xmath47 to include @xmath59 given by ( [ opt ] ) , then @xmath116 to interpret this identity , let us introduce @xmath117\ ] ] for random variable @xmath7 .
then , @xmath118 and so , @xmath119 note that @xmath120 where @xmath121\ ] ] for random variable @xmath7 .
this functional @xmath122 is known as the aggregated utility under optimal risk allocation ( see e.g. , barrieu and el karoui @xcite ) .
the above computation shows that the optimal risk allocation between the rls and the rld is achieved by a dynamic trading strategy @xmath59 if @xmath109 is deterministic , while some utility can be lost otherwise .
+ since the optimal strategy @xmath59 of the rld is given by ( [ opt ] ) , the efficient price and the eipu are respectively @xmath123 if @xmath83 is differentiable .
the latter can rewritten as @xmath124 since the first term is a martingale , the second term is interpreted as the _ risk premium_. on intervals where @xmath109 is constant , we have @xmath125 since @xmath83 is concave , the risk premium is nonnegative if the aggregated initial endowment @xmath126 is nonnegative . a positive ( resp .
negative ) endowment shock @xmath127 induces a nonpositive ( resp .
nonnegative ) jump of @xmath128 and a nonnegative ( resp .
nonpositive ) jump of the risk premium .
the efficient price convexity is computed as @xmath129 if @xmath83 is twice differentiable .
this implies that the illiquidity level strongly depends on the convexity of @xmath83 and so , the distribution tail of @xmath130 .
if @xmath131 is one more times differentiable and the lvy measure @xmath72 is spectrally negative , then @xmath132 and the skewness of @xmath130 is negative . in this case , since @xmath133 is increasing , a positive endowment shock @xmath134 induces a positive jump of the infinitesimal illiquidity , in other words , a wider bid - ask spread , in addition to a drop of @xmath128 .
here , we assume the filtration @xmath135 is generated by a @xmath136-dimensional standard brownian motion @xmath87 .
we solve the problem of the rld under a completeness condition specified below .
we set @xmath137 for brevity .
let @xmath138 be the set of random variables @xmath7 such that for all @xmath139 , @xmath140<\infty$ ] .
note that @xmath138 is a linear space .
we assume that @xmath141 and @xmath130 are elements of @xmath138 .
let @xmath142 be a continuous modification of @xmath143 and @xmath144 for @xmath145 .
the key assumption of the section is the existence of a map @xmath146 \times
\mathbb{r}^d \to \mathbb{r}^k\ ] ] with the following properties : 1 . for each @xmath27 $ ]
, the restriction @xmath147 } : \omega \times [ 0,t ] \times \mathbb{r}^d \to \mathbb{r}^k\ ] ] is @xmath148 ) \times \mathcal{b}(\mathbb{r}^d)$]-measurable .
2 . there exists @xmath149 with @xmath150 such that @xmath151 for all @xmath152 \times \mathbb{r}^d$ ] .
3 . there exists @xmath149 with @xmath150 such that @xmath153 is continuous in @xmath34 for each @xmath154 $ ] .
4 . there exists a map @xmath155 \times
\mathbb{r}^k \to \mathbb{r}^d\ ] ] such that the restriction @xmath156 } : \omega \times [ 0,t ] \times \mathbb{r}^k \to \mathbb{r}^d\ ] ] is @xmath148 ) \otimes \mathcal{b}(\mathbb{r}^k)$]-measurable for each @xmath27 $ ] and there exists @xmath149 with @xmath150 such that @xmath157 for each @xmath158 \times \mathbb{r}^k$ ] .
we call this assumption the _ completeness condition _ hereafter
. + the terminology is justified by considering the risk - neutral case ; when @xmath15 and @xmath25 and @xmath130 are square - integrable , then @xmath159 + y \mathbb{e}[s|\mathcal{f}_t ] = \int_0^t ( g^\prime_t + y s^\prime_t)\mathrm{d}w_t\ ] ] almost surely by the it@xmath160 representation theorem , where @xmath161 and @xmath162 are progressively measurable processes which are @xmath163-valued and @xmath164-valued respectively . in this case , the asset price is linear in volume and @xmath165 since @xmath166 the completeness condition is satisfied if @xmath167 has rank @xmath136 as a @xmath168 matrix and admits a left inverse matrix which is progressively measurable as a process . +
another elementary example is the case @xmath169 and @xmath170 with @xmath171 , @xmath172 , @xmath173 and @xmath174 . since @xmath175 the completeness condition is satisfied if and only if @xmath176 has rank @xmath136 .
+ [ ext ] for @xmath56 , we have @xmath177 where @xmath178 . _
proof : _ by the completeness condition , we identify @xmath179 .
let @xmath56 . denote by @xmath180 the stopping times when @xmath48 jumps .
we have @xmath181 for sufficiently large @xmath182 almost surely .
remember that @xmath183 is @xmath184-measurable .
note that @xmath185 where @xmath186 with @xmath187 .
as is well - known , @xmath188 is the unique solution of the backward stochastic differential equation ( bsde hereafter ) @xmath189 to see this , it is enough to observe that @xmath190 is a local martingale with @xmath191 and then , apply the it@xmath160 representation theorem .
this bsde representation implies that @xmath192 and so , the result follows . ////
+ by lemma [ ext ] , we can extend the definition of @xmath193 onto the set @xmath194 of the progressively measurable processes @xmath48 with @xmath195 almost surely and @xmath196 the problem is then to find an optimal strategy @xmath197 .
+ [ ph ] for any @xmath198 , there exists @xmath197 such that @xmath199 _ proof : _ this is a perfect hedging problem and amounts to finding the solution @xmath200 of the bsde @xmath201 as before , this quadratic bsde admits an explicit solution given by @xmath202 by the completeness condition , we can define a progressively measurable process @xmath59 by @xmath203 to have @xmath204 //// + by lemma [ ph ] , any claim @xmath205 is perfectly replicated by a strategy @xmath197 under the completeness condition .
the _ replication price _ of @xmath205 is given by @xmath206 this is convex in @xmath205 , which can be interpreted as a diversification of the liquidity risk .
+ [ thm : ph ] if @xmath39 , then @xmath207 where @xmath59 is given by ( [ yast ] ) .
_ proof : _ recall a well - known representation @xmath208 + \frac{1}{c } \mathbb{e}[\frac{\mathrm{d}q}{\mathrm{d}\mathbb{p}}\log \frac{\mathrm{d}q}{\mathrm{d}\mathbb{p } } ] \right\}\ ] ] for @xmath209 .
take the limit @xmath40 , to have @xmath210 = \mathrm{ess.inf }
\ f\ ] ] for any bounded random variable @xmath7 ( see e.g. , delbaen @xcite ) .
this means that the rld is extremely risk - averse and tries to offset her initial endowment @xmath41 almost surely . by lemma [ ph ]
, @xmath211 is constant and
so , @xmath212 suppose that there exists @xmath213 such that @xmath214 since @xmath215 we have @xmath216 by the uniqueness of the solution of the bsde . from this and
( [ deny ] ) , we deduce a contradiction as follows : @xmath217 //// here we extend theorem [ thm : ph ] to the case @xmath218 .
first we state a version of the result referred as borch s theorem in barrieu and el karoui @xcite .
[ borch ] if @xmath218 , then for any @xmath219 , @xmath220 where @xmath221 and @xmath122 is defined by ( [ uast ] ) . _
proof : _ since @xmath222 is concave , it suffices to observe that @xmath223 for all @xmath224 . ////
+ [ thm : opt ] if @xmath218 , @xmath225 where @xmath59 is given by @xmath226 and @xmath227 is the continuous modification of @xmath228 .
_ proof : _ by definition , @xmath229 since @xmath230 is the solution of the bsde @xmath231 we have @xmath232 it follows then , @xmath233 it remains to show that @xmath234 as before , we have ( [ pig ] ) for all @xmath197
. therefore , @xmath235 by lemma [ borch ] . ////
+ since @xmath236 and so , @xmath237 as @xmath40 , theorem [ thm : opt ] extends theorem [ thm : ph ] . as mentioned in the previous section , @xmath238 is the aggregated utility of the aggregated initial endowment .
the above result implies that an optimal risk allocation ( pareto allocation ) is achieved by the dynamic trading strategy @xmath59 for general @xmath219 under the completeness condition .
here we assume in addition that @xmath239 with borel functions @xmath240 and @xmath241 to extract a more tractable structure of the optimal strategy .
define functions @xmath242
\times \mathbb{r}^k \to \mathbb{r}$ ] and @xmath243 \times \mathbb{r}^k \times \mathbb{r}^d \to \mathbb{r}$ ] as @xmath244,\\ & p(t , w , y ) = - \frac{1}{\gamma } \log \mathbb{e}\left[\exp\left\{-\gamma(g - ys ) \right\ } \big|w_t = w\right ] .
\end{split}\ ] ] by the assumption on the filtration , we have @xmath245 as is well - known and easily checked , @xmath246 and @xmath247 are the solutions of the partial differential equations ( pde hereafter ) @xmath248 where @xmath249 the completeness condition can be restated as follows : there exists a measurable function @xmath250\times \mathbb{r}^k \to \mathbb{r}^d$ ] such that @xmath251 for each @xmath252 \times \mathbb{r}^k$ ] .
the optimal strategy given by theorems [ thm : ph ] and [ thm : opt ] is @xmath253 the efficient price , the efficient price per unit and the efficient price convexity are given by @xmath254 here we study through a simple example how the volatility of the eipu is determined by model parameters .
let @xmath255 , @xmath256 , @xmath257 and @xmath258 for constants @xmath259 .
we assume @xmath260 and @xmath261 .
although @xmath41 does not belong to @xmath138 , it is not difficult to see the preceding results are extended to this specific model . in this case , @xmath130 follows a normal distribution and @xmath262 it is straightforward to have @xmath263 it follows then that @xmath264 the completeness condition is therefore satisfied and we have @xmath265 the efficient price , the eipu and the efficient price convexity are given by @xmath266 the volatility of @xmath128 is therefore given by @xmath267 the volatility monotonically depends on @xmath268 , the convexity of @xmath205 as a function of @xmath130 . in case @xmath268
is negative , or equivalently , if the rld has to hedge an european payoff @xmath205 which is a convex function of @xmath130 , then the volatility is larger in case @xmath269 than in the risk - neutral case @xmath15 .
this means that economic signals are amplified in a market where hedgers of convex payoffs are dominant .
each of the pdes in ( [ kpz ] ) is a kpz equation and by differentiating in @xmath270 , we obtain ( backward ) navier - stokes equations for irrotational flow @xmath271 for @xmath272 and @xmath273 when @xmath274 , @xmath275 and @xmath276 are differentiable . in case @xmath277 , they are burgers equations @xmath278 a nontrivial explicit solution for the burgers equation is known ; it is easy to see @xmath279 is a solution with @xmath280 where @xmath281 and @xmath268 can be arbitrary constants .
the solution is unique ; see hopf @xcite .
remark that if @xmath282 , then @xmath283 as @xmath284 .
therefore in case @xmath285 and @xmath286 with constants @xmath287 , the problem of rld is to hedge a payoff which is close to a call option payoff : @xmath288 the completeness condition is satisfied and @xmath289 it follows then that @xmath290 in case that @xmath291 is large , the function @xmath292 has a steep slope around @xmath293 .
this means that when @xmath294 approaches to @xmath281 to cross it from below , the eipu @xmath128 exhibits a drastic drop .
see figure [ fig1 ] for a sample path .
this can be interpreted as an asset price crash invoked by a propagated endowment shock .
+ the burgers equation models fluid dynamics and is known to form a shock wave .
more precisely , an inviscid burgers equation can form a shock wave with a singularity .
a viscous burgers equation like ( [ burg ] ) does not form a singularity but can form a steep shape because as @xmath295 , the equation tends to be inviscid .
as in the above explicit case , the eipu will crash when the factor process @xmath87 meets a shock wave .
the shock wave does not have a singularity but has a steep slope when @xmath296 is large , or equivalently , when both the rls and rld are strongly risk averse .
a more detailed analysis on this shock wave model would be worth to be done ; it remains for future research .
here we show a version of dynamic programming principle holds in our problem .
let @xmath297 with an integer @xmath298 and denote by @xmath299 the set of strategies @xmath56 such that @xmath300 only if @xmath301 .
in this section , we consider the maximization of @xmath302 subject to @xmath303 . as the following lemma shows , this problem coincides with the original one if the filtration @xmath135 is discrete in the sense that @xmath304/n}$ ] for all @xmath5 .
this means that information is updated only at a discrete time set .
_ proof : _ for @xmath306 , putting @xmath307 , @xmath308 which means that buying @xmath309 units at time @xmath310 and @xmath34 units at time @xmath32 is equivalent to buying @xmath311 units at time @xmath310 . ////
_ proof : _ first note that @xmath316 is constant on @xmath317 for each @xmath318 and @xmath319 . since @xmath320 for any random variable @xmath7 and @xmath321 for any @xmath303 and @xmath322
, we obtain @xmath323 on the other hand , since @xmath42 is compact , there exists a sequence @xmath324 adapted to @xmath325 such that @xmath326 with @xmath327 } y^\ast_j\ ] ] by the measurable selection theorem ( see parthasarathy @xcite ) . ////
as a direct application of the above results , let us consider a special case that there exists @xmath340 such that @xmath341 this holds , for example , both @xmath25 and @xmath41 are proportional to @xmath130 and @xmath42 is sufficiently large .
assume @xmath25 , @xmath41 and @xmath130 are bounded and @xmath42 is compact . applying the preceding results , we have @xmath342 let us observe that the supremum is attained by @xmath343 , \ \
y_0 = 0.\ ] ] this means that the rld buys @xmath344 units of the securities at time @xmath11 and holds them until time @xmath2 .
the supremum in @xmath345 is attained by @xmath340 because @xmath346 is concave and @xmath347 for all @xmath348 .
further we have @xmath349 \\ = & \frac{c+\gamma}{c } \pi_{(n-1)/n}\left(\frac{c}{c+\gamma}(g+h)\right ) . \end{split}\ ] ] then we go to the next step .
we have @xmath350 where @xmath351 and @xmath138 is the linear space spanned by @xmath352 , @xmath353 .
the function @xmath354 is concave for each @xmath355 and its derivative in @xmath356 at @xmath357 vanishes due to that @xmath358 .
therefore @xmath359 and the upper bound is attained when @xmath360 .
consequently we get @xmath361\ ] ] and apparently this argument can be repeated to conclude that the optimal strategy to hold @xmath344 units from the beginning .
+ consequently , the optimal strategy is @xmath362 for @xmath363 , which implies @xmath364 this extends ( [ alloc ] ) and means that the optimal risk allocation is achieved .
denote by @xmath365 the probability measure defined by @xmath366 } , \ \
\gamma = \exp\left\ { -\frac{c\gamma}{c+\gamma}(g+h ) \right\}.\ ] ] then , @xmath367 } { \mathbb{e}[\exp\{-\gamma(g - y^\ast s)\}|\mathcal{f}_t ] } = \mathbb{e}^q[s | \mathcal{f}_t].\ ] ] further , the efficient price convexity coincides with the conditional covariance matrix of @xmath130 under @xmath365 : @xmath368 - \mathbb{e}^q[s^i|\mathcal{f}_t ] \mathbb{e}^q[s^j |\mathcal{f}_t].\ ] ] we may have these simple expressions for @xmath128 and @xmath369 because the optimal strategy is of buy - and - hold type .
+ the boundedness condition on @xmath130 , @xmath25 and @xmath41 can be relaxed . due to ( [ emm ] ) , standard models of mathematical finance are then supported in this framework .
for example , if @xmath370 and @xmath371 with a standard brownian motion @xmath87 and constants @xmath372 and @xmath373 , then we obtain the bachelier model @xmath374 if @xmath370 and @xmath375 with constants @xmath376 and @xmath377 , then we have @xmath378 and obtain the black - scholes model @xmath379 here we reconsider the framework of section 4.4 in terms of the dynamic programming principle .
our aim here is to recover theorem [ thm : opt ] by letting @xmath380 .
the filtration @xmath135 is generated by a @xmath136-dimensional @xmath135-standard brownian motion @xmath87 .
we suppose that @xmath42 is a compact set and @xmath381 with functions @xmath382 , @xmath383 and @xmath384 . here
we assume in addition the boundedness and the uniform continuity for the derivatives @xmath385 for @xmath247 and @xmath246 defined in section 4.4 .
_ proof : _ by ( [ markov ] ) , @xmath387 by taylor s theorem , @xmath388 \\ = & p(t+1/n , w , y ) \\ & -\frac{1}{\gamma } \log \biggl\ { 1 -\frac{\gamma}{2n } \delta p(t+1/n , x , y ) + \frac{\gamma^2}{2n}\left| \nabla p ( t+1/n , w , y)\right|^2 + o(1/n ) \biggr\ } \\ = & p(t+1/n , w , y ) + \frac{1}{2n } \delta p(t+1/n , w , y ) - \frac{\gamma}{2n}\left| \nabla p ( t+1/n , w , y)\right|^2 + o(1/n ) .
\end{split}\ ] ] here and hereafter , the estimate @xmath389 is uniform in @xmath390 .
similarly , @xmath391 \\ = & v(t+1/n , w)- p(t+1/n , w , y ) + \frac{1}{2n } \left(\delta v(t+1/n , w ) - \delta p(t+1/n , w , y ) \right ) \\ & - \frac{c}{2n}\left| \nabla v(t+1/n , w ) -
\nabla p(t+1/n , w , y)\right|^2 + o(1/n ) .
\end{split}\ ] ] therefore , @xmath392 then the result follows from the completeness condition . ////
_ proof : _ put @xmath394 by the previous lemma , @xmath395 as @xmath396 . for @xmath397 , put @xmath398 then , by the cash - invariance property of @xmath399 , @xmath400 then , by proposition [ repsup ] , @xmath401 as @xmath380 . //// | we construct an utility - based dynamic asset pricing model for a limit order market .
the price is nonlinear in volume and subject to market impact .
we solve an optimal hedging problem under the market impact and derive the dynamics of the efficient price , that is , the asset price when a representative liquidity demander follows an optimal strategy .
we show that a pareto efficient allocation is achieved under a completeness condition .
we give an explicit representation of the efficient price for several examples .
in particular , we observe that the volatility of the asset depends on the convexity of an initial endowment .
further , we observe that an asset price crash is invoked by an endowment shock .
we establish a dynamic programming principle under an incomplete framework .
* keywords : * utility indifference principle ; market impact ; efficient price ; + quadratic backward stochastic differential equation ; burgers equation . +
* jel classification : * g12 ; g13 . *
msc2010 :* 91b25 ; 91g80 . |
early investigators studied the spatial distribution of galaxies because they hoped to learn about the structure of the universe on the largest scales .
their influential work was superseded , in the end , by its competition .
problems began with the demonstration that galaxies contained only a small fraction of the matter in the universe .
galaxy formation remained too poorly understood to quell doubts about how faithfully galaxies traced underlying distribution of dark matter .
other observations improved gravitational lensing , peculiar velocities , intergalactic absorption lines , and so on and seemed easier to relate to matter fluctuations .
computers became fast enough to predict the evolution of the large - scale matter distribution from the the initial conditions that microwave - background missions were measuring with increasing precision . as it became clear that the simulations and observations agreed remarkably well , most researchers concluded that the large - scale structure of the universe could be understood completely as the product of gravitational instability amplifying small inflationary perturbations .
galaxies , once believed to be the primary constituent of the universe , came to be seen as small test particles swept into ever larger structures by converging dark matter flows .
the spatial distribution of galaxies remains interesting because it can teach us about galaxy formation .
galaxy formation must be closely related to larger process of gravitational structure formation , since the formation of a galaxy begins with gas streaming into a massive potential well and ends with stars drifting in the cosmic flow .
lessons from 20 years of numerical investigations into structure formation should therefore carry over to the analysis of galaxy clustering .
one example is the known positive correlation between clustering strength and mass for virialized dark matter halos .
since galaxies reside in dark matter halos , their clustering strength provides an indication of the mass of the halos that contain them .
the resulting mass estimate depends on the assumptions that the microwave background and other observations have given us reliable estimates of cosmological parameters and of the initial matter power - spectrum , that numerical simulations can correctly trace the evolution of the matter distribution , at least for moderate densities , and that no process can significantly separate baryons from dark matter on mpc scales assumptions that are at least as plausible as those behind competing techniques for mass estimation .
another example is the evolution in the clustering strength of a population of galaxies once it has formed .
this is driven solely by gravity and is easy to predict from numerical simulations .
comparing the clustering of ( say ) galaxies in the local universe and those at high redshift can therefore suggest or rule out possible links between them .
few other methods are as useful for unifying into evolutionary sequences the galaxy populations we observe at different look - back times . an excellent review of the history of these techniques has been written by giavalisco ( 2002 ; pp 620 - 624 ) .
this paper has two aims .
the first is to present measurements of the clustering of uv - selected star - forming galaxies in a redshift range @xmath2 that is only partially explored .
section [ sec : data ] describes the way we obtained our data , [ sec : methods ] describes and justifies the techniques we used to estimate the spatial clustering in our galaxy samples , and
[ sec : results ] presents our estimates of the galaxy correlation function at redshifts @xmath30 , @xmath31 , and @xmath32 .
the survey analyzed here is several times larger than its predecessors ; the surveyed area of @xmath3 square degrees is roughly 700 times larger than the area analyzed by arnouts et al .
( 2002 ) and 4 times larger than the areas analyzed by giavalisco & dickinson ( 2001 ) and ouchi et al .
the second aim is to discuss what our measurements imply about the galaxies and their descendants . in
[ sec : correspondence ] we show that the galaxies correlation functions are indistinguishable from those of virialized dark matter halos with mass @xmath33 . in [ sec : evolution ] we show that the galaxies , dragged by gravity for billions of years , caught in the press of structure formation , would by redshift @xmath28 have a correlation function that is indistinguishable from that of the elliptical galaxies that surround us . our results are summarized in [
sec : summary ] .
the data we analyzed were drawn from our ongoing surveys of high - redshift star - forming galaxies .
a brief description of the surveys follows ; see steidel et al .
( 2003 , 2004 ) for further details .
deep , multi - hour @xmath7 images of 21 fields scattered around the sky were obtained with various 4m - class telescopes ( table [ tab : fields ] ) .
tens of thousands of objects were visible in these images . for the analysis of this paper we ignored all but the subset ( @xmath34 20% ) with ab magnitude @xmath35 and ab colors satisfying the `` lbg '' selection criteria of steidel et al .
( 2003 ) , @xmath36 the `` bx '' selection criteria of adelberger et al .
( 2004 ) , @xmath37 or the `` bm '' selection criteria of adelberger et al .
( 2004 ) , @xmath38 in this range of @xmath14 magnitudes , the colors are characteristic of galaxies at @xmath39 .
( the restriction to @xmath40 helps eliminate most interlopers ; see adelberger et al .
2004 and steidel et al . 2004 . ) by obtaining spectra of more than @xmath8 galaxies with these colors , steidel et al .
( 2003 , 2004 ) established that the mean redshift and @xmath41 range of galaxies in the three samples is @xmath42 ( these values exclude galaxies with @xmath43 or @xmath44 as well as the handful of low - redshift `` interloper '' galaxies with @xmath45 . ) redshift histograms are shown in figure [ fig : zhistos ] .
we will use the term `` photometric candidates '' to describe the objects with @xmath46 whose colors satisfy one set of the selection criteria presented above , and the term `` spectroscopic sample '' to describe the subset of photometric candidates that had a spectroscopic redshift measured by steidel et al .
( 2003 ) or steidel et al .
although the spectroscopic sample is sizeable , it contains only a small fraction ( @xmath47% ) of the photometric candidates ( see figure [ fig : specfrac ] and table [ tab : fields ] ) .
lrrrrrrr 3c324 & @xmath48 & 11/49 & 0/166 & 0/126 & 0.0035 & 0.0034 & 0.0042 + b20902 & @xmath49 & 31/65 & 1/207 & 0/189 & 0.0036 & 0.0035 & 0.0043 + cdfa & @xmath50 & 34/99 & 0/336 & 0/280 & 0.0029 & 0.0028 & 0.0036 + cdfb & @xmath51 & 20/120 & 0/316 & 0/273 & 0.0028 & 0.0028 & 0.0035 + dsf2237a & @xmath51 & 39/100 & 1/367 & 0/328 & 0.0028 & 0.0028 & 0.0035 + dsf2237b & @xmath52 & 44/161 & 1/516 & 0/309 & 0.0028 & 0.0028 & 0.0035 + hdf & @xmath53 & 54/187 & 128/735 & 37/587 & 0.0022 & 0.0022 & 0.0028 + q0201 & @xmath54 & 18/90 & 4/339 & 0/285 & 0.0029 & 0.0029 & 0.0036 + q0256 & @xmath55 & 45/126 & 1/346 & 0/243 & 0.0030 & 0.0029 & 0.0037 + q0302 & @xmath56 & 46/824 & 0/1778 & 0/749 & 0.0018 & 0.0018 & 0.0023 + q0933 & @xmath57 & 63/192 & 0/435 & 0/273 & 0.0028 & 0.0028 & 0.0035 + q1307 & @xmath58 & 16/483 & 47/1352 & 9/936 & 0.0017 & 0.0017 & 0.0023 + q1422 & @xmath59 & 96/253 & 1/728 & 0/491 & 0.0024 & 0.0024 & 0.0030 + q1623 & @xmath60 & 6/462 & 189/1220 & 2/847 & 0.0016 & 0.0016 & 0.0022 + q1700 & @xmath61 & 15/406 & 62/1456 & 1/948 & 0.0018 & 0.0018 & 0.0024 + q2233 & @xmath62 & 44/76 & 1/267 & 0/181 & 0.0028 & 0.0028 & 0.0035 + q2343 & @xmath63 & 10/385 & 148/938 & 8/541 & 0.0016 & 0.0016 & 0.0022 + q2346 & @xmath64 & 1/362 & 34/1142 & 1/754 & 0.0016 & 0.0017 & 0.0022 + ssa22a & @xmath65 & 42/151 & 10/360 & 1/253 & 0.0029 & 0.0028 & 0.0036 + ssa22b & @xmath65 & 29/73 & 5/281 & 1/308 & 0.0029 & 0.0028 & 0.0036 + westphal & @xmath66 & 172/270 & 43/724 & 20/632 & 0.0018 & 0.0018 & 0.0024 + total & 2907 & 836/4934 & 676/14009 & 80/9533 & 0.0021 & 0.0021 & 0.0029 + [ tab : fields ]
all redshifts were measured with the low - resolution imaging spectrograph ( lris ; oke et al .
1995 ) on the keck telescopes .
the number of redshifts in each field was determined by the number of clear nights that were allocated .
photometric candidates were selected for spectroscopy more - or - less at random , but in one way the selection was far from random : spectroscopic objects in each field were constrained to fit together in a non - interfering way on one of a small number of multislit masks .
this introduced artificial angular clustering to the spectroscopic samples , particularly in fields where our image s size significantly exceeded the spectrograph s @xmath67 field - of - view . in some cases the artificial angular clustering
was increased by our desire to obtain particularly dense spectroscopic sampling in some parts of an image , e.g. , near a background qso .
table [ tab : fields ] lists the number of bm , bx , and lbg photometric candidates and spectroscopic galaxies in each field .
the field - to - field variations in the number of photometric candidates per square arcminute were caused primarily by differences in exposure times , seeing , sky brightness , telescope plus instrumental throughput , and so on , from one run to the next .
recall that we used many different telescopes and cameras during our imaging survey . the expected variations in intrinsic surface density ( also shown in the table , and calculated for galaxies with bias @xmath68 as described in [ sec : angularmethods ] below )
are significantly smaller .
we estimated the angular correlation functions from the lists of photometric candidates . inferring a comoving correlation length @xmath69 from the measured angular clustering required an estimate of the objects redshift distribution .
for this we took the measured redshift distributions of the spectroscopic samples .
since the spectroscopic samples are large
several hundred for the bx and lbg criteria , nearly 100 for bm
random fluctuations are unlikely to have given redshift distributions to them that are significantly different from those of the parent photometric samples .
we were able to measure a redshift for only @xmath70% of the objects we observed spectroscopically , however , and it is therefore possible that various systematics ( e.g. , difficulties measuring spectroscopic redshifts for galaxies in certain redshift ranges ) could have caused the spectroscopic and photometric samples to have somewhat different redshift distributions .
repeated observations of a subset of the initial spectroscopic failures show that these objects have the same redshift distribution as the initial successes , implying that any systematics are not severe . to help us
interpret our observations , we refered at times the gif - lcdm numerical simulation of structure formation in a cosmology with @xmath71 , @xmath72 , @xmath73 , @xmath74 , and @xmath75 .
this gravity - only simulation contained @xmath76 particles with mass @xmath77 in a periodic cube of comoving side - length @xmath78 mpc , used a softening length of @xmath79 comoving kpc , and was released publicly , along with its halo catalogs , by frenk et al .
astro - ph/0007362 .
further details can be found in jenkins et al .
( 1998 ) and kauffmann et al .
( 1999 ) . since the gif - lcdm cosmology is consistent with the wilkinson microwave - anisotropy probe results ( spergel et al .
2003 ) , and since modeling the gravitational growth of perturbations on large ( @xmath34 mpc ) scales is not numerically challenging , we will assume that the growth of structure found in this simulation closely mirrors the growth of structure in the actual universe . for our purposes
the most interesting aspect of the simulation is the spatial distribution of virialized `` halos '' , or overdensities with @xmath80 , since these deep potential wells are the sites where galaxies can form from cooling gas .
the public halo catalogs were created , by the gif team , by running a halo - finding algorithm at various time steps in the simulation .
we will say that all the halos identified by the algorithm at time - step @xmath81 ( say ) had @xmath81 as their time of identification .
correlation functions for halos at the time of identification were calculated directly from the public halo catalogs . in subsequent time - steps these halos
were progressively displaced by gravity .
some grew ; others were destroyed as they were subsumed into larger structures .
any galaxies within the halos would be likely to survive intact , however , and it is interesting to trace the expected evolution in their correlation function over time . to do this , we assumed that the galaxies in a halo would be displaced by gravity by the same amount , and in the same direction , as the halo s most bound particle .
if @xmath82 denotes the set of particles that were the most bound particle in a halo at @xmath83 , we assumed that the correlation function at time @xmath84 of the galaxies that lay within halos identified at time @xmath83 would be roughly equal to the correlation function of particles @xmath82 at time @xmath84 . at spatial separations that are large
compared to the typical halo radius , the expected evolution of the galaxy correlation function is insensitive to the details of this procedure .
these are the only spatial separations we will consider .
our focus on large spatial scales also justifies our ignoring the possibility of galaxy mergers .
although mergers can strongly affect the correlation function on small scales , on large scales the effect is more subtle .
it can be understood as follows . at time @xmath84 , the galaxies identified at time @xmath83 will be found in halos with a range of masses , and their large - scale correlation function will be a weighted average of the correlation functions of the halos . since the weighting depends on typical number of descendant galaxies in halos of each mass , and since halos with different masses have different correlation functions , the large - scale correlation function of descendant galaxies will be altered by mergers if the merger frequency depends on halo mass . in practice , however , the difference in correlation functions between the more massive and less massive halos that host the descendants is not enormous , and as a result the merger frequency would have to be an implausibly strong function of halo mass to alter the descendant correlation functions on large scales in a significant way .
two approaches will be used to estimate the clustering strength . the first approach , which is standard , relies almost exclusively on the angular positions of the galaxies .
the second relies almost exclusively on the galaxies measured redshifts .
these approaches exploit different aspects of our data and are subject to different systematics .
the level of agreement between them provides an important test of our conclusions robustness .
this section describes and justifies the two approaches .
readers interested primarily in our scientific results may wish to skip ahead to [ sec : results ] .
the observed clustering of galaxies on the plane of the sky is related to the galaxies three - dimensional correlation function in a straightforward way .
let @xmath85 denote a galaxy s redshift and @xmath86 denote its angular position .
@xmath86 is written in bold face because two numbers ( e.g. , right ascension and declination ) are required to specify the galaxy s angular position .
if @xmath87 is the probability that a galaxy at known position @xmath88 has a neighbor at position @xmath89 , then elementary identities show that the probability that a galaxy at angular position @xmath90 will have a neighbor at angular position @xmath91 is @xmath92 observations indicate that the reduced correlation function is well approximated by an isotropic power law , @xmath93,\ ] ] where @xmath69 and @xmath94 parametrize the shape of the power law and @xmath95 is the distance between @xmath89 and @xmath88 .
this implies , in the circumstances of interest to us , that the reduced angular correlation function will also be a power law , @xmath96 with @xmath97 .
if the angular separation @xmath98 between the galaxies is small , @xmath99 , and if the comoving correlation length @xmath69 does not change significantly from the front to the back of the survey , then @xmath100 ^ 2 \label{eq : alimber}\end{aligned}\ ] ] ( see , e.g. , totsuji & kihara 1969 ) where @xmath101 is the survey selection function , @xmath102 is the beta function in the convention of press et al .
( 1992 ) , @xmath103 is the change in comoving distance with redshift , @xmath104 is the change in comoving distance with angle , and @xmath105 is the angular diameter distance .
this follows from the relationship @xmath106^{\gamma/2 } = r_0^\gamma r^{1-\gamma } b(1/2,(\gamma-1)/2)$ ] . our first approach to estimating the three dimensional clustering strength will be to measure the parameters @xmath107 and @xmath108 of the reduced angular correlation function @xmath109 , then infer values for @xmath69 and @xmath94 using the relationships above .
our estimates of @xmath110 in different angular bins will be based on the landy - szalay ( 1993 ) estimator @xmath111 where @xmath112 is the observed number of unique galaxy pairs with separation @xmath113 , @xmath114 is the number of unique pairs with separation in the same range between the observed galaxy catalog and a galaxy catalog with random angular positions , and @xmath115 is the number of unique pairs in the random catalog with separations in the same range . in practice
we reduce the noise in the random pair counts by creating random catalogs with many times more objects than the data catalogs ( @xmath116 ) , calculating @xmath117 and @xmath118 , then multiplying @xmath117 and @xmath118 by @xmath119 and @xmath120 $ ] , respectively .
unless fluctuations on the size of our typical field - of - view are negligible , the number of detected galaxies in any field will be somewhat higher or lower than in a fair sample of the universe , and the number of galaxies in the field s ideal random catalog would therefore be lower or higher than the observed number . as a result
the values @xmath117 and @xmath118 that we calculate with our approach will be incorrect to some degree . in a single field
this can make the clustering appear stronger or weaker than it truly is , but when many fields are averaged it tends to make the observed clustering appear artificially weak .
this can be shown as follows .
assume that the observed mean density in a field differs from the global average by the unknown factor @xmath121 , i.e. , @xmath122 , and let @xmath117 and @xmath118 be the pair counts calculated from scaling the random catalogs to the observed density .
one might guess that the estimator of equation [ eq : landyszalay ] ought to have @xmath117 and @xmath118 replaced by values corrected to the true mean density , i.e. , by @xmath123 and @xmath124 , and indeed hamilton ( 1993 ,
3 ) has shown that the estimator @xmath125 is equal to the true angular correlation function on average , @xmath126 this equation does not help us directly , since we do not know @xmath127 and can not calculate @xmath128 , but it does show that the estimator @xmath129 must be biased : @xmath130 where @xmath131 and the approximation assumes the weak clustering ( @xmath132 ) limit .
it is therefore customary to estimate @xmath110 by adding a constant @xmath133 to the calculated values @xmath134 .
the constant @xmath135 depends on the unknown values of @xmath127 in the observed field or fields and can not be calculated exactly . if @xmath136 , so that field - to - field fluctuations are in the linear regime and have a nearly gaussian distribution , and if our data are drawn from @xmath137 independent fields with measured pair counts @xmath138 , @xmath139 , and @xmath140 , then the value of @xmath135 appropriate to a given angular bin in our data set will have a variance of @xmath141 \\ & \simeq & \frac{1}{rr_{\rm tot}^2}\sum_{i=1}^n 2\sigma_i^4 dd_i^2 \label{eq : var_i}\end{aligned}\ ] ] around its expectation value @xmath142 here @xmath143 is the sum over all fields of the random pair counts in the chosen angular bin . in practice @xmath144 and @xmath145 depend very weakly on which bin is chosen . below we will take @xmath144 at @xmath146 as our best guess at the correction @xmath135 .
when it matters we will discuss the effect of the uncertainty in @xmath135 .
we use two approaches to estimate the size of the uncertainty @xmath147 in the mean galaxy density of the @xmath148th field .
since @xmath149 it can be estimated numerically as @xmath150 if @xmath151 is known ( infante 1994 ; roche & eales 1999 ) .
unfortunately the iterative approach suggested by equations [ eq : actualestimator ] and [ eq : rocheeales ] can be unstable , at least when the correlation function slope is allowed to vary : a large value of @xmath134 will imply a large correction @xmath152 , which implies an even larger @xmath134 and even larger correction , and so on .
the instability is undoubtedly worse for large images , where the estimate of the integral constraint correction for one iteration is completely dominated by the assumed correction from the previous .
we were unable to use equation [ eq : rocheeales ] as anything other than a consistency check .
a more robust estimate of @xmath152 follows from theoretical considerations .
since matter fluctuations will still be in the linear regime on the large scales of our observations , the relative variance of mass from one surveyed volume to the next can be estimated from the linear cold - dark matter power - spectrum @xmath153 ( bardeen et al .
1986 ; we adopt the parameters @xmath154 , @xmath75 , @xmath155 ) with parseval s relationship @xmath156 where @xmath157 is the fourier transform of a survey volume .
the shape of the observed volume in any one of our fields is reasonably approximated in the radial direction by a gaussian with comoving width ( rms ) @xmath158 and in the transverse directions by a rectangle with comoving dimensions @xmath159 . in this case @xmath160 \frac{\sin(k_xl_x/2)}{k_xl_x/2}\frac{\sin(k_yl_y/2)}{k_yl_y/2}.\ ] ] the implied value of @xmath161 for each sample in each field is shown in table [ tab : fields ] ; the values assume the powerspectrum normalization required for the r.m.s .
fluctuation as a function of redshift in spheres of comoving radius @xmath162 mpc to obey @xmath163 where @xmath164 and @xmath165 is the linear growth factor to redshift @xmath85 .
the desired corrections @xmath152 are then given by @xmath166 where @xmath167 , the galaxy bias , is calculated from the ratio of galaxy to matter fluctuations in spheres of comoving radius @xmath162 mpc : @xmath168 here the galaxy variance @xmath169 ( peebles 1980 eq .
59.3 ) can be derived from the fit to the galaxy correlation function .
this approach also requires an iterative solution , since the correction @xmath152 to @xmath110 depends on @xmath110 , but the advantage is that the assumed size of large - scale fluctuations is anchored in reality by our requirement that the slope of the correlation function match other observations on very large scales .
the discussion so far assumes that we will know the precise shape of the selection function @xmath101 .
in fact this is not true , and uncertainty in the true shape of our selection function is a source of error in the derived values of @xmath69 . means that errors in @xmath101 alter the inferred value of @xmath94 as well .
we will neglect this small effect . ]
a larger width for the selection function means that projection effects are stronger , and therefore implies a larger value of @xmath69 for given angular clustering ( see equation [ eq : alimber ] ) . if the selection function is a gaussian with mean @xmath170 and standard deviation @xmath171 , and if the weak redshift variations of @xmath172 and @xmath173 can be ignored , then the constant @xmath107 in equation [ eq : alimber ] is proportional to @xmath174 , and the implied value of @xmath69 is proportional to @xmath175
. measuring @xmath137 redshifts drawn from this selection function determines @xmath170 to a precision @xmath176 and @xmath177 to a relative precision @xmath178 . excluding interlopers with @xmath179 , we have measured roughly 800 , 700 , and 80 redshifts for galaxies in the lbg , bx , and bm samples , and the selection function width is @xmath180 for each .
the relative uncertainty in @xmath171 is therefore approximately @xmath181% for the lbg and bx samples and @xmath182% for the bm sample , which implies @xmath183% uncertainty in @xmath69 for the lbg and bx samples and @xmath184% uncertainty in the bm sample .
variations of the selection function from one field to the next ( owing , for example , to differences in the depth of the data or to systematic errors in our photometric zero points ) are another source of concern , especially at the redshifts @xmath185 where galaxies colors are insensitive to redshift and small color errors mimic large redshift differences .
suppose for simplicity that all fields have the same number of photometric candidates , let the rms width of the selection function in the @xmath148th field be written @xmath186 , where @xmath187 is the mean width among all fields , and let the mean redshift of the selection function be written @xmath188 where @xmath170 is the mean redshift among all fields .
then the rms width of the total selection function @xmath189^{1/2 } = s[1+{\rm var}(\delta)+{\rm var}(\epsilon)]^{1/2},\ ] ] exceeds the value @xmath187 that should be used in determining @xmath107 .
since we will ( by necessity ) use the total selection function in estimating @xmath69 , our estimates will be biased high .
systematic errors in our zero points are unlikely to be larger than @xmath190 , and variations in photometric depth will at most change our characteristic color uncertainties from @xmath191 to @xmath192 magnitudes .
the measured variations in galaxy redshift with @xmath7 color ( see , e.g. , adelberger et al .
2004 ) imply ( a ) that zero point errors with @xmath190 will shift the mean redshifts of galaxies that satisfy the lbg , bx , and bm selection criteria by @xmath193 , @xmath194 , and @xmath195 , respectively , and ( b ) that increasing the photometric uncertainty from @xmath196 to @xmath197 will increase the widths of the lbg , bx , and bm selection functions by @xmath198 , @xmath199 , and @xmath199% .
the upper limits on @xmath200 and @xmath201 are therefore @xmath202 ( @xmath203 ) and @xmath204 ( @xmath205 ) , respectively , for the lbg ( bx , bm ) sample .
the required reduction in @xmath69 is negligible for the lbg sample but could be as large as @xmath206% for the other two .
we will account for uncertainties in the selection function by decreasing the best - guess value of @xmath69 for the bm and bx samples by 3.5% and increasing the uncertainty in quadrature by @xmath207 .
as figure [ fig : zhistos ] shows , some fraction of the objects in the bx and bm samples will be low redshift interlopers .
we correct for the resulting dilution in the clustering strength by using the full selection function , starting at @xmath28 , in our estimate of @xmath69 from equation [ eq : alimber ] .
this is the optimal correction only if the interlopers have the same comoving correlation length @xmath208 mpc ( see below ) as the galaxies in the primary samples .
this should be nearly true , since budavri et al .
( 2003 ) estimate @xmath209 mpc for the blue star - forming galaxy population at @xmath210 from which our interlopers are drawn . in any case , since the correction itself is small eliminating the tail with @xmath179 from @xmath101 alters the inferred values of @xmath69 for the bm and bx samples by only @xmath198%errors in it should not have an appreciable effect on our estimates of @xmath69 .
we face three significant obstacles in trying to estimate the clustering strength from the spectroscopic catalogs .
\(1 ) the objects in a given field that were selected for spectroscopy were not distributed randomly across the field , but were instead constrained to lie on one of a small number of multislit masks . since only a small fraction of the galaxies were observed spectroscopically in the typical field , the finite size of the masks coupled with the need to avoid spectroscopic conflicts produced significant artificial clustering in the angular positions of sources in the spectroscopic catalog .
the effect was worsened in some fields by our decision to obtain particularly dense spectroscopy near background qsos .
( 2 ) because galaxies @xmath7 colors change slowly with redshift near @xmath185 , the expected redshift distribution @xmath101 of our bm and bx color - selected samples depends sensitively on the quality of the photometry .
the larger color errors from noisy photometry will lead to a broader @xmath101 , while relatively small systematic shifts in the photometric zero points can significantly alter the mean of @xmath101 .
adelberger et al .
( 2004 ) and [ sec : selfn_uncertainties ] of this paper discuss this point in more detail , but the upshot is that we can not estimate the selection function @xmath101 with great precision for the bx and bm samples .
( 3 ) peculiar velocities and redshift uncertainties render imprecise our estimate of each galaxy s position in the @xmath85 direction .
this limits the accuracy of our estimate of the distance from one galaxy to its neighbors , complicating our efforts to measure the correlation function on small spatial scales .
effects ( 1)(3 ) are usually compensated with the aid of detailed simulations .
although this approach should work in principle , in practice it is hard for outsiders to evaluate whether the simulations were flawed .
the remainder of this section describes the alternate approach that we adopt .
it is based on analyzing observable quantities that are not affected by systematics ( 1)(3 ) .
the spurious angular clustering signal can be eliminated if we take the angular positions of spectroscopic galaxies as given and estimate the clustering strength solely from their redshifts .
let @xmath211 be the comoving distance to redshift @xmath85 , and let @xmath212 be the transverse comoving separation implied by the angular separation @xmath213 between a galaxy and some reference position , e.g. , the center of the observed field . according to elementary probability identities , if we know that one galaxy has position @xmath214 , then the probability that a second galaxy at transverse position @xmath215 has radial position @xmath216 is @xmath217 and the expected distribution of radial separations @xmath218 for galaxies with transverse separation @xmath219 is @xmath220\times\nonumber\\ & & \quad \int_0^{\infty } dz_2\,\frac{p(z_2)p(z_2+z_{12})}{1+p(z_2)r_0^\gamma r_{12}^{1-\gamma}\beta(\gamma ) } \label{eq : pzgivenr}\end{aligned}\ ] ] where we have used results from the previous section and adopted the shorthand @xmath221 for the beta - function given above , @xmath222 .
( equations [ eq : pzgivenrrz ] through [ eq : pzgivenr ] assume that the quantity @xmath223 is constant with redshift , an approximation that is valid for the small separations @xmath224 comoving mpc between the galaxy pairs we will use in this analysis . ) equation [ eq : pzgivenr ] shows that the observable quantity @xmath225 is sensitive to the clustering strength but independent of angular variations in the spectroscopic sampling density .
unfortunately the correlation function @xmath226 can be estimated from @xmath225 with equation [ eq : pzgivenr ] only if we have a reasonably accurate estimate of the selection function shape @xmath227 .
this can be seen more clearly by taylor - expanding the integral in equation [ eq : pzgivenr ] around @xmath228 and approximating the selection function as a gaussian with standard deviation @xmath171 that is centered many standard deviations from @xmath28 .
one finds @xmath229\times\nonumber\\ & & \quad\quad\bigl[\frac{a_0}{\sigma_{\rm sel } } + \frac{a_2}{2\sigma_{\rm sel}^3}z_{12}^2+\frac{a_4}{24\sigma_{\rm sel}^5}z_{12}^4\bigr ] \label{eq : pzgivenrapprox}\end{aligned}\ ] ] where @xmath230 @xmath231 , @xmath232 , @xmath233 , @xmath234 and @xmath235 .
the coefficients @xmath236 all have similar sizes since the integrals are dominated by contributions from @xmath237 where the integrands are of the same order . in the angular clustering case
above , inaccuracies in the adopted width @xmath171 of the selection function affected the inferred amplitude of the correlation function but not its shape . here
they affect both .
moreover errors in @xmath171 are multiplied not by @xmath226 , but by @xmath238 , which implies that they can easily dominate the true clustering signal when @xmath239 .
equation [ eq : pzgivenrapprox ] shows that one must be careful estimating the strength of redshift clustering when the shape of the selection function is poorly known .
our solution exploits the fact that @xmath240 for our survey , which implies @xmath241 for all separations where @xmath226 is large enough to measure . as long as @xmath241
, the terms proportional to @xmath242 and @xmath243 can be neglected in equation [ eq : pzgivenrapprox ] , and @xmath225 will be very nearly equal to @xmath244 $ ] , with @xmath245 a function that does not depend on @xmath246 . the function @xmath245 does depend on the unknown selection function , but it can be eliminated by taking ratios of pair counts in a manner we discuss below
. ratios of @xmath225 at fixed @xmath247 and different @xmath246 will therefore be the basis of our estimate of the clustering strength in the spectroscopic sample ; they are nearly immune to systematics from the irregular spectroscopic sampling and from the unknown selection function shape .
the final complication is the significant uncertainty @xmath248 in each object s radial position @xmath211 from peculiar velocities and redshift uncertainties .
this uncertainty can be treated in various ways .
we will follow the standard approach and estimate the value of the correlation function only within bins whose radial size @xmath249 is large compared to @xmath248 .
we are now ready to present the estimator that we adopt . letting @xmath250 denote the observed number of galaxy pairs with transverse separation @xmath251 and redshift separation @xmath252 , the discussion of the preceding paragraphs shows that the expected total number of pairs with radial separation @xmath252 ( and any transverse separation ) is @xmath253 \label{eq : expntot}\ ] ] where @xmath254 as long as @xmath255 is large enough that @xmath256 the ratio of pair counts @xmath257 will have expectation value @xmath258 regardless of angular selection effects , of uncertainties in the selection function , is independent of uncertainties in the selection function width .
the expectation value @xmath259 _ will _ be affected by errors in the mean redshift of the selection function if these errors are large enough to significantly alter the mapping of redshifts and angles to distances .
] of peculiar velocities , and of redshift measurement errors , provided @xmath260 , @xmath261 , @xmath262 , @xmath263 , the selection function does not have strong features on scales smaller than @xmath171 , and @xmath223 varies slowly with @xmath85 . here
@xmath264 is given by equation [ eq : expntot ] and @xmath265.\nonumber\ ] ] the second approximate equality in equation [ eq : exp_kb ] exploits the fact that @xmath266 is a very weak function of @xmath267 in realistic situations .
we estimate the correlation function from the spectroscopic sample by finding the parameters required to match the observed ratio @xmath268 . in principle
@xmath268 could be calculated separately for pairs in different bins of transverse separation @xmath267 , producing an estimate of the function @xmath269 and allowing one to estimate both @xmath69 and @xmath94 from the data . in practice a much larger sample
is needed to fit for both @xmath69 and @xmath94 , so we hold @xmath94 fixed and estimate @xmath69 only .
fortunately , as we will see , the best fit value of @xmath69 hardly changes as @xmath94 is varied across the range allowed by the galaxies angular clustering .
the dependence of this estimator on the clustering strength is easy to understand intuitively .
if the galaxies were unclustered ( @xmath270 ) , we would observe the same number of pairs at every separation and @xmath268 would be equal , on average , to the ratio of the bin sizes @xmath271 .
correlation functions that peak near @xmath272 will produce more pairs in bins at smaller separations , driving @xmath268 away from @xmath273 .
the difference between @xmath268 and @xmath273 is sensitive to the strength of the clustering , and therefore can be used to estimate it .
adelberger ( 2005 ) uses monte carlo simulations to analyze the behavior of @xmath268 in more detail .
figure [ fig : lsraw ] shows the raw ( integral - constraint correction @xmath274 ) values of the landy - szalay estimator @xmath134 ( equation [ eq : landyszalay ] ) as a function of angular separation for galaxies in the three samples .
we limited these data , and our subsequent fits , to angular separations @xmath275 , since at larger scales the weak angular - clustering signal could be swamped by various low - level systematics .
the uncertainty @xmath147 in each bin was taken to be the larger of @xmath276 ( peebles 1980 , 48 ) and the observed standard deviation of the mean of @xmath277 among the different fields in the survey .
typically the two were comparable .
numerical @xmath278 minimization produced the power - law fits shown with dashed lines .
the correlation function parameters implied by the lbg fit , @xmath279 comoving mpc , @xmath280 , agree well with the estimates of giavalisco & dickinson ( 2001 ) which also assumed @xmath274 .
it is clear , however , that these parameters can not be correct .
substituting them into equations [ eq : s2_from_r0 ] , [ eq : bias_from_sigma ] , and [ eq : ic_cdm ] shows that a significant correction @xmath135 should have been applied to account for fluctuations on scales larger than the field - of - view .
( porciani & giavalisco 2002 reached a similar conclusion , and derived a result for lbgs that agrees well with the integral - constraint - corrected result we present below . ) figure [ fig : wthet_converg ] shows how our best - fit estimates of @xmath69 and @xmath94 change as the correction @xmath135 is applied . in our first iteration , described above , we assumed @xmath274 and calculated the correlation function @xmath281 .
for the second iteration we assumed the value of @xmath135 implied by @xmath282 ( equations [ eq : exp_i ] , [ eq : s2_from_r0 ] , [ eq : bias_from_sigma ] , and [ eq : ic_cdm ] ) and estimated @xmath283 . for the third iteration we calculated @xmath135 from @xmath283 .
the process continued in this way until convergence .
it settled on the same final parameters if we initially assumed a value for @xmath135 that was too large .
as figure [ fig : wthet_converg ] shows , the applied integral constraint corrections were comparable for each of the three samples .
this is because the increase in @xmath135 implied by the longer correlation lengths at lower redshifts happened to be cancelled by a decrease in @xmath135 that resulted from the lower - redshift samples greater comoving depths . to check the plausibility of our adopted values for @xmath135
, we inserted into equation [ eq : rocheeales ] the best power - law fits to @xmath110 from each sample s final iteration .
the equation returned @xmath284 , @xmath285 , and @xmath286 as the empirical estimates of @xmath135 for bm , bx , and lbg samples .
these values differ somewhat from the ones we adopt ( figure [ fig : wthet_converg ] ) , because the empirical and cdm approaches ( see [ sec : integralconstraint ] ) make different assumptions about the behavior of @xmath110 on the scales @xmath287 where we can not measure it , but they are consistent within their large @xmath288 uncertainties and small changes ( @xmath289 ) to the best - fit values of @xmath94 would make them agree perfectly .
readers may also be reassured to recall that our estimates of @xmath69 and @xmath94 agree well with those of porciani & giavalisco ( 2002 ) , who corrected for the integral constraint in a completely different way .
we estimated the random uncertainty in @xmath69 and @xmath94 in two ways .
first , we analyzed many alternate realizations of our @xmath110 measurements that were generated under the assumption that the uncertainties were uncorrelated . to create a single alternate realization
, we added to each measured value @xmath151 a gaussian random deviate with standard deviation equal to its uncertainty @xmath147 . after creating numerous alternate realizations , we calculated and tabulated the values of @xmath69 and @xmath94 implied by each .
our @xmath288 confidence interval on @xmath69 was defined as the range that contained 68.3% of the measured values of @xmath69 among the alternate data sets .
the @xmath94 confidence interval is defined in the same way .
we found @xmath290 mpc , @xmath291 ( lbg ) , @xmath292 mpc , @xmath293 ( bx ) , and @xmath294 mpc , @xmath295 ( bm ) .
these numbers assume uncorrelated error bars and neglect the uncertainty in our selection functions . the uncertainty in @xmath135 is also neglected , since each alternate realization had the same integral constraint correction .
our second approach was to extract random subcatalogs from our full galaxy catalog , estimate @xmath69 and @xmath94 for each with the iterative solution for @xmath110 described above , measure how the r.m.s .
dispersion in best - fit parameter values depended on the number of sources in the subcatalog , and extrapolate to the full catalog size .
in fact we created our random subcatalogs in pairs , with both subcatalogs in a pair containing a random fraction @xmath296 of the sources in the full catalog and no sources in common between them , and estimated the uncertainty in @xmath69 at a given value of @xmath172 as @xmath297 times the r.m.s .
difference in @xmath69 among pair members .
this prevented us from underestimating the random uncertainty in @xmath69 as @xmath298 , when random subcatalogs could otherwise contain nearly the same galaxies . with this approach
we estimate @xmath299 mpc , @xmath300 ( lbg ) , @xmath301 mpc , @xmath302 ( bx ) , and @xmath303 mpc , @xmath304 ( bm ) .
these numbers neglect the uncertainty in our selection function and do not fully account for the uncertainty in @xmath135 .
they do not assume uncorrelated error bars , however , and we will therefore assume that they are more accurate than the numbers from the preceding paragraph .
the uncertainty in @xmath135 is not negligible . according to equation [ eq : var_i ] the @xmath288 uncertainty in @xmath135 is @xmath305@xmath306% as large as @xmath135 itself for our 21 fields . as @xmath135 varies over its @xmath288 allowed range , the best - fit parameters @xmath69 and @xmath94 change by roughly as much as the uncertainties quoted above . adding these changes in quadrature to the random uncertainties above , and making the minor corrections for the selection function uncertainties discussed in
[ sec : angularmethods ] , we arrive at the following estimates : @xmath307 mpc , @xmath308 ( lbg ) , @xmath309 mpc , @xmath310 ( bx ) , and @xmath311 mpc , @xmath312 ( bm ) . other investigators ( e.g. , giavalisco & dickinson 2001 ; foucaud et al .
2003 ) have claimed that at redshift @xmath313 bright galaxies cluster more strongly than faint galaxies .
our data support this conclusion .
figure [ fig : clust_seg ] shows that in the bx and lbg samples the correlation lengths of galaxies with @xmath314 ( @xmath315 mpc ) exceed those of galaxies with @xmath316 ( @xmath317 mpc ) by a significant amount . if we split the bx and lbg samples into two halves at random , rather than by apparent magnitude , the difference in correlation lengths between the two halves is this large only about @xmath318% ( bx ) to @xmath319% ( lbg ) of the time .
the situation is less clear for the bm sample at @xmath30 , where the uncertainties are larger owing to the poor determination of @xmath101 from the small number of measured redshifts , but the data do not seem to suggest stronger clustering for brighter galaxies . on the one hand it makes sense that uv - brightness should become less associated with strong clustering as redshift decreases ,
since uv - bright galaxies are known to be weakly clustered at @xmath28 and @xmath29 . on the other
, the overall clustering of the bm sample is still quite strong , stronger than one expects for typical collapsed objects at @xmath30 ( see below ) , and so it seems that the numerous objects too faint to satisfy our selection criteria must be less clustered than the bright objects in our sample .
we will wait for additional spectroscopic observations of bm galaxies before commenting further .
the dependence of clustering strength on luminosity can produce a false impression of a change in @xmath69 with redshift , since lower redshift samples will tend to reach fainter absolute luminosities .
truncating the samples at a fixed absolute luminosity does not seem a good solution to us , however , since the bright end of the uv luminosity function rises rapidly towards higher redshifts ( e.g. , adelberger & steidel 2000 ) and one would therefore be comparing rare objects at lower redshifts to common objects at higher redshifts . a better approach is to compare galaxy samples of roughly the same comoving number density . since selection with a constant apparent magnitude limit @xmath320 happens to produce similar comoving number densities for the three samples
( see equations [ eq : nlbg ] and [ eq : nbxbm ] and the related discussion ) , we will continue to use the constant apparent magnitude limits of equation [ eq : maglimits ] for our samples in the remainder of the paper .
readers should be aware that the reported value of each sample s correlation length is somewhat arbitrary for this reason .
it reflects the characteristics of the sample as defined here , not of the general galaxy population at high redshift . for our estimator of the redshift clustering strength we took @xmath321 , the ratio of the number of galaxy pairs with comoving radial separation @xmath322 to those with comoving radial separation @xmath323 .
since @xmath79 mpc is significantly larger than the uncertainty in each galaxy s radial position ( @xmath324 comoving mpc ) , and since @xmath325 mpc is significantly smaller than the selection functions widths ( @xmath326 comoving mpc ) , the expected value of @xmath321 should be given by equation [ eq : exp_k ] .
we limited our analysis to pairs with transverse separations @xmath327 , equivalent to @xmath328 comoving mpc at @xmath329 , to reduce the sensitivity of our results to any deviations of the correlation function from a @xmath330 power - law on large scales .
only the bx and lbg spectroscopic samples were large enough to allow meaningful measurements of @xmath321 .
for @xmath330 , the right - hand side of equation [ eq : exp_k ] is equal to the observed ratio @xmath321 when @xmath331 ( lbg ) or @xmath332 ( bx ) comoving mpc .
the values change by roughly @xmath333% as @xmath94 is varied from 1.45 to 1.65 . when analyzed with this technique , mock galaxy catalogs from the gif simulation ( [ sec : simulated_data ] ) with sizes similar to
our observed catalogs show a @xmath41 dispersion in @xmath69 around the true mean of @xmath334 ( lbg ) and @xmath335 ( bx ) comoving mpc , so we adopt @xmath336 and @xmath337 mpc as the best fit values to @xmath69 for our spectroscopic catalogs .
the results do not change significantly if we eliminate pairs with @xmath338 from the analysis , showing that we have measured genuine large - scale clustering and not merely the clumping of objects within individual halos .
figure [ fig : kclust ] presents the result in a more graphical way .
we divided our lists of galaxy pairs into bins according to transverse separation @xmath267 , then calculated @xmath321 separately for each bin .
points with error bars show the values we found .
the solid lines show the values predicted by the @xmath330 correlation function described in the preceding paragraph .
the plot shows that the derived correlation function parameters provide a reasonable fit to the clustering of the galaxies in the spectroscopic sample .
we presented two independent estimates of the correlation function for each of our galaxy samples .
the estimates were consistent with each other , but the first , based on the galaxies angular clustering , had somewhat smaller uncertainties .
this resulted from the larger size of the photometric sample , and was accentuated by the serious systematics in the spectroscopic sample that made us throw much of our data away .
we will adopt the angular clustering results for the remainder of the paper .
on small scales , smaller than roughly the typical radius @xmath339 of a virialized halo , the spatial clustering of galaxies is difficult to predict or interpret .
it depends on the ease with which nearby galaxies merge with each other , on the ability of a galaxy to maintain its star - formation rate as it orbits within a larger potential well , on the possible impact of a galaxy s feed - back on its surroundings , and so on . on larger scales these baryonic complications have little effect and the correlation function of galaxies should be virtually identical to the correlation function of the halos that host them . to see that this is true ,
consider the galaxy correlation function in a simplified situation where every galaxy is associated with a single halo and the probability that a galaxy lies a distance @xmath340 from its halo s center is @xmath341 . in this case
the galaxy distribution will be a poisson realization of the continuous density field @xmath342 that is created when the discrete halo distribution is convolved by @xmath172 , and the galaxy correlation function @xmath343 will be equal to the correlation function of @xmath344 ( see , e.g. , peebles 1980 , eq .
33.6 ) . since the halo powerspectrum is @xmath345 ( peebles 1980 , eq .
41.5 ) , where @xmath346 is the halo correlation function and @xmath137 is the halo number density , since the powerspectrum of @xmath344 is equal to @xmath347 , where @xmath348 is the fourier transform of @xmath172 , and since the powerspectrum and correlation function are fourier - transform pairs , the galaxy and halo correlation functions will be related through @xmath349 where @xmath350 denotes convolution and @xmath351 .
now @xmath352 when @xmath340 is greater than some maximum separation @xmath353 ; galaxies can not be located arbitrarily far from the center of their halo .
the first term will therefore be zero for @xmath354 .
the second term will be almost exactly equal to @xmath355 at the same large separations , because plausible correlation functions do not not change significantly from @xmath340 to @xmath356 when @xmath357 .
this shows that @xmath358 for large @xmath340 .
although it was derived for a simplistic model , the result is more general . as long as galaxies are associated in some way with dark matter halos , and
as long as there is some maximum separation @xmath353 between each galaxy and the center of its halo , galaxies will have the same correlation function as the halos that host them for @xmath359 .
we will accordingly focus our attention solely on separations @xmath360 comoving mpc that are many times larger than the typical virial radius . readers who are skeptical of the claimed similarity between galaxy and halo correlation functions at @xmath359 may wish to consider figure [ fig : xi_halo_0z3 ] , which compares observed galaxy correlation functions at redshifts @xmath361 , @xmath362 , @xmath363 to the halo correlation functions in the gif - lcdm simulation outputs at the same redshifts ( see [ sec : simulated_data ] ) .
the agreement is excellent .
the implied association of galaxies with massive potential wells is hardly surprising .
the interesting result is the characteristic mass of the virialized halos that contain the galaxies .
this can be estimated since more massive halos cluster more strongly .
figure [ fig : xi_halo ] compares the correlation functions at @xmath364 for galaxies in the bm , bx , and lbg samples to the correlation functions of virialized dark matter halos above various mass thresholds in the gif - lcdm simulation ( see
[ sec : simulated_data ] ) .
the agreement is best if galaxies in the bm , bx , and lbg samples are associated with halos of mass @xmath365 , @xmath366 , and @xmath367 , respectively .
uncertainties in our measured correlation functions lead to uncertainties in the estimated characteristic masses .
the size of these uncertainties can be gauged most cleanly by comparing our measured angular correlation function to the angular correlation functions of halos above different mass thresholds .
we calculated halo angular correlation functions numerically from equation [ eq : limber0 ] after substituting in the observed redshift distributions for the different samples ( fig .
[ fig : zhistos ] ) and the gif - lcdm halo correlation functions at the mean redshift of each sample .
typical results are shown in figure [ fig : wthet_halo ] .
the halo angular correlation functions fall significantly below the data on small scales , since ( by definition ) a halo can not have another halo as a neighbor within the radius @xmath339 . as argued above , however , it is the larger scales @xmath360 comoving mpc ( i.e. , @xmath368 ) that are relevant for comparing to galaxies , and here the agreement is good . the uncertainty in the galaxies implied mass scale is dominated by the uncertainty in the integral constraint correction @xmath135 that was applied to the data , since this moves all points up or down together .
rough @xmath288 limits on the threshold masses of the hosting halos are @xmath369 readers unfamiliar with the idea of threshold masses may wish to see footnote 6 in [ sec : summary ] , below , for further discussion of how to interpret them .
one implication of figure [ fig : wthet_halo ] is that some halos are occupied by more than one of the galaxies in our samples ; the data on small scales are strongly inconsistent with the correlation functions that assumed one object per halo .
( this was first pointed out by wechsler et al .
( 2001 ) and later denied by porciani & giavalisco ( 2002 ) .
our analysis agrees far better with that of wechsler et al . )
the fraction of lbgs that reside in the same halo as another lbg can be calculated as follows .
if there were never more than one lbg per halo , the expected number of lbgs within @xmath370 of a randomly chosen lbg would be @xmath371 where @xmath372 is the surface density of lbgs , @xmath373 , and @xmath374 is halo angular correlation function that fits our data best for @xmath375 .
the actual number of lbgs within @xmath370 , @xmath376 , is given by the same equation with the halo correlation function @xmath374 replaced by the galaxy correlation function @xmath377 .
the expected number of additional lbgs in a halo that is known to contain one lbg , @xmath378 , is equal to the difference between @xmath376 and @xmath379 .
numerically integrating the angular correlation functions for observed galaxies and simulated halos , and multiplying by the lbg surface density from table [ tab : fields ] , we estimate @xmath380
. the numbers for the other samples , estimated with a similar approach , are @xmath381 and @xmath382 .
_ some _ galaxies must share the same halo to explain the data , but the required number is small .
note that we are referring solely to galaxies that satisfy our color and magnitude selection criteria ; we obviously can not say anything about the spatial distribution of galaxies that are not in our samples
. having established a rough characteristic mass for the halos that contain our galaxies , we can compare the galaxy and halo number densities and estimate what fraction of the most massive halos at redshifts @xmath383 do not contain a galaxy that is detectable with our techniques . according to adelberger & steidel ( 2000 ) the comoving number density of lbgs brighter than @xmath384 is @xmath385 for @xmath71 , @xmath72 . combining the bm and bx completeness coefficients in table 3 of adelberger et al .
( 2004 ) with the surface densities in table [ tab : fields ] of this paper , we estimate @xmath386 where the assigned uncertainties of 50% are approximate guesses that will be refined later with monte carlo simulations .
number densities for the population of halos that contain the galaxies can be estimated from the gif - lcdm simulation given the range of halo masses ( equation [ eq : mass_scales ] ) that are compatible with the galaxies clustering strength .
figure [ fig : nhalo ] shows that the number density of galaxies in the bm , bx , and lbg samples is comparable to the number density of halos that can host them . as we will discuss in
[ sec : summary ] , below , this implies that the duty cycle of star - formation in the galaxies must be of order unity and shows that our surveys can not be severely incomplete .
similar arguments have been made by adelberger et al .
( 1998 ) , giavalisco & dickinson ( 2000 ) , martini & weinberg ( 2001 ) , and others .
the spatial distribution of a population of galaxies evolves in an easily predictable way as it responds to the gravitational pull of dark matter .
we used a simple approach to estimate this evolution from the lcdm - gif simulation .
after connecting the observed galaxies to halos with a range of masses ( equation [ eq : mass_scales ] ) , we measured the evolution in the clustering of those halos in the simulation and assumed that the galaxies clustering would evolve in the same way .
[ sec : simulated_data ] . figure [ fig : r0_vs_z ] shows the implied change in correlation length @xmath69 with time for the galaxies in our samples . by @xmath185 galaxies in the lbg sample
will have a correlation length similar to measured correlation lengths of galaxies in the bm and bx samples .
by @xmath27 their correlation length will be similar to the correlation length of early - type ( i.e. , `` absorption line '' ) galaxies in the sample of coil et al .
by @xmath210 their correlation length will be equal to the observed correlation length of ellipticals ( budavri et al . 2003 ) .
the evolution of @xmath69 for galaxies in the bm and bx samples is similar .
figures [ fig : z1gals ] and [ fig : z0gals ] present a more detailed view of the possible relationships between the descendants of galaxies in our samples and various galaxy populations at lower redshift .
figure [ fig : z1gals ] shows that at @xmath29 the descendants clustering strength will significantly exceed that of average galaxies in optical magnitude - limited surveys . since these surveys are dominated by star - forming ( `` emission - line '' ) galaxies , we can conclude that the typical descendant is no longer forming stars by @xmath27 . emission - line sample .
( our estimate of the number density in the emission - line sample , shown in figure [ fig : nhalo ] , is from a. coil , private communication . ) ] a similar point was made by adelberger ( 2000 ) and coil et al .
the stronger clustering of redder and brighter sub - populations at @xmath27 is more compatible with the descendants expected clustering , but the match is best for the sub - population with early - type spectra .
this is especially true for descendants of the brightest galaxies in the bx and lbg samples .
although the observed number density of early - type galaxies at redshift @xmath27 , roughly @xmath387 ( chen et al . 2003 ) , is consistent with the idea that most had a bm / bx / lbg galaxy as a progenitor , we can not rule out the idea that some had multiple merged bm / bx / lbg progenitors and others had none .
figure [ fig : z0gals ] is an analogous plot for redshift @xmath210 .
correlation lengths for various populations in the sloan digital sky survey ( sdss ) were taken from budavri et al .
number densities were calculated assuming a surveyed volume of @xmath388 mpc@xmath389 with an uncertainty of @xmath390% ( t. budavri 2004 , private communication ) .
the expected @xmath361 clustering strength of typical lbg descendants ( darker shaded box ) agrees best with the clustering of galaxies with early - type seds in the budavri et al .
( 2003 ) sample .
the galaxies in our high - redshift samples are roughly as numerous as these early - type galaxies , though the possibility of merging prevents us from estimating the number density of their bm / bx / lbg descendants at @xmath210 . the descendants of brighter lbgs will have a correlation length closer to that of bright ellipticals , though there are probably not enough descendants to account for the entire bright elliptical population .
the first part of the paper was concerned with measuring the spatial clustering of large samples of star - forming galaxies at redshifts @xmath30 , @xmath5 , and @xmath6 .
we fit a three - dimensional correlation of the form @xmath391 to the galaxies angular clustering with standard techniques and to the galaxies redshift clustering with a new estimator .
the new estimator , @xmath13 , is insensitive to many of the possible systematic biases in our spectroscopic surveys .
we reached consistent conclusions about the correlation function with the two approaches , but adopted the angular results since their random uncertainties were somewhat smaller .
as given in
[ sec : angularresults ] , the best - fit correlation - function parameters from the angular clustering are @xmath392 where bm , bx , and lbg are the names we have given the @xmath7 color - selection criteria that we used to find galaxies at @xmath30 , @xmath5 , and @xmath6 ( [ sec : observeddata ] ) .
the quoted @xmath288 errors include random uncertainties , uncertainties in the integral constraint corrections , and uncertainties in the shapes of the selection functions .
since the value of @xmath69 depends on the apparent - magnitude limit of the samples , at least in the two higher - redshift bins ( figure [ fig : clust_seg ] ) , the reported values of @xmath69 are somewhat arbitrary .
we chose to limit each sample to a fixed range of apparent magnitudes , @xmath393 , on the grounds that this resulted in a similar comoving density in each redshift bin .
different magnitude limits would have resulted in different correlation lengths .
readers should be aware that the numbers we give are appropriate to the samples as we have defined them , not to the general galaxy population at high redshifts .
the second part of the paper was based on the proposition that wmap ( spergel et al .
2003 ) and other experiments have given us reliable measurements of the cosmological parameters and of the shape of the dark matter power - spectrum .
this implies that we know what sorts of virialized dark - matter halos existed at different epochs in the past and how their spatial distribution evolved over time .
since galaxies reside within dark - matter halos , they will have the same correlation function as the halos on large scales ( equation [ eq : xigxih ] ) .
the galaxies clustering should therefore tell us what sort of halo they reside within .
we found a good match ( figures [ fig : xi_halo ] and [ fig : wthet_halo ] ) between the correlation functions of the galaxies and of halos with threshold masses ranging from @xmath394 ( lbg ) to @xmath395 ( bm ) .
, we mean the subset of halos with mass @xmath396 .
the median mass of this subset is @xmath397 in the gif - lcdm simulation at @xmath398 .
halo subsets can be defined with schemes more elaborate than our simple mass threshold ( e.g. , kauffmann et al . 1999 , bullock , wechsler , & somerville 2002 ) , but the differences between the possible schemes are too small to affect our analysis when the subsets they produce are constrained to have the same clustering on large scales . ]
equation [ eq : mass_scales ] gives rough @xmath288 limits on the halos total masses .
similar masses for lyman - break galaxies have been derived with the same approach by jing & suto ( 1998 ) , adelberger et al .
( 1998 ) , giavalisco & dickinson ( 2000 ) , porciani & giavalisco ( 2002 ) , and others .
although the estimated masses were derived solely from the galaxy clustering , they seem reasonable on other grounds
. they can not be much higher .
the number density of halos would be lower than the number density of lbgs , for example , if the halo mass were greater than @xmath399 .
such large halo masses would be possible only if significantly more than one lbg resided in the typical halo , and that is something that our observations rule out ( [ sec : correspondence ] ) . nor
can they be much lower .
the halos that contain lbgs would not have enough baryons to form the median lbg stellar mass of roughly @xmath400 ( shapley et al .
2001 times lower than the value shapley et al . calculated for an imf with a salpeter slope between @xmath401 and @xmath402 .
their assumed salpeter imf is probably unrealistic since the imf in the solar neighborhood is known to flatten near @xmath403 and eventually turn over at lower masses .
see leitherer ( 1998 ) or renzini ( 2004 ) for further discussion . ] ) unless their total mass were greater than about @xmath404 .
we should mention in passing that our best - fit halo masses seem to imply that only a small fraction of the baryons in the halos are associated with the observed galaxies .
for example , the best - fit mass threshold of @xmath394 for lbgs corresponds to a median total mass of @xmath405 and median baryonic mass of @xmath406 ( for @xmath407 , spergel et al .
2003 ) , roughly ten times larger than the observed stellar masses of lbgs . since the @xmath408 supernovae that explode during the assembly of the typical lbg s stellar mass will eject roughly @xmath409 of metals ( e.g. , woosley & weaver 1995 ) , enough to enrich at most @xmath410 of gas to lbgs typical metallicities of @xmath411 ( pettini et al . 2002 ) , their observed interstellar gas can not contain a large fraction of the remaining baryons .
these baryons need not be associated with other objects in the halo , however .
they may be locked in dim stars that formed in previous episodes of star - formation ( e.g. , papovich , dickinson , & ferguson 2001 ) , or may have been heated by various processes to undetectably high temperatures .
the latter is presumably the case for nearby galaxies , whose ratios of mass in stars and gas to total mass are usually also smaller than the wmap value @xmath407 . the milky way , for example , has a total mass of @xmath412 ( zaritsky 1999 ; wilkinson & evans 1999 ) and a mass in gas and stars of only @xmath413 ( k. freeman 2004 , private communication ) , yet few would assert that its missing baryons belong to another galaxy in its halo .
after establishing plausible total masses for the halos associated with the galaxies , we considered some of the implications .
our arguments were not new ( see , e.g. , moustakas & somerville 2002 , martini & weinberg 2001 , adelberger et al .
they seemed worth revisiting only because our knowledge of the cosmogony , of the local universe , and of high - redshift galaxies has improved so much in the last few years .
we began by estimating the completeness of our surveys from a comparison of the galaxies number densities to the number densities of halos with similar clustering strength ( figure [ fig : nhalo ] ) .
similar number densities would imply that almost all of the most massive halos contained a galaxy that satisfied our selection criteria ; a much lower galaxy number density would imply that most of the galaxies in massive halos are missed by our survey . defining @xmath273 as the ratio of galaxy to halo number density
, we found rough @xmath288 limits of @xmath414 , @xmath415 , and @xmath416 .
these limits were derived from the clustering at radii @xmath360 comoving mpc .
the clustering on smaller scales , sensitive to the possible presence of more than one galaxy in a halo , implies that the upper limits on @xmath417 and @xmath418 should be revised downwards to @xmath419 .
the data appear consistent with the claim of franx et al .
( 2003 ) that our selection criteria find roughly half of the most massive galaxies at @xmath185 .
a completeness of order @xmath306% seems plausible to us for other reasons as well .
shapley et al .
( 2001 ) estimate a lifetime for the typical lbg of @xmath420 yr , for example , which implies that the typical lbg will be bright enough for us to detect for only about half of the time that elapsed between the survey selection limits of @xmath421 and @xmath422 .
we considered next the way the clustering of the galaxies would evolve ( figure [ fig : r0_vs_z ] ) .
analysis of the gif - lcdm simulation suggested that the correlation length of lbg descendants would be similar by @xmath31 to the correlation length of galaxies in the bx sample and by @xmath30 to the correlation length of galaxies in the bm sample .
the spatial clustering is therefore consistent with the idea that we are seeing the same population at all three redshifts , though the selection criteria s @xmath423% incompleteness leaves room for the populations to be distinct and the difference in stellar masses between the lbg ( @xmath400 ) and bx ( @xmath424 ; steidel et al .
2005 , in preparation ) samples may not be consistent with continuous star - formation at observed rates through the elapsed time . turning our attention to lower redshifts
, we found that at @xmath27 our descendants clustering would most closely match the observed clustering of galaxies that are red and bright and have early - type spectra ( figure [ fig : z1gals ] ) .
by @xmath210 the estimated clustering of the descendants suggested elliptical galaxies as the most likely counterparts ( figure [ fig : z0gals ] ) .
the correspondence is especially hard to dispute for the descendants of the brightest and most strongly clustered galaxies in the high - redshift samples .
one conclusion seems difficult to escape : the descendants of the galaxies in our samples must have significantly larger stellar masses than their high - redshift forebears .
only @xmath425% of the total stellar mass in the local universe is found in galaxies with stellar masses smaller than @xmath426 ( kauffmann et al .
2003 ) , similar to the values in our high - redshift samples , and these faint galaxies are too weakly clustered to have descended from the galaxies we find at @xmath2 .
only elliptical galaxies have a spatial distribution consistent with our expectations for the descendants , and the characteristic stellar mass of ellipticals is @xmath404 ( padmanabhan et al .
the increase in stellar mass from @xmath185 to @xmath26 could have been produced by ongoing star formation or by mergers .
our results at redshift @xmath27 may favor the latter , but in any case our findings are strongly inconsistent with traditional notions of monolithic collapse
. it would be unfair to close without mentioning one population of high - redshift galaxies that we have ignored completely .
these are the bright ( @xmath427 ) near - infrared - selected galaxies . their reported star - formation rates ( @xmath428/yr ) , correlation lengths ( @xmath429 mpc , daddi et al .
2004 ) and stellar masses ( @xmath430 ; van dokkum et al .
2004 ) are extraordinary , far larger than the corresponding values for typical galaxies in our samples .
see , e.g. , shapley et al . 2004 and adelberger et al .
2005 ] galaxies with similarly extreme properties are not a negligible component of the high redshift universe .
the shape of the @xmath431 m background implies that up to a third ( cowie , barger , & kneib 2002 ) of all stars could have formed in objects with star - formation rates greater than @xmath432 ; halos at @xmath313 with the large masses @xmath433 implied by @xmath434@xmath435 mpc contain in total almost @xmath199% as many baryons as the more numerous and smaller halos with @xmath436 that contain lbgs ; objects with stellar masses @xmath437 contain nearly @xmath318% of all stars in the local universe ( kauffmann et al .
2003 ) and @xmath199% of the stars in local elliptical galaxies ( padmanabhan et al .
no treatment will be entirely complete if it neglects galaxies similar to those found in near - ir surveys .
the galaxies we studied are neither the most massive , nor the most rapidly star - forming , nor the most clustered galaxies in the high redshift universe , but it is precisely this that makes them plausible progenitors for the early - type galaxies that surround us today .
we are grateful to the virgo consortium for its public release of the gif simulation data and to g. kauffmann for bringing the data to our attention .
a. coil and t. budavri responded helpfully to our questions about their data .
m. giavalisco , the referee , gave us an insightful report .
kla , aes , and nar were supported by fellowships from the carnegie institute of washington , the miller foundation , and the national science foundation .
dke and ccs were supported by grant ast 03 - 07263 from the national science foundation .
adelberger , k.l . ,
steidel , c.c . ,
giavalisco , m. , dickinson , m. , pettini , m. & kellogg , m. 1998 , apj , 505 , 18 adelberger , k.l .
2000 , in `` clustering at high redshift , '' asp conf series 200 , eds .
mazure , le fvre , le brun , p13 adelberger , k.l . & steidel , c.c .
2000 , apj , 544 , 218 adelberger , k.l . ,
steidel , c.c . ,
shapley , a.e . ,
hunt , m.p . ,
erb , d.k . ,
reddy , n.a . , & pettini , m. 2004 , apj , 607 , 226 adelberger , k.l . ,
erb , d.k . , et al .
2005 , apj , submitted arnouts , s. , moscardini , l. , vanzella , e. , colombi , s. , cristiani , s. , fontana , a. , giallongo , e. , matarrese , s. , & saracco , p. 2002 , mnras , 329 , 355 baldry , i.k .
& glazebrook , k. 2003 , apj , 593 , 258 budavri , t. et al .
2003 , apj , 595 , 59 bullock , j.s . ,
wechsler , r.h .
, & somerville , r.s .
2002 , mnras , 329 , 246 chen , h .- w . ,
marzke , r.o . ,
mccarthy , p.j . , martini , p. , carlberg , r.g . ,
persson , s.e . ,
bunker , a. , bridge , c.r . , & abraham , r.g . , 2003 ,
apj , 586 , 745 coil , a.l .
et al . , 2004 , apj , in press ( astro - ph/0305586 ) cowie , l.l . , barger , a.j . , & kneib , j .-
2002 , aj , 123 , 2197 daddi , e. , rttgering , h.j.a .
, labb , i. , rudnick , g. , franx , m. , moorwood , a.f.m . ,
rix , h.w . ,
van den werf , p.p . , &
van dokkum , p.g .
2003 , apj , 588 , 50 foucaud , s. , mccracken , h.j . ,
le fvre , o. , arnouts , s. , brodwin , m. , lilly , s.j . , crampton , d. , & mellier , y. 2003 , a & a , 409 , 835 franx , m. et al .
2003 , apjl , 587 , 79 giavalisco , m. 2002 , araa , 40 , 579 giavalisco , m. & dickinson , m. 2001 , apj , 550 , 177 hamilton , a.j.s .
1993 , apj , 417 , 19 infante , l. 1994 , a & a , 282 , 353 jenkins , a. , frenk , c.s . , pearce , f.r . ,
thomas , p.a . ,
colberg , j.m . ,
white , s.d.m .
, couchman , h.m.p . , peacock , j.a . ,
efstathiou , g. , & nelson , a.h .
, 1998 , apj , 499 , 20 jing , y.p . & suto , y. 1998 , apjl , 494 , l5 kauffmann , g. , colberg , j.m . ,
diafero , a. , & white , s.d.m . , 1999 ,
mnras , 303 , 188 kauffmann , g. et al .
2003 , mnras , 341 , 33 landy , s.d . &
szalay , a.s
. 1993 , apj , 412 , 64 leitherer , c. 1998 , in `` dwarf galaxies and cosmology , '' eds .
thuan , c. balkowski , v. cayatte , & j.t.t .
van ( gif - sur - yvette : editions frontieres ) , 235 martini , p. & weinberg , d.h .
2001 , apj , 547 , 12 moustakas , l.a . &
somerville , r.s .
2002 , apj , 577 , 1 oke , j.b .
1995 , pasp , 107 , 3750 ouchi , m. et al .
2001 , apj , 558 , 83 padmanabhan , n. et al . , 2004 , new astron , 9 , 329 papovich , c. , dickinson , m. , & ferguson , h.c . 2001 , apj , 559 , 620 peebles , p.j.e . , 1980 , the large - scale structure of the universe , ( princeton , princeton university press ) pettini , m. , rix , s.a . , steidel , c.c . ,
adelberger , k.l . , hunt , m.p . , &
shapley , a.e .
2002 , apj , 569 , 742 porciani , c. & giavalisco , m. 2002 , apj , 565 , 24 press , w. h. , flannery , b. p. , teukolsky , s. a. , & vetterling , w.t .
1992 , `` numerical recipes in c '' , ( cambridge : cambridge university press ) renzini , a. , in `` imf@50 : the initial mass function 50 years later , '' conference proceedings , in press roche , n. & eales , s.a . 1999 ,
mnras , 307 , 703 shapley , a.e . , steidel , c.c . ,
adelberger , k.l . ,
dickinson , m. , giavalisco , m. , & pettini , m. 2001 , apj , 562 , 95 shapley , a.e .
2004 , apj , in press spergel , d.n .
2003 , apjs , 148 , 175 steidel , c.c . ,
adelberger , k.l . ,
shapley , a.e . , pettini , m. , dickinson , m. , & giavalisco , m. 2003 , apj , 592 , 728 steidel , c.c . ,
shapley , a.e . ,
pettini , m. , adelberger , k.l . ,
erb , d.k . ,
reddy , n.a .
, & hunt , m.p .
2004 , apj , 604 , 534 totsuji , h. & kihara , t. 1969 , pasj , 21 , 221 van dokkum , p.g .
2004 , apj , in press ( astro - ph/0404471 ) wechsler , r.h . ,
somerville , r.s . ,
bullock , j.s . ,
kolatt , t.s . ,
primack , j.r . ,
blumenthal , g.r .
, & dekel , a. 2001 , apj , 554 , 85 wilkinson , m.i . &
evans , n.w .
1999 , mnras , 310 , 645 woosley , s.e .
& weaver , t.a .
1995 , apjs , 101 , 181 zaritsky , d. 1999 , in the third stromlo symposium : the galactic halo , eds .
gibson , t.s .
axelrod , & m.e .
putman , asp conf ser 165 , 34 | we analyzed the spatial distribution of @xmath0 photometrically selected galaxies with magnitude @xmath1 and redshift @xmath2 in 21 fields with a total area of @xmath3 square degrees .
the galaxies were divided into three subsamples , with mean redshifts @xmath4 , @xmath5 , @xmath6 , according to the @xmath7 selection criteria of adelberger et al .
( 2004 ) and steidel et al .
( 2003 ) . combining the galaxies measured angular clustering with redshift distributions inferred from @xmath8 spectroscopic redshifts , we find comoving correlation lengths at the three redshifts of @xmath9 , @xmath10 , and @xmath11 mpc , respectively , and infer a roughly constant correlation function slope of @xmath12 .
we derive similar numbers from the @xmath8 object spectroscopic sample itself with a new statistic , @xmath13 , that is insensitive to many possible systematics .
galaxies that are bright in @xmath14 ( @xmath15@xmath16 ) cluster more strongly than fainter galaxies at @xmath17 and @xmath18 but not , apparently , at @xmath19 .
comparison to a numerical simulation that is consistent with recent wmap observations suggests that galaxies in our samples are associated with dark matter halos of mass @xmath20@xmath21 ( @xmath17 ) , @xmath22@xmath23 ( @xmath18 ) , @xmath24@xmath25 ( @xmath19 ) , and that a small fraction of the halos contain more than one galaxy that satisfies our selection criteria . adding recent observations of galaxy clustering at @xmath26 and @xmath27 to the simulation results , we conclude that the typical object in our samples will evolve into an elliptical galaxy by redshift @xmath28 and
will already have an early - type spectrum by redshift @xmath29 .
we comment briefly on the implied relationship between galaxies in our survey and those selected with other techniques . |
classical cepheids have been key objects in the long - lasting efforts to measure the extragalactic distance scale and to probe the predictions of stellar evolution and stellar pulsation theory ( e.g. freedman and madore 2010 , caputo et al . 2005 ) . given the enormous importance of cepheids for the determination of the cosmic distance scale and cosmological parameters , it is of great importance to fully understand these stars astrophysically .
one of the most nagging problems in cepheid research has been the difficulty to reliably determine their masses .
christy ( 1968 ) and stobie ( 1969 ) were the first to notice that cepheid masses calculated from stellar pulsational theory were about 20 % smaller than the corresponding masses estimated from their evolutionary tracks on the hertzsprung - russell diagram . in spite of the considerable progress in understanding the physics of cepheid variable stars over the years
, the `` cepheid mass discrepancy problem '' has remained unsolved for more than 40 years ( keller and wood 2002 , keller 2008 , evans et al .
2008 , caputo et al .
2005 , neilson , cantiello and langer 2011 and references therein ) . the obvious solution to the problem comes from an independent and accurate measurement of the _ dynamical masses _ of a number of cepheids with a range of pulsation periods , and there have been a number of efforts to find binary cepheids allowing such a precise mass determination ( evans et al .
1997 , 2006 and 2008 ) . however , the few cepheids found in binary systems for which masses have been estimated all occur in single - lined , non - eclipsing systems .
this has limited the accuracy of the best cepheid dynamical mass measurements to less than 15 % , not sufficiently accurate to resolve the mass discrepancy problem .
recently , several eclipsing binary systems in the lmc have been discovered from ogle project data which contain candidates for classical cepheid components ( udalski et al .
1999 , soszynski et al .
our group , as part of the _ araucaria project _
dedicated to the improvement of stellar distance indicators ( gieren et al .
2005 ) , has been obtaining spectroscopic observations of these candidates to confirm the true binary nature of these systems and the cepheid nature of the intrinsically variable components .
accurate high resolution spectra collected by our group for ogle - lmc - cep-0227 ( hereinafter cep-0227 ) resulted in the first confirmed discovery of a classical cepheid variable in an eclipsing system , perfectly suited for an accurate determination of its physical parameters ( pietrzynski et al .
the analysis of our data led to the determination of the dynamical mass of this 3.8-day cepheid with an unprecedented accuracy of 1 % .
this study provided strong constraints on theoretical pulsation and evolutionary models and favored the mass predictions for cepheids from pulsation theory .
cassisi and salaris ( 2011 ) have analyzed evolutionary models of this cepheid finding good agreement between its evolutionary and dynamical mass when extra mixing is included .
neilson et al . (
2011 ) find that the mass discrepancy can be accounted for if the evolutionary models include convective mass overshooting and pulsation - driven mass loss .
the models also predict that the size of the cepheid mass discrepancy depends on the cepheid mass , with the size of the mass discrepancy expected to rise to lower periods ( caputo et al .
. on the other hand keller ( 2008 ) showed that the predicted behavior of the mass discrepancy depends on the assumed mass - luminosity relation .
therefore it is clearly desirable to measure accurate dynamical masses for several additional cepheids spanning a range in period in order to fully resolve this problem and to understand the structure , physics and evolution of cepheid variables . in this paper
we present the analysis of the second confirmed classical cepheid in a double - lined eclipsing binary system , ogle - lmc - cep-1812 ( hereinafter cep-1812 ) , and derive its dynamical mass with an accuracy of 1.5 % .
our new binary cepheid has a shorter pulsation period of 1.31 days compared to 3.80 days for cep-0227 , and its measured physical properties therefore present a valuable complement to those derived for cep-0227 .
the optical phtometry of cep-1812 was obtained with the warsaw 1.3 m telescope at las campanas observatory in the course of the third phase of the ogle project ( e.g. soszynski et al .
a total of 883 @xmath0-band epochs spanning a period of 1076 days were secured .
the data were reduced with the image - subtraction technique ( udalski 2003 , wozniak 2000 ) .
the instrumental data were calibrated onto the standard system using observations of several landolt fields over several photometric nights .
the estimated zero point errors are about 0.01 mag ( udalski et al .
2008 ) . for more details about the instrumental system , observing ,
reduction and calibration procedures adopted in the course of the ogle project the reader is referred to the references cited above .
high resolution spectra of the cep-1812 system were collected with the las campanas observatory magellan clay 6.5 m telescope and the mike echelle spectrograph ( bernstein et al .
2003 ) , with the eso vlt kueyen 8.2 m telescope and the uves echelle spectrograph ( dekker et al .
2000 ) , and with eso 3.6 m telescope and the harps fiber - fed echelle spectrograph ( mayor et al .
2003 ) . in the case of the mike observations
a 0.7 arcsec slit was used giving a resolution of about 40,000 .
the spectra were reduced with pipeline software developed by kelson ( 2003 ) .
exposure times ranged from 2400 sec to 3200 sec depending on observing conditions , and a typical resulting s / n ratio was 10 at a wavelength of 4500 .
the uves observations were obtained with an exposure time of 2700 sec and a 0.7 arcsec slit resulting in a spectral resolution of about 50,000 and a s / n ratio at 4500 of better than 15 .
the uves data were reduced with the eso pipeline .
the harps observations were obtained at a resolution of 60,000 and a s / n at 5000 of 4 for one hour integrations , and were reduced with the data reduction software deweloped by the geneva observatory . in total 88 high quality spectra were collected with the three spectrographs .
cep-1812 is located in a dense region in the lmc and even based on inspection of the ogle template images with a spatial resolution of 0.8 arcsec ( soszynski et al .
2008 ) one can see several faint stars located very close to the star .
simple cross correlation measurements showed the presence of a third velocity component .
we therefore decided to measure velocities for cep-1812 with the broadening function ( bf ) formalism ( rucinski 1992 ) , which is widely recognized as an effective technique for radial velocity determination from complex spectra ( rucinski 2003 ) .
the bfs were calculated on the wavelength interval 4350 to 6100 using a template interpolated from the coelho library of synthesized echelle resolution spectra ( coelho et al .
the bfs revealed clear peaks for both components of cep-1812 in all spectra .
in addition , the third signal was detected in some 40 % of our spectra at a constant radial velocity of 273 @xmath1 5 km / s .
the bf signal of this star was used to calculate its light contribution to the total i - band brightess of the system to be about 10 % .
we found no significant offsets between the velocity systems of the three instruments .
based on the photometric data the following ephemerides were derived : @xmath2 days @xmath3 + @xmath4 days @xmath5 + adopting these ephemerides the spectroscopic orbit ( systemic velocity , velocity amplitudes , eccentricity , periastron passage and mass ratio ) plus a fourier series of order six ( which approximates the pulsations of the cepheid primary component ) were fitted to the radial velocity data .
the orbit solution and the pulsational radial velocity curves of the cepheid component are shown in figure 1 .
figure 2 shows the pulsational @xmath0-band light curve of the cepheid . adopting the obtained spectroscopic mass ratio of 0.705 @xmath1 0.015 , and an i - band third light contribution of 10 %
, we modeled the ogle - lmc - cep-1812 system with the wilson devinney code ( wilson and devinney 1971 , van hamme and wilson , 2007 ) in an iterative way removing the intrinsic brightness variations of the cepheid component .
the parameters corresponding to our best model together with their uncertainties estimated from extensive monte carlo simulations are presented in table 1 .
unfortunately , at the present moment without near infrared photometry , we are not in a position to derive @xmath6 of the components .
figure 3 presents the light curve of the eclipsing system with the intrinsic brightness variations of the cepheid removed , together with our best fit wd model for the system .
although the relative distance between components is very large the eclipses are total with the secondary transiting over the cepheid disk during the primary eclipse .
it is worth mentioning that the presence of the third component only marginally affects the mass determination . neglecting the third light leads to an inclination of 89.5 degrees .
the cepheid in ogle - lmc - cep-1812 is a classical cepheid , and not a type ii low - mass cepheid .
the evidence comes from its mass of 3.74 @xmath1 0.06 @xmath7 which agrees well with the predicted pulsational mass for a classical cepheid of this short period ( bono et al .
2001 ) , and from its radius ( 17.4 @xmath1 0.9 @xmath8 ) which again is in good agreement with the radius of a classical cepheid predicted from period - radius relations ( e.g. gieren et al .
the position of the cepheid in the period - mean magnitude plane for lmc cepheids shown in figure 4 also proves beyond any doubt that cep-1812 is a classical cepheid .
based on the analysis of the fourier parameters soszynski et al .
( 2008 ) firmly classified this star as a fundamental mode pulsator .
the system ogle - lmc - cep-1812 is thus the second known double - lined eclipsing binary system with a classical fundamental mode cepheid component .
the secondary component is a less massive , smaller and cooler stable giant star , a configuration which is significantly different from the ogle - lmc - cep-0227 system whose stable secondary giant star has the same mass , and a slightly larger diameter than the cepheid in that system .
the configuration is also quite unusual from an evolutionary point of view because the system consists of two well separated stars in a relatively short stage of common giant phase evolution in spite of the large mass difference .
according to the basti evolutionary tracks ( pietrinferni et al .
2004 ) a star with mass 3.74 @xmath9 will enter the instability strip from the giant branch after approximately 190 myr of evolution , whereas a star of mass 2.64 @xmath9 will reach a radius of 12.1 @xmath10 on the subgiant branch after approximately 369 myr of evolution . while detailed evolutionary models of the two stars in the binary will be needed
, these estimates call into question the assumption that the stars are members of a coeval binary system . in order to calculate the pulsational mass of cep-1812 we adopted a period - mass - relation based on nonlinear , convective cepheid models constructed for the typical chemical composition of lmc cepheids ( z(metals)=0.008 , y(helium)=0.256 ) ( bono et al
. 2000 , luck et al . 1998 ) .
we note that the calculation of the pulsation mass depends neither on the assumed distance nor on the reddening .
the resulting pulsation mass ( 3.27 @xmath1 0.64 @xmath7 ) agrees very well with the dynamical mass of the star .
pietrzynski et al .
( 2010 ) came to the same conclusion for cep-227 .
therefore , we have now strong observational evidence that the pulsation mass of a cepheid variable is indeed correctly measuring its true , current mass .
the dynamical mass determinations for both stars in ogle - lmc - cep-1812 are accurate to about 1.5% , adding them to the very short list of evolved massive stars with very accurate mass determinations . in a forthcoming study
, we will discuss evolutionary models for the cepheid in the system which will further add to the solution of the cepheid mass discrepancy problem , and to a deeper understanding of the physics and evolution of classical cepheid variable stars .
we gratefully acknowledge financial support for this work from the chilean center for astrophysics fondap 15010003 , and from the basal centro de astrofisica y tecnologias afines ( cata ) pfb-06/2007 .
support from the polish grant n203 387337 , and the focus and team subsidies of the fundation for polish science ( fnp ) is also acknowledged .
ibt acknowledges the support of nsf grant ast-0507325 .
the ogle project has received funding from the european research council under the european community s seventh framework programme ( fp7/2007 - 2013 ) / erc grant agreement no .
it is a pleasure to thank the support staff at eso - paranal and at las campanas observatory for their expert help in obtaining the observations .
we also thank the eso opc and cntac for allotting generous amounts of observing time to this project .
bernstein , r. , shectman , s. a. , gunnels , s. m. , mochnacki , s. & athey , a. 2003 , spie , 4841 , 1694 bono , g. , castellani , v. , & marconi , m. 2000 , , 529 , 293 bono , g. , gieren , w. , marconi , m. , fouqu , p. , & caputo , f. 2001 , , 563 , 319 caputo , f. , bono , g. , fiorentino , g. , marconi , m. , & musella , i. 2005 , , 629 , 1021 cassisi , s. , & salaris , m. 2011 , , 728 , l43 christy , r.f .
1968 , qjras , 9 , 13 coelho , p. , barbuy , melendez , j. , sciavon , r.p . , & castilho , b.v .
2005 , , 443 , 735 decker , h. et al .
2000 , spie , 4008 , 534 evans n.r . , bohm - vitense , e. , carpenter , k. , beck - winchaz , b. , & robinson , r. 1997 , , 109 , 789 evans , n.r . ,
massa , d. , fullerton , a. , sonneborn , g. , & iping , r. 2006 , , 647 , 1387 evans , n.r . ,
schaefer , g.h . , bond , h.e . ,
bono , g. , karovska , m. , nelan , e. , sasselov , d. , & mason , b.d .
2008 , , 136 , 1137 freedman , w.l . , & madore , b.f .
2010 , araa , 48 , 673 gieren , w. , fouqu , p. , & gomez , m. 1998 , , 496 , 17 gieren , w. , pietrzyski , g. , bresolin , f. , et al .
2005 , eso messenger , 121 , 23 keller , s.c .
2008 , , 677 , 483 keller , s.c . , & wood , p.r .
2002 , , 578 , 144 kelson , d. d. 2003 , , 115 , 688 luck , r.e . ,
moffett , t.j .
, barnes , t.g . , & gieren , w. 1998 , , 115 , 605 mayor , m. et al .
2003 , the messenger , 114 , 20 neilson , h.r . , cantiello , m. , & langer , n. 2011 , , 529,l9 pietrinferni , , a. , cassisi , s. , salaris , m. , & castelli , f. 2004 , , 612 , 168 pietrzyski , g. , thompson , i.b.,graczyk , d. , gieren , w. , udalski , a. , szewczyk , o. , minniti , d. , kolaczkowski , z. , bresolin , f. , & kudritzki , r.p .
2009 , , 697 , 862 pietrzyski , g. , thompson , i.b . , gieren , w. , graczyk , d. , bono , g. , udalski , a. , soszyski , i. , minniti , d. , & pilecki , b. 2010 , nature , 468 , 542 rucinski , s.m .
1992 , , 104 , 1968 rucinski , s.m .
2003 , proceedings of iau symposium no .
215 , held 11 - 15 november , 2002 in cancun , yucatan , mexico , 2004 , p. 17
soszyski , i. , poleski , r. , udalski , a. , et al .
2008 , acta astron . , 58 , 163 stobie , r.s .
1969 , , 144 , 511 udalski , a. , szymanski , m. , kubiak , m. , pietrzynski , g. , soszynski , i. , wozniak , p. , & zebrun , k. 1999 , acta astron . , 49 , 201 udalski , a. , soszynski , i. , szymanski , m. , kubiak , m. , pietrzynski , g. , wyrzykowski , . ,
szewczyk , o. , ulaczyk , k. , & poleski , r. 2008 , acta astron . , 58 , 4632 udalski , a. 2003 , acta astron . , 53 , 291 van hamme , w. , & wilson , r.e .
2007 , , 661 , 1129 wilson , r.e . , & devinney , e.j . 1971 , , 166 , 606 woniak , p. 2000 , acta astron . , 50 , 421 cc p ( orbital ) [ days ] & 551.797 @xmath1 0.010 + p ( pulsational ) [ days ] & 1.312904 @xmath1 0.000003 + i [ deg ] & 90.0 @xmath1 0.4 + a [ @xmath8 ] & 524.5
@xmath1 1.1 + e & 0.129 @xmath1 0.012 + @xmath12 [ deg ] & 144.9 @xmath1 4.7 + @xmath13 & 0.705 @xmath14
0.015 + @xmath15 [ km / s ] & 253.6 @xmath16 + @xmath9 ( cepheid ) & 3.74 @xmath1 0.06 + @xmath9 ( secondary component ) & 2.64 @xmath1 0.04 + @xmath10 ( cepheid ) & 17.4 @xmath1 0.9 + @xmath10 ( secondary component ) & 12.1 @xmath1 2.3 + l2/l1 ( i ) & 0.28 @xmath1 0.10 + l3 ( i ) & 0.1 * ( l1+l2 ) + | we have analyzed the double - lined eclipsing binary system ogle - lmc - cep-1812 in the lmc and demonstrate that it contains a classical fundamental mode cepheid pulsating with a period of 1.31 days .
the secondary star is a stable giant .
we derive the dynamical masses for both stars with an accuracy of 1.5% , making the cepheid in this system the second classical cepheid with a very accurate dynamical mass determination , following the ogle - lmc - cep-0227 system studied by pietrzynski et al .
( 2010 )
. the measured dynamical mass agrees very well with that predicted by pulsation models .
we also derive the radii of both components and accurate orbital parameters for the binary system .
this new , very accurate dynamical mass for a classical cepheid will greatly contribute to the solution of the cepheid mass discrepancy problem , and to our understanding of the structure and evolution of classical cepheids . |
analysis of the interaction between the electric field @xmath1 applied perpendicularly to the plane @xmath12 and the robin boundary condition ( bc ) at it @xcite @xmath13 has shown that the spectrum of the one - dimensional quantum particle of the mass @xmath14 moving on the half - line @xmath15 and satisfying the schrdinger equation for the wave function @xmath16 and energy @xmath17 , @xmath18 , @xmath19 for the finite nonzero extrapolation length @xmath0 is determined from the transcendental equation @xcite @xmath20^{1/3}\!\!e_n\right)-\frac{1}{\lambda}\,{\rm ai}\!\left(-\left[\frac{2m}{(e\hbar\mathscr{e})^2}\right]^{1/3}\!\!e_n\right)=0.\ ] ] here , @xmath21 is an inward unit normal to the interface @xmath12 , @xmath22 is an electrostatic potential @xmath23 creating the field @xmath24 that pushes the particle ( for definiteness , we will talk about the electron with its negative charge @xmath25 ) closer to the surface , @xmath26 is airy function @xcite and prime denotes its derivative with respect to argument @xmath27 . for @xmath28 ( @xmath29 ) the bc from eq .
simplifies to the dirichlet ( neumann ) requirement @xmath30 [ @xmath31 with the energy spectrum directly following from eq . :
@xmath32 where the negative @xmath33 and @xmath34 are @xmath35th roots , @xmath36 , of the equations @xmath37 or @xmath38 , respectively @xcite .
for the small nonzero parameter @xmath39 the taylor expansion of eq .
shows that the first - order correction to the dirichlet spectrum is equal to the voltage drop across the robin distance : [ asymptoticenergies1 ] @xmath40,\nonumber\\ \label{asymptoticenergies1_dirichlet0 } & \frac{2me\mathscr{e}|\lambda|^3}{\hbar^2}\ll1.\\ \intertext{in the opposite limit the energies are : } \label{asymptoticenergies1_neumann } & e_n^{r\mp}=-a_{n+1}'\left(\frac{e^2\hbar^2\mathscr{e}^2}{2m}\right)^{1/3}\left[1\mp\frac{1}{{a_{n+1}'}^{\!\!\!\!\!\!\!\!2}\,\,\,\,}\left(\frac{\hbar^2}{2me\mathscr{e}|\lambda|^3}\right)^{1/3}\right],\nonumber\\ & n=0,1,2,\ldots,\quad\frac{\hbar^2}{2me\mathscr{e}|\lambda|^3}\ll1.\end{aligned}\ ] ] eq .
manifests that the energy admixture to the slightly distorted neumann wall is of the order @xmath41 .
hence , the varying electric field changes the energy spectrum in a very wide range .
this is especially clearly seen after introducing the dimensionless scaling where all distances are measured in units of the magnitude of the robin length @xmath42 , energies in units of @xmath43 , and electric fields in units of @xmath44 . then , the spectrum of the attractive robin wall at the low voltages consists of the densely populated positive part [ asymptoticenergies2 ] @xmath45 this highly nonuniform distribution of the levels on the energy axis leads , as shown below , to the spectacular thermodynamic properties , such as , for example , the gigantic growth of the heat capacity @xmath46 , which are calculated on the basis of the knowledge of the spectrum @xmath17 .
similar to the tilted quantum well ( qw ) with miscellaneous combinations of the dirichlet and neumann bcs at the opposite walls @xcite , the analysis below addresses both canonical , sec .
[ sec_canonical ] , and two types of the grand canonical ensembles , sec .
[ sec_grandcanonical ] .
previous research on the bc influence on the thermodynamics of the low - dimensional nanostructures concentrated on the calculations of the bosonic systems only and was limited to the flat qws either with periodic bcs or at @xmath28 and/or @xmath29 relevant for the study of liquid helium in pore geometries and thin film superconductors @xcite . in a recent contribution ,
the author has shown @xcite that for the canonical distribution the permutation of the dirichlet and neumann bcs of a qw at @xmath47 has a significant impact on its thermal properties ; for example , the heat capacity of the pure neumann configuration exhibits on the temperature axis a conspicuous maximum accompanied by the adjacent minimum that are absent for any other bc geometry .
its physical explanation invoked an analysis of the corresponding structures of the associated spectra what mathematically reduced to correct handling of theta functions .
a remarkable feature of the fermionic qw is , at any bc combination in zero fields , a salient maximum of the specific heat for one particle and its absence for any other number of corpuscles .
it was shown that the voltage applied to the qw leads to the increase of the heat capacity and formation of the new or modification of the existing extrema what is qualitatively described by the influence of the associated electric potential .
the main message of the present research is to show on the example of one wall that the impact of the field for either type of the statistical ensemble has much more drastic consequences for the robin bc , especially for that with @xmath2 .
below , in addition to the dimensionless units introduced above , we will measure the heat capacity , if not specified otherwise , in units of the boltzmann constant @xmath48 , and particle kinetic momentum @xmath49 in units of @xmath50 .
the basic quantity in the description of the system that is in the thermal equilibrium with the much larger bath is the partition function defined as @xmath51 with the summation running over all possible quantum states , the parameter @xmath52 being ( in regular , unnormalized units ) @xmath53 , and @xmath3 is the thermodynamic temperature of the bath .
for example , the probability @xmath54 of finding particle in the state @xmath35 is a function of the temperature and the energy @xmath55 , which , in turn , depends on the parameter @xmath1 , as discussed in the introduction : @xmath56 and the mean value @xmath57 of any physical quantity @xmath58 is calculated as @xmath59 the mean energy [ canonicalmeanenergy1 ] @xmath60 while the heat capacity @xmath46 , i.e. , a work that has to be done to change the temperature of the system by one degree [ heatcapacity1 ] @xmath61 for @xmath8 noninteracting corpuscles in the system , the right - hand sides of eqs .
have to be multiplied by @xmath8 .
zero - field canonical mean energy @xmath62 ( solid line , left axis ) and heat capacity @xmath46 ( dashed curve , right axis ) of the attractive robin wall as a function of temperature @xmath63 .
inset shows the peculiarity of the heat capacity at the large temperatures . ] before discussing the electric field influence on the thermodynamics of the system , it does make sense to analyze first a zero - voltage case . for the nonnegative extrapolation lengths characterized by the positive - energy continuous spectrum only the summation in the partition function , eq
, is replaced by the integration over the momentum @xmath49 : @xmath64 what leads to the same expressions of the mean energy and temperature independent specific heat as for the free space @xcite : @xmath65 situation is different for the attractive robin wall .
the existence of the bound state modifies the partition function as @xmath66 what results in the following expression for the mean energy : [ canonicalmeanenergyzerofieldrminus1 ] @xmath67 allow with the help of eq . to calculate the specific heat @xmath46 in the same limits as : [ heatcapacityzerofielddnrminus1 ] @xmath68 the variation of the mean energy and specific heat with the temperature @xmath63
is shown in fig .
[ heatcapacitycanonicalnofieldsfig1 ] .
it is seen that the initial warming of the extremely cold wall has an exponentially negligible effect on both parameters , in accordance with eqs . and .
this is explained by the smallness of the thermal energy @xmath69 with respect to the gap between the bound state and the positive energy continuum .
subsequent growth of the temperature causes a monotonic increase of the mean energy that at @xmath70 approaches asymptotically the value @xmath69 of the free particle , as eq
. also demonstrates .
the avalanche rise of the heat capacity starts at the smaller temperatures as compared to that of the mean energy .
the specific heat after the steep surge reaches at @xmath71 ( @xmath72 ) the maximum of @xmath73 .
its following decrease terminates at @xmath74 ( @xmath75 ) where the minimum @xmath76 is located .
the higher temperatures lead to the smooth nearing of the heat capacity to the free - electron configuration of @xmath77 .
observe that these extrema and their positions can be with the decent accuracy extracted from the asymptotic expansions from eqs . ; for example , a zeroing of the @xmath52-derivative of eq . yields
the approximate location of the minimum @xmath78 at @xmath79 which coincide quite well with the values provided above .
universal curves of canonical ( a ) ratio @xmath80 and ( b ) heat capacity @xmath46 versus the parameter @xmath81 for the dirichlet ( dashed lines ) and neumann ( solid lines ) walls . ] thus , the formation by the attractive robin wall of the split - off negative - energy bound state strongly modifies its thermodynamic properties in comparison with the interfaces with the nonnegative @xmath0 ; for example , it leads to the nonmonotonic dependence of the heat capacity on the temperature , as just shown above .
the difference is greatly enhanced upon application of the electric field .
first , let us note that the classical `` potential '' partition function @xcite @xmath82 for the configuration from eq
. becomes : @xmath83 while a thermally averaged value of the potential energy @xmath84 for the robin wall in the electric field is just the heat quantum : @xmath85 then , the potential contribution to the heat capacity @xmath86 turns to unity what means that the total specific heat at the high temperatures when the classical expressions from eqs .
are only applicable to the quantum case is equal to @xmath87 .
expressions for the dirichlet and neumann mean energies @xmath88 with the associated partition functions @xmath89 and coefficients @xmath90 show that the corresponding heat capacities @xmath91 and the values @xmath80 are functions of one variable @xmath92 only . in particular ,
for the large values of this parameter , @xmath93 , one has : [ asymptotecanonicaldnlargey1 ] @xmath94.\end{aligned}\ ] ] to find these functions performance in the opposite limit of @xmath95 , one employs the behavior of the airy zeros coefficients at the large index @xcite @xmath96^{2/3},\quad n\gg1,\ ] ] which produce a reasonably good accuracy even for @xmath97 @xcite , and the asymptotic expansion of the infinite series [ asymptoticseries1 ] @xmath98 with @xmath99 being @xmath100-function @xcite . as a result , one obtains the following expressions for the heat capacities at @xmath95 : @xmath101 observe that the neumann specific heat approaches from above the infinite temperature limit of @xmath87 what means that it has a maximum on the @xmath102 axis .
this qualitative conclusion is confirmed by the exact calculations shown in fig .
[ canonicalenergyheatcapdirneufig1 ] , which exhibits the universal dependencies of the heat capacity and the ratio @xmath80 on the variable @xmath102 for the dirichlet and neumann surfaces .
the specific heat @xmath103 reaches a shallow maximum of @xmath104 at @xmath105 .
it is seen that at the large @xmath102 the neumann heat capacity fades faster since the corresponding difference @xmath106 in eq . is greater than its dirichlet counterpart @xmath107 @xcite .
the dirichlet mean energy lies above the neumann one since , for example , in the same limit of @xmath93 they are mainly determined , as eq .
demonstrates , by the first airy zeros @xmath108 and @xmath109 , respectively .
canonical specific heat @xmath110 of the attractive robin wall in terms of the temperature @xmath63 and electric field @xmath1 . ] the most dramatic feature of the heat capacity of the negative robin wall shown in fig .
[ heatcapacitycanonicalfig1 ] is its colossal increase at the weak fields and low voltage - dependent temperatures . to explain this phenomenon from a mathematical point of view
, it suffices to keep in eqs . and of the corresponding energy spectrum the leading terms only .
next , assuming that , in addition to the condition @xmath111 , the requirement @xmath112 also holds ( its validity for our range of interest will be justified below ) , one invokes the expression for the coefficients @xmath33 from eq . and the expansions from eqs . .
as a result , the mean energy in these limits reads : @xmath113 first , for the large temperatures it simplifies to @xmath114 note that the last expression remains intact if the terms linear in the field are retained in eqs . and .
this voltage - dependent nearing from above to the high - temperature limit is clearly seen in fig .
[ heatcapacitycanonicalfig1 ] .
next , the mean energy turns to zero at the temperature @xmath115 , which is found from equation : @xmath116 its solution is expressed via the lambert @xmath117 function @xcite @xmath118 asymptotics of the lambert function @xcite @xmath119 shows that at the very weak ( but nonzero ) fields the temperature at which the zero mean energy is achieved , is inversely proportional to the logarithm of the applied voltage : @xmath120 eq .
manifests that the mean energy for the decreasing @xmath1 becomes a steeper function of the low temperature .
this is exemplified in panel ( a ) of fig .
[ energyheatcapacitynegativelambdacanonicalsmallfieldsfig1 ] that shows the thermally averaged energies and heat capacities for several ultra weak electric intensities as functions of the temperature . rapid growth of @xmath121 means , according to eq .
, larger values of the heat capacity whose expression in the same limit reads @xmath122 its asymptotics for the large temperatures , which approaches , as expected , the value of @xmath87 , was derived above , eq . .
taking a derivative @xmath123 of eq . , equating it to zero and keeping only the largest power of @xmath52 leads to the following expression for the temperature @xmath124 of the maximum of the specific heat : @xmath125 invoking the asymptotics from eq .
produces : @xmath126 what essentially is the same dependence ( with the different proportionality coefficient ) as that from eq . .
note that in this way our starting assumption of @xmath127 is automatically satisfied .
corresponding peak of the heat capacity @xmath128 is calculated as @xmath129 eq . and show that the maximum of the specific heat is proportional basically to @xmath4 .
this growth of @xmath130 and its shift to the lower temperatures are seen in figs .
[ heatcapacitycanonicalfig1 ] and [ energyheatcapacitynegativelambdacanonicalsmallfieldsfig1](b ) .
they also prove that the half width of this resonance narrows at the smaller electric intensities .
canonical ( a ) mean energy @xmath121 and ( b ) specific heat @xmath110 of the attractive robin wall as functions of the low temperature @xmath63 at several extremely weak electric fields where solid lines are for @xmath131 , dotted curves for @xmath132 , dashed lines for @xmath133 , and dash - dotted curves are for @xmath134 . ] a physical reason of this giant enhancement of the heat capacity at the weak fields and extremely low voltage - dependent temperature stems from the peculiarities of the energy spectrum and its modification by the applied electric force . as discussed in the introduction , at the weak intensities @xmath1 the lowest lying level is split off from the quasi continuum of the positive - energy states .
the difference @xmath135 between the first excited , @xmath97 , and ground state is in this regime practically the same as for many other higher lying levels with the ratio @xmath136 turning to unity for the vanishing fields : @xmath137 accordingly , the very low temperatures can promote the electron from its ground position not only to the first excited level but with about the same probability to a huge number of its counterparts what results in a steep increase of the heat capacity with its slope being more vertical for the weaker field .
further tiny warming of the structure can not sustain this very fast growth since the latter has to involve the levels with such enormously large indices @xmath35 for which the ratio @xmath138 considerably deviates from the unity what makes the transitions to them more difficult at this temperature .
consequently , the specific heat drops forming a characteristic sharp resonance with its peak increasing for the decreasing @xmath1 .
the higher applied voltage splits wider the positive energy levels with the corresponding change of the ratio @xmath139 ; for example , in the opposite limit of the large electric intensities it becomes a field independent quantity that noticeably grows with the quantum number @xmath35 : @xmath140 as it follows from eq . .
for that reason , in this case the low temperatures propel the electron mainly to the first excited state with the transitions to the higher levels being strongly suppressed due to their large energy separation from the one with @xmath97 .
they gradually can be populated one by one at the increasing temperatures .
this change of the energy spectrum by the electric field smooths out the heat capacity and eventually at the growing @xmath1 the resonance is transformed into the shallow neumann maximum discussed above .
as fig . [ heatcapacitycanonicalfig1 ] demonstrates , the peak is strongly subdued already at @xmath141 . to underline the crucial role of the zero - field negative - energy bound state in the formation of the specific heat resonance , we provide in fig .
[ heatcapacitycanonicalpositivelambdafig1 ] the mean energy and the heat capacity of the repulsive robin wall , which at any field is characterized by the positive spectrum only ; for example , at the weak electric intensity the corresponding ratio @xmath142 @xmath143 similar to the high - field limit from eq . , is a salient function of the index @xmath35 . as a result ,
the heat capacity in the whole range of the electric forces is a monotonic function of the temperature , as panel ( b ) of fig .
[ heatcapacitycanonicalpositivelambdafig1 ] demonstrates .
for the weaker voltages the high - temperature limit @xmath87 of the specific heat is achieved at the lower temperatures since the lesser energy differences @xmath144 permit to occupy all the levels at the smaller @xmath63 . however , since the ratio from eq .
is strongly @xmath35 dependent , the heat capacity remains at these fields a monotonic function of the temperature too .
canonical ( a ) mean energy @xmath145 and ( b ) specific heat @xmath146 of the repulsive robin wall as functions of the temperature @xmath63 and electric field @xmath1 . ]
the quantum system that , in addition of being in the thermal balance with the external reservoir , can exchange the particles as well as energy with it , is said to be in the chemical equilibrium with the heat bath .
since the number @xmath8 of the constituent corpuscles varies , the properties of the structure strongly depend on it ; for example , the chemical potential @xmath147 that is defined as the work that has to be done to change the number of the particles by one @xmath148 is determined from @xmath149 where the upper sign corresponds to the fermi - dirac ( fd ) distribution while the lower one describes bose - einstein ( be ) systems .
knowledge of the mean energy @xmath150 allows one with the help of eq . to calculate the associated heat capacity @xmath151 as @xmath152 ^ 2}\,e^{(e_n-\mu)\beta},\ ] ] where the chemical potential , which for the former ensemble is also frequently called the fermi energy ,
is calculated from eq .
, as just stated above , and its partial derivative with respect to the inverse temperature is found from the same expression with the help of the rule of differentiating of the implicit functions and is given as @xcite @xmath153 ^ 2}\,e^{(e_n-\mu)\beta}}{\sum_n\frac{1}{\left[e^{(e_n-\mu)\beta}\pm1\right]^2}\,e^{(e_n-\mu)\beta } } .\ ] ] mathematical difference of the opposite signs in eq . has its physical consequence in the fact that each quantum level is occupied by at most one fermion while the arbitrary number of bosons can coexist in the same state .
the fd distribution is applied for the system of the particles with the half - integer spin such as , for example , the electron in the free space or in the crystal lattice , while the boson statistics is used for the constituents with the integer spin , e.g. , photons or cooper pairs in superconductors .
thus , our remark from the introduction that we consider the electron does not apply for the be consideration where we will assume the abstract negatively charged particle . also , for the half - integer spin fermions , say , the electron with its spin of @xmath77 , the orbital states with the same energy can be occupied by the particles with the opposite projections of the spin @xmath154 .
however , this fact is neglected in the discussion below where it is assumed that the number of the fermions for each energy @xmath17 is not larger than one .
fd chemical potential @xmath155 of the attractive robin wall as a function of the temperature @xmath63 for several numbers @xmath8 of fermions shown in each panel and different weak electric fields where solid lines are for @xmath156 , dashed curves for @xmath157 , dotted lines for @xmath131 , and dash - dotted ones are for @xmath132 .
note different upper limits of each vertical axis . ] for the dirichlet and neumann wall , utilizing asymptotic expansions of the series [ asymptoticseries2 ] @xmath158 with @xmath159 being a polylogarithm @xcite @xmath160 one transforms eqs . and to @xmath161 valid at @xmath127 .
here , @xmath162 . assuming that the chemical potential is large and negative , the following ensemble independent expressions are obtained : @xmath163 observe that the specific heat per particle @xmath164 derived from eq .
exactly coincides with the one from eq . for the canonical distribution
thus , the well - known general result @xcite of the independence at the high temperatures of the thermodynamic properties on the ensemble is confirmed again @xcite . for the attractive robin wall at the weak fields , eq
. turns to @xmath165 where also a smallness of the product @xmath81 has been assumed . besides , considering the coefficient @xmath166 as a tiny parameter too , one gets : @xmath167 this equation has a logarithmic solution for the chemical potential @xmath147 but its explicit form is too unwieldy ; therefore , to simplify our qualitative analysis , we disregard the last right - hand side term to obtain : [ asymptotefdchempoten5 ] @xmath168 \end{array}\right .. \end{aligned}\ ] ] the corresponding mean energy under the same assumptions reads : @xmath169 the associated heat capacity is calculated from this expression with the help of eq . .
heat capacity per particle @xmath170 of the attractive robin wall as a function of the temperature @xmath63 for different weak electric fields shown in each panel and several numbers @xmath8 of fermions where solid lines are for @xmath9 , dashed curves for @xmath171 , dotted lines for @xmath172 , dash - dotted ones for @xmath173 , and dash - dot - dotted curves are for @xmath174 .
insets of each panel show enlarged views at the small temperatures where the formation of the plateaus from eq . is demonstrated .
] consider first the case of the fd distribution .
for one fermion , @xmath9 , at the attractive robin wall its specific heat reaches the maximum of @xmath175 at the inverse temperature @xmath176 that is a solution of equation @xmath177 these expressions show that , similar to the canonical ensemble , the heat capacity unrestrictedly grows as @xmath4 with the vanishing field .
[ fermifdnegativelambdafig1 ] depicts the fermi energies of the attractive surface as functions of the temperature for several numbers @xmath8 and weak electric intensities @xmath1 .
it was shown before @xcite that for the structures for which the differences between the several adjacent levels are of the same order , in particular , @xmath178 , the fermionic chemical potential @xmath179 after its rapid increase from @xmath180 at @xmath181 exhibits a temperature - independent plateau @xmath182 whose width on the @xmath3 axis is a function of the differences @xmath183 , @xmath184 .
it is a consequence of the interaction at the small temperatures between these two neighboring levels only , which , at @xmath181 were the highest occupied and lowest unoccupied orbitals . reminiscences of these flat parts can be seen in the figure for the stronger voltages and/or larger number of fermions . the plateaus are completely absent at the extremely weak fields and @xmath9 as the upper left panel of fig . [ fermifdnegativelambdafig1 ] demonstrates . here
, the split - off level with the growth of the temperature interacts not only with the lowest unoccupied at @xmath181 state but , in addition to this , with the huge number of its adjacent counterparts . as a result
, the fermi energy has a sharp peak @xmath185 after which the occupation of the positive - energy levels pushes it downward with the steepness of the curve increasing for the smaller @xmath1 . at the large temperatures ,
it is described by the dirichlet dependence from eq . .
corresponding heat capacities are shown in fig .
[ heatcapacityfdnegativelambdafig1 ] . as discussed above , a specific heat of the attractive robin interface possesses at the weak fields a conspicuous maximum whose magnitude grows with the decreasing electric intensity @xmath1 while its location shifts to the lower @xmath63 .
a physical explanation of this phenomenon is the same as for the canonical ensemble discussed in the previous section .
quantitatively , at the fixed voltage the extremum for one fermion is smaller and located at the higher temperatures , as compared to the canonical distribution , with some data provided in table [ table1 ] .
the increasing number of the electrons in the system subdues the peak and moves it to the warmer temperatures since the higher lying at @xmath181 particles impede the interaction of their lowest counterpart with the levels from the positive part of the spectrum .
mathematically , this can be shown from the asymptotic expansion of the solution of eq .
, which reads for the large @xmath8 : @xmath186 keeping the first two largest terms only leads to the following expression for the heat capacity per particle : @xmath187 for the case of the fermions , this equation shows that the specific heat rapidly flattens for the larger number of the electrons in the system .
this is exemplified in fig .
[ heatcapacityfdnegativelambdafig1 ] and table [ table1 ] .
another feature worth mentioning is the fact that the maximum is preceded on the temperature axis by the flat plateau with its value equal to @xmath188 it is explained by the statistical interplay between the fermions ; namely , at the zero temperature @xmath189 particles occupy closely located on the @xmath190 axis positive - energy states , eq . ,
and the remaining one has the negative energy from eq . .
as the density of states at @xmath191 is very large , each of @xmath189 electrons attains the classical heat capacity of @xmath87 at the very small temperature that decreases logarithmically with the field .
further warming of the system does not cause any appreciable change of the specific heat as the lowest level is separated from its positive counterparts by the unit energy gap .
only when the thermal quantum is strong enough to promote the fermion to the positive energies , it does start to contribute to the specific heat .
similar to the canonical distribution , for the smaller electric intensities there are more closely packed positive - energy states in the same energy interval ; accordingly , the peak of the heat capacity gets higher for the weaker fields .
the formation of the plateaus is clearly seen in the insets of fig .
[ heatcapacityfdnegativelambdafig1 ] .
the widths of these temperature - independent flat parts shrink at the lower voltages .
[ table1 ] bosonic heat capacity per particle @xmath170 of the attractive robin wall as a function of the normalized temperature @xmath192 for ( a ) @xmath157 , ( b ) @xmath156 , ( c ) @xmath193 and ( d ) @xmath194 and several numbers @xmath8 of bosons where thick solid lines are for @xmath9 , dashed curves for @xmath174 , dotted lines for @xmath195 , dash - dotted curves for @xmath196 , dash - dot - dotted ones are for @xmath197 , and thin solid lines for @xmath198 .
note different upper limits for the lower and upper panels .
] a remarkable property of the collective behavior of this type of particles is a formation of the condensate when the overwhelming majority of bosons reside in the ground level .
impressive experimental discovery @xcite of this state of matter predicted in 1925 @xcite ( the history of its theoretical derivation and subsequent development is described in , e.g. , ref .
@xcite ) reignited a huge interest in the topic @xcite .
relevant to our research , let us mention a theoretical analysis of the influence of the different forms of the confining potential @xcite and/or bc @xcite on the properties of the ideal bose gas . a quantitative measure of the be condensation is provided by the ground state occupation @xmath199 , i.e.
, a ratio of the number of bosons @xmath200 in the ground state @xmath201 to the total number of corpuscles in the system : @xmath202 as this fraction is a function of @xmath63 , we , following ketterle and van druten @xcite , define and will use the critical temperature @xmath203 as the largest temperature at which the be condensation can still be observed what describes the configuration with the chemical potential equal to the ground - state energy , @xmath204 , and with no particles dwelling in the lowest level , @xmath205 : @xmath206 critical temperature is an increasing function of the number of bosons in the system @xmath8 and electric field @xmath1 ; for example , for its small values , @xmath207 , one has @xmath208 ground - state occupation @xmath209 of the attractive robin wall as a function of the normalized temperature .
the same conventions as in fig .
[ boseheatcapacityneglambdafig1 ] are used . ]
bosonic ( a ) heat capacity per particle @xmath170 and ( b ) ground - state occupation @xmath209 of the attractive robin surface as a function of the normalized temperature for several electric fields and fixed number of bosons @xmath210 .
solid lines denote dependencies for @xmath132 , dashed curves for @xmath131 , dotted ones - for @xmath211 , and dash - dotted lines are for @xmath157 . ] fig .
[ boseheatcapacityneglambdafig1 ] draws be heat capacity per particle of the attractive surface as a function of temperature ( in units of its critical counterpart ) for several fields and numbers of bosons .
it exemplifies that for the one particle the @xmath6 behavior is qualitatively the same as for the fermion ; namely , as the applied voltage lowers , the extremum of the specific heat grows and is shifted to the colder temperatures .
some field - specific values provided in table [ table1 ] show that the bosonic peak is the largest among the three ensembles and is achieved at the highest @xmath3 .
note that , depending on the electric intensity , with the growth of the moderate @xmath8 the maximum can increase , as panel ( a ) and data from the table demonstrate . as the number of bosons becomes sufficiently large , the domination of the statistical effects starts to reveal itself leading to the formation from the resonance of an asymmetric cusp - like structure with its right side leaning more vertically with the increasing @xmath8 .
observe that this rapid descent of the heat capacity occurs at the temperature closer and closer coinciding with the critical one .
simultaneously , the ground - state population becomes a steeper function of the temperature almost turning to zero at @xmath203 , as fig .
[ bosegroundpopulneglambdafig1 ] depicts . at zero temperature ,
all bosons stay on the ground level , @xmath212 .
the heating of the wall pushes them away from the lowest state with the decrease of @xmath213 at @xmath214 ( @xmath215 ) being flatter ( more precipitous ) at the smaller fields . for the infinite number of bosons , @xmath216 ,
the ground orbital will be completely depopulated at @xmath217 what means a full destruction of the be condensate by the warming of the structure but due to the finiteness of @xmath8 a tiny field - dependent fraction @xmath213 persists for the temperatures above the critical one . to emphasize the role of the varying field in the process of
be condensate evolution and destruction , fig .
[ boseneglambdafig1 ] depicts heat capacity per particle and ground - state occupation at the fixed number of bosons @xmath198 and several very low voltages .
it is seen that the decreasing electric intensity causes , similar to the canonical and fd ensembles , an increase of the peak value of the heat capacity with the lowest level depopulation taking place at the higher normalized temperatures @xmath218 . in the limit of the vanishing field
the ground - state occupation turns into the step function @xmath219 : [ asymptotebose1 ] @xmath220{}h(\beta_{cr}/\beta ) , \intertext{while the critical temperature itself approaches the absolute zero , as it follows from eq.~\eqref{criticaltemp2 } : } \label{asymptotebose1_tcr } t_{cr}^{r-}&\xrightarrow[\mathscr{e}\rightarrow0]{}0.\end{aligned}\ ] ] the expression for @xmath221 from eq . , which was derived directly from eq . under the assumptions @xmath222 and @xmath111 ,
is obtained also by zeroing the denominator in eq .
that , in addition to the above two conditions , is valid for the large @xmath8 .
the infinite heat capacity at the critical temperature that follows from the latter formula is due to the neglect of the higher - order terms , which , if included into the consideration , smooth out the heat capacity peak to the finite field - dependent magnitude , as exact calculations depicted in fig . [ boseneglambdafig1](a ) demonstrate . the cusp - like structure of the heat capacity that was predicted theoretically for a number of the confining potentials @xcite and observed experimentally for , e.g. , the dilute bose gas of @xmath223 atoms @xcite , for the infinite number of particles , @xmath216 , turns at @xmath224 into the discontinuity that is a manifestation of the phase transition ; in our case , it is a transition from the be condensate to the normal phase of the noninteracting particles in the linear potential .
this justifies the definition of the critical temperature from eq . .
there has been a lively discussion of the influence of the shape and dimensionality of the confining potential on the existence , form and evolution of this feature @xcite .
as our results show , the sharpness of the resonance can be effectively controlled by the voltage applied to the attractive robin surface . for comparison ,
[ boseposlambdafig1 ] shows the heat capacity and ground - level occupation of the repulsive wall at the quite strong and very weak fields .
it is seen that in terms of the scaled temperature @xmath218 the shape of the curves is practically independent of the electric intensity .
the increasing number of bosons leads to the formation of the maximum of the specific heat at @xmath225 but , contrary to the attractive surface , this extremum does not grow to infinity rather saturating to the value lying below 2 .
contrasting the results of the negative and positive robin walls shows once again the crucial role of the split off level in the dramatic evolution of the thermodynamic properties .
upper ( lower ) panels show bosonic heat capacities per particle @xmath226 ( ground - state occupations @xmath227 ) of the repulsive robin surface as a function of the scaled temperature where left subplots [ ( a ) and ( c ) ] are for @xmath157 while the right ones [ ( b ) and ( d ) ] correspond to @xmath228 . the same line convention as in fig . [ boseheatcapacityneglambdafig1 ]
is used . ]
a theoretical analysis of the thermodynamic properties of the robin wall in the perpendicular electric field @xmath1 discovered that the heat capacity of the attractive surface for the canonical and either type of the grand canonical ensemble exhibits a pronounced maximum with its peak increasing at the vanishing voltage as @xmath4 while the corresponding temperature goes to zero as @xmath229 .
this phenomenon that is absent for the repulsive interface is explained by the peculiarities of the associated energy spectrum ; namely , at @xmath230 it is characterized by the quasi continuum of the positive - energy states and , additionally , the split off level with @xmath231 @xcite . at the cold temperatures ,
the heating of structure enforces the transitions from this orbital to a huge number of the higher lying states leading in this way to the giant enhancement of the specific heat @xmath46 . increasing number @xmath8 of the fermions decreases the extremum while for the be distribution for the quite large @xmath8 the resonance changes into the cusp - like structure , which is a manifestation of the phase transition into the condensate regime with the critical temperature being strongly @xmath1- and @xmath8-dependent .
thus , the applied voltage allows to effectively control the thermodynamics of the robin quantum wall .
having seen the dramatic evolution of the thermodynamic properties predicted above , one might wonder about their experimental verification . in the previous analysis of the attractive field - free robin wall @xcite
it was conjectured that this model approximates the piecewise continuous potentials that can be grown by the modern semiconductor technologies .
consider , for example , the term @xmath22 in eq . of the form ( below we return to the regular units ) : @xmath232 with positive @xmath233 and @xmath234 .
it is elementary to show that , in addition to its positive spectrum , this asymmetric qw can support bound states with their number depending on the interrelation between @xmath234 and @xmath233 ; for example , each new localized level @xmath35 emerges from the positive continuum at the following threshold magnitude of @xmath234 : @xmath235 it means that for @xmath234 satisfying the condition @xmath236 only one negative - energy state does exist . as a result , this potential forms zero - field energy spectrum complying with all the requirements of the present research .
accordingly , applying to it appropriately directed voltage , it is possible to check the heat capacity enhancement discussed above .
100 k. gustafson and t. abe , math . intell . * 20*(1 ) , 63 ( 1998 ) .
o. olendski , ann .
( berlin ) * 528 * , 865 ( 2016 ) .
handbook of mathematical functions , edited by m. abramowitz and i. a. stegun ( dover , new york , 1964 ) .
o. valle and m. soares , airy functions and applications to physics ( world scientific , hackensack , nj , 2010 ) .
o. olendski , ann .
( berlin ) * 527 * , 278 ( 2015 ) .
o. olendski , ann .
( berlin ) * 527 * , 296 ( 2015 ) .
d. a. krueger , phys . rev . * 172 * , 211 ( 1968 )
. e. b. sonin , zh .
. fiz . * 56 * , 963 ( 1969 ) [ sov .
- jetp * 29 * , 520 ( 1969 ) ] .
s. greenspoon and r. k. pathria , phys .
rev . a * 9 * , 2103 ( 1974 ) . c. s. zasada and r. k. pathria , phys .
a * 14 * , 1269 ( 1976 ) .
d. f. goble and l. e. h. trainor , can .
* 44 * , 27 ( 1966 ) .
d. f. goble and l. e. h. trainor , phys . rev . * 157 * , 167 ( 1967 ) .
d. f. goble and l. e. h. trainor , can .
* 46 * , 1867 ( 1968 ) .
s. grossmann and m. holthaus , z. phys .
b * 97 * , 319 ( 1995 ) .
r. k. pathria , phys .
a * 5 * , 1451 ( 1972 ) .
s. greenspoon and r. k. pathria , phys .
rev . a * 8 * , 2657 ( 1973 ) . m. n. barber and m. e. fisher , phys . rev . a * 8 * , 1124 ( 1973 ) .
z. r. hasan and d. f. goble , phys .
rev . a * 10 * , 618 ( 1974 ) .
n. dalarsson , m. dalarsson , and l. golubovi , introductory statistical thermodynamics ( academic , amsterdam , 2011 ) .
o. olendski , ann .
( berlin ) , to be published ; doi : 10.1002/andp.201600144 ; arxiv:1611.00197 [ cond-mat.mes-hall ] .
r. m. corless , g. h. gonnet , d. e. g. hare , d. j. jeffrey , and d. e. knuth , adv . comput . math . * 5 * , 329 ( 1996 ) .
d. veberi , comput .
. commun . * 183 * , 2622 ( 2012 ) .
a. houari , eur . j. phys .
* 34 * , 695 ( 2013 ) .
l. lewin , polylogarithms and associated functions ( elsevier , new york , 1981 ) . m. h. anderson , j. r. ensher , m. r. matthews , c. e. wieman , and e. a. cornell , science * 269 * , 198 ( 1995 ) .
c. c. bradley , c. a. sackett , j. j. tollett , and r. g. hulet , phys .
lett . * 75 * , 1687 ( 1995 ) . k. b. davis , m .- o .
mewes , m. r. andrews , n. j. van druten , d. s. durfee , d. m. kurn , and w. ketterle , phys .
lett . * 75 * , 3969 ( 1995 ) .
s. n. bose , z. phys .
* 26 * , 178 ( 1924 ) .
s. v. gaponenko , j. nanophoton .
* 8 * , 087599 ( 2014 ) .
f. dalfovo , s. giorgini , l. p. pitaevskii , and s. stringari , rev .
phys . * 71 * , 463 ( 1999 ) .
l. pitaevskii and s. stringari , bose - einstein condensation ( clarendon , oxford , 2003 ) .
a. j. leggett , rev .
phys . * 73 * , 307 ( 2001 ) . c. j. pethick and h. smith , bose - einstein condensation in dilute gases ( cambridge , cambridge , 2004 ) . w. ketterle and n. j. van druten , phys .
a * 54 * , 656 ( 1996 ) . n. j. van druten and w. ketterle , phys . rev .
* 79 * , 549 ( 1997 ) .
r. napolitano , j. de luca , v. s. bagnato , and g. c. marques , phys . rev .
a * 55 * , 3954 ( 1997 ) .
w. j. mullin , j. low temp .
phys . * 106 * , 615 ( 1997 ) .
v. bagnato , d. e. pritchard , and d. kleppner , phys .
a * 35 * , 4354 ( 1987 ) .
v. bagnato and d. kleppner , phys .
a * 44 * , 7439 ( 1991 ) .
s. grossmann and m. holthaus , phys .
a * 208 * , 188 ( 1995 ) .
h. haugerud , t. haugset , and f. ravndal , phys .
a * 225 * , 18 ( 1997 ) .
t. haugset , h. haugerud , and j. o. andersen , phys .
a * 55 * , 2922 ( 1997 ) .
s. goswami , t. k. das , and a. biswas , phys .
a * 84 * , 053617 ( 2011 ) .
j. r. ensher , d. s. jin , m. r. matthews , c. e. wieman , and e. a. cornell , phys .
lett . * 77 * , 4984 ( 1996 ) .
p. eba , lett .
* 10 * , 21 ( 1985 ) .
v. pama and p. prenajder , eur
. j. phys .
* 10 * , 35 ( 1989 ) .
t. flp , t. cheon , and i. tsutsui , phys .
a * 66 * , 052102 ( 2002 ) .
b. belchev and m. a. walton , j. phys .
a * 43 * , 085301 ( 2010 ) .
o. georgiou , g. gligori , a. lazarides , d. f. m. oliveira , j. d. bodyfelt , and a. goussev , europhys . lett . * 100 * , 20005 ( 2012 ) . | a theoretical analysis of the thermodynamic properties of the robin wall characterized by the extrapolation length @xmath0 in the electric field @xmath1 that pushes the particle to the surface is presented both in the canonical and two grand canonical representations and in the whole range of the robin distance with the emphasis on its negative values which for the voltage - free configuration support negative - energy bound state . for the canonical ensemble ,
the heat capacity at @xmath2 exhibits a nonmonotonic behavior as a function of the temperature @xmath3 with its pronounced maximum unrestrictedly increasing for the decreasing fields as @xmath4 and its location being proportional to @xmath5 . for the fermi - dirac distribution ,
the specific heat per particle @xmath6 is a nonmonotonic function of the temperature too with the conspicuous extremum being preceded on the @xmath3 axis by the plateau whose magnitude at the vanishing @xmath1 is defined as @xmath7 , with @xmath8 being a number of the particles .
the maximum of @xmath6 is the largest for @xmath9 and , similar to the canonical ensemble , grows to infinity as the field goes to zero . for the bose - einstein ensemble ,
a formation of the sharp asymmetric feature on the @xmath6-@xmath3 dependence with the increase of @xmath8 is shown to be more prominent at the lower voltages .
this cusp - like dependence of the heat capacity on the temperature , which for the infinite number of bosons transforms into the discontinuity of @xmath10 , is an indication of the phase transition to the condensate state .
some other physical characteristics such as the critical temperature @xmath11 and ground - level population of the bose - einstein condensate are calculated and analyzed as a function of the field and extrapolation length .
qualitative and quantitative explanation of these physical phenomena is based on the variation of the energy spectrum by the electric field . |
the possible use of gaas detectors in the future atlas experiment at the lhc imposes severe demands on the radiation hardness of the gaas detectors .
semi - insulating , undoped gaas detectors have been shown to withstand doses up to 100mrad of @xmath3co gamma rays @xcite . their sensitivity to neutrons , protons and pions
are reported in this paper .
the irradiations were performed at the isis facility @xcite for the neutron exposure , with a spectrum strongly peaked at 1mev , the cern ps for 24gev / c protons @xcite and the psi , villigen , for 300mev / c pions @xcite .
all irradiations were performed at room temperature .
all the detectors ( except for two described in section [ sec : indoped ] ) were fabricated with si u gaas substrates @xcite , with thicknesses varying from 115@xmath4 m to 500@xmath4 m .
the substrates were lapped and polished at glasgow university to obtain the desired thickness .
various thicknesses were used so that the dependence of the leakage current and charge collection efficiency on this parameter could be studied .
all the detectors had 3 mm diameter circular pad contacts with a 200@xmath4 m wide guard ring separated from the pad by 10@xmath4 m .
the schottky contact on one face of the wafer was metallized with ti / pd / au and the ohmic contact on the other face was pd / ge annealed at 300@xmath2c .
the leakage current characteristics before and after the irradiations were obtained with a keithley 237 voltage source measurement unit .
a @xmath5sr source was used to determine the response of the detectors to minimum ionising particles with an amplifier shaping time of 500ns .
the charge collection efficiencies of either electrons or holes created in the first 20@xmath4 m of the detector thickness were obtained with the use of an @xmath6am alpha particle source placed on the schottky or ohmic contact , respectively .
an amplifier with a 500ns shaping time was used . the current characteristics and charge collection efficiencies from the @xmath5sr source
were obtained at both 20@xmath2c and -10@xmath2c .
the -10@xmath2c temperature was chosen as this is the proposed operating temperature of the atlas semiconductor tracker .
the effect of all three types of irradiation on the reverse current characteristics of the gaas detectors is small .
figure [ fig : iv ] shows four reverse current characteristics with increasing proton fluence for a 210@xmath4 m thick detector measured at 20@xmath2c . the plateau current after a fluence of 8.9 @xmath7 10@xmath8 p/@xmath1 is only 2.2 times larger than the pre - irradiated value . before irradiation the detector is fully depleted at a bias , v@xmath9 , of 250v ; after only 1.56 @xmath7 10@xmath8 p/@xmath1 this has reduced to 160v .
the value remains unchanged with higher fluence .
for an applied bias greater than v@xmath9 the signal created by irradiating the ohmic contact with an alpha source ( the hole signal ) is non - zero , implying that the detector is active across the entire bulk . as the value of v@xmath9
is reduced with fluence the extension of the high field region in the detector changes after irradiation .
the value of the reverse current for biases greater than v@xmath9 reduces with increased fluence and the reverse current characteristic becomes more resistive in nature .
although the reverse current at v@xmath9 increases with dose up to 1 @xmath7 10@xmath0 p/@xmath1 , the majority of the increase occurs by 3 @xmath7 10@xmath8 p/@xmath1 . from 1 @xmath7
10@xmath0 p/@xmath1 up to 2.4 @xmath7 10@xmath0 p/@xmath1 ( the maximum fluence tested with this sample ) the current at this bias remains approximately constant ( see figure [ fig : ivdose ] ) .
reverse current characteristics of two detectors of different thickness , 160@xmath4 m and 500@xmath4 m , both of which were irradiated to 6 @xmath7 10@xmath8 p/@xmath1 are shown in figure [ fig : thickiv ] .
it can be seen that the value of the leakage current up to v@xmath9 is independent of the detector thickness .
this is not the case in silicon where the leakage current , i@xmath10 , has a dependence on fluence , @xmath11 , given by @xcite @xmath12 where @xmath13 is the reverse current damage constant and vol is the volume of the detector .
no varriation in this effect on the leakage current has been observed for detectors that were biased during irradiation .
the forward current characteristics also change with increasing fluence .
the effect of protons on a 160@xmath4 m thick detector , shown in figure [ fig : forwardiv ] , illustrates this .
a plateau appears that extends to higher bias values with increasing fluence while the magnitude of the plateau current falls .
this allows the detector to be operated under forward bias .
the charge collection efficiencies for 200@xmath4 m thick detectors as a function of neutron , proton and pion fluence are shown in figures [ fig : cceneutrons ] , [ fig : cceprotons ] and [ fig : ccepions ] .
these were measured using minimum ionising particles ( mips ) from a @xmath5sr source at an applied bias of 200v .
the reduction in cce with fluence occurs at a greater rate for pions and protons than for neutrons .
a 10000 electron signal response to the @xmath5sr source , ( @xmath14 40% cce ) , is obtained after 1.4 @xmath7 10@xmath0 n/@xmath1 , 6.0 @xmath7 10@xmath8 p/@xmath1 , and only 3.0 @xmath7 10@xmath8 @xmath15/@xmath1 .
the cce falls rapidly initially and then at a slower rate for higher fluences with the position of the change in gradient depending upon the particle type . in the atlas experiment at the lhc a particle flux of 0.12 @xmath7
10@xmath0 @xmath15/@xmath1/year at the position of the gaas wheels has been calculated @xcite for full luminosity .
this means that after only two years at full luminosity the pion fluence would be sufficient to reduce the signal from a 200@xmath4 m thick detector to approximately 10000 electrons .
the neutron flux is the only other significant contribution and this has been calculated to be a factor of 3 to 10 times less than the pions .
the exact value depends upon the moderator used in the calorimeters .
after the 10 years expected operational lifetime of atlas the neutron fluence will not be enough to reduce the signal to less than 10000 electrons .
irradiated 200@xmath4 m thick detectors biased in excess of 200v demonstrate an increase in cce .
for example for a 160@xmath4 m thick detector irradiated to a fluence of 1.0 @xmath7 10@xmath0 p/@xmath1 the signal increased from 8500 electrons at 200v to 12000 electrons at 400v .
the reason for the increase in signal can be understood from the electron and hole collection data . before irradiation ,
the signal due to the holes is greater than that due to the electrons , implying that the electrons are trapped more than the holes .
the signal due to the holes however falls at a faster rate with fluence than that due to the electrons .
this means that both electron and hole traps are introduced with irradiation , while the electron trapping is the less important after irradiation contrary to what is observed in unirradiated detectors .
the electron signal , which is constant for a bias up to v@xmath9 , rises at greater applied biases as shown in figure [ fig : alphaehdata ] .
the hole signal remains almost constant with increasing bias after its initial steep rise at v@xmath9 . from this
it is possible to conclude that field - enhanced electron detrapping may be present .
as the mip signal for a gaas detector irradiated in excess of 1.0 @xmath7 10@xmath0 p/@xmath1 appears to be independent of substrate thickness , even at an applied bias much larger than v@xmath9 , it is proposed that the carriers have a mean free path less than or equal to the thickness of the thinnest detector tested , that is 115@xmath4 m .
the charge deposited by a mip , q@xmath16 is given by @xmath17 where w is the width of the detector and de / dx is the energy loss per unit length of detector traversed .
if the carrier has a mean free path , @xmath18 , then the charge collected , q@xmath19 , is @xmath20 thus @xmath21 the fitted values of @xmath18 , which is a function of the fluence , for proton and pion irradiation are given by equations ( [ equ : lambdaprotons ] ) and ( [ equ : lambdapions ] ) , and show a similar fluence dependence .
the change in the reverse current of a detector irradiated to 1.0 @xmath7 10@xmath0 p/@xmath1 has been measured over a period of time greater than 1 year .
the sample was stored at room temperature , which varied throughout the year from 18@xmath2c to 25@xmath2c .
all the current measurements were carried out at 20@xmath2c with the same keithley 237 voltage source measurement unit .
the first current measurement was performed 20 days after the irradiation period due to the initial activity of the sample and holder preventing earlier handling . the value of the reverse current , i@xmath10 , at v@xmath9 reduced with time ( t ) , in hours , with a logarithmic dependence given by @xmath24 the value of v@xmath9 remained constant over this period .
some samples have shown an increase in cce over a similar period of annealing , however this represented at most an improvement of 8% of the initial cce .
elevated temperature annealing , for periods ranging from 15 minutes to 33 hours , was performed on six irradiated detectors in an oven at 210@xmath2c in air without bias .
the reverse current and charge collection efficiency variations for a 180@xmath4 m thick detector irradiated to 1.0 @xmath7 10@xmath0 p/@xmath1 are shown in figures [ fig:210cann ] and [ fig:210cce ] .
the reverse current decreased after an 33 hour anneal from 0.22@xmath4a to 0.14@xmath4a ( corresponding to 30na / mm@xmath25 to 20na / mm@xmath25 ) where the unirradiated value was 0.10@xmath4a ( 14na / mm@xmath25 ) .
the full depletion voltage remained unaffected by the annealing .
for an applied bias of 100v the cce increased from 23% to 30% for annealing times up to 5 hours .
as the bias increased , the increase in cce for the annealed detector was at a slower rate than in the pre - annealed case , resulting in a lower cce for the annealed detectors at biases in excess of 400v . for longer anneal times
the cce decreased .
for an anneal time of 33 hours , for example , the cce was only 15% at 100v rising to 22% for a maximum applied bias of 500v .
the arsenic antisite defect in gaas is known to anneal at temperatures in excess of 450@xmath2c @xcite . as
the irradiation could produce such antisite defects the effect of annealing at 450@xmath2c was investigated .
a flash - lamp anneal was used to illuminate the detectors for either 30,60,90 , or 120 seconds .
the effect was shown to be independent of the detector side which was illuminated , demonstrating that the whole bulk was being heated .
the plateau reverse leakage current was reduced but so was the full depletion voltage . for an applied bias greater than v@xmath9
the annealed current was greater than before annealing with the gradient of this section of the current characteristic being quite large .
the mip cce of the detector was increased for applied biases less than v@xmath9 , however due to the large increase in leakage current above v@xmath9 it was not always possible to measure the charge collection at a higher applied bias .
the increase in mip cce was due to an increase in electron mean free path with anneal time while the hole mean free path remained unchanged .
a bias of 300v , which could be applied to a detector annealed at 450@xmath2c for 120s after a fluence of 1.7 @xmath7 10@xmath0 @xmath15/@xmath1 , gave a mip cce of 71% .
the pre - irradiation cce was 95% and the post - irradiation , pre - anneal value was 26% .
[ sec : indoped ] two detectors were fabricated with 500@xmath4 m thick si gaas doped with indium @xcite .
these were tested before and after an exposure of 1.4 @xmath7 10@xmath0 24gev / c protons/@xmath1 .
the reverse current of the detectors before irradiation was 30na / mm@xmath25 , less than for the 500@xmath4 m thick si - u gaas detectors . after the irradiation the value of v@xmath9
was reduced to only 220v , compared to 340v for the undoped sample .
the current at the applied bias of v@xmath9 was 0.4@xmath4a ( 57na / mm@xmath25 ) which was higher than that of the undoped sample which was 0.22@xmath4a .
the charge collection efficiency for the indium doped samples was less that that of the undoped samples before irradiation by 20% .
after the irradiation the signal fell to 10% ( 6650 electrons ) compared to 17% ( 11305 electrons ) , respectively , for an applied bias of 1000v .
the changes in reverse current and charge collection efficiency for si - u gaas detectors have been studied after irradiations with 1mev neutrons , 24gev / c protons and 300mev / c pions .
while the leakage current variations are slight and pose little problem to the operation of the detectors , the fall in cce gives major cause for concern . at a given high fluence ,
the reduction is greater for charged hadrons than for neutrons ( this difference is discussed in detail in reference @xcite ) .
the cce falls rapidly in the first 3 @xmath7 10@xmath8 p(@xmath15)/@xmath1 .
the reduction has been shown to be due to both electrons and holes being trapped , with the hole mean free path affected much more dramatically .
for an applied bias greater than v@xmath9 the cce increases due to the increased signal from the electrons .
room temperature annealing has been identified in the reverse current characteristics and a slight improvement has also been seen in the cce .
annealing at elevated temperatures has shown beneficial effects in both leakage current and cce .
the effect of proton irradiation on indium - doped gaas detectors has shown the material to be no more radiation - hard than si - u gaas .
further tests on different samples should be performed .
the authors would like to thank f.mcdevitt , a.meikle and f.doherty for technical support and all those at the isis , cern ps and psi facilities during the irradiation runs .
one of us ( r.bates ) gratefully acknowledges the support received through a case postgraduate studentship from ral .
the results obtained within the rd8 collaboration are from work partly funded by pparc ( uk ) , infn ( italy ) , and the bmfd ( germany ) . | semi - insulating , undoped , liquid encapsulated czochralski ( si - u lec ) gaas detectors have been irradiated with 1mev neutrons , 24gev / c protons , and 300mev / c pions .
the maximum fluences used were 6 , 3 , and 1.8 10@xmath0 particles/@xmath1 respectively . for all three types of irradiation
the charge collection efficiencies ( cce ) of the detector are reduced due to the reduction in the electron and hole mean free paths .
pion and proton irradiations produce a greater reduction in cce than neutron irradiation , with the pions having the greatest effect .
the effect of annealing the detectors at room temperature , at 200@xmath2c and at 450@xmath2c with a flash lamp have been shown to reduce the leakage current and increase the cce of the irradiated detectors .
the flash - lamp anneal produced the greatest increase in the cce from 26% to 70% by increasing the mean free path of the electrons .
two indium - doped samples were irradiated with 24gev / c protons and demonstrated no improvement over si u gaas with respect to post - irradiation cce .
glas - ppe/970427 august 1997 |
given a convex body @xmath3 and some subspace @xmath4 , how to compute the volume of the intersection @xmath5 ?
how to find the maximal or minimal sections ?
these questions have been considered for various convex bodies .
a first example is the unit cube intersected with central hyperplanes .
explicit formulas were already found by laplace . the question of the minimal and the maximal section were answered by d. hensley @xcite resp . k. ball @xcite . since then
several other bodies and modified questions have been considered .
we mention a few examples : @xmath6-balls @xcite , @xcite , complex cubes @xcite ; also general @xmath1-dimensional sections @xcite and non - central sections @xcite as well as taking other than lebesgue measures @xcite have been investigated . here
we are interested in the regular simplex .
s. webb @xcite gave a formula for central hyperplane sections .
he also proved that the maximal central section is the one containing @xmath2 vertices and the centroid .
the question of the minimal central hyperplane section is not completely solved yet .
p. filliman stated that his methods can be used to prove that the section parallel to a face is minimal @xcite .
but he gave no precise arguments .
p. brzezinski proved a lower bound which differs from the conjectured minimal volume by a factor of approximately @xmath7 @xcite .
in this paper we consider @xmath1-dimensional and also non - central sections .
we extend s. webb s formula and show that , for hyperplanes for fixed small distances from the centroid to the hyperplane , the one containing @xmath2 vertices still gives maximal volume . concerning the minimal section we give a result that supports the conjecture . for dimensions up to @xmath8 we prove the conjecture .
using brascamp - lieb inequality we also give a bound on the volume of @xmath1-dimensional sections . in the final chapter we consider irregular simplices and construct an interesting example , using an idea by walkup @xcite .
it has the property that all its faces have smaller volume than some central section .
this is not true for the regular simplex and for any simplex in dimensions up to @xmath8 .
we use the following notations .
let @xmath9 @xmath10 is the embedded regular @xmath0-simplex with @xmath11 vertices and side length @xmath12 . the euclidean norm is denoted by @xmath13 , the standard scalar product by @xmath14 .
the distance of two sets is given by @xmath15 , especially @xmath16 .
if @xmath4 is a @xmath1-dimensional ( affine ) subspace and @xmath17 , the @xmath1-volume of @xmath18 is the standard induced lebesgue volume of the subspace , denoted by @xmath19 . for @xmath20 with @xmath21 and @xmath22 , let @xmath23 be a translated hyperplane , especially @xmath24 . for @xmath25
the hyperplane @xmath26 is still well defined .
we denote two special directions . @xmath27
the volume of the corresponding sections can be computed elementary .
@xmath28 the hyperplane @xmath29 is parallel to one of the faces of the simplex .
the hyperplane @xmath30 contains @xmath2 vertices and the midpoint of the remaining two vertices . due to the symmetries of the simplex the volume is invariant under permutations of the coordinates of @xmath31 and multiplying @xmath31 by @xmath32 .
our results are stated in the following theorems . the general volume formula that we find is [ thm : general formula ] let @xmath4 be a k - dimensional subspace of @xmath33 and @xmath34 , @xmath35 some orthonormal basis of @xmath36
. then @xmath37 for non - central hyperplane sections we get a result on the maximal volume .
[ thm : maximal non central sections ] let @xmath38 . for all @xmath39 with @xmath40 and @xmath41
we have @xmath42 with equality for @xmath43 . concerning small central sections we prove [ thm : small central sections ] ( i ) for all @xmath44 the volume of the section @xmath45 is locally minimal , more precisely for all @xmath46 with @xmath47 and @xmath48 , we have @xmath49 ( ii ) for dimensions
@xmath50 and @xmath8 the volume of the section @xmath45 is globally minimal , i.e. for all @xmath46 with @xmath47 we have @xmath49 also for @xmath1-dimensional sections we find an upper bound .
[ thm : large k dimensional sections ] let @xmath4 be a @xmath1-dimensional subspace of @xmath33 that contains the centroid of @xmath10 .
then we have @xmath51 if additionally @xmath52 for all @xmath53 , then @xmath54 holds . in this case
the estimate is sharp .
we conclude the paper with a new proof of a result on irregular simplices .
[ ex : example deformed simplex ] for odd dimension @xmath55 there exist irregular simplices whose largest hyperplane section is not one of its faces .
instead of considering the simplex @xmath10 we first look at , a right - angled @xmath56-simplex .
this is a proper @xmath56-dimensional body in @xmath33 .
the intersection @xmath57 is a pyramid with base @xmath58 and with its top in the origin . knowing the volume of the intersection @xmath59 we can find the volume of @xmath58 by the formula for the volume of a pyramid .
we start with the height of the pyramid @xmath60 which is given by @xmath61 , where @xmath62 .
-simplex , scaledwidth=35.0% ] [ lem : distance to 0 ] let @xmath4 be a k - dimensional subspace of @xmath33 and @xmath34 , @xmath35 some orthonormal basis of @xmath36 .
then @xmath63 we minimize @xmath13 under the constraints @xmath64 for all @xmath65 , and @xmath66 .
define the lagrange function @xmath67 for the derivative with respect to @xmath68 we find @xmath69 , so for a critical vector @xmath70 we have @xmath71 summing ( [ eq : lemma distance 1 ] ) over @xmath72 and using @xmath66 leads to @xmath73 multiplying ( [ eq : lemma distance 1 ] ) with @xmath68 and then summing over @xmath72 gives @xmath74 multiplying ( [ eq : lemma distance 1 ] ) with @xmath75 for a fixed @xmath76 and then summing over @xmath72 we find @xmath77 this implies that @xmath78 is a necessary condition for an extremum .
obviously this critical condition gives a minimum .
the claim follows taking the square root .
we compute the volume of @xmath59 following an approach by m. meyer and a. pajor .
they derived a formula for sections of the @xmath79-ball @xcite .
the @xmath56-simplex @xmath80 corresponds to one orthant of the @xmath79-ball .
the formulas look similar .
note that the @xmath79-ball is origin - symmetric , whereas the simplex @xmath80 is not .
meyer s and pajor s proof has to be modified slightly since their lemma requires symmetry of the body .
[ lem : integral formula 1 ] let @xmath4 be a @xmath81dimensional subspace of @xmath33 .
then @xmath82 let @xmath4 be defined by some orthonormal basis of @xmath36 , say @xmath83 .
let @xmath84 .
for @xmath85 we define @xmath86 \big\}.\ ] ] the set @xmath87 and therefore the computations below depend on the choice of the orthonormal basis .
but finally we consider limits for @xmath88 .
these are independent of the choice of @xmath89 .
the @xmath1-volume of @xmath90 is given by @xmath91 .
for @xmath92 we have @xmath93 we consider the following integral with respect to the lebesgue measure on @xmath33 : @xmath94 then by integration on the level sets of @xmath95 resp . by fubini s theorem
we find @xmath96 furthermore we have @xmath97 by homogeneity of @xmath98 we get @xmath99 since @xmath100 and by ( [ eq : good subset ] ) we have @xmath101 for @xmath88 this tends to @xmath102 . for @xmath103 let @xmath104 be the subset of @xmath105 such that @xmath106 . since the following inclusion holds @xmath107 we get by ( [ eq : good subset ] ) : @xmath108 note that @xmath109 .
so for @xmath88 we have @xmath110 since we may interchange the limit and the integral due to dominated convergence .
therefore @xmath111 to get an explicit formula it is classical to use the fourier transformation and the fourier inversion theorem , e.g. @xcite , @xcite .
[ lem : integral formula 2 ] let @xmath4 be a k - dimensional subspace of @xmath33 and @xmath34 , @xmath35 some orthonormal basis of @xmath36 .
then @xmath112 first rewrite the integral @xmath113 and define a map @xmath114 by @xmath115 this function is integrable , since @xmath116 .
we apply the fourier transform : @xmath117 this function is continuous and integrable . by the fourier inversion formula
we get @xmath118 now we put together lemmas [ lem : distance to 0 ] , [ lem : integral formula 1 ] and [ lem : integral formula 2 ] and use the formula for the @xmath1-volume of a pyramid @xmath119 to obtain the formula from theorem 1 . for hyperplane sections
this formula can be evaluated further by using the residue theorem .
s. webb did this for central sections and this method also works for general sections .
note that the restriction on @xmath31 is not an essential one , since the function @xmath120 is continuous .
[ cor : volume formula simplex ] let @xmath121 with @xmath21 such that all @xmath122 are pairwise distinct and there is at least one @xmath122 and one @xmath123 .
then we have @xmath124 s. webb s formula required @xmath125 . in this case
the centroid of @xmath10 lies in the hyperplane .
a different way of representing non - central sections is translating central sections .
so for suitable @xmath126 , @xmath127 with @xmath128 and @xmath129 we have @xmath130 the two representations can be converted into each other : given @xmath131 with @xmath21 , @xmath132 and @xmath127 set @xmath133 for @xmath53 .
then @xmath134 has norm @xmath135 .
let @xmath136 .
then @xmath137 is equivalent to @xmath138 .
so @xmath139 on the other hand , given @xmath140 with @xmath141 set @xmath142 for @xmath53 .
then @xmath31 has norm 1 and @xmath132 .
for @xmath136 we have @xmath143 .
therefore the condition @xmath144 is equivalent to @xmath137 .
so we have @xmath145 due to ( [ eq : hyperplane convert2 ] ) we also know @xmath146
we investigate the volume of a hyperplane section with fixed distance to the centroid . for a normal vector @xmath31 set @xmath147 .
due to ( [ eq : dist c hcaps ] ) , fixed distance means @xmath90 is fixed .
we may assume @xmath148 , since @xmath31 and @xmath149 determine the same hyperplane .
by continuity of @xmath120 we may assume that all @xmath150 .
furthermore we assume that @xmath151 with @xmath152 for @xmath153 and @xmath123 for @xmath154 .
this does not change the volume , since permutations of the coordinates correspond to the symmetries of the simplex . according to corollary [ cor : volume formula simplex ]
we have to estimate @xmath155 as long as all positive @xmath156 are pairwise distinct .
the estimates rely on the following inequalities , valid for @xmath157 and @xmath158 .
these inequalities are bernoulli s inequality resp . of the arithmetic - geometric - mean inequality
: for all @xmath159 and @xmath158 we have @xmath160 the idea for our estimates is to modify a given vector @xmath31 such that the sum of the coordinates , their square sum and their signs do not change . preserving the sign geometrically means the polytopal structure of the section is preserved ( see subsection [ sec : small sections ] below ) . preserving
the sum of the coordinates means preserving the distance of the hyperplane to the centroid .
[ lem : upper bound ] let @xmath38 , @xmath39 with @xmath21 and @xmath41 .
then we have @xmath161 with equality for @xmath162 . for a given @xmath31
we define @xmath163 with @xmath164 and @xmath165 such that @xmath166 and @xmath167 .
then we have @xmath168 we show that @xmath169 and @xmath170 in @xmath171 $ ] with these properties exist : + equation ( [ eq : rescale 1 ] ) describes a line @xmath172 in the @xmath173-plane with positive slope , and values @xmath174 and @xmath175 , since @xmath176 . equation ( [ eq : rescale 2 ] ) describes an ellipse . solving this for the positive part of the ellipse , we get @xmath177 and @xmath178
. therefore @xmath179 and @xmath180 .
the two functions @xmath181 and @xmath182 are continuous on @xmath183 . due to the intermediate value theorem they intersect ,
so there are with the desired properties .
we compare @xmath184 and @xmath185 and find @xmath186 the last inequality is due to the left - hand side of ( [ eq : real inequality ] ) , i.e. bernoulli s inequality , and the fact that @xmath187 $ ] .
more precisely , for all @xmath188 , we have @xmath189 we do the same trick for @xmath154 .
so consider a vector of the form @xmath190 , with @xmath191 for @xmath153 and @xmath192 .
we define @xmath193 such that @xmath166 and @xmath194 . with a similar argument as in the first step , additionally using @xmath195 , we find that @xmath196 $ ] exist .
we compare @xmath197 and @xmath198 .
note that the function @xmath199 now only has one summand .
the estimates are analog to the first step .
so @xmath199 attains its maximum at a vector of the form @xmath200 . using @xmath21 and @xmath41
, we conclude that @xmath201 and @xmath202 .
using proposition [ lem : upper bound ] we immediately determine the maximal section for a fixed distance close to the centroid .
close means precisely that the distance between the hyperplane and the centroid is at most the distance from a face to the centroid .
this maximal section contains @xmath2 of the vertices of the simplex .
this includes the previous result of s. webb .
if we are only interested in central section , i.e. @xmath203 , we do not have to introduce @xmath169 and @xmath170 and the proof of the lemma is just by bernoulli s inequality .
therefore we gave a different and shorter proof for webb s result on maximal central sections .
a similar lemma can be obtained for an estimate in the converse direction .
instead of concentrating the coordinates we balance them .
this estimate is weaker since we can only balance the negative half of the coordinates of @xmath31 , whereas we could concentrate the positive and the negative coordinates in the above lemma .
[ lem : lower bound ] let @xmath204 , @xmath39 with @xmath21 and @xmath41 .
then we have @xmath205 where @xmath206 , with @xmath207 such that @xmath166 , @xmath208 and @xmath209 . as in the last lemma
we modify the vector @xmath31 to obtain an estimate for @xmath199 .
this time we choose @xmath210 such that for @xmath211 the equations @xmath212 hold .
equation ( [ eq : rescale 3 ] ) is the same as ( [ eq : rescale 1 ] ) from the previous proof with the same implications @xmath213 and @xmath175 .
note that the slope of @xmath181 is larger than @xmath135 .
equation ( [ eq : rescale 4 ] ) again defines an ellipse .
with @xmath214 we find that @xmath215 .
furthermore @xmath216 for @xmath217 . by the intermediate value theorem
, there are @xmath218 and @xmath219 with ( [ eq : rescale 3 ] ) and ( [ eq : rescale 4 ] ) . since the slope of the line defined by ( [ eq : rescale 3 ] ) is greater than @xmath135 , also @xmath220 .
we compare @xmath185 and @xmath184 and obtain @xmath221 now we use @xmath222 and the right inequality of ( [ eq : real inequality ] ) , which is the agm inequality : @xmath223 finally , we find that @xmath224 one might conjecture that the minimum of @xmath185 without changing the signs of @xmath156 is attained in a vector of the form @xmath225 , for some @xmath226 .
but for a vector of this form the formula from corollary [ cor : volume formula simplex ] and the estimates from the previous lemma do not work anymore .
the example in lemma [ lem : minimum small dimension ] shows that this conjecture is even false for higher dimensions . from proposition [ lem : lower bound
] we get the local result , i.e. theorem [ thm : small central sections ] ( i ) .
this also immediately shows that @xmath227 is a global minimum for @xmath228 . if we find all the local minima , in the sense of a fixed distribution of signs for the vector @xmath31 , we can prove the global result . for dimension @xmath229 we derive a different formula in the next section , which allows us to compute the remaining local minima .
we take a closer look at the geometric structure of @xmath230 . for general @xmath31
we can not say too much about @xmath230 .
the section is some polytope without obvious regularity .
but if we assume @xmath31 to have the form @xmath231 for some @xmath232 , as in proposition [ lem : lower bound ] , then we have some regularity and we get an additional volume formula . at least for small dimensions
this leads to a solution of the minimal section problem .
let @xmath233 with @xmath234 and @xmath235 .
@xmath236 is the number of positive coordinates , @xmath209 is the number of negative coordinates .
the points @xmath237 with @xmath238 and @xmath239 are the intersection points of @xmath240 with the edges @xmath241 $ ] of the simplex @xmath10 .
they are of the form @xmath242 where the @xmath243-th and the @xmath244-th coordinate are non - zero .
the section polytope @xmath240 is the convex hull of these @xmath245 points @xmath237 .
let us further decompose the section polytope @xmath240 .
it is the convex hull of polytopes of the form @xmath246 for @xmath247 .
for the case @xmath248 , @xmath249,scaledwidth=100.0% ] now let @xmath250 and @xmath232 , see figures [ fig : balanced section 3d ] and [ fig : balanced section 4d ] .
in this case the @xmath251 are regular simplices with @xmath252 vertices and side length @xmath253 , since for fixed @xmath243 and for all @xmath254 we have @xmath255 the @xmath256-volume of these simplices is @xmath257 two simplices
@xmath258 lie in parallel subspaces , since their edges @xmath259 $ ] and @xmath260 $ ] are parallel .
the distance of @xmath251 and @xmath261 is given by the distance of their centroids , which is @xmath262 note that @xmath263 and @xmath264 are independent from the scaling of @xmath31 .
the value @xmath169 is determined by the constraints for the normal vector , up to scaling .
we have @xmath265 . two simplices @xmath251 and
@xmath261 constitute a truncated pyramid , if @xmath248 , resp . a frustum with a regular @xmath256-simplex as its base and another regular @xmath256-simplex as its top .
let @xmath249 .
the section @xmath240 is the convex hull of two parallel @xmath256-simplices , i.e. a frustum .
let @xmath266 , @xmath267 be the @xmath256-volume of the top resp . bottom and @xmath268 , @xmath269 the relative heights . by homogeneity we know @xmath270 .
set @xmath271 , then we have @xmath272 .
the volume @xmath273 of @xmath240 is computed as @xmath274 note that for @xmath275 , @xmath157 the equation @xmath276 holds .
therefore @xmath277 so the volume of the section can be expressed by @xmath278 with @xmath279 . under the assumption @xmath249 , the normal vector @xmath31 only depends on one variable . since the formulas are independent of scaling the vector @xmath31 , it is sufficient to consider @xmath31 of the form @xmath280 with @xmath281 .
so the volume is given by @xmath282 for @xmath283 and @xmath284 the geometric arguments do not work , since the involved simplices become degenerated .
however , this function is still well defined .
note that for @xmath283 and @xmath284 the vector @xmath285 corresponds to the vector @xmath286 with @xmath287 .
the values of @xmath288 and @xmath289 equal the volume of the section for such @xmath285 , computed in ( [ eq : volmin ] ) .
therefore the formula may be extended to @xmath171 $ ] .
summarizing we proved let @xmath290 , @xmath21 and @xmath132 .
additionally let @xmath291 for some @xmath232 and @xmath292 . then with the definition ( [ eq : volume formula v ] ) : @xmath293 for normal vectors of this special form we now determine the minimal section volume .
[ lem : minimum small dimension ] let @xmath273 be as defined above . then for @xmath294 and @xmath295 $ ] @xmath296 but for @xmath297 we have @xmath298 .
the function @xmath273 is differentiable .
so one finds the extrema by finding the zeros of its derivative .
the calculations are elementary and we just give the results . note that it is sufficient to consider @xmath299 $ ] due to the symmetry of the function . for @xmath294
the derivative of @xmath273 has one zero in @xmath300 and one more zero in @xmath301 .
the other zeros are complex or not in @xmath302 . comparing the function values at the critical points
we know that @xmath273 has a global minimum in @xmath300 . for @xmath297
we obtain a different behavior .
@xmath273 still has a minimum in @xmath300 , but this is not the global minimum anymore : @xmath303 recall that multiplying the normal vector by @xmath32 or permuting the coordinates does not change the volume of the section . if @xmath131 with @xmath304 and @xmath305 , by proposition [ lem : lower bound ] we have @xmath306
. for @xmath228
this is already sufficient .
let @xmath307 and @xmath308 and @xmath309 .
then by lemma [ lem : minimum small dimension ] with @xmath310 we know @xmath311 this solves the case @xmath307 .
finally let @xmath248 .
for @xmath308 and @xmath312 lemma [ lem : minimum small dimension ] with @xmath313 yields @xmath314 with @xmath315 . for @xmath6-balls with @xmath316 the hyperplane section volume
is given by @xmath317 where @xmath318 is some constant depending on @xmath319 and @xmath0 and @xmath320 is the fourier transform of the function @xmath321 on @xmath322 , see ( * ? ? ?
koldobsky used the fact that @xmath323 is convex on @xmath183 to find the minimum and the maximum .
the maximum is attained if one coordinate is @xmath135 and the rest is zero .
the minimum is attained if all coordinates are equal .
this technique can not be used for the simplex .
the analogue to @xmath324 is the function @xmath325 .
this function is complex - valued .
furthermore the example in lemma [ lem : minimum small dimension ] shows that for the simplex the _ balanced _ vector can not be minimal in general .
the maximal central hyperplane section of the simplex is the one that contains @xmath2 vertices and the midpoint of the opposite edge .
the natural generalization of this hyperplane section to lower dimensional sections is @xmath326 where @xmath327 is a suitable @xmath1-dimensional subspace .
this polytope contains @xmath328 vertices and the centroid of the opposite face .
it has @xmath329-volume @xmath330 , which is computed by the elementary volume formula for cones . with an extra condition
we prove that this is indeed maximal . without additional conditions we prove a bound that differs from this conjectured maximum by some factor depending on the dimension resp .
codimension of the intersecting subspace . the main tool for the proof is the brascamp - lieb inequality in a normalized form @xcite .
let @xmath331 be unit vectors in @xmath327 , @xmath4 a @xmath1-dimensional subspace and @xmath332 satisfying @xmath333 . then for integrable functions @xmath334 we have @xmath335 let @xmath236 be the orthogonal projection onto @xmath4 , @xmath336 and @xmath337 . then @xmath338 .
furthermore @xmath339 is just the trace resp .
the rank of the projection @xmath236 , so @xmath340 . with the operator norm of the projection we observe @xmath341 and
if we require @xmath4 to contain the centroid @xmath342 we have by pythagoras @xmath343 note that @xmath344 .
so we get the conditions @xmath345 now we consider the additional assumption of theorem [ thm : large k dimensional sections ] .
let for all @xmath53 .
for each @xmath346 we have @xmath347 where @xmath18 is the matrix whose columns are the vectors @xmath348 .
+ if @xmath349 then @xmath350 .
therefore @xmath351
now we apply the brascamp - lieb inequality . for @xmath352
we have @xmath353 and therefore @xmath354 we maximize @xmath355 under the constraints given in ( [ eq : constraints k - dim ] ) .
we consider the equivalent problem to maximize @xmath356 under the constraints @xmath357 the function @xmath358 is convex on @xmath359 .
therefore also @xmath360 is convex on @xmath361 . the set @xmath362^{n+1 } \mid \sum_{j=1}^{n+1 } x_j = k , \
\frac{1}{n+1 } \leq x_j \leq 1\bigg\}\ ] ] is also convex , so the maximum is attained in some extremal point of the set .
the extremal points are permutations of the point @xmath363 with @xmath328 coordinates equal to @xmath135 and @xmath364 coordinates equal to @xmath365 .
+ for @xmath366 this means @xmath367 with @xmath328 coordinates equal to @xmath135 and @xmath364 coordinates equal to @xmath368 .
+ therefore the maximum is @xmath369 . with the additional condition ( [ eq : constraints k - dim2 ] )
we find that @xmath370 ( with @xmath328 coordinates equal to @xmath135 and @xmath371 coordinates equal to @xmath372 ) is maximal and @xmath373 .
\(i ) fix @xmath374 .
then the quotient of the proven bound ( [ eq : k dim bound nonopt ] ) and the conjectured optimal bound ( [ eq : k dim bound opt ] ) from theorem [ thm : large k dimensional sections ] tends to @xmath135 for @xmath375 .
so this is asymptotically optimal .
\(ii ) now we fix the codimension of the section , i.e. fix @xmath376 and let @xmath377 .
then for @xmath375 the quotient tends to @xmath378 .
\(iii ) in general , brascamp - lieb is a sharp inequality and its application gives sharp estimates .
for example in the consideration of sections of @xmath6-balls one gets sharp results @xcite , @xcite .
the reason why we do not get a sharp bound here is the additional restriction @xmath379 .
if we do not assume the subspace @xmath4 to contain the centroid or even fulfill some extra condition , there is no lower bound on @xmath366 as in ( [ eq : constraints k - dim ] ) or ( [ eq : constraints k - dim2 ] ) .
then the integral is simply bounded by @xmath135 .
this bound is attained by @xmath380 .
the corresponding section of @xmath10 is a @xmath329-face which has distance @xmath381 from the origin .
for the regular simplex the maximal hyperplane section is a face , if the distance to the centroid is not prescribed .
this can be proven by a theorem of fradelizi about isotropic convex bodies .
he proves that the maximal section of a cone in isotropic position is its base ( * ? ? ?
* corollay 3 ( 2 . ) , p. 169 ) .
this is still true for deformed and therefore non - isotropic simplices in dimensions @xmath382 and @xmath8 .
the case @xmath228 is obvious .
the case @xmath307 was considered by @xcite . the first proof that this phenomenon does not generalize to all dimensions appears in @xcite .
the full solution , i.e. @xmath248 affirmative and @xmath383 negative , was given by philip @xcite .
he uses walkup s construction for @xmath384 and concluded by induction .
we modify our volume formula from corollary [ cor : volume formula simplex ] to be applicable to irregular simplices .
then we give a direct proof that for odd dimensions larger than @xmath385 there exist simplices such that not some of the faces is the maximal section .
the constructed example uses walkup s idea .
every simplex is an affine image of the regular simplex .
here we denote the regular simplex by @xmath386 and a general simplex by @xmath10 .
the formula follows by an application of the transformation theorem .
[ thm : deformed simplex ] let @xmath387 be an arbitrary simplex in @xmath388 without loss of generality we may assume that @xmath389 for all @xmath390 , i.e. all vertices lie in the affine hyperplane defined by the regular simplex .
let @xmath391 be the linear transformation that maps @xmath10 to the regular simplex @xmath386 .
let @xmath46 with @xmath21 .
then we have @xmath392 where @xmath393 .
the transformation @xmath391 maps @xmath10 to the regular simplex bijectively .
therefore @xmath394 is given by @xmath395 where the @xmath396 are column vectors .
we analyze the behavior of the @xmath397-volume under @xmath391 .
we only sketch the proof .
for @xmath56-dimensional subsets we have the transformation theorem in @xmath33 .
the set @xmath398 is @xmath397-dimensional .
we enlarge this set in two directions .
we thicken @xmath398 in direction @xmath31 , see ( [ eq : thicken ] ) , and we take the convex hull of this thickened set and @xmath399 .
then we consider the image of this @xmath56-dimensional set and use the transformation theorem to obtain the formula from this proposition .
we show that for odd dimension @xmath55 there is a simplex whose largest hyperplane section is not one of its faces .
we consider the regular simplex with @xmath11 vertices , where @xmath11 is even .
then we take a hyperplane such that on both sides of the hyperplane the vertices build a @xmath400-simplex parallel to the hyperplane .
now we compress the simplex along the normal vector of this hyperplane .
the intersection with the simplex remains the same but the faces of the compressed simplex become smaller . with the formula from proposition [ thm : deformed simplex ] we compute the volumes and show that for @xmath383 the compressed simplex has the desired property for a certain degree of compression .
we make this construction explicit . for @xmath401 let @xmath402
the columns the matrix @xmath394 define the vertices of a deformed simplex @xmath403 .
note that @xmath404 and @xmath405 is a degenerate simplex which is a @xmath397-dimensional set with @xmath0-dimensional volume equal to @xmath406 .
with @xmath407 , we may also write @xmath408 . using basic matrix theory we know that @xmath409 and @xmath410 . so @xmath411 let @xmath412 we compute the vectors @xmath413 and @xmath414 according to proposition [ thm : deformed simplex ] .
we find @xmath415 and @xmath416 .
the intersections @xmath417 resp .
@xmath418 are the central sections described at the beginning of the proof .
the intersection @xmath419 is a face of the regular simplex .
therefore @xmath420 is a face of the simplex @xmath403 , since @xmath391 maps faces to faces .
all faces of the simplex are of this form and therefore have the same volume .
it remains to compute the ratio of the section volumes using the formulas from theorem [ thm : general formula ] and analyze the behavior for @xmath421 .
we find @xmath422 for odd @xmath383 this quotient is larger than @xmath135 .
so for some @xmath423 the simplex @xmath403 has the desired property .
this work is part of my phd thesis .
i thank my advisor hermann knig for his support and advice .
my research was partly supported by dfg ( project ) . , _ on the maximal measure of sections of the @xmath0-cube _ , geometric analysis , mathematical relativity , and nonlinear partial differential equations , contemp .
599 , amer .
soc . , providence , ri , 2013 , 123155 . ,
_ a remark on the extremal non - central sections of the unit cube _ , asymptotic geometric analysis .
proceedings of the fall 2010 fields institute thematic program , new york , ny : springer ; toronto : the fields institute for research in the mathematical sciences , 2013 , 211228 . | we state a general formula to compute the volume of the intersection of the regular @xmath0-simplex with some @xmath1-dimensional subspace .
it is known that for central hyperplanes the one through the centroid containing @xmath2 vertices gives the maximal volume .
we show that , for fixed small distances of a hyperplane to the centroid , the hyperplane containing @xmath2 vertices is still volume maximizing .
the proof also yields a new and short argument for the result on central sections . with the same technique
we give a partial result for the minimal central hyperplane section .
finally , we obtain a bound for @xmath1-dimensional sections . |
this notes represent the written version of lectures i gave in mini - courses at universidade de braslia ( april 3 - 6 , 2013 ) , universidad central de venezuela ( may 23 - 27 , 2016 ) and at the villa de leyva summer school `` geometric , topological and algebraic methods for quantum field theory '' ( july 15 - 27 , 2013 ) .
they were mainly intended as an introduction to some aspects of operator algebras , emphasizing the prominent role they play in quantum physics . as the audience consisted of students of both physics and mathematics at different stages of their studies ,
my choice was to focus on the most basic structures and examples , in the hope that a good grasp of these would motivate them to go deeper into the subject .
now , something that to a physicist may appear as completely familiar ( as , say , an experimental set - up with polarizers , or the distinction between a classical and a quantum field ) to a mathematician may not
. the same could be said of the proof that the spectrum of any element in a @xmath0-algebra is never empty : it is a standard result in analysis , but may look quite awkward to many physics students .
therefore , the emphasis of these lecture notes will be on explaining _ why _ certain mathematical structures may be useful for the study of quantum theory .
this will be illustrated by means of several examples that include a discussion of bipartite entanglement , algebraic and geometric aspects of quantum phase transitions in spin chains , quasi - free states in fermionic systems and applications to quantum field theory .
our starting point will be based on the sharp distinction between _ classical probability _ and the probability theory inherent to quantum mechanics .
this will provide a physical motivation to the various mathematical concepts we will be dealing with . after introducing the basic concepts about @xmath0-algebras
, we will show how an algebraic approach to entanglement can lead to a resolution of certain discrepancies appearing when we deal with systems of identical particles .
then we will focus on the study of certain specific models ( quantum ising and xy models ) that turn out to be ideal in order to illustrate how entanglement , geometry and the theory of car algebras - algebras used to model fermionic systems ] are interrelated .
our first objective will be to understand what are the main structural differences between _ classical _ and _ quantum _ physics , when regarded as probability theories .
this will , by the way , provide a physical motivation for some of the mathematical notions we will consider in the next sections .
let us recall that classical physics is usually modeled on a configuration ( or phase ) space , with a dynamics governed by , _
e.g. _ , hamilton s principle . on the other hand , quantum mechanics
is modeled on a hilbert space .
so the first issue we want to explore is : why do we have to use quantum mechanics to describe the microscopic world ? indeed ,
simple experiments with light polarizers make it clear that there is no way we can describe certain phenomena using classical physics or , better said , classical _
probability_. let us then explore some elementary polarization phenomena , following the presentation in @xcite .
first let us recall that light is just made of electromagnetic waves , their behavior being governed by maxwell s equations : @xmath1 where @xmath2 ( resp .
@xmath3 ) stands for the electric ( resp . magnetic ) field , @xmath4 for the charge density and @xmath5 for the current density .
[ ex : intro1 ] show that maxwell s equations ( [ eq : intro - maxwell ] ) in vacuum ( @xmath6 ) lead to plane - wave solutions for the electromagnetic field propagating at a speed @xmath7 and such that @xmath2 and @xmath3 are always perpendicular to each other and to the propagation direction . what is the relation between the _ intensity _ of the wave ( defined as @xmath8 ) and the energy content of the fields ? in simple terms , a _
polarizer _ is a filter that only allows the transmission of light waves which have a specific polarization angle .
if we let unpolarized light go through an ideal polarizer , the intensity @xmath9 of the transmitted wave will be found to be half the intensity @xmath10 of the incident wave : @xmath11 ( malus law ) . after passing through the polarizer ,
the light is said to be linearly polarized .
let us now suppose that we have a beam of linearly polarized light and we let it go through a second polarizer , such that its polarization axis has been rotated by an angle @xmath12 with respect to the axis of the first polarizer .
then , experiment tells us that the intensity of the transmitted light will be @xmath13 .
this is all fine if we are working with classical electromagnetic waves , which are described by maxwell s equations . in this case
we just need to consider the projection of the field onto the direction singled out by the polarization axis .
but we know that light is actually made of photons and , if their number is small , we are led to regard the @xmath14 term as a kind of `` expectation value '' . in a very influential paper
@xcite , einstein , podolski and rosen presented a criticism of quantum theory in what is now known as the _ epr paradox_. this led to the development of alternative , so - called
hidden variable " theories that aimed at explaining physical phenomena using classical probability models .
it was only until bell proved his famous inequalities , and aspect s experiments proved the former were violated , that the controversy could be resolved , showing that quantum theory provided the correct description of the phenomena .
an experimental set - up , of the type studied by aspect , consists of a source ( ca atom ) located in the middle that emits simultaneously a pair of photons .
one of them goes to the right , the other to the left .
there are two detectors , one at each extreme .
there is also a polarizer in - between each detector and the source , as depicted in fig .
[ fig : intro - ca - atom ] . , @xmath15 ) .
there is a polarizer between the ca atom and each of the detectors . ]
let us now consider the following two propositions : + @xmath16``left photon passes through ( is detected by @xmath17 ) when polarizer s angle is @xmath12 '' .
+ @xmath18``right photon passes through ( is detected by @xmath15 ) when polarizer s angle is @xmath19 '' .
+ then , what we learn from experiment is that the _ joint probability _ for both photons passing through the polarizers , thus being detected , is @xmath20 before exploring why classical probability is in conflict with this result , let us recall how the principles of quantum mechanics allow us to predict it .
let @xmath21 denote a hilbert space which , for simplicity , will be considered to be finite dimensional . in quantum mechanics ,
probability distributions are obtained from state vectors @xmath22 or , more generally , by density matrices @xmath4 , that is , self - adjoint positive operators of trace one .
on the other hand , observables are described by self - adjoint operators .
let @xmath23 be such an observable .
let @xmath24 denote its spectrum .
we can then consider its spectral decomposition @xmath25 if the state of the system is given by a density matrix @xmath4 , the _ expectation value _ of @xmath23 is defined as @xmath26 in the case of a _ pure _ state , the density matrix is a rank - one projector , of the form @xmath27 , @xmath28 , and so ( [ eq : intro < a > ] ) reduces to @xmath29 . in this way
we obtain a probability distribution over @xmath30 , with @xmath31 in fact , it follows from ( [ eq : intro - speca ] ) and ( [ eq : intro < a > ] ) that @xmath32 and @xmath33 .
let us use this to give a mathematical description of the experiment described above .
the polarization state of a photon can be described using a 2-dimensional hilbert space .
let @xmath34 denote an orthonormal basis , that can be used to describe , say , horizontal and vertical polarization states .
since we are considering a system consisting of two photons , the hilbert space of the system can be taken to be @xmath35 ( as the photons are supposed to be far from each other at the time of detection , the symmetrization postulate can be ignored ) .
the 2-photon state , as produced by the emission from an excited state of the ca atom , can be described by the following state vector : @xmath36 we can use @xmath37 , defined as the _ projector _ onto @xmath38 .
since we have two polarizers , we introduce the following projectors ( _ cf . _ fig .
[ fig : intro - ca - atom ] ) : @xmath39 it is important to remark that these two operators correspond to compatible observables , in the sense that the measurement of one of them does not affect the result of the other , _ i.e. _ , they are commuting observables : @xmath40=0.\ ] ] show that @xmath41 also show that , for the _ joint _ measurement of the two polarization states , one obtains @xmath42 the previous exercise shows that the predictions of quantum mechanics for this type of experiment are in accordance with what is actually measured in the laboratory . in spite of its extreme simplicity , there is an intriguing feature of this result : by appropriately arranging the polarization angles @xmath19 and @xmath12 , we can obtain a _ total anticorrelation _ for the joint measurements . as we will see , this is due to the fact that the state ( [ eq : intro-2-photon - state ] ) is an _ entangled state_. that these kind of correlations can not be obtained from a ( local , realistic ) classical theory is at the core of the original epr controversy . for an interesting discussion of these issues in the context of an actual experimental situation , we recommend @xcite .
our immediate aim will therefore be to understand where exactly classical probability fails at describing the results of such experiments .
for this purpose we will assume an approach to ( classical ) probability based on propositions and degrees of plausibility , following jaynes @xcite . in this setting ,
the objects to be considered are the following : * a set of propositions : @xmath43 , each one of which can take on ( only ) two values : true or false . * logical operations , that can be performed on the set of propositions : * * conjunction , or logical product ( and ) : @xmath44 .
it is true if and only if both @xmath23 and @xmath3 true . * * disjunction , or logical sum ( or ) : @xmath45 .
it is true if at least one of them true .
* * negation ( not ) : @xmath46 , with opposite truth value as @xmath23 .
these logical operations are assumed to satisfy the defining rules of a _
boolean algebra _ : * idempotency : @xmath47 * commutativity : @xmath48 * associativity : * * @xmath49 * * @xmath50 * distributivity : * * @xmath51 * * @xmath52 * duality ( de morgan s laws ) : @xmath53 show that the proposition `` @xmath54 '' is equivalent to `` @xmath55 '' .
show that ( d2 ) follows from the other rules .
now , given a proposition @xmath23 , a probability ( or `` degree of plausibility '' , _ cf .
_ @xcite ) @xmath56 can be assigned to it , under the following basic assumptions : * let @xmath57 denote the probability , or degree of plausibility , of a given proposition @xmath23 .
then we assume that : * * @xmath58 , for any logical function @xmath59 of the propositions @xmath60 * * using the notation @xmath61 for the conditional probability that @xmath23 is true , given @xmath3 is true , we assume that @xmath62 whenever @xmath63 more plausible than @xmath64 .
* given a proposition @xmath65 , let us suppose that we improve our state of knowledge , obtaining a new proposition @xmath66 .
if as a result @xmath23 becomes _ more _ plausible , _ i.e. _ @xmath67 , but @xmath68 , then @xmath69 should hold ( this is dubbed the _ common sense _ assumption by jaynes @xcite )
. from i and ii above we may obtain , under very general assumptions ( like consistency ) , the following two basic rules : * product rule:@xmath70 * sum rule : @xmath71 the reader is invited to consult jaynes for a comprehensive exposition of this point of view , including many illustrative derivations and examples .
now , in order to return to our experiment , let us define the following `` coincidence '' function @xmath59 , for two given propositions @xmath23 and @xmath3 : @xmath72 given 4 propositions @xmath73 , use the product and sum rules to show that the following proposition is true : @xmath74 since @xmath75 whenever @xmath54 holds ( product rule ) we obtain , from the previous exercise , + `` bell s inequality '' : @xmath76 referring back to figure [ fig : intro - ca - atom ] , let us consider the following propositions ( @xmath77 ) : + @xmath78 `` left photon passes through ( is detected by @xmath17 ) when polarizer s angle is @xmath79 '' .
+ @xmath80 `` right photon passes through ( is detected by @xmath15 ) when polarizer s angle is @xmath81 '' .
the inequality ( [ eq : intro - bell - chsh ] ) should hold true for all choices of @xmath82 . but
this then implies : @xmath83 for the choice @xmath84 , this means : @xmath85 @xmath86 @xmath87 this contradiction means that the assumptions we have considered above ( which are at the basis of classical probability ) do not apply in the quantum realm .
so , definitely , quantum theory leads to a very different type of probability theory .
in the next section we will explore some of the more notorious differences between classical and quantum probabilities .
until now we have avoided any mention of propositions in terms of set theory , which is the basis of the kolmogorov axiomatics . in the finite dimensional case , at least , they turn out to give equivalent structures .
let us then consider the main properties of a ( classical ) probability theory , formulated in terms of set theory and , for simplicity , in the finite dimensional context .
let us consider a finite set @xmath88 , regarded here as the _ sample space_. the _ event space _
@xmath89 is a certain collection of subsets of @xmath90 , that must contain the empty set and be closed under complements and unions .
the set of events forms a boolean algebra , under the standard set theoretic operations : * and : @xmath91 , * or : @xmath92 , * not : @xmath93 . a probability distribution
is then defined as a map @xmath94\subset \mathds r$ ] , such that * @xmath95 @xmath96 , * @xmath97 for @xmath98 pairwise disjoint .
notice that the space of probability distributions is a convex set , for if @xmath99 and @xmath100 are two probability distributions , then @xmath101 is again a probability distribution , provided @xmath102 .
but not only is the space of probability distributions a convex set , it is a special kind of convex set : a _ simplex_. to see this , all we have to do is to consider the following `` extremal '' distributions , @xmath103 , defined by @xmath104 it is clear that any @xmath105 can be written as a _
unique _ convex combination of these extremal distributions , as we have @xmath106 with @xmath107 .
a random variable ( or `` observable '' ) is a function @xmath108 .
we call the space of all such functions @xmath109 .
notice that @xmath109 forms a commutative algebra .
summarizing , in the finite - dimensional case we have the following structure : * the event space ( @xmath110 ) forms a boolean algebra . *
the space of probability distributions is _ convex _ and furthermore has the structure of a simplex : every probability distribution @xmath105 can be uniquely written as convex combination of extremal " distributions . *
the space of observables ( random variables ) has the structure of a commutative algebra . in cases where the sample space is
not a finite set ( like in classical mechanics ) we need a suitable generalization of the above definitions .
this is afforded by measure theory , in the following way : sample and event spaces are now replaced by a measurable space @xmath111 , where @xmath89 is a suitable @xmath112-algebra of @xmath90 .
probability distributions are then defined as _ normalized , positive measures_. in particular , we can now handle countable additivity : @xmath113 finally , the space of observables is still a commutative algebra , for the product and sum of two measurable functions is a measurable function .
a similar remark applies to the convex structure of the space of probability distributions .
how do these structures show up in classical physics ?
in classical mechanics , for example , the dynamics of a system can be described in terms of canonical variables `` position '' ( @xmath114 ) and `` momentum '' ( @xmath115 ) , that give rise to the _ phase space _ of the system .
it follows from hamilton s variational principle that if @xmath116 is the hamiltonian of the system , the canonical variables will evolve along solutions to hamilton s equations : @xmath117 usually the phase space is the cotangent bundle @xmath118 of some configuration space @xmath119 .
the canonical variables @xmath120 are then local coordinates on @xmath118 and the _ observables _ are smooth functions on phase space : @xmath121 .
since time evolution is given by hamilton s equations or , at the level of observables by @xmath122 , with @xmath123 denoting the poisson bracket , one would think that there is no much space for a probabilistic model here since , given a set of initial conditions , by solving the equations of motion we are able , in principle , to predict the position and momenta of all the particles .
but , as we learn from classical statistical physics , for @xmath124 ( @xmath125 ) we certainly need a statistical approach ! therefore we are forced to introduce probability distributions , expressed in terms of probability densities @xmath126 , so that the average value of an observable @xmath127 is given by @xmath128 where @xmath129 is the liouville measure on phase space .
for instance , for the _ canonical ensemble _
( used to describe a subsystem embedded in a thermal bath at temperature @xmath130 ) , we have @xmath131 .
now we can return to the case of a few point particles and interpret the state of the system at a given time , usually defined as just a point @xmath132 in phase space , as an extremal probability distribution ( a `` pure state '' ) , for which the probability density is just a dirac delta distribution : @xmath133 . in any case
, we see that classical physics can be regarded as a special type of probability theory , where the three properties of classical probability highlighted above still hold , _
i.e. _ , every classical system of point particles can be understood as a probability theory , in the kolmogorov sense . what about quantum mechanics ?
as discussed above , probability distributions are obtained from state vectors @xmath134 in a hilbert space @xmath21 , or more generally from density matrices , @xmath4 , whereas observables are described by self - adjoint operators .
consider the spectral decomposition of a self - adjoint operator @xmath23 , as in ( [ eq : intro - speca ] ) . assuming a non - degenerate , discrete spectrum
, we notice that the projectors @xmath135 , having as spectrum the set @xmath136 , can be regarded as `` indicators of events '' .
now , there is a correspondence between projections @xmath137 and subspaces @xmath138 .
the partial order that we naturally obtain by inclusion then gives rise to the structure of an `` orthocomplemented lattice of proposition '' , where the corresponding operations are defined as @xmath139 an important feature of this system is that it _ does not _ give rise to a boolean algebra structure .
the reason for this is that the orthogonal complement is not the only possibility for a complement in this lattice .
this in turn implies the breakdown of the distributive law , which is part of the definition of a boolean algebra . as in the classical case ,
the state space ( here the space of density matrices ) is a convex space .
but , in contrast to the classical case , this space is not a simplex .
a consequence of this is that the representation of a density matrix as a convex sum of pure states is ( highly ) non - unique . finally , in quantum theory
the space of observables forms a non - commutative algebra , in contrast to the classical case , where the algebra is commutative .
it is now clear that , in principle , both approaches ( classical and quantum ) can be considered in order to describe physical phenomena in probabilistic terms .
in fact , epr - like arguments are in favor of a `` local - realistic '' point of view according to which even quantum phenomena should be explained in terms of classical probability . that this is not the case is proved by bell s inequalities and their violation , experimentally verified by aspect in the 80 s .
so quantum theory can be regarded as a kind of `` non - commutative '' probability theory .
one of the points of these lectures is that , when formulated in the language of operator algebras , both quantum and classical physics can be described in a unified way .
we will also take advantage of this formulation to discuss a similar phenomenon , occurring in topology and geometry : in non - commutative geometry @xcite the generalization of topological / geometric notions to the non - commutative setting has been mainly achieved by first expressing them in algebraic terms and then realizing that dropping the commutativity assumption allows for vast generalizations .
let @xmath21 be a separable hilbert space . recall that
, given a linear operator @xmath140 , @xmath130 is said to be bounded if there is some @xmath141 such that @xmath142 for all @xmath143 in @xmath21 , where the norm is the one induced by the inner product : @xmath144 . if @xmath130 is a bounded operator , we define its norm as follows : @xmath145{\mbox{sup } } { } \frac{\|t(x)\|}{\|x\|}.\ ] ] one then checks that for @xmath130 and @xmath146 bounded the inequalities @xmath147 are satisfied . from the completeness of @xmath21
it follows that the space @xmath148 is a complete normed space .
it is also an algebra and has an involution `` @xmath149 '' given by the adjoint : @xmath150 .
we thus may call @xmath151 the `` @xmath149-algebra of bounded operators on @xmath21 '' show that for @xmath152 we have : @xmath153 we now abstract these notions and make them independent of any underlying hilbert space . as we will see below
, a lot will be gained from this , since we will then be able to study _ representations _ of the ( abstract ) operator algebras , and the equivalence / inequivalence of these representations will have a deep physical meaning .
a _ banach space _ is a normed vector space @xmath154 which is complete with respect to the ( metric induced by ) @xmath155 .
a _ banach algebra _ is a banach space @xmath156 which is also an algebra , with the property that @xmath157 ( which in turn implies that multiplication is a continuous operation ) .
an _ involution _ on a ( complex ) algebra @xmath158 is a map @xmath159 such that , for any @xmath160 and @xmath161 : * @xmath162 * @xmath163 * @xmath164 notice that this is , basically , an abstraction of the adjoint operation on @xmath151 . a _
@xmath0-algebra _ is a banach @xmath149-algebra @xmath165 with the fundamental property @xmath166 [ example : hausdorff ] let @xmath167 be a compact , hausdorff topological space and set @xmath168 , the space of continuous complex functions on @xmath167 and , for @xmath169 , set @xmath170 , and @xmath171{\mbox{sup } } { } |f(x)|$ ] . then @xmath172 is a @xmath0-algebra . let @xmath21 be a hilbert space .
then , as we expect , @xmath173 is a @xmath0-algebra , where the norm is the operator norm and the involution is given by the adjoint operation .
later we will see that the list of examples of @xmath0-algebras is basically exhausted by the previous two examples , a remarkable fact . a motivation to work with algebras of bounded operators
comes from physics since , as can be shown , the _ canonical commutation relations _
( ccr ) @xmath174 = i\hbar \mathds 1\ ] ] can not be implemented by means of bounded operators @xmath175 and @xmath176 .
then , although these commutation relations have a very clear meaning from the physical point of view , mathematically they correspond to _ unbounded _ operators and hence issues like self - adjointness , domains , etc .
come into play , which make them more difficult to deal with .
one can , nevertheless , replace the ccr by their exponentiated ( or weyl ) form , as follows .
first we define operators @xmath177 and @xmath178 , for @xmath179 , acting on wave functions as follows : @xmath180 by stone s theorem , it follows that @xmath181 and @xmath182 .
[ ex : weyl - ccr ] show that the operators @xmath177 and @xmath178 satisfy the following commutation relations : @xmath183 as we will see later on , these operators give rise to a @xmath0-algebra .
in fact , we will see that to any symplectic vector space we can canonically associate a @xmath0-algebra ( its weyl @xmath0-algebra ) .
the unitary representations of these algebras , in the case of infinite dimensional symplectic vector spaces , play a prominent role in the study of quantum field theory on curved spacetimes . consider a ( complex , involutive , with unit @xmath184 ) algebra @xmath158 generated by elements @xmath185 , subject to the following _ canonical anticommutation relations ( car ) _ : @xmath186 if we want to define a @xmath0 norm on this algebra , it has to be such that the @xmath0-property ( [ eq : basics - cstar - ppty ] ) is satisfied . but from ( [ eq : basics - car1 ] ) we obtain @xmath187 , which along with ( [ eq : basics - cstar - ppty ] ) implies @xmath188 .
the only option we have , then , is to define @xmath189 for every @xmath190 .
[ ex : car - algebra - finite - dim ] generalize the previous example to the case of a finite dimensional hilbert space @xmath191 , in the following way .
let @xmath192 denote the inner product .
for any pair of vectors , @xmath193 , consider generators @xmath194 satisfying the following relations ( car ) : @xmath195 where @xmath196
. how must @xmath197 be defined in this case ? the ( mathematical ) notion of _ state _ for a @xmath0-algebra is also very close to the notion of quantum state . to appreciate this ,
consider a physical system modeled by the algebra @xmath198 of bounded operators on a hilbert space @xmath21 .
as mentioned in the previous section , in physics we distinguish between two kinds of states : * pure states : these are described by normalized vectors @xmath28 ( actually , by rays on hilbert space ) .
if @xmath23 is an observable , then its expectation value is defined as @xmath199 * mixed states : they are described by operators @xmath200 such that @xmath201 , @xmath202 .
the expectation value is given , in this case , by @xmath203 show that the condition for a mixed state ( described by a density matrix @xmath4 ) to be pure is @xmath204 .
thus , all states can be described by density matrices .
the space of all density matrices is naturally a convex space .
the extremal elements then turn out to be the pure ones .
check this assertion in the simple case @xmath205 .
we therefore see that a state can be regarded as a mapping @xmath206 from the observable algebra to the complex numbers .
it must be possible , then , to express the properties @xmath207 and @xmath202 only in terms of this mapping .
this leads us to the general definition of state for a @xmath0-algebra .
[ def : basics - state ] let @xmath158 denote a ( unital ) @xmath0-algebra , with unit @xmath184 . a positive linear functional @xmath208 such that @xmath209 is called a _ state_. we will denote the set of all states on @xmath158 with @xmath210
. in this context , positivity of @xmath211 means that @xmath212 for all @xmath213 .
if @xmath158 is not unital , we can replace the normalization condition by the condition @xmath214 .
for unital algebras , both definitions coincide . having in mind the example of @xmath151
, it then makes sense to have a notion of spectrum for an abstract @xmath0-algebra .
this leads , in turn , to the spectral theorem for @xmath0-algebras @xcite .
let @xmath215 be an element of a unital @xmath0-algebra @xmath158 . then
its _ spectrum _ is defined as the set @xmath216 let @xmath215 be an element of a unital @xmath0-algebra @xmath158
. then its _ spectral radius _ is defined as @xmath217 show that the space @xmath218 of @xmath219 matrices with complex entries is a @xmath0-algebra with respect to the norm defined as @xmath220 where @xmath221 is the eigenvalue of @xmath222 with the largest absolute value .
show also that this norm coincides with the operator norm .
the previous exercise illustrates the remarkable fact that , for a @xmath0-algebra ( @xmath223 ) , the norm of any element normal @xmath213 coincides with its spectral radius : @xmath224 this is an important fact , as it links topological properties of the algebra with algebraic ones .
in particular , it implies that the norm in a @xmath0-algebra is _
unique_. the following two properties will be very useful : [ ex : cauchy - schwarz ] let @xmath225 be a positive linear functional on a @xmath0-algebra .
prove that * @xmath226 . * @xmath227 ( `` cauchy - schwarz inequality '' ) hint : consider @xmath228 as a quadratic form on @xmath229 .
as an alternative , you may try deriving these relations just by thinking of @xmath230 as `` @xmath231 '' .
another consequence of positivity is continuity : let @xmath232 be a positive linear functional on a unital @xmath0-algebra @xmath158 . then @xmath232 is a continuous linear functional , and @xmath233 .
if @xmath215 is positive ( @xmath234 and @xmath235 ) then we have , from the spectral radius formula , that @xmath236 $ ] .
this means that @xmath237 for all @xmath238 .
this can be restated as follows .
define @xmath239 then @xmath240 and @xmath241 .
but then it follows ( from the continuous functional calculus ) that @xmath242 , so that @xmath243 or , in other words , that @xmath244 is a positive operator .
since @xmath232 is positive and linear , this implies @xmath245 .
now , for an arbitrary @xmath246 , apply the cauchy - schwarz inequality from exercise [ ex : cauchy - schwarz ] with @xmath247 and @xmath248 , and use the @xmath0 property of the norm to obtain @xmath249 .
it follows that @xmath232 is continuous , with @xmath250 .
but we also have @xmath251 so that @xmath233 .
thus , for unital @xmath0-algebras we conclude that the set of states over @xmath158 is a convex subset of the continuous dual @xmath252 of @xmath158 .
we gather from all this that it makes sense to formulate quantum physics in terms of @xmath0-algebras .
the general philosophy will be to describe a given physical system in terms of its `` algebra of observables '' , which will be here taken to be a @xmath0-algebra @xmath253
. then we will consider the following two `` dual '' notions : * a _ quantum state _ will be defined as a state ( in the sense of definition [ def : basics - state ] ) on the algebra @xmath158 .
* an _ observable _ will be an element of @xmath158 .
furthermore , the `` pairing '' between an observable @xmath215 and a state @xmath211 will be given the physical interpretation of an _ expectation value _ , and we will write @xmath254 regarding the distinction between pure and mixed states , we can now call a state _ pure _ if it can not be written as a convex combination of other states ; otherwise we will call it a _ mixed _ state . before discussing physical applications of the notions introduced in the previous section
, we will take the opportunity to briefly review the characterization of @xmath0-algebras due to gelfand , naimark and segal .
the characterization of commutative @xmath0-algebras ( the gelfand - naimark theorem ) is of fundamental importance , in particular because it leads to the notion of a _ `` noncommutative topological space '' _ @xcite . also , the so - called gns - construction discussed below is relevant not only because of its role in the characterization problem for noncommutative @xmath0-algebras , but also because of its striking consequences for the study of quantum systems ( equilibrium states , symmetry breaking , inequivalent vacua , among others ) @xcite . a _ character _ of a @xmath0-algebra @xmath158 is a @xmath149-homomorphism @xmath255 .
let us denote with @xmath256 the set of all characters of @xmath158 .
now , it is easy to see that if @xmath257 is a character , then @xmath258 .
this , together with the spectral radius formula ( [ eq : spectral - radius - formula ] ) , implies that @xmath259 , so that @xmath260 , where @xmath261 is the unit ball in the dual space . since ( by the banach - alaoglu theorem @xcite ) this ball is compact in the weak-@xmath149 topology ( _ i.e. _ the one induced by the family of seminorms on @xmath252 : @xmath262 , @xmath246 ) then @xmath256 becomes a _
compact topological space _ in the subspace topology inherited from @xmath252 .
this is quite interesting in view of example [ example : hausdorff ] as this is providing the converse statement .
in fact , for a commutative , unital @xmath0-algebra @xmath158 , we define the _ gelfand transform _ @xmath263 as the map @xmath264 where @xmath265 .
[ thm : gelfand - naimark]the gelfand transform is an isometric @xmath149-isomorphism @xmath266 rather than reviewing the proof of this theorem , we will consider some basic examples that should provide a good intuition about the correspondence between commutative @xmath0-algebras and ( locally compact , hausdorff ) topological spaces .
consider the unital , commutative @xmath0-algebra @xmath158 generated by a unitary element @xmath267 ( _ i.e. _ @xmath268 ) , with norm fixed by the condition @xcite that @xmath269 for all @xmath270 .
theorem [ thm : gelfand - naimark ] states that this algebra must be the function algebra of a topological space which , as a set , is precisely the set @xmath256 of all characters of @xmath271 .
since the algebra is generated by @xmath267 , a character @xmath257 is fixed by its value on @xmath267 . but @xmath272 , and so every character is of the form @xmath273 , _
i.e. _ @xmath274 . in order to see that actually @xmath275 , we proceed as in @xcite by noticing that from example [ example : hausdorff ] and from the uniqueness of the norm , it follows that @xmath276 , so that we must have @xmath277{\mbox{sup } } { } |1+e^{-i\alpha}\mu(u)|=2.\ ] ] but this can only happen if for every @xmath270 there is a character @xmath257 with @xmath278 .
the 2-sphere @xmath279 can also be easily characterized in terms of its algebra of continuous functions . in this case , the algebraic structure is well - known , as @xmath280 is generated by the spherical harmonics @xmath281 , their product being given as in the usual clebsch - gordan decomposition .
more details on how @xmath279 can be obtained as the character space of such an algebra ( this involves an appropriate definition of the norm , as in the previous example ) can be found in @xcite .
take @xmath282 as above ( with the sup norm ) and consider the subalgebra @xmath283 consisting of all even functions on @xmath158 , that is , functions @xmath59 such that @xmath284 .
then @xmath285 .
this fact has been exploited in studies of the spin - statistics connection in quantum mechanics @xcite .
the isomorphism provided by theorem [ thm : gelfand - naimark ] is actually an equivalence between the categories of commutative @xmath0-algebras and locally compact , hausdorff topological spaces .
this means , among other things , that all information regarding the topology of @xmath167 is encoded in the algebra @xmath286 : an open set in @xmath167 can be equivalently described in terms of a ideal in @xmath286 , closed sets are described by quotient algebras , metrizability of @xmath167 amounts to separability of @xmath286 , and so on @xcite .
this is the reason why the gelfand - naimark theorem is considered one of the main sources for the development of noncommutative geometry @xcite .
now we turn to the gns ( gelfand - naimark - segal ) construction .
let @xmath158 be a @xmath0-algebra , and @xmath211 a state thereon .
the gns construction furnishes a representation of @xmath158 on some ( @xmath211-dependent ) hilbert space .
the basic idea is to use multiplication in @xmath158 in order to obtain a linear action of @xmath158 on a vector space .
so we begin by regarding @xmath158 as a vector space @xmath287 ( _ i.e. _ we just `` forget '' multiplication ) . even though the underlying spaces are equal , it will be convenient to distinguish the _ algebra _ , with elements @xmath213 , from the underlying _ vector space _
, with elements @xmath288 .
recalling the cauchy - schwarz inequality ( exercise [ ex : cauchy - schwarz ] ) , we realize that it makes sense to introduce the following sesquilinear form on @xmath289 : @xmath290 this is `` almost '' an inner product , as it may happen that @xmath291 for some @xmath292 . but ignoring that fact for a moment , we notice that the product on @xmath158 can be used to make @xmath158 act as a linear operator on @xmath289 : @xmath293 now , in order to have a hilbert space representation , we have to `` fix '' the problem with ( [ eq : w(a*b ) ] ) .
this is done as follows .
define @xmath294 , and show that it is a closed left - ideal of @xmath158 .
we therefore obtain a hilbert space @xmath295 given by the completion of @xmath296 with respect to the inner product @xmath297 | [ b]\rangle_\omega:= \omega ( a^*b)$ ] , where we use @xmath298\rangle$ ] to denote the equivalence class of @xmath215 in ( the completion of ) @xmath299 . with this
we have obtained a @xmath149-representation of @xmath158 by bounded operators acting on @xmath295 : @xmath300 where @xmath301\rangle:= |[ab]\rangle$ ] .
obtain the following inequality : @xmath302 although the representation @xmath303 is a @xmath149-homomorphism , it is not isometric .
but it can be shown that for every @xmath213 there is a state @xmath211 such that @xmath304 . then ,
if we consider the direct sum @xmath305 where in each sequence @xmath306 only countable many elements are different from zero .
the representations @xmath303 then give rise to a representation @xmath307 on @xmath21 for which @xmath308 .
the details of this construction can be found _
e.g. _ in @xcite .
the importance of this result is that _ any _ @xmath0-algebra is isometrically isomorphic to the @xmath0-algebra of bounded operators in some hilbert space . in this section
, our aim will be to study certain quantum correlations that arise as a result of _ entanglement_. this is a vast subject , and here we will only consider elementary examples , corresponding to bipartite systems .
although very simple from the mathematical point of view , these examples already contain the essence that will allow us to distinguish between classical and genuinely quantum correlations .
we will also discuss a recent application of the algebraic formalism to the study of entanglement for systems of identical particles @xcite , a topic of current interest for quantum information , condensed matter , atomic physics , and quantum optics .
we thus start by considering a composite system that is the result of coupling two subsystems 1 and 2 .
we will first consider quantum systems described directly in terms of hilbert spaces .
but , as will become clear , the algebraic approach based only on observable algebras and states will allow us to reformulate these ideas in more generality .
this will be important for the applications to systems of identical particles .
let @xmath309 and @xmath310 denote the corresponding ( finite dimensional , for simplicity ) hilbert spaces , and assume that the subsystems are described in terms of observable algebras @xmath311 , @xmath312 .
subsystems 1 and 2 are then coupled to form a composite system that will be described by @xmath313 .
let now @xmath314 be a state of the composite system .
consider `` partial '' , or `` local '' measurements performed on each subsystem .
this leads us to consider the restriction of @xmath211 to @xmath315 , @xmath312 : @xmath316 in the present context of a bipartite composite system , if @xmath211 is a pure state on @xmath158 , we say it is a _ separable _
state if @xmath317 for all @xmath318 , @xmath319 .
a ( pure ) state that is not separable in the above sense is called an _ entangled _ state .
let @xmath320 , with basis @xmath321 and @xmath322 , with basis @xmath323 .
let @xmath324 and @xmath325 .
the composite system is then described by the matrix algebra @xmath326 .
pick now a vector state @xmath327 and set @xmath328 then , the following three conditions are equivalent : * the vector state @xmath329 is of the form @xmath330 .
* @xmath331 , ( @xmath312 ) where @xmath332 stands for the _ von neumann entropy _ of the reduced density matrix @xmath333 ( see definitions below ) . *
the state @xmath334 is separable . to see where these equivalences come from ,
let us expand @xmath329 as @xmath335 then @xmath336 , so we may perform a singular value decomposition and write it in the form @xmath337 , with @xmath338 a diagonal matrix in @xmath339 , @xmath340 and @xmath341 .
the elements of @xmath338 are @xmath342 , where the @xmath343 , called the schmidt coefficients " , are the eigenvalues of @xmath222 .
if we define vectors @xmath344 from this expression , it is clear that ( @xmath345 ) holds precisely when there is only one non - vanishing schmidt coefficient .
furthermore , from @xmath346 we obtain @xmath347 , with @xmath348 we can , therefore , regard @xmath349 as a probability distribution , and compute its ( shannon ) entropy , defined as @xmath350 notice that this function vanishes precisely when @xmath329 is a separable state .
we can relate it to the von neumann entropy of the restricted states @xmath351 as follows .
[ ex : partial - trace ] consider the restriction @xmath351 of @xmath334 to @xmath352 ( @xmath312 ) .
find density matrices @xmath333 ( `` reduced density matrices '' ) acting on @xmath353 and such that @xmath354 and @xmath355 .
the _ von neumann entropy _ of a density matrix @xmath4 is defined as @xmath356 show that @xmath357 and @xmath358 coincide and are exactly equal to @xmath359 from ( [ eq : h(lambda ) ] ) .
we therefore consider the quantity @xmath332 ( @xmath312 ) as a _ measure _ of the `` amount of entanglement '' of the state @xmath334 . in order to really appreciate the meaning of this assertion
, we have to understand the ( very surprising , and interesting ) features of quantum correlations .
this can be considered as one of the starting points of quantum information theory . in quantum information theory
, there are orderly ways to pose and study these type of problems where , for instance , in the ( pure ) bipartite case the von neumann entropy of the reduced density matrix can be obtained as ( basically ) the unique entanglement measure of the state @xmath329 , when entanglement is regarded as a physical resource and defined in operational terms @xcite .
the general problem , that of multipartite entanglement , is much more complicated and involves many mathematically as well as physically interesting problems .
however , our interest here will be restricted to the ( pure ) bipartite case .
this will be enough to illustrate one of our points , which is that the algebraic framework leads to a unified description of physical systems , no matter whether they are classical or quantum . also , the `` detachment '' from an _ a priori _ given hilbert space in the quantum case allows for a more clear understanding of phenomena .
as it turns out , the noncommutativity of the observable algebra can be seen as the main source of ( truly quantum ) correlations .
this fits nicely with the relation between commutative and noncommutative spaces , as hinted above .
the next example illustrates the previous remarks .
let us consider the composition of two classical systems .
by this we mean that the `` coupling '' of two systems described by ( commutative ) observable algebras of the form @xmath360 and @xmath361 is effected by the tensor product @xmath362 .
tensor products of @xmath0-algebras have to be treated carefully , but assuming we have defined the appropriate tensor product , let us assume that @xmath363 ( a detailed discussion about tensor products can be found in @xcite ) .
hence , the composite system will be assumed to be described by the observable algebra @xmath364 .
consider now a state @xmath211 on @xmath158 .
it is not difficult to see that it must be given by a probability distribution @xmath365 in such a way that for @xmath366 , @xmath367 restriction to @xmath368 or @xmath369 in this case leads to marginal distributions @xmath370 and @xmath371 , which in turn define states @xmath372 on @xmath373 . from the above definition we have that @xmath211 is separable if , and only if , @xmath374 but from the properties of the gelfand transform it follows that any pure state on @xmath375 is of the form @xmath376 , _
i.e. _ , they are given by point measures : @xmath377 .
it follows immediately that @xmath378 .
this illustrates the fact that entanglement entails purely quantum correlations , as all pure states of a classical composite ( bipartite ) system are separable .
the previous example was formulated in rather loosely terms .
provide the necessary details to turn it into a rigorous proof that works for locally compact spaces .
[ example : psi_lambda ] on @xmath379 consider the following family of states : @xmath380 and the following types of observables , acting on @xmath381 : @xmath382 with @xmath383 a _ unit _ vector in @xmath384 and @xmath385 ( pauli matrices ) .
straightforward computations give @xmath386 and @xmath387 whereas for @xmath388 we obtain separable states , for @xmath389 we have entangled states . in particular , for @xmath390 we obtain a maximally entangled state ( bell state ) , for which @xmath391 in this case , the marginal distributions give @xmath392 whereas for the joint probabilities we obtain ( @xmath393 ) : @xmath394 in particular , for the choice @xmath395 , we obtain _
total anticorrelation_. what is so special about the previous example ?
consider the clauser - horne - shimony - holt ( chsh ) inequality @xcite : @xmath396 where @xmath397 denotes the expectation value of an observable @xmath398 , and where for the involved observables it is assumed that @xmath399 .
the inequality can be obtained if we assume some underlying ( classical ) probability space @xmath400 , for which the correlations above take the form @xmath401 or , stated in terms of algebras of observables : let @xmath158 be a _ commutative _ @xmath0-algebra , @xmath211 a pure state thereon , and @xmath402 elements of @xmath158 , all of them with norm less or equal to one . show that @xmath403 but , in contrast to this , in example [ example : psi_lambda ] ( with @xmath404 ) we obtain @xmath405 now , it is easy to find an arrangement for the vectors @xmath406 such that the chsh inequality is violated .
the importance of these type of inequalities , of which bell s inequality was the first one , can not be overemphasized , as they have allowed for definitive experimental tests of quantum mechanics .
the chsh inequality is also relevant in the context of hidden - variable models , and related no - go theorems @xcite . in the next section
we will consider the problem of entanglement for systems of identical particles , for which the tools developed in the last sections will prove very useful .
when combined , two of the most intriguing features of quantum theory -the intrinsic indistinguishability of identical particles and quantum entanglement- lead to formidable conceptual issues that have been addressed for years , and for which no generally accepted framework exists . the main source of problems comes from the fact that entanglement and related concepts have been studied mainly in cases where a subsystem decomposition coincides with a tensor product decomposition of the hilbert space .
but precisely in the case of identical particles , the hilbert space is the antisymetric / symmetric subspace obtained from the action of the permutation group on a tensor product space .
so precisely in this case decomposition into subsystems does not correspond to tensor product decomposition .
this makes the use of partial trace a doubtful operation , leading to a clash with standard notions of entanglement . in @xcite , we have shown how , using the representation theory of operator algebras , it is possible to obtain a generalized notion of entanglement .
the universality of the approach , which can be applied to particles obeying any kind of statistics including bosons , fermions , parafermions and particles obeying braid statistics , provides a unifying view of entanglement , with many potential applications .
the key idea is that , since subsystems can always be described in terms of subalgebras , the use of partial trace will be superseded by the more general notion of restriction of a quantum state @xmath211 to a subalgebra .
then one can make use of the gns construction in order to find a hilbert space representation @xmath303 , as above .
[ ex : entropy - irreducibility ] prove that the gns representation @xmath303 is irreducible precisely when the state @xmath211 is pure . from this exercise
, it follows that the condition for irreducibility of the representation @xmath303 is precisely that the von neumman entropy of the state @xmath211 vanishes and so , given a subsystem described in terms of a subalgebra @xmath407 , one may consider the restriction of the state @xmath211 to @xmath408 .
this is a generalization of the notion of partial trace .
hence , a generalized notion of entanglement emerges , based on the von neumann entropy of the restricted state .
the properties of this entanglement measure are very closely tied to the gns - representation of the subalgebra and the restricted state . in other words
, one finds that entanglement depends _ both _ on the state and the subsystem of the full system .
this formalism , then , lends itself to the study of problems where partial trace loses its meaning , as is the case with systems of identical particles .
let us consider a few illustrative examples .
[ example:2x2matrix ] consider the algebra @xmath409 of @xmath410 matrices with complex entries , and let @xmath411 be a state on @xmath158 .
of course , this can only happen if @xmath102 .
let us then see how the gns representation is constructed , and how we can associate an entropy to the state @xmath412 . for this purpose ,
let us consider the matrix units @xmath413 @xmath414 .
they generate the algebra , and provide a basis for the underlying vector space @xmath289 .
they also fulfill the relations @xmath415 let us now check whether there will be null vectors .
writing a general element @xmath213 as @xmath416 we compute @xmath417 from this expression we recognize that , if @xmath418 , then @xmath419 , whereas for @xmath420 or @xmath421 we obtain @xmath422 .
in fact , if @xmath229 is different from zero or one , we must have both @xmath423 and @xmath424 .
it follows that @xmath425 . from @xmath426\rangle = |[ab]\rangle\ ] ] and ( [ eq : e_ij ] ) we see that the representation is reducible , with 2 invariant subspaces generated , respectively , by @xmath427\rangle,|[e_{21}]\rangle\rbrace$ ] and @xmath428\rangle,|[e_{22}]\rangle\rbrace$ ] . from exercise [ ex : entropy - irreducibility ]
we conclude that , in this case , the state has to be a mixed state . the cases @xmath420 and @xmath429 lead to a non - trivial null space , and from ( [ eq : null - states-2x2 ] ) we conclude that the dimension of these irreducibles is , in both cases , equal to 2 .
it is left to the reader to obtain a pure state on the algebra @xmath430 such that @xmath412 is the restriction to @xmath431 .
the next question we need to address is that of how to compute the entropy of the reduced state . since our algebras will in general be unital , there is a simple way to acomplish this , namely , given a ( unital ) @xmath0-algebra @xmath158 and a state @xmath211 on it , it is always possible to find a _ density matrix _ @xmath432 acting on the gns space @xmath295 , and such that @xmath433 for all @xmath215 in @xmath158 . from the decomposition of the gns space into irreducibles ,
@xmath434 we obtain _ projectors _
@xmath435 , with @xmath436 . from the definition of the inner product in @xmath295 we have @xmath437 | \pi_\omega ( a ) |
[ \mathds 1_{\mathcal a}]\rangle$ ] .
but then , using an orthonormal basis @xmath438 on @xmath439 , we can write : @xmath440 | \pi_\omega ( a ) | [ \mathds 1_{\mathcal a}]\rangle\nonumber\\ & = & \langle [ \mathds 1_{\mathcal a } ] | \sum_k p^{(k)}\pi_\omega ( a ) | [ \mathds 1_{\mathcal a}]\rangle\nonumber \\ & = & \langle [ \mathds 1_{\mathcal a } ] | \sum_k p^{(k)}\pi_\omega ( a)p^{(k ) } | [ \mathds 1_{\mathcal a}]\rangle\nonumber\\ & = & \langle [ \mathds 1_{\mathcal a } ] | \sum_k p^{(k)}\pi_\omega ( a ) \sum_n |n\rangle \langle n| p^{(k ) } | [ \mathds 1_{\mathcal a}]\rangle\nonumber\\ & = & \tr_{\mathcal h_\omega } ( \rho_\omega \pi_\omega(a)),\nonumber\end{aligned}\ ] ] with @xmath432 given by @xmath441 \rangle \langle [ \mathds 1_{\mathcal a}]| p^{(k)}.\ ] ] consider now a composite , bipartite system for which the hilbert space is of the form @xmath442 . if @xmath334 is a state on the algebra of bounded operators on @xmath443 that is obtained from a vector state @xmath444 , then its restriction to the subalgebra @xmath368 generated by elements of the form @xmath445 gives a state @xmath446 where @xmath447 is the _ reduced _ density matrix , defined through partial trace ( _ cf .
_ exercise [ ex : partial - trace ] ) , @xmath448 we therefore see that , for these kind of systems , partial trace equals restriction ( to some subalgebra ) .
compute de von neumann entropy of the density matrix @xmath449 corresponding to the gns space of example [ example:2x2matrix ] and provide a physical interpretation .
the real usefulness of the algebraic approach becomes apparent only when we consider situations where the hilbert space does not have a simple tensor product structure .
this is what happens , _
e.g. _ , with systems of identical particles .
because of the symmetrization postulate , the hilbert space contains only the symmetric ( for bosons ) or the antisymmetric ( for fermions ) subspaces of the many - particle hilbert space . as a consequence
, one finds that states that from a physical point of view should not _ a priori _ be considered to be entangled , will have reduced density matrices with non - vanishing entanglement entropy @xcite . in approaches like those of @xcite or @xcite , the description of subsystems is given by specifying suitable subalgebras .
then , the restriction of a given state to the subalgebra provides a physically sensible generalization of the notion of partial trace . applying the gns construction to the restricted state
, it is possible to study the entropy emerging from restriction and use it as a generalized measure of entanglement .
let us briefly discuss how this entanglement measure can be computed in concrete cases .
for this purpose , consider a hilbert space @xmath450 , assumed to correspond to the space of 1-particle states of a fermionic system .
the full hilbert space is then the antisymmetric fock space @xmath451 obtained from @xmath452 .
it decomposes as a direct sum of spaces of fixed number of particles .
the @xmath453-particle hilbert space @xmath454 is just the antisymmetrized @xmath453-fold tensor product of @xmath455 .
this is a kind of `` toy model '' for a fermionic quantum field theory , but many important features of a quantum field theory can be seen to appear already at this level .
one of these features is the connection to the representation theory of clifford algebras , as explained in full detail in @xcite .
as is well - known , the clifford algebra of @xmath452 acts naturally on the exterior algebra @xmath456 which , in turn , is related to the fock space construction in the following way .
let @xmath457 denote an orthonormal basis for @xmath452 , and denote with @xmath458 the corresponding annihilation ( creation ) operators ( _ cf .
_ exercise [ ex : car - algebra - finite - dim ] ) . then , there is a vector space isomorphism between @xmath451 and @xmath459 , furnished by the correspondence @xmath460 this correspondence is behind the famous quote by e. nelson : _ ``
first quantization is a mystery , but second quantization is a functor''_. in fact , given a self - adjoint operator @xmath23 on @xmath455 ( that is , a 1-particle observable ) , we obtain ( by functoriality ) an operator @xmath461 acting on fock space , whose restriction to the @xmath453-particle sector @xmath454 is given by @xmath462 taking the correspondence ( [ eq : nelson - correspondence ] ) into account , we obtain the following expression for @xmath461 in terms of creation / annihilation operators , @xmath463 where @xmath464 . in the physics literature , the operator @xmath465 is referred to as the _ second quantization _ of @xmath23 . in the present quantum - mechanical context , where the number of particles is kept fixed
, we want to focus our attention on the operator @xmath466 .
one of the properties of this operator is that it preserves the symmetries of @xmath454 .
furthermore , the map @xmath467 allows us to study subalgebras of 1-particle observables .
the simplest example we can consider in order to illustrate entanglement issues for systems of identical particles is that of just two fermions .
let us then consider , as done in @xcite and in @xcite , a 2-fermion system , where each fermion can be in a linear superposition of 4 basic states which , for the sake of physical interpretation , will be divided into internal and external degrees of freedom .
so we describe 1-particle states in terms of a set of ( fermionic ) creation / annihilation operators @xmath468 , where @xmath215 stands for _ `` left '' _ , @xmath469 for _ `` right '' _ ( the external degrees of freedom ) and @xmath470 for spin up and down ( the internal degrees of freedom ) .
hence , in this case we have @xmath471 and , for the 2-fermion space , @xmath472 . an orthonormal basis for @xmath473
is given by the vectors @xmath474 the two - particle algebra @xmath158 of observables is thus isomorphic to the matrix algebra @xmath475 . for @xmath476 , the corresponding state @xmath477 is given by latexmath:[$\omega_{\theta } ( \alpha)= \langle \psi_{\theta } we choose the subalgebra @xmath408 to be the one generated by @xmath479 , @xmath480 , @xmath481 and @xmath482 .
physically , it corresponds to the subalgebra of one - particle observables corresponding to measurements at the left location . consider the restriction of @xmath477 to @xmath408 and study the gns representation corresponding to this choice . for @xmath483 you should obtain vanishing entropy , in contrast to the result @xmath484 obtained via partial trace for states with slater rank one @xcite .
the hamiltonian for the ( quantum ) ising chain in a transverse field is given by @xmath485 an important aspect when solving the model and studying its solution is the type of boundary conditions considered .
we will be interested in both open as well as periodic boundary conditions .
the first step in the solution of this model is the so - called wigner - jordan transformation , that allows us to express all spin operators in terms of _ fermionic _ creation / annhilation operators : as can be easily checked , the operators defined by @xmath486 where @xmath487 , are fermionic operators . [ ex:1.1 ] check that the operators defined above indeed satisfy the car algebra : @xmath488 , @xmath489 .
the inverse transformation is given by @xmath490 [ ex:1.2 ] check that ( [ eq : wigner - jordan - inverse ] ) is the inverse transformation to ( [ eq : wigner - jordan - def ] ) . with this
we obtain , for the interaction terms of the hamiltonian , @xmath491 and so @xmath492 the takes the form @xmath493 the term containing the external field @xmath229 is diagonal in this basis .
the constant term @xmath494 coming from the last sum is usually disregarded , because its only effect is to shift the energy spectrum .
we will nevertheless keep all terms , in order to be able to compare with numerical solutions in the spin basis , for small values of @xmath495 . expanding all terms in ( [ eq : h - ising2 ] ) and moving all creation operators to the left ,
we obtain ( _ cf .
( 10.14 ) in @xcite ) : @xmath496 the point of using the wigner - jordan transformation for this model is that the hamiltonian becomes `` almost '' diagonal . by this
we mean that it is a sum of local , _ quadratic _ expressions in the creation and annihilation operators .
as we will see , such models can be exactly solved .
we now write @xmath492 in a suggestive matrix notation . arranging all creation and annihilation operators in rows and columns
, we can write @xmath492 as a kind of quadratic form . for example
, we have , for @xmath497 , @xmath498 the expression corresponding to the same term for @xmath499 is then @xmath500 for arbitrary @xmath495 , we may define the following @xmath501 matrices : @xmath502 the matrix elements of these two matrices can be written as follows : @xmath503 hence , the hamiltonian takes the following form : @xmath504 -\lambda n.\ ] ] for several reasons , including the study of edge states , it is instructive to explore the explicit solution of this model for _ open _ boundary conditions .
we start with the ising hamiltonian written in the form ( [ eq : h - ising3 ] ) . any model that can be written as a quadratic form
can be expressed in this way , the `` only '' difference being the explicit form of the matrices @xmath23 and @xmath3 ( notice that , in order for @xmath492 to be hermitian , @xmath23 has to be symmetric and @xmath3 antisymmetric ) . in principle , thus , the method presented below ( following the work of lieb , schultz and mattis @xcite ) can be applied to any such model . for periodic boundary conditions , it is usually more convenient to take into account translation invariance and hence to introduce fourier transformed operators .
but for open boundary conditions , translation invariance is `` broken ''
and then it is a good idea to start right away with a bogoliubov transformation , as explained below .
recall that the operators @xmath505 defined in ( [ eq : wigner - jordan - def ] ) obey fermionic canonical anti - commutation ( car ) relations ( _ cf .
_ exercise [ ex:1.1 ] ): @xmath506 in general , a bogoliubov transformation is a mapping induced by a change of basis on the one - particle hilbert space ( unitary transformation ) , its main effect being to provide a new set of creation / annihilation operators for which the hamiltonian ( [ eq : h - ising3 ] ) becomes diagonal .
consider , then , a new set of operators given by @xmath507 where @xmath508 and @xmath509 are @xmath501 matrices to be chosen so that * the new operators satisfy the same car algebra : @xmath510 * the hamiltonian becomes _ diagonal _ in the new basis : @xmath511 ( with @xmath257 some constant ) .
[ ex:1.3 ] show that the requirement ( [ eq : bogol - car ] ) leads to the following conditions : @xmath512 [ ex:1.4 ] compute the trace of @xmath492 in two different ways , in order to show that the constant term in ( [ eq : h - ising - diagonal ] ) is given by @xmath513 the second condition above , eq .
( [ eq : h - ising - diagonal ] ) , will lead to an eigenvalue problem for @xmath508 and @xmath509 , the solution of which amounts -in principle- to the solution of the full problem .
now we compute the commutator @xmath514 $ ] in two different ways , once using ( [ eq : h - ising3 ] ) and once using ( [ eq : h - ising - diagonal ] ) .
this leads to the following set of equations : @xmath515 in order to solve this eigenvalue problem , it proves convenient to introduce new matrices @xmath516 and @xmath517 , as follows : @xmath518 if we now define for each @xmath453 a vector @xmath519 , the @xmath520 component of which is given by @xmath521 , and similarly for @xmath517 , we find that ( [ eq : ga - hb ] ) can be written as follows : @xmath522 or , equivalently , as @xmath523 with this we have reduced our problem from the diagonalization of a @xmath524 matrix to that of diagonalizing two @xmath501 ones . as mentioned before , this would be a very easy task if we would have chosen periodic boundary conditions .
the reason being that for periodic boundary conditions the matrices @xmath525 are _
toeplitz_. but for open boundary conditions the matrices @xmath23 and @xmath3 are given by ( [ eq : ab - open ] ) , so that @xmath526 and @xmath527 in order to solve ( [ eq:(a - b)(a+b ) ] ) , we propose the following ansatz : @xmath528 with @xmath529 and @xmath530 constants to be determined .
the eigenvalue equation will then give 3 independent equations .
the first one is obtained by equating the @xmath531 component of @xmath532 with @xmath533 , for @xmath534 .
all these choices of @xmath5 yield the same equation ( by enforcing the vanishing of the coefficients of @xmath529 and @xmath530 ) , namely , @xmath535 this is ( almost ) the spectrum of our problem . we still need to find what are the allowed values of the label `` @xmath453 '' .
this is done by considering the two other cases ( @xmath536 and @xmath537 ) , that give a system of equations for @xmath529 and @xmath530 : @xmath538 the non - trivial solution @xmath539 is obtained provided the determinant of this matrix vanishes .
this condition is equivalent to @xmath453 being solution of the following transcendental equation , for which @xmath540 has to be assumed : @xmath541 notice that , since we must have @xmath495 eigenvectors , we expect this equation to have @xmath495 roots .
the behavior of these roots as a function of @xmath229 is quite relevant ; comparison with the periodic chain allows for the recognition of edge states . in the case of periodic boundary conditions
, we extend the sums in ( [ eq : h - ising - transv ] ) to @xmath542 , adopting the convention that @xmath543 ( @xmath544 ) .
this generates a boundary term that couples the first spin operator to the last one : @xmath545 we already know how to write the first two terms of this hamiltonian in terms of creation and annihilation operators .
let us therefore consider the last term : @xmath546 .
here it is convenient to consider the _ parity _ operator , @xmath547 , where @xmath548 is the _ number _ operator . using the identities @xmath549 and ( [ eq : wigner - jordan - inverse ] )
we obtain : @xmath550 with this we can write @xmath492 in terms of fermionic operators : @xmath551 an important fact ( that can be easily checked ) is that the parity operator commutes with @xmath492 : @xmath552 = 0 . \nonumber\ ] ] thus , it is possible to diagonalize the hamiltonian separately in sectors of even and odd numbers of `` particles '' .
the eigenvalues of @xmath547 are of the form @xmath553 , with the plus sign for states with an _ even _ number of particles and the minus sign for states with an _ odd _ number of particles .
we can , as in the previous case , write @xmath492 as a quadratic form in the fermion operators ( _ cf . _ ( [ eq : h - ising3 ] ) ) , the only difference being the explicit form of the matrices @xmath23 and @xmath3 , for which we now get : @xmath554 in contrast to ( [ eq:(a - b)(a+b ) ] ) or ( [ eq:(a+b)(a - b)matrix ] ) , the matrix @xmath555 is now toeplitz : @xmath556 the eigenvalue problem ( [ eq:(a - b)(a+b ) ] ) can again be solved using the ansatz ( [ eq : psi_kl ] ) .
the eigenvalues are again given by ( [ eq : lambda_k ] ) , with the difference that the allowed values of @xmath453 can now be explicitly given ) .
] , as follows from the following exercise . [ ex:1.5 ]
show that , in the even parity sector ( @xmath557 ) , the allowed values of @xmath453 are given by @xmath558 whereas in the odd parity sector ( @xmath559 ) they are given by @xmath560 [ ex:1.6 ] explain how the @xmath453 s can be made to take positive and negative values and how then we obtain an explicit solution for the matrix @xmath516 , of the form @xmath561 it should by now be clear that the solution of the eigenvalue problem is much easier in the periodic case .
we have chosen to discuss the open chain mainly because of the role that boundary effects play in connection to symmetry breaking and also for studies of surface effects @xcite .
now we are going to take advantage of the translational symmetry in the periodic case . as discussed below ,
if instead of applying a bogoliubov transformation right after transforming the spin operators to fermion operators via the jordan - wigner transformation , we perform a fourier transformation as an intermediate step , we quickly obtain a much more elegant solution , that can be easily generalized to treat the limiting case of an infinite chain . here
we will again make use of the fact that the hamiltonian commutes with the parity operator .
looking back at ( [ eq : h - with - bdry - term ] ) , we realize that it is possible to express @xmath492 in a more compact form if we introduce boundary conditions of the fermion operators , depending on the parity sector we are interested in .
in fact , notice that if we choose @xmath562 , then the boundary term can be included in the first sum , just by setting @xmath563 .
but if we want to do the same in the sector @xmath564 , we will need @xmath565 . in view of our previous discussion , it is clear that the choice @xmath562 is more convenient computationally ( the matrices @xmath525 are circulant in that case ) .
but if we are interested in the _ ground state _ , then we must consider the other choice ( @xmath566 ) .
this can be seen if we compare the ground state energies of the fermionic hamiltonian in each sector . as can be easily checked
, the lowest energy eigenvalue indeed comes from the sector @xmath566 .
the difference between them decreases with increasing @xmath495 and eventually disappears in the thermodynamic limit .
it is for this reason that the boundary term is usually disregarded .
we nevertheless feel that the price to pay if we keep all terms is actually very low , and in return we can have complete control over the spectrum and its degeneracies @xcite .
[ ex:1.7 ] making use of the solution of the eigenvalue problem for both periodic and antiperiodic boundary conditions , plot the lowest energy eigenvalues for @xmath567 as functions of @xmath229 and @xmath495 .
we therefore define @xmath568 in this way the hamiltonian becomes @xmath569 notice that the sum now goes over @xmath570 .
so we see that we can describe _ both _ sectors using the _ same _ quadratic form , the choice of parity being now encoded in the boundary conditions for the fermion operators .
it should be remarked that when we consider anti - periodic boundary conditions for the fermions we are still considering periodic boundary conditions for the _ spin chain_. anti - periodicity for the fermions in this context is only related to the choice of a negative eigenvalue for the parity operator .
let us now introduce the fourier transformed operators @xmath571 the choice of the phases @xmath572 must be made in such a way that the car algebra is preserved , but in addition it has to imply the boundary condition @xmath573 . therefore , the phases @xmath572 will also depend on the parity sector .
[ ex:1.8 ] show that in the _ even _ parity sector ( @xmath574 ) , the phases are given by @xmath575 whereas in the _ odd _ parity sector ( @xmath576 ) they are given by @xmath577 a word of caution is perhaps in order : in principle , @xmath453 is an integer that takes values in the range @xmath578 .
nevertheless , due to the translational symmetry , @xmath453 actually belongs to a `` brillouin zone '' , meaning with this that there is no difference between @xmath453 and @xmath579 .
so , for instance , for @xmath566 we have that @xmath580 whenever @xmath581 is an integer or , in other words , when @xmath582 for this reason , even if the allowed values of @xmath453 are originally in the range @xmath583 , we will allow @xmath453 to take any integer value , provided we interpret this ( for @xmath566 ) in the sense of ( [ eq : mod - n ] ) . with this convention
we then obtain , for example : @xmath584 a similar remark applies to the case @xmath562 .
for the remaining part of this section we will confine ourselves to the even parity sector and so @xmath566 will be tacitly assumed .
we can now go back to the hamiltonian ( [ eq : h - fermions - periodic ] ) and insert there the new operators defined by ( [ eq : fourier - dk ] ) . as a result
, we obtain the following form of the hamiltonian : @xmath585.\ ] ] this form of the hamiltonian is indeed very convenient as it suggests that a diagonalization by means of a bogoliubov transformation that only mixes @xmath586 with @xmath587 should be possible . for this reason we propose the following transformation : @xmath588 with @xmath589 ( possibly complex ) coefficients to be found . imposing the condition
@xmath590 we readily obtain @xmath591 .
furthermore , it is easy to check that the choices @xmath592 enforce @xmath593 . inverting ( [ eq : c_k ] ) and inserting the result in ( [ eq : h - en - terminos - de - d_k ] ) , we obtain : @xmath594 + \mbox{h.c . }
\right ) { } } \nonumber\\ & & { } + 2 \sum_{k=1}^n c_k^\dagger c_k \big [ ( \lambda -\cos \phi_k ) ( |\alpha_k|^2-|\beta_k|^2 ) +
i \sin\phi_k ( \bar{\alpha}_k \beta_k - \alpha_k \bar{\beta}_k ) \big ] { } \;\;\;\nonumber\\ & & { } + \sum_{k=1}^n\big ( 2(\lambda -\cos\phi_k ) |\beta_k|^2 + i \sin \phi_k ( \alpha_k \bar{\beta}_k-
\bar{\alpha}_k\beta_k ) -\lambda \big ) . { } \end{aligned}\ ] ] vanishing of the coefficients in the first sum leads to @xmath595 recalling that @xmath591 , we see that the right hand side of this expression simplifies if we parameterize @xmath596 and @xmath597 as follows : @xmath598 in fact , with this choice ( [ eq : rl1 ] ) simplifies to @xmath599 we can now introduce a normalization factor @xmath600 we will multiply @xmath492 by a factor 1/2 , in order to simplify some expressions . ] by means of @xmath601 this leads to @xmath602 with these definitions , the remaining terms in ( [ eq : rl2 ] ) simplify further yielding the desired diagonal form for the hamiltonian : @xmath603 [ ex:1.9 ] verify all the computations leading from ( [ eq : h - fermions - periodic ] ) to ( [ eq : rl3 ] ) . * the ground state . *
having brought the hamiltonian to the diagonal form ( [ eq : rl3 ] ) , we now proceed to find an expression for the ground state . since
@xmath600 has been chosen to be positive ( or zero ) for all values of @xmath453 , the ground state @xmath604 will be given by the condition @xmath605 .
let @xmath606 be the `` vacuum '' state for the operators @xmath607 , defined by the condition @xmath608 .
observe now that if we define @xmath609 then we get @xmath610 for all @xmath453 .
since @xmath611=0 $ ] , one would think that @xmath612 does the job , but we must take into account that @xmath613 whenever @xmath614 .
if @xmath495 is even , this means that _ each _ @xmath615 appears _
twice _ in @xmath616 and then an easy calculation shows that @xmath617 does not vanish .
hence , we must be careful to include in the product only one instance of each @xmath615 . to do this
, we distinguish two cases : @xmath495 even and @xmath495 odd . restricting the domain of @xmath572 to the interval @xmath618 we get , for @xmath495 even , @xmath619 for @xmath620 ( also for @xmath621 ) and @xmath622 for @xmath623 .
now , for each @xmath453 such that @xmath619 , there is _ exactly _ one @xmath624 with @xmath625 and such that @xmath626 .
in fact , let @xmath453 be such that @xmath619 and put @xmath627 .
it follows that @xmath625 and @xmath628 , so that @xmath626 .
we therefore see that , _ for even _
@xmath495 , the ground state is given by @xmath629 writing @xmath630 , we may as well consider @xmath453 to be such that @xmath631 .
the ground state , then , takes the form @xmath632 for the case @xmath495 odd , write @xmath633 . in this case
all the @xmath615 appear twice in @xmath616 , _ except _ when @xmath634 .
when @xmath634 we have @xmath635 .
this , together with the fact that @xmath636 , implies that @xmath637 , so we must exclude it from the product . in analogy to the previous case
, for each @xmath453 for which @xmath638 we can find exactly one @xmath624 such that @xmath639 and @xmath626 , namely @xmath640 .
now , for @xmath641 , we have @xmath642 , whereas for @xmath643 , we have @xmath644 .
the result is that , _ for odd _
@xmath495 , the ground state is given by @xmath645 the operator @xmath646 has to be included in ( [ eq : gs - ising - odd ] ) in order to ensure that @xmath647 ( recall that in this case @xmath648 ) .
having obtained the exact spectrum of the model , as well as the ground state both for open and periodic boundary conditions , it is now possible -in principle- to obtain all relevant correlation functions and relate them to physical observables like magnetization , susceptibilities , and so on . as is well - known , the quantum ising chain does not present any phase transition at finite temperature , but it certainly is one of the paradigmatic examples of a system displaying a quantum critical point .
in fact , in the thermodynamic limit there is a quantum phase transition occurring when the external field @xmath229 approaches the critical value @xmath649 . in this notes we will review an approach to the study of this quantum phase transition from the point of view of _ entanglement_. we will also be interested in certain geometric and topological properties of the ground state that are relevant for this quantum phase transition .
let us start by considering the model ( [ eq : h - ising - transv ] ) for an open , two - site chain ( @xmath497 ) .
putting @xmath650 , we find that the ground state is given in the spin basis by @xmath651 taking the partial trace with respect to one of the sites , we obtain @xmath652 which in terms of the basis @xmath653 takes the following simple matrix form : @xmath654 figure [ fig:1.1 ] displays a plot of the eigenvalues of @xmath4 as a function of @xmath229 . ] as we shall see , the behavior of this _ entanglement spectrum _ as the size of the chain increases will provide important information regarding the quantum phase transition in this model .
[ ex:1.10 ] follow the steps outlined in section [ sec : openboundaryconditions ] to show that for @xmath497 the matrices @xmath508 and @xmath509 introduced in ( [ eq : g - h - defined ] ) are given by@xmath655 use this in order to show that the condition @xmath656 leads exactly to ( [ eq : gs - ising - two - sites ] ) .
as we have seen , in this simple case ( @xmath497 ) the density matrix for the reduced state can be easily computed . in spite of the fact that the model can be exactly solved , the computation of correlation functions ( as those involved in the computation of the reduced density matrix ) for arbitrary values of @xmath495 is a non - trivial task .
we will therefore present , following @xcite , a method that allows us to obtain the spectrum of @xmath4 for arbitrary values of @xmath495 and that can be used for any quadratic hamiltonian . as a preparation for the general case
, we first explain the idea using our very simple example of a two - site chain , where all calculations can be explicitly carried out .
let @xmath158 be the car algebra generated by the two fermionic operators @xmath657 and the unity @xmath184 .
we consider the `` fock representation '' where @xmath658 acts on the state @xmath659 .
this state is cyclic and therefore we can write the ground state ( [ eq : gs - ising - two - sites ] ) in terms of it ( _ cf .
_ exercise [ ex:1.10 ] ) : @xmath660 as discussed in section [ sec : algebras - and - states ] , we can use it to define a state @xmath412 in the _ algebraic sense _
: @xmath661 this is , of course , a pure state .
nevertheless , if we _ restrict _ it to the subalgebra @xmath368 that is generated by @xmath184 and @xmath662 ( `` a half - chain '' ) , then we find that in general the resulting state @xmath663 we expect that for any @xmath664 @xmath665 will hold , with @xmath666 @xmath667 , both traces being taken on the hilbert space corresponding to the half - chain ( which in this case has a basis given by @xmath668 ) .
finding @xmath669 amounts to finding the spectrum of @xmath4 and this , in turn , can be used to compute the entanglement entropy .
the specific form of @xmath4 as the exponential of a quadratic form is to be expected in view of the fact that the ground state is gaussian . in order to find @xmath669
, we notice that it is possible to compute the left hand side of ( [ eq : a1-n=2 ] ) .
the only relevant case is @xmath670 , so let us define @xmath671 . using ( [ eq : gs - n=2 ] ) , we get : @xmath672 where we still keep the notation @xmath673 . on the other hand
@xmath674 it follows that @xmath675 or , equivalently , @xmath676 since @xmath4 is already diagonal in the basis @xmath668 , we obtain : @xmath677 which coincides with ( [ eq : rho - n=2 ] ) .
notice , in passing , that @xmath678 can also be obtained directly from @xmath509 .
in fact , a small computation shows that @xmath679 .
this last remark gives a hint towards the solution for the general case : as we shall see , the spectrum of @xmath4 in the general case can be extracted from an eigenvalue problem that uses the matrices @xmath509 and @xmath508 as input .
let us , then , start by considering the algebra @xmath158 generated by @xmath184 and by all operators @xmath680 , with @xmath681 .
again , the ground state @xmath604 gives rise to a state @xmath682 .
we are interested in the restriction of this state to the subalgebra @xmath683 generated by @xmath684 . by the same argument as before
, we expect this new state @xmath685 to be of the form @xmath686 ( @xmath687 ) , with @xmath688 and @xmath689 for the ising model , the matrices @xmath690 and @xmath167 ( yet to be determined ) will be real , so that @xmath690 has to be symmetric , and @xmath167 antisymmetric .
now we define @xcite , for @xmath691 the correlation functions @xmath692 using ( [ eq : g - h - defined ] ) one shows that @xmath693 and @xmath694 . let us first consider the simpler case @xmath695 . in this case @xmath696 can be diagonalized by a simple transformation of the form @xmath697 where the matrix @xmath698 is such that @xmath699 . solving the eigenvalue problem @xmath700 and introducing the convention @xmath701 ( @xmath702 in the @xmath520 entry and @xmath703 everywhere else ) , we set @xmath704 , _
i.e. _ , the rows of @xmath698 are the eigenvectors @xmath705 .
we then obtain @xmath706 it follows that @xmath707 , so that @xmath708 . using ( [ eq : b - gamma - a ] )
we then obtain @xmath709 , which is equivalent to the matrix identity @xmath710 .
this can also be written in the following form : @xmath711 if @xmath712 , we replace ( @xmath713 ) by @xmath714 where now we require ( _ cf .
_ exercise [ ex:1.3 ] ) : @xmath715 once again , the matrices @xmath698 and @xmath716 should be such that @xmath4 becomes diagonal , _
i.e. _ , @xmath717 should hold .
the eigenvalue problem to be solved is @xmath718 using these eigenvectors as rows for the matrices @xmath719 and @xmath12 and putting @xmath720 we obtain the desired diagonal form for @xmath696 .
as in the previous case , we can now write all correlation functions that involve the operators @xmath680 in terms of correlation functions that only involve the operators @xmath721 . [ ex:1.11 ] obtain the following two relations : @xmath722 and use them in order to prove the following identity : @xmath723 where @xmath65 and @xmath724 are the @xmath725 matrices with components given by ( [ eq : c_ij - f_ij ] ) . from this exercise
we obtain , making use of ( [ eq:(k+m)(k - m ) ] ) , @xmath726 sites as a function of the external field @xmath229 ] we will now make use of this expression in order to obtain the entanglement spectrum and the entanglement entropy of the ground state of the ising chain when @xmath728 , for various values of @xmath495 and @xmath229 . in fig .
[ fig:1.2 ] we plot the entanglement spectrum ( in logarithmic scale ) for a chain of @xmath729 sites , when the ground state is restricted to a half - chain .
we notice that , in contrast to the case @xmath497 , the gap between the first two eigenvalues is now closed for almost all values of @xmath229 below @xmath649 . in fig .
[ fig : subfigures ] we compare the entanglement spectrum for an ising chain of @xmath729 sites against the spectrum of a chain of @xmath730 sites .
it is apparent that for @xmath730 the gap near @xmath731 has almost closed . and @xmath730 , showing how the gap at @xmath421 closes as the number of sites increases.,scaledwidth=150.0% ] this gap , when evaluated at the critical value @xmath731 , bears the name of _ schmidt gap_. we will use the notation @xmath732 for the schmidt gap .
de chiara and collaborators have shown , using finite - size scaling , that the schmidt gap can be interpreted as an _ order parameter _ @xcite .
in fact , from a finite size scaling analysis , one can in fact obtain numbers @xmath733 and @xmath734 such that when we plot @xmath735 versus @xmath736 , all points collapse to a single curve , irrespective of the value of @xmath495 chosen .
this result was obtained in @xcite using a dmrg algorithm .
figure [ fig : catalina ] , obtained by c. rivera using the same dmrg technique @xcite depicts the resulting curve , for which points corresponding to different lengths of the chain ( ranging from @xmath737 to @xmath738 ) are seen to collapse to a single curve .
the values of @xmath733 and @xmath734 for which this is achieved give values for the critical exponents which are very close to the actual values ( @xmath739 and @xmath740 ) . the entanglement spectrum can also be used to compute the entanglement entropy and its dependence on the size of the subsystem chosen .
a numerical evaluation of this entropy gives the result expected from conformal field theory , that is , at the critical point , the entanglement entropy follows a logarithmic behavior , of the form @xmath741 where @xmath742 denotes the central charge of the corresponding cft . in this case
( ising chain ) , numerical evaluation of the entropy for different chain sizes gives a value of @xmath743 , in accordance with the predictions of cft .
the xy model is a generalization of the ising model .
the model also represents a spin chain with nearest - neighbor interactions .
its hamiltonian is given by @xmath744 there are two parameters in this model . as in the ising case , @xmath229 represents an external field .
the parameter @xmath698 is an anisotropy parameter .
notice that if @xmath745 , we recover , up to an overall factor of @xmath746 , the ising hamiltonian ( [ eq : h - ising - transv ] ) .
our interest in this model lies in the fact that according to the values of the parameters @xmath698 and @xmath229 , we may study critical regions corresponding to _ different _ universality classes : if @xmath747 , there is a critical line determined by @xmath748 that corresponds to the ising universality class , with central charge @xmath743 . on the other hand , for @xmath749 the whole line @xmath750 corresponds to the @xmath751 universality class , with central charge @xmath752 .
the solution of this model can be obtained in the same fashion as we did in the case of the ising model .
the first step , then , consists in using the wigner - jordan transformation to bring the hamiltonian to a quadratic form in fermionic operators .
the result one obtains is @xmath753 with @xmath754 , the fermionic number operator . since we are interested in the ground state , we consider the even parity sector and proceed , as with the ising model , with a fourier transformation , followed by a bogoliubov transformation . for odd @xmath495 ( @xmath755 ) and with @xmath756
we obtain : @xmath757 we can now consider a bogoliubov transformation , as in ( [ eq : c_k ] ) , @xmath758 and obtain the following diagonal form for the hamiltonian : @xmath759 with @xmath760 defining @xmath761 we obtain the following explicit form for the ground state @xmath762 : @xmath763 where @xmath606 is the vacuum state with respect to the operators @xmath607 .
the xy model has a rich phase diagram , and this provides an opportunity to check the behavior of entanglement entropy at the critical point and its relation to the central charge . in the previous section we focused on a finite - size scaling analysis for the ising model , which in turn brought up the difficulty of having to compute correlation functions for finite size chains . in this section
, we will only be interested in properties in the thermodynamic limit , _
i.e. _ , we will discuss the behavior of the entanglement entropy of a sub - chain of length @xmath764 , considered as a subsystem of a chain of infinite length , or the geometric phase of the ground state associated to closed loops around the critical point .
calculations in the thermodynamic limit are greatly simplified because discrete sums become integrals that can be readily evaluated .
let us now very briefly describe how to obtain the entanglement entropy of a sub - chain of size @xmath764 in the thermodynamic limit .
we will follow the approach of vidal _ et al . _
@xcite , which amounts to the real ( _ e.g. _ majorana ) version of the method discussed in section [ sec : entanglement - properties - gs ] .
we begin by replacing the fermion operators ( @xmath680 ) appearing in ( [ eq : h_xy ] ) by new , majorana fermion operators ( @xmath765 ) , defined as follows : @xmath766 these majorana operators satisfy commutation relations of the form @xmath767 .
now , consider the correlation matrix @xmath768 , where the expectation value is taken with respect to the ground state .
the matrix @xmath769 can be expressed as a block matrix , with the block @xmath770 being given by the @xmath410 matrix @xmath771 ( @xmath345 and @xmath5 range from @xmath703 to @xmath772 , @xmath495 being the size of the full chain ) , where @xmath773 in the thermodynamic limit ( @xmath774 ) the coefficient function @xmath775 takes the following compact form : @xmath776 the restriction of the correlation matrix to the sites belonging to a sub - chain of size @xmath764 is then a @xmath725 matrix @xmath777 which is obtained just by deleting the entries of @xmath769 that do not correspond to sites in the sub - chain . following the method of section [ sec : entanglement - properties - gs ] as well as vidal et al .
@xcite , show that the entanglement entropy corresponding to the restriction of the full chain to a sub - chain of @xmath764 sites is given by the formula @xmath778 where @xmath492 denotes the shannon entropy as defined in ( [ eq : h(lambda ) ] ) and @xmath779 are the positive eigenvalues of @xmath780 .
figure [ fig : vidal ] displays the behavior of the entropy as a function of @xmath764 . one of the curves ( dashed ) shows the results for the xx chain , which corresponds to @xmath749 . from the logarithmic dependence of the entropy
we extract the value @xmath752 for the central charge in this model .
the continuous curve shows the entropy for the critical case @xmath781 , which belongs to the ising universality class . in this case , the value obtained for the central charge is @xmath743 .
the dots correspond to the non - critical case @xmath782 .
we can see that in this case the entropy saturates with @xmath764 .
thus , the results confirm the expected behavior of the entropy as predicted by cft .
is represents the size of the sub - chain . for model parameters corresponding to critical phases ,
we observe a logarithmic behavior for the entanglement entropy . outside the critical point , the entropy saturates . ]
in section [ sec:2 ] we mentioned that the _ canonical commutation relations _ ( ccr ) lead to a @xmath0-algebra , whose representation theory plays a very important role in quantum field theory . let us recall that the operators @xmath177 and @xmath178 defined in ( [ eq : u(a)v(b)def ] ) satisfy the so - called weyl form of the ccr ( _ cf . _
exercise [ ex : weyl - ccr ] ) .
the first thing we want to do is to recognize that any implementation of the ccr in their weyl form corresponds in fact to the construction of a representation of a certain group ( the heisenberg group ) .
the heisenberg group of @xmath783 , whose underlying set is @xmath784 , is the group defined by the following operation : @xmath785 the key point regarding these commutation relations is actually their relation to the ( canonical ) symplectic structure of the underlying classical phase space , @xmath790 .
in fact , if for @xmath791 we define @xmath792 we obtain the following identity : @xmath793 where @xmath267 and @xmath794 denote elements of the _ symplectic vector space _ @xmath790 , and @xmath112 the standard symplectic form . notice that we have the following relations between all these operators : @xmath795 ( _ cf .
_ @xcite ) let @xmath789 be a real vector space and @xmath796 a symplectic form ( _ i.e. _ a bilinear , skew - symmetric , non - degenerate form ) .
a @xmath149-algebra @xmath797 is called a weyl @xmath149-algebra of @xmath798 if there is a family @xmath799 of `` generators '' such that in @xcite it is proved that every symplectic vector space @xmath798 determines uniquely a weyl @xmath149-algebra , up to @xmath149-isomorphism . also , if @xmath804 is a symplectomorphism between two symplectic vector spaces , then the corresponding weyl @xmath149-algebras are isomorphic .
a most important fact is that @xmath797 can be completed to a @xmath0-algebra , the weyl @xmath0-algebra .
it is a fundamental result , due to stone and von neumann , that when the dimension of @xmath789 is finite ( and hence necessarily even ) , there is essentially only one representation of the ccr , which can be taken to be the standard schrdinger representation @xcite .
but in infinite dimensions uniqueness is lost , and so _ inequivalent _ representations do exist .
this fact is closely related to the non - uniqueness of a vacuum state for a free quantum field in a curved spacetime background .
the following example illustrates how the construction of a ( free , bosonic ) quantum field runs in parallel to the previous discussion of ccr , the `` only '' ( ! ) difference here being that the symplectic vector space we are dealing with is infinite dimensional .
let @xmath805 denote a globally hyperbolic spacetime , and consider the klein - gordon equation on that background : @xmath806 let @xmath807 denote a fixed cauchy surface , to which the initial data for the solutions of the klein - gordon equation will be referred and consider the following vector space , a _ space of solutions of the k - g equation _ : @xmath808 it is a most remarkable fact that the classical field equation comes equipped with a natural symplectic structure .
in fact , if @xmath809 is any space - like ( cauchy ) hypersurface , and @xmath810 denotes the unit normal , then on the vector space @xmath811 defined above we can define the following symplectic form : @xmath812 for @xmath813 , let @xmath23 and @xmath814 denote the advanced and retarded fundamental solutions of the k - g equation .
that is , for @xmath815 we have @xmath816 where the support of @xmath817 lies in the causal past of the support of @xmath59 ( and analogously for @xmath818 ) .
then @xmath819 gives a map @xmath820 the ccr for the quantum field ( in its _ smeared form _ ) then take the form @xmath821=-i \sigma(ef , eg ) .
\nonumber\ ] ] just as in non - relativistic quantum mechanics , the operators obeying ccr are unbounded , and so it is convenient to go to the weyl form , which leads us back to the weyl @xmath0-algebra , this time the one corresponding to the infinite dimensional vector space @xmath822
. representation spaces for the weyl algebra of the previous example are bosonic fock spaces . for the classical dirac field
, the transition to the corresponding quantum description can be performed in a completely analogous way , only that in this case , instead of an antisymmetric form , a symmetric one is used which , again , comes `` for free '' as part of the structure of the ( vector space ) of solutions of the classical field equation .
it is an interesting fact that the existence of a natural bilinear antisymmetric form in the scalar field case ( and of a symmetric one in the case of the dirac field ) leads to quantum fields that obey the physically correct relation between spin and statistics .
in very general terms , quantization of free fields can be formulated as follows .
suppose we are given a real vector space @xmath789 together with a non degenerate ( anti-)symmetric bilinear form @xmath823 .
if in addition we choose a complex structure @xmath824 which is compatible with @xmath825 ( meaning @xmath826 holds ) , we can complexify and obtain a complex vector space @xmath827 .
the bilinear form @xmath825 can then be used to define an inner product on @xmath827 , making it a hilbert space . according to
whether the bilinear form is symmetric or antisymmetric , one then considers the antisymmetric or the symmetric subspace of the tensor algebra of @xmath827 , call it @xmath828 .
endomorphisms of @xmath827 may be lifted to @xmath829 in a natural ( _ i.e. _ functorial ) way .
these liftings are then interpreted as the `` second quantization '' of observables ( self - adjoint operators in @xmath827 ) and symmetries ( unitary operators ) .
+ in this section we will focus on the fermionic case , following the beautiful treatment of the subject presented in @xcite , which makes strong use of clifford algebras and spin group representations . in the next section
we will explore the connection between second quantization ( as presented here ) and quantum criticality in spin chains ( as discussed previously ) .
let us start by first considering the simple case where @xmath830 .
if @xmath789 is a finite dimensional real vector space , and @xmath508 a positive definite symmetric bilinear form on it , we can construct the corresponding clifford algebra @xmath831 .
as shown in @xcite , it is possible to obtain a concrete realization of the complexified clifford algebra @xmath832 as a subalgebra of @xmath833 , but the corresponding representation is not irreducible .
the situation changes if we assume @xmath789 has even dimension , say @xmath834 .
then we may choose an orthogonal complex structure , that is , a linear operator @xmath835 such that @xmath836 and @xmath837 for all @xmath838 , and regard @xmath789 as a complex vector space @xmath827 of dimension @xmath839 .
the scalar product on @xmath827 is given by @xmath840 it then turns out that the exterior algebra @xmath841 is an irreducible clifford module .
we call this module _ fock space _ and use the notation @xmath842 for it .
the action of @xmath832 on @xmath842 is obtained by defining creation " and annihilation " operators @xmath843 acting on @xmath842 as follows : @xmath844 from these definitions , we obtain the anti - commutation relations @xmath845 from which @xmath846 follows .
hence , the map @xmath847 may be extended to @xmath848 by making use of the universal property of the clifford algebra .
the scalar product @xmath849 on @xmath827 induces one on @xmath842 , which on basis elements is given by @xmath850 thus making @xmath842 a ( finite dimensional ) hilbert space .
it may be checked that @xmath851 is really the adjoint of @xmath852 , with respect to this scalar product .
our interest is to lift ( or quantize ) operators @xmath853 to operators acting on fock space . in the present context
, this is done by means of maps @xmath854 defined as the graded operators obtained from @xmath855 and @xmath856 given any @xmath859 and @xmath860 any _ unitary _ vector , we see that @xmath861 , _
i.e. _ the twisted conjugation by @xmath794 using the grading automorphism @xmath862 is a reflection on the hyperplane perpendicular to @xmath794 . but reflections generate the orthogonal group @xmath863 so that the map @xmath232 , being defined for all invertible elements of the clifford algebra , when restricted to the subgroup generated by all products of an even number of unitary vectors ( @xmath864 ) , reduces to a homomorphism onto @xmath865 .
this homomorphism may be restricted to @xmath866 .
let us recall that there is a canonical vector space isomorphism between the clifford algebra and the exterior algebra , given explicitly in terms of an orthonormal basis by : @xmath867 direct computation then shows that @xmath868 coincides with @xmath869 and hence @xmath870 , the isomorphism being given by @xmath871\nonumber\end{aligned}\ ] ] let @xmath872 be an irreducible representation of @xmath848 . since @xmath868 is realized as a subspace of @xmath848 , we can compose @xmath742 with the isomorphism @xmath873 to obtain a representation , called the _ infinitesimal spin representation _ , of @xmath874 on @xmath146 . setting @xmath875 : @xmath876 _ i.e. _ , @xmath877 the infinitesimal spin representation satisfies the following basic relation : @xmath878=c(bv).\ ] ] comparing ( [ basic_rel ] ) with the identity from exercise [ ex:[dlambda a , v]=av ] , one realizes that @xmath857 and @xmath879 must be closely related . in order to express this relation , note that any real linear operator @xmath814 on @xmath789 can be written as @xmath880 , with @xmath881 linear in @xmath827 and @xmath882 antilinear in @xmath827 . for an element @xmath883 , @xmath884 is skewadjoint and @xmath885 antilinear and skewsymmetric ( in @xmath886 ) .
then , it can be shown that for any @xmath883 , the following relation holds : @xmath887 where @xmath888 and @xmath889 ( with @xmath890 any orthonormal basis in @xmath827 ) @xcite .
this equation plays an important role for the definition of the quantization map in the infinite dimensional case , discussed below . in the infinite dimensional context ,
we begin with a separable real vector space @xmath789 on which a positive definite symmetric bilinear form @xmath508 is defined .
it is assumed that @xmath789 is complete in the metric induced by @xmath508 , so that @xmath891 is a separable hilbert space , where @xmath892 .
the clifford algebra is constructed as follows .
consider the algebra @xmath893 obtained from the union of all algebras @xmath894 , where @xmath895 runs through all finite dimensional subspaces of @xmath789 .
there is a unique trace @xmath896 on @xmath893 , inherited from the traces on each @xmath894 .
the scalar product induced by the trace makes @xmath893 a prehilbert space .
its completion @xmath897 allows one to represent @xmath893 as a subalgebra of @xmath898 via the gns construction .
the closure of this algebra is then a @xmath0-algebra , which is defined to be the clifford algebra @xmath848 .
the universal property remains valid in this context and in particular an orthogonal map
@xmath899 extends to a @xmath0-algebra automorphism @xmath900 of @xmath848 ( bogoliubov automorphsim ) . in order to construct a representation of the clifford algebra on fock space we need , as before , an orthogonal complex structure @xmath824 .
the fock space @xmath842 is now defined as the completion of the exterior algebra @xmath841 with respect to the scalar product ( [ eq : scalar - prod ] ) .
the action of @xmath848 on the exterior algebra is defined as in the finite dimensional case , its extension giving rise to an irreducible representation @xmath901 . as in the finite dimensional case , the action of the orthogonal group @xmath902 on the set of orthogonal complex structures is transitive , for given two complex structures @xmath824 and @xmath690 , there is a unitary map @xmath903 such that @xmath904 . as can be shown ,
if the bogoliubov automorphism @xmath900 is implementable by a unitary operator on fock space , then @xmath905 and @xmath906 are equivalent . in general
however , they will be inequivalent .
the converse also holds : implementability of @xmath900 follows from equivalence of the representations .
necessary and sufficient conditions on @xmath509 for @xmath900 to be implementable are afforded by the shale - stinespring theorem ( _ cf .
_ @xcite , thm .
6.16 ) , namely that the antilinear part @xmath907 of @xmath509 be hilbert - schmidt . in terms of the representations
, it says that @xmath905 and @xmath906 are unitarily equivalent if and only if @xmath908 is hilbert - schimdt
. for the construction of quantized currents ( like the charge or the number operator ) we need a map taking self - adjoint operators on @xmath909 to self - adjoint operators on @xmath910 .
the explicit construction of this map , which can be found in @xcite , will not be presented here .
instead , we will try to describe some of the motivations behind the construction .
recall that @xmath902 acts transitively on the set of orthogonal complex structures , @xmath911 .
since we have chosen a fixed complex structure @xmath824 to construct the representation @xmath905 on fock space , then any other complex structure will be given in terms of @xmath824 and of an element @xmath912 ( @xmath913 ) .
recall also that an element of the unitary group @xmath914 is just an orthogonal map @xmath912 that commutes with @xmath824 . a unitary transformation @xmath915 is implemented in fock space by a map @xmath916 such that : @xmath917 now , the observables of the 1-particle theory correspond to elements of the lie algebra @xmath918 of the restricted orthogonal group , which is defined as @xmath919\ ; \mbox{is hilbert - schmidt}\rbrace$ ] . if an element @xmath920 is such that @xmath921=0 $ ] , then it follows that @xmath922 is self - adjoint on @xmath827 and it can be quantized by means of @xmath923 .
the question of ( unitary ) equivalence of the quantization obtained by means of @xmath824 on one side and by means of @xmath913 on the other , may be formulated as follows .
the complex structure @xmath924 is determined by @xmath509 .
the map @xmath925 extends to an automorphism of the clifford algebra and the question is then whether this automorphism is implementable on fock space .
so , given @xmath912 , we are looking for a unitary operator @xmath926 on fock space such that : @xmath927 in view of the inclusion @xmath928 , we see that we are looking for an extension of @xmath916 to @xmath902 . if @xmath929 $ ] is hilbert - schmidt , that is , if @xmath930 , then the map @xmath257 can be constructed ( it is the pin representation of @xmath931 on fock space ) .
its restriction to @xmath932 is the spin representation , from which an infinitesimal version @xmath879 can be defined on a dense domain of fock space and turns out to be given by ( [ eq : mu = dlambda+tr ] ) without the trace term and , therefore , it is not a lie algebra homomorphism .
this gives rise the so - called _ schwinger term _ , defined as @xmath933-\dot \mu ( [ a , b ] ) .
\nonumber\ ] ] the schwinger term is a cocycle that gives an obstruction for @xmath934 to be a lie algebra homomorphism .
as explained in @xcite , the schwinger term is related to anomalies in quantum field theory and also to noncommutative geometry .
it appears naturally when studying the virasoro algebra , and therefore is also related to the notion of _ central charge _ in conformal field theory .
as we have seen , central charges do also appear when we study entanglement entropy in spin chains that display critical behavior .
our aim in the next section is to give an idea of how all these concepts are interrelated .
this will allow us to close the circle by providing a new topological interpretation of quantum criticality , for which the underlying geometry is not anymore the geometry of the space of external parameters , but a geometry in the sense of noncommutative geometry .
this geometric interpretation of anomalies , central charges , schwinger cocycles _ etc .
_ , has been known for several decades ( to the best of my knowledge the first person to realize this was araki @xcite ) .
but the connection to _ quantum critical phenomena _ does not seem to have been exploited .
previously in these notes we discussed the gelfand - naimark theorem , which serves as a motivation for the notion of a noncommutative space .
it turns out that the schwinger term described in the previous section also has a nice geometric interpretation , in the context of noncommutative geometry @xcite . in order to give at least a glimpse of how this interpretation appears we will present , following @xcite , a very basic ( yet instructive ) example , namely the gauss - bonnet theorem , in the language of noncommutative geometry . but
before that we quickly review some aspects of the geometry of surfaces , following the beautiful presentation of the subject by pressley @xcite .
we thus start with a given 2-dimensional surface , described locally through a parametrization @xmath935 at each point , the tangent plane is generated by @xmath936 and @xmath937 and we can define the normal unit vector as @xmath938 the infinitesimal length element on the surface is given by @xmath939 where @xmath940 are the components of the _ first fundamental form _ : @xmath941 recall that the first and second fundamental forms enter the definition of gaussian curvature of a surface , which we now recall , starting from the curvature of a curve . for a plane curve , parametrized by @xmath942 with unit speed ( @xmath943 ) ,
let @xmath944 denote the unit normal at @xmath945 .
then the ( signed ) curvature @xmath946 is defined through the relation @xmath947 .
the curvature gives us an idea of how much the curve deviates from a straight line @xcite .
in fact , we may compute @xmath948 now we repeat the same reasoning , but with a surface .
that is , we want to compute @xmath949 , where now @xmath810 denotes the unit normal to the surface . from a second order taylor expansion and using the fact that @xmath950 , we get @xmath951 where @xmath952 , @xmath953 and @xmath954 are the coefficients of the _ second fundamental form _ : @xmath955 now , for any smooth curve @xmath698 we have a decomposition of the form @xmath956 where @xmath957 and @xmath958 denote , respectively , the _ normal _ and _ geodesic _ curvatures of @xmath698 . to define the curvature of the surface we need to consider only curves with @xmath959 , for which the normal curvature takes the form @xmath960 the value of @xmath957 changes depending on the direction of the curve at the evaluation point , and it is clear it must attain maximum and minimum values .
they are the _ principal curvatures _
, denoted as @xmath961 and @xmath962 .
the gaussian curvature is defined as @xmath963 .
we can obtain several useful formulas for the curvature using the fundamental forms .
the principal curvatures @xmath961 and @xmath962 are obtained from an extremization problem , where we look for extremal values of the principal curvature as we change the direction of the curve -given by @xmath964- subject to the unit - speed restriction @xmath965 .
so , with @xmath966 , @xmath967 , we want to extremize @xmath968 subject to the restriction @xmath969 .
but we also have @xmath970 solve the extremization problem posed above , subject to the constraint @xmath969 , and show that the solution is given by the solution of the eigenvalue problem @xmath971 where the eigenvalue @xmath229 is the lagrange multiplier of the problem .
show furthermore that , for a given solution @xmath972 , the corresponding normal curvature coincides precisely with @xmath229 and conclude that the gaussian curvature can be expressed as follows : @xmath973 there are other ways we can characterize @xmath690 .
one of them uses the gauss map : for any point @xmath105 in the surface , consider the unit normal , @xmath974 .
regarding this vector as a point in the unit sphere , we obtain a map @xmath975 from the surface to the unit sphere , the gauss map .
as we go around a small loop centered at @xmath105 , the gauss map produces an image loop on the unit sphere .
the quotient of the areas enclosed by the loops converges , in the limit where the original loop shrinks to a point , to the gaussian curvature . yet
another way to obtain the gaussian curvature is by means of the _
weingarten map_. first notice that we have @xmath976 , @xmath977 and @xmath978 .
this can be easily checked ; for instance , the first relation is obtained by taking the partial derivative of @xmath979 with respect to @xmath267 .
the others are obtained in the same way .
the weingarten map @xmath980 is most easily defined in the basis @xmath981 , where it has a matrix given by @xmath982 .
in fact , if we define @xmath983 and write @xmath984 then it follows that @xmath985 now we compute @xmath986 from which the following identity is obtained : @xmath987 now , as the components of the unit normal @xmath988 are smooth functions of @xmath989 , we can compute their exterior derivatives @xmath990 in order to obtain @xmath991 this leads to the following formula for the euler characteristic : @xmath992 this formula is particularly well suited to explain the noncommutative viewpoint , to which we now turn , following the example presented by connes in @xcite .
let us consider an algebra @xmath158 ( without assuming other properties , for the moment being ) together with a trilinear map @xmath993 such that @xmath994 and @xmath995 now let @xmath996 be a ( continuous ) family of idempotents , @xmath997
. then it can be shown that the quantity @xmath998 does not depend on @xmath999 , providing an `` invariant '' of the algebra . a quick way to see what is going on
is to assume that we can differentiate with respect to @xmath999 . in this case
we have @xmath1000,\nonumber\ ] ] with @xmath1001 $ ] .
using this fact and ( [ eq : cond(i ) ] ) , we obtain @xmath1002 , e_t , e_t).\nonumber\ ] ] but the last expression vanishes exactly . to see this , we put @xmath1003 and @xmath1004 in ( [ eq : cond(ii ) ] ) and use @xmath1005 .
now , coming back to gauss - bonnet , if we take the algebra to be @xmath1006 , we see that the trilinear map @xmath1007 satisfies ( [ eq : cond(i ) ] ) and ( [ eq : cond(ii ) ] ) above , provided @xmath1008 .
but , as explained in @xcite , in this algebra there are no interesting idempotents
. this can be remedied by adding a bit of noncommutativity " to this algebra , that is , by replacing it by the following one : @xmath1009 now , recalling the gauss map , we realize that we can use the normal to the surface in order to obtain an idempotent , namely @xmath1010 where @xmath1011 , a `` vector '' whose components are the pauli matrices .
then , if we define -following @xcite- @xmath1012 we obtain ( an algebraic version of ) the euler characteristic .
what we learn from the previous example is that it seems plausible to encode geometry and topology in algebraic structures related to a manifold @xmath167 .
this is precisely the way noncommutative geometry works . after establishing such algebraic and operator - theoretic characterization of certain geometric or topological structures , one can try to keep the algebraic version , but now dropping the commutativity assumption . such is the case with differential calculus .
let us then quickly list the main properties of de rham cohomology and see whether they can be formulated for more general algebras .
given a compact manifold @xmath167 , the space of 1-forms @xmath1015 consists of elements of the form @xmath1016 , where @xmath59 and @xmath508 are smooth functions on @xmath167 , and @xmath1017 is the exterior derivative of @xmath508 .
differential forms of order @xmath453 , @xmath1018 , are generated locally by expressions of the form @xmath1019 so that , in general , the exterior derivative @xmath1020 can be defined through @xmath1021 .
the exterior derivative satisfies the leibniz rule @xmath1022 and fulfills the basic identity @xmath1023 , from which we obtain the ( de rham ) cohomology `` groups '' : @xmath1024 that contain important information about the topology of @xmath167 .
finally , we also have stoke s theorem : @xmath1025 now suppose we are given a unital algebra @xmath158 .
thinking of it as given by some space of functions on a `` virtual '' ( _ i.e. _ noncommutative ) space , we may try to define a differential calculus satisfying properties as close as possible to the ones recalled above .
one possible construction goes as follows @xcite : in order to define the space of 1-forms , we consider the map @xmath1026 defined as @xmath1027 .
its usefulness comes from the fact that it acts as a derivation : @xmath1028 but @xmath1029 is `` too big '' to be regarded as a space of 1-forms .
we want to have a closer resemblance to @xmath1015 .
recalling that 1-forms can always be put in the form @xmath1030 , we consider the restriction to the subspace of @xmath1029 generated by elements of the form @xmath1031 . noticing that @xmath1032 we arrive at the following definition : @xmath1033 where @xmath1034 is the multiplication map @xmath1035 .
then , starting from @xmath1036 we can go on and define @xmath1037 for @xmath1038 .
the exterior derivative , then , is simply given by @xmath1039 from these definitions we get , as can be easily checked : this means that in any complex algebra we can find an analog of the space of differential forms ( _ i.e. _ , a graded differential algebra ) .
the importance of this construction lies in the fact that it satisfies a universal property , as any derivation of @xmath158 into a bimodule factors through a bimodule morphism from @xmath1042 .
we also want to have a notion of integration . in this context
this is achieved through the introduction of _ cycles_. an @xmath810-dimensional cycle is a complex graded differential algebra ( @xmath1043 ) , along with an `` integral '' , _
i.e. _ a linear map @xmath1044 , such that : @xmath1045 the third requirement is tantamount to stokes theorem for the case of spaces without boundary .
now , given an algebra @xmath158 , a _ cycle _ over @xmath158 is a cycle @xmath1046 together with a homomorphism @xmath1047 @xcite .
a basic example is afforded by the de rham complex on a compact manifold without boundary . given a cycle over
@xmath158 we may now define , for @xmath1048 , @xmath1049 as shown in @xcite , this map satisfies ( the generalization to @xmath810 of ) the relations ( [ eq : cond(i ) ] ) and ( [ eq : cond(ii ) ] ) , _ i.e. _ , it is a _
cyclic cocycle_. given a complex algebra @xmath158 , an _
odd _ fredholm module over @xmath158 is given by a representation @xmath1050 on the algebra of bounded operators on a hilbert space , together with a self - adjoint operator @xmath1051 , such that @xmath1052 and such that @xmath1053 $ ] is a compact operator for each @xmath215 in @xmath158 . an _ even _ fredholm module contains , in addition to the structure of an odd fredholm module , an additional grading of @xmath21 , say @xmath698 , such that @xmath1054 and @xmath1055=0 , a\in \mathcal a$ ] . on a fredholm module
one can introduce the structure of a graded differential algebra defining the differential as @xmath1056.\ ] ] letting @xmath1057 be spanned by elements of the form @xmath1058 , and so on , we arrive at the formula @xmath1059_{\pm}$ ] as a suitable definition for @xmath1060 , where the bracket denotes a supercommutator ( depending on the degree of @xmath211 ) .
an important class of examples for cyclic - cocycles is afforded by the so - called schwinger terms ( as defined in the previous section ) in quantum field theory . in the context of charged fermionic fields ,
we start with a real vector space @xmath789 ( given as a certain space of solutions to the classical field equation ) , a symmetric bilinear form @xmath508 on @xmath789 and a compatible ( orthogonal ) complex structure @xmath824 .
physically , @xmath824 is determined by the spectrum of the hamiltonian . from the data @xmath1061
one then defines a complex hilbert space which is the complexification of @xmath789 using @xmath824 as the complex structure .
so we get a hilbert space @xmath1062 , where the inner product is given by @xmath1063 we also have a representation of the field algebra on fock space , @xmath1064 now , the point is that the schwinger term @xmath1065 -\dot \mu[a , b ] = \frac{i}{8}\tr \left(j[j , a][j , b]\right),\ ] ] defines a cyclic 1-cocycle . a geometric interpretation of this cocycle is then made possible in the context of non - commutative geometry described above .
when discussing the entanglement properties of spin chains , we found that the entanglement entropy of a sub - chain of length @xmath764 , at the critical point , scales logarithmically with @xmath764 .
we mentioned that the constant in front of the logarithm was related to _
central charge _ of the corresponding conformal field theory .
these central charges can also be understood as anomalies appearing in the quantization process of the effective field theories describing the spin chains . as an example , let us consider the ising chain : @xmath1066 using the wigner - jordan transformation and then defining the ( majorana ) operators @xcite @xmath1067 we obtain @xmath1068 we can go to the continuum limit by assuming the sites in the chain are separated by a unit length @xmath1069 and defining @xmath1070 .
then , to first order in @xmath1069 , we have : @xmath1071 defining the `` chiral '' components @xmath1072 and the corresponding spinor @xmath1073 one obtains , using heisenberg s equations of motion , @xmath1074 where @xmath1075 is such that @xmath1076 . in the massless case , ( [ eq : dirac ] ) can be obtained from the lagrangian @xmath1080 .
considering the spatial coordinate @xmath1081 to be defined on @xmath1082 ( which amounts to imposing periodic boundary conditions on the spin chain ) , expand the chiral fields in fourier modes and then express the hamiltonian density @xmath1083 , and the energy function @xmath1084 in terms of the fourier modes . as the previous calculations show , when @xmath1078 the two chiral fields are completely decoupled from each other .
thus , we may study them separately .
let us consider @xmath1085 , for which the mode expansion must be of the form @xmath1086 imposing canonical anticommutation relations to the field ( which is real ) , we obtain : @xmath1087 from the previous exercise , we see that the contribution of @xmath1085 to the ( second quantized ) hamiltonian will be @xmath1088 where normal ordering has been introduced .
now , how do we define normal ordering depends on what we regard as the `` vacuum state '' .
this has important consequences for the commutation relations of the ( fourier components of the ) stress - energy tensor . decomposing the stress - energy tensor @xmath1089 in light - cone components @xmath1090 and @xmath1091 ,
show that @xmath1092 , and that the fourier components ( @xmath1093 ) of @xmath1094 satisfy the following commutation relations ( virasoro algebra ) : @xmath1095 = ( m - n ) l_{m+n } + \frac{c}{12 } ( m^3-m)\delta_{m+n,0},\ ] ] with @xmath743 .
the correct solution of the previous exercise depends crucially in a careful handling of the normal ordering prescription .
this in turn is related to the `` filling - up of the dirac sea '' which , in mathematical terms , can be reduced to the problem of quantizing a linear system with an appropriately chosen complex structure .
in fact , writing the hamiltonian density in terms of the spinor @xmath719 , we obtain @xmath1096 this means that the 1-particle hamiltonian ( which is defined on the hilbert space @xmath1097 ) is given by @xmath1098 .
as we have seen , the splitting of the 1-particle hilbert space into positive and negative energy states gives rise to a complex structure , which is the one that should be used for quantization .
when this is done , one obtains the anomalous term of the virasoro algebra as an anomaly ( schwinger term ) given precisely as a cyclic 1-cocyle of the form ( [ eq : schwinger - term ] ) .
now , the relevance of this fact in the context of quantum phase transitions is that there are already some geometric characterizations of the critical point in terms of , _
e.g. _ , berry phases @xcite .
this has also been related to the behavior of certain chern numbers associated to the parameter space of the spin chain @xcite .
now , using araki s self - dual formalism , it should also be possible to compute the cocycle outside the critical point .
the behavior of the cocycle as a function of the model s external parameters may provide a new geometric characterization of the critical point .
h. maassen , quantum probability and quantum information theory , in _ quantum information , computation and cryptography _ , eds . f. benatti , m. fannes , r. floreanini and d. petritis , lecture notes in physics , vol . 808 ( springer berlin heidelberg , 2010 ) pp .
65108 . d a guzmn , l j uribe , a valencia , f j rodrguez , and l quiroga , contrasting classical probability concepts with quantum mechanical behavior in the undergraduate laboratory , _
european journal of physics _
* 36 * ( 2015 ) , no . 5 , 055039 .
e. langmann , _ geometric methods for quantum field theory : proceedings of the summer school : villa de leyva , colombia , 12 - 30 july 1999 _ ( world scientific , 2001 ) , ch . quantum theory of fermion systems : topics between physics and mathematics , pp .
170237 .
h. contreras and a. reyes - lega , berry phases , quantum phase transitions and chern numbers , _ physica b : condensed matter _ * 403 * , 1301 ( 2008 ) , proceedings of the international conference on strongly correlated electron systems . | motivated by the sharp contrast between classical and quantum physics as probability theories , in these lecture notes i introduce the basic notions of operator algebras that are relevant for the algebraic approach to quantum physics .
aspects of the representation theory of c*-algebras will be motivated and illustrated in physical terms .
particular emphasis will be given to explicit examples from the theory of quantum phase transitions , where concepts coming from strands as diverse as quantum information theory , algebraic quantum physics and statistical mechanics agreeably converge , providing a more complete picture of the physical phenomena involved . |
computers are physical systems and information has to be stored and transmitted using physical devices .
this trivially sounding statement has led to very important insights as soon as one starts to ask for the fundamental physical limits of computation .
the most known statement is probably landauer s principle : erasing a data set with information content ( i.e. , entropy ) @xmath0 causes a minimum heat dissipation of @xmath1 if @xmath2 is the inverse temperature of the surrounding environment .
this was first formulated by landauer in 1961 @xcite . more generally , it was argued by bennett @xcite and others @xcite , that each logically irreversible operation ( as , e.g. , information erasure ) must be accompanied with a corresponding heat dissipation whereas _
each logically reversible operation can be implemented at least in principle in an energetically neutral way . _
the heat flow of such computers , if operated slowly enough , is then equal to the change in their shannon entropy ( times @xmath3 ) demonstrating the usefulness of the shannon entropy to describe thermodynamic processes .
although the question about the relation between logical and thermodynamical reversibility sounds like a purely academic question , it is indeed of practical importance because one of the limitations of todays computers lies in the heat which they produce during computation and which is hard to be drained off quickly enough .
in addition , it is well - known that the thermodynamics of computation can be successfully applied to resolve the famous maxwell demon paradox @xcite . today
, it seems that most physicists accept the thermodynamics of computation as a well established field and also feynman notes ( page 160 in @xcite ) : `` i see nothing wrong with his [ bennett s ] arguments .
[ ... ] i concluded that there was no minimum energy [ consumption of computers ] . '' indeed , however , criticism was raised against bennett s exorcism of maxwell s demon and the thermodynamics of computation already some time ago @xcite , which was subsequently defended by bub and bennett again @xcite , but criticism and controversies still prevail for different reasons , mainly ( but not only ) on the philosophical side @xcite .
an important class of physical models used to illustrate the thermodynamics of computation are inspired by biochemical processes as , e.g. , dna replication @xcite . indeed , copying a dna strand can be regarded as a simple computational task where the dna strand represents a certain input signal , which is manipulated by enzymes to produce an identical copy of the input .
energetic barriers in the computational path , i.e. , barriers between two logical states of the computation , can be overcome by the random thermal motion of the molecules involved and a bias in chemical potentials can be used to drive the computation in a desired direction . for a small enough chemical bias the average dissipation of energy per step can be made arbitrarily small and thus , one usually concludes that computation can be carried out thermodynamically reversibly . in a more general frame , systems which use the random thermal motion of its components to perform a computation
are usually called _
brownian computers_. however , in addition to the arguments presented above , a detailed and general mathematical treatment of such computers seems to be missing in the literature .
indeed , the authors of refs .
@xcite based their reasoning largely on ingenious arguments instead of detailed calculations . if one finds more detailed ( yet not very general calculations ) in the literature
@xcite , then they seem to contradict the statement that brownian computers are thermodynamically reversible . in the present contribution we will therefore treat the subject of brownian computers in general , i.e. , without having any specific computational task or physical problem in mind
. we will start by considering an arbitrary turing machine ( tm ) , which is known to be a model for a general purpose computer .
this means that for each computable function ( or algorithm ) there exists a tm which can implement it .
moreover , it is even possible to construct a special tm , called a universal tm , which is able to simulate any other tm and hence , tms are said to be computationally universal ( an introduction to tms can be found in refs .
@xcite ) . based on the ideas of bennett
we will then show how to construct a logically reversible tm and in addition , we show how to model this tm by a continuous - time markov process , i.e. , by a markovian master equation ( me ) .
the resulting model is then able to compute in a stationary regime where it transforms a string of incoming symbols ( the inputs of the computation ) into a corresponding string of outgoing symbols ( simply called the outputs ) .
the big advantage of a description in terms of a me is that its thermodynamic behaviour is well understood since many years @xcite , and also their stochastic behaviour can be treated within a consistent thermodynamic formalism , which is known as stochastic thermodynamics @xcite .
indeed , there has been a large interest recently in using small autonomous machines describable by , e.g. , a markovian me , to address questions of information processing as , e.g. , sensing , feedback or adaptation , in a thermodynamic context .
such machines were studied for abstract models @xcite , more general settings @xcite , in a biochemical context @xcite or in the field of artificial nanostructures @xcite .
furthermore , we have reached the realm where experiments are performed at the landauer limit @xcite . however , to the best of our knowledge , there is no reference which has treated the question of how to describe a _ general _
computer within a me framework and which has worked out its thermodynamic consequences in detail .
we will find out that the entropy production rate of a brownian computer can be made arbitrarily small while the total entropy production grows logarithmically with the number of computational steps , thereby resolving a part of the controversy between bennett , norton and others .
_ outline : _ because the treatment of tms is no standard subject taught in physics , we will give an introduction to it in sec .
[ sec tms ] to make the paper as self - contained as possible .
then , in sec .
[ sec stochastic tm ] we will first show how to build a tm in a logically reversible way before we demonstrate how to map it to a me .
because the details of a logical reversible tm are a bit technical but not of major importance for a general understanding of the rest of the paper , we will shift them to appendix [ app logical reversible tm ] . finally , after we have discussed the structure of the me in sec . [ sec stochastic tm ] , we discuss the thermodynamics of brownian computation in sec .
[ sec thermodynamics brownian computation ] . a last section is then devoted to a summary of our results and an outlook on interesting future work .
simple sketch of a tm : the machine , specified by a state @xmath4 , has access to an infinite tape , which is divided into equal squares . the machine scans with its head always only one square on the tape , on which it finds a symbol @xmath5 ( which might be also the blank symbol @xmath6 ) written .
, scaledwidth=25.0% ] a tm @xmath7 is an idealized machine to model or simulate problems in computer science ( see fig .
[ fig tm sketch ] ) . in the standard treatment @xmath7 has two parts : first , we have the machine itself , which can be in some state @xmath8 where @xmath9 denotes the finite set of internal machine states .
second , the machine has access to an external storage medium ( usually called the _ tape _ ) , which is divided into equal _ squares _ and each square contains either a special symbol @xmath10 ( where @xmath11 is again a finite set ) or it contains a blank @xmath6 . the machine has a _ head _ with which it is coupled to one and only one square of the tape at each point of time . at each time step
, the machine in state @xmath4 reads the symbol @xmath5 written on the square and subsequently it changes its own state to @xmath12 , writes a new symbol @xmath13 on the square ( which can be the same as the old one ) and either shifts the tape one square to the left or to the right or stays where it is .
mathematically , these rules can be defined by three functions @xmath14 and @xmath0 : @xmath15 where @xmath16 encodes the direction of movement of the tape with @xmath17 meaning move the tape right , do not move it , move it left .
these three maps ( together with an agreement about the initial state , see below ) _ completely define _ the action of @xmath7 .
often one writes these maps in terms of a so - called quintuple @xmath18 or simply as @xmath19 a computation of @xmath7 is then defined as follows : the machine starts initially in a special state r @xmath20 ( we use r for `` ready '' ) and scans by convention the first blank symbol to the left of a finite number of input symbols @xmath21 written on the tape .
note that we demand that there is no blank symbol in between the input symbols and that the input string is finite .
then , @xmath7 starts to move right and reads the first input symbol @xmath22 .
it then proceeds according to the rules above [ eq . ( [ eq tm standard map ] ) ] . after some time
the tm might be done with the computation .
we then assume that it shifts to the first blank symbol to the right of the remaining string of symbols @xmath23 written on the tape and then changes to a special final state h ( h for `` halt '' ) .
the string @xmath24 is called the output or the _ result _ of the computation , which is again finite but not necessarily of the same length as the input ( i.e. , @xmath25 is possible ) . in short ,
we will also write a computation as @xmath26 or @xmath27 .
hence , we see that the idea of a tm is very simple : given @xmath28 , i.e. , a tm @xmath7 with input @xmath29 in standard format as above , it follows the rules ( [ eq tm standard map ] ) until it is done with the computation . especially , we note that there is only a _
finite _ set of quintuples or rules ( [ eq tm standard map ] ) because @xmath9 and @xmath11 were assumed to be finite sets . introducing @xmath30 and @xmath31 ( with `` @xmath32 '' denoting the cardinality of a set )
, we see that any tm is completely specified by @xmath33 many quintuples .
note that we also need a rule for the machine if it scans a blank symbol , hence the factor @xmath34 . in view of these facts
it seems very remarkable that , first of all , tms are capable of universal computation and , second , that they can show an incredibly complex behaviour .
the first property is related to the church - turing thesis which has to be taken for granted though which states that every intuitvely computable function can be computed by a tm .
the second property is reflected , for instance , in the fact that it is _ impossible _ to design a tm @xmath35 , which tells us for an arbitrary given tm @xmath7 and input @xmath29 whether @xmath28 will halt or not .
this is the famous halting problem .
thus , it might be that the computation defined above never reaches the state h and goes on forever . in this case
, the computation simply has no result .
thus , the incredible power of tms ( namely their computational universality ) has a serious drawback ( namely their in general unpredictable behaviour ) .
a much more detailed account of tms can be found in refs .
sketch of the general setup , which allows us to analyze any abstract computational process in terms of thermodynamic quantities .
the upper tape corresponds to the output tape , which is initially blank , and the lower tape corresponds to the input tape . note that we will always assume that blank tapes are for free , i.e.
, the machine can have as many blank space on which it can write as it requires .
furthermore , the machine itself has access to two additional internal tapes ( not sketched ) , see sec .
[ sec logical reversible tm ] .
, scaledwidth=45.0% ] examining the thermodynamics of computation can be done in many different ways . here ,
as stressed in the introduction , we want to capture three main features : first , we want to look at a general computational problem and not one specific task ; second , we are interested in a logically reversible computer ; and third , we want to model the computation stochastically , i.e. , as a brownian computer .
the general thermodynamic picture is hence as sketched in fig .
[ fig td comp ] .
some machine ( with a very complex interior in general , see below ) is coupled to a thermal reservoir at inverse temperature @xmath2 and a work reservoir .
the work reservoir can be used to drive the computation in a certain direction whereas the thermal reservoir is equipped with a well - defined notion of heat and entropy .
the task of our machine is to compute .
it therefore receives input signals and transforms them to output signals corresponding to the result of the computation .
logical reversibility demands that we use two separate tapes for the inputs and outputs ( henceforth called the input and output tapes , respectively ) .
in fact , if we do not keep the initial inputs but simply overwrite them with the output ( as the tm from sec .
[ sec tms ] would do ) , our machine will be in general irreversible . and @xmath6 , which are given as their binary equivalent on the input string .
clearly , knowing only the result @xmath36 of the computation , i.e. , @xmath37 , does not allow us to infer the values of @xmath38 and @xmath6 and hence , our machine would be logically irreversible . ] more specifically , a logically reversible computer is in principle able to unambiguously retrace its _ computational path _
( i.e. , the sequence of logical states visited so far ) back to the initial state .
in fact , most tms as introduced in sec .
[ sec tms ] are logically irreversible , for instance , already due to the fact that the machines usually do not remember from which direction they were coming from . but even if it would remember this ( for instance by writing the direction @xmath39 on an additional tape ) , the machine might still be logically irreversible .
consider for example that there exists a pair of states and symbols @xmath40 and @xmath41 such that @xmath42 , @xmath43 and @xmath44 .
then , given the state @xmath45 [ or even @xmath46 , the machine has two possible predecessor states and it can not know from which it was coming .
hence , it is logically irreversible .
this situation is sometimes called the merging of two computational paths , see fig . [ fig comp paths ]
. however , even a logically reversible computer still proceeds deterministically step by step along its computational path .
this deterministic behaviour unambigously defines a _ computational direction _ , see fig .
[ fig comp paths ] again ( we remark that the computational direction does not necessarily coincide with the `` physical '' direction of the movement of the tape , which can as we have seen be shifted either way ) .
in contrast , we also want to look at a stochastic machine , which makes random transitions in both directions , i.e. , it is also allowed to jump back to a previous computational state . in order to assure that we can control the direction of computation _ on average _ , we demand that we can change the potential energy externally , for instance , by using a work source or by adjusting chemical potentials appropriately . in the next section we will present the main ideas of how to obtain a logically reversible tm from an irreversible tm as discussed in sec . [ sec tms ] sparing the mathematical details to the appendix .
then , given that we have a logically reversible tm , we will show how to associate a me to it in sec .
[ sec stochastic tm subsec ] .
a computational path is defined by the way the machine proceeds from state to state through a high - dimensional state space ( each state of the machine including the tapes is symbolized by a black circle ) . here
, we sketched two different paths ( solid red and dashed blue lines ) . for standard machines ( which compute only from left to right ) each path
is uniquely determined by the input signal string and the initial machine state . for a logically irreversible machine ,
however , it might happen that two different paths merge at some point ( as shown on top ) making it impossible to find the inverse of a state in general ( that is to say there is a unique way to go from left to right in the diagram but not from right to left ) .
in contrast , this can not happen for logically reversible machines ( as shown in the middle ) , but this feature comes at the cost of introducing additional states and tapes making the state space even larger . finally , whereas the standard logically reversible tm still proceeds deterministically from left to right , a stochastic tm jumps randomly according to some rates [ e.g. , @xmath47 and @xmath48 as used in eqs .
( [ eq me one step process ] ) and ( [ eq help 1 ] ) ] and hence , it can move both ways ( bottom ) .
, scaledwidth=45.0% ] the procedure to make a computation of a tm logically reversible was explained in detail in a famous publication by bennett @xcite .
he explicitly showed given an arbitrary tm @xmath7 as described by eq .
( [ eq tm standard map ] ) how to construct a machine @xmath49 which is computationally equivalent to @xmath7 but always logically reversible . to accomplish this , bennett introduced a finite number of new machine states and two additional tapes , which record the previous computational steps taken .
furthermore , the computation is broken up into three _ stages _ ( each stage in general consists of many single computational steps ) and in total , the logically reversible tm needs approximately four times as many steps as the irreversible one @xcite . before we proceed , we remark that the construction given by bennett is , of course , not unique ( as he discusses as well ) but seems to be very convenient .
it is also noteworthy that tms acting on @xmath50 tapes are not more powerful ( in terms of what they can compute ) than a tm described by eq .
( [ eq tm standard map ] ) because one can be mapped onto the other @xcite .
in addition to the treatment of bennett , who considered only a single computation @xmath26 , we want to construct a machine which continuously processes a stream of incoming input strings of the form @xmath51 .
thus , we imagine an infinite input tape with different input strings @xmath52 separated by blank symbols to mark the beginning and the end of each single input string .
the output tape then contains the results of the different computations , i.e. , it looks like @xmath53 . in this picture
our machine resembles the devices from refs .
@xcite in which also an external tape is manipulated but mainly to extract work and not for computational purposes though . [
cols="<,<,<,<,<,<",options="header " , ] our logically reversible tm thus will have in total four tapes and one computational cycle proceeds in five stages .
the four tapes are called the input , working , history and output tape . whereas the input and output tape are supplied externally ( see fig .
[ fig td comp ] ) , the working and history tape belong to the machine itself . by this
we especially want to emphasize that we require them to be blank at the end of the computation such that they are ready for usage again .
more specifically , one computational cycle consists of the following five stages ( also see table [ table ] ) : 1 . _
copy input onto working tape : _ a new input arrives at the machine on the input tape and the machine copies this input onto its working tape leaving it there in standard format at the end of the first stage .
compute : _ in this stage the actual computation is performed which finally maps the input to the output . furthermore , a history tape keeps track of the intermediate steps such that the computer would be able to retrace each step .
copy output to output tape : _ if the computation halts , the output on the working tape is copied to the output tape and the working tape is reset to its position as at the end of stage 2 .
retrace computation : _ the computer retraces all its computation such that the output on the working tape becomes the input and the history is blank again .
this stage is the inverse of stage 2 .
_ erase working tape : _ we erase the working tape with the help of the input tape such that the working tape is blank again .
note that this erasure step is logically reversible because we have an identical copy of the input on the input tape .
finally , we use an additional symbol ( @xmath54 ) to mark that we already performed a computation for the current input and the machine moves on to the next input on the input string .
stage 2 , 3 and 4 were already treated by bennett in ref .
in addition , we require stages 1 and 5 because we want that our machine works continuously and not only once .
the reader who is curious about the details of each step is refered to appendix [ app logical reversible tm ] . otherwise , instead of one big machine @xmath7 doing a computation in five stages
, it might also help to imagine five small machines @xmath55 .
our big machine @xmath7 is then nothing else than a composition or concatenation of these small machines , i.e. , @xmath56 ( similar to the composition of different functions ) .
the action of @xmath7 on the state of the four tapes ( in the order of the input , working , history and output tape , respectively ) can be written as @xmath57 eq . ( [ eq computational cycle ] ) can be regarded as a short form of table [ table ] . here , @xmath58 denotes a blank tape and @xmath59 denotes the history tape at the end of stage 2 . clearly ,
if all the machines @xmath55 are logically reversible , then also @xmath7 is . in this section
we will show how to use a continuous - time markov process to model any logically reversible tm , which we will simply call a stochastic tm .
we start , however , by repeating what a me is and what we need for a consistent thermodynamic interpretation . a continuous - time markov process describing the dynamics of a system @xmath60 corresponds to a set of states @xmath61 and an associated probability @xmath62 to be in a state @xmath63 , which changes according to a markovian first order differential equation called the me @xcite : @xmath64 here , the rate matrix @xmath65 has real - valued entries and fulfills @xmath66 for all @xmath67 .
this guarantees that probability is conserved throughout the evolution , i.e. , @xmath68 for all @xmath69 .
if we want to equip the me ( [ eq me generic ] ) with a thermodynamic interpretation @xcite , we have to associate to each state @xmath70 an energy @xmath71 and the rate matrix has to additionally fulfill a property called _ local detailed balance _ , which states that @xmath72 = -\beta(e_x - e_{x'})$ ] where @xmath2 is the inverse temperature of the environment to which the system @xmath60 has contact . note
that this automatically implies that , if @xmath73 , then also @xmath74 , i.e. , for each transition @xmath75 the reversed one @xmath76 must also exist .
this framework can also be extended to more general situations involving , e.g. , multiple environments at different temperatures @xcite , but we do not need more than the things just mentioned . now , associating a me to a logically reversible deterministic tm is in fact very easy .
all we have to do is to change the ( unidirectional ) deterministic updating rules from appendix [ app logical reversible tm ] into ( bidirectional ) probabilistic transition rules , i.e. , we allow for transitions in the computational forward direction as well as transitions which just undo the last computational step ( backward direction ) .
it is worth pointing out that the underlying state space @xmath61 of the me is in general infinite , but this is not necessarily related to the size of the input tape .
remember that due to the halting problem a computation might go on forever even if it only received a finite input .
in fact , even if the computation halts , there is no general way to give a reasonable estimate of the size of @xmath61 in advance @xcite .
however , on the other hand , the structure of the me is very simple and this is related to the fact that we build the machine in a logically reversible way .
in fact , each state @xmath63 has only two adjacent states , namely its logical predecessor and its logical successor state .
hence , our me describes a simple one - step or birth - and - death process @xcite or equivalently , according to schnakenberg @xcite , we could say that the topology of the underlying network is trivial .
in fact , if there were any branchings or loops in the underlying network , the computation would not be logically reversible anymore because then a state could have multiple predecessors or successors .
hence , quite generally we can put the final me into the form @xmath77 here , of course , @xmath78 is a multi - index denoting the entire machine and tape configuration .
let us discuss the general structure of the me a little further .
first of all , it is important to note that only the current squares of the tapes can change stochastically whereas the rest of the tapes , which is not coupled to the machine , remains fixed .
in fact , although there is an infinite number of possible different states , not all states @xmath63 are coupled with each other .
which states are coupled to each other is determined by the rules from appendix [ app logical reversible tm ] and by the input strings @xmath79 because they single out a unique computational path through the `` labyrinth '' of states in @xmath61 .
in addition , the number of transition rules is always _ finite _ and fixed as expressed in appendix [ app logical reversible tm ] .
this is true independently of the number of computations or the lengths of the input strings .
although there seem to be quite a lot of rules , note that they suffice to build a _
universal logically reversible computer_. of course , things become much easier if we relax some of the requirements . hence , in a more pictorial language we could say that the hardware ( i.e. , the set of transitions rules with the associated rates ) of our machine remains fixed , but the software ( i.e. , the inputs determining the computational path ) can change . thus , if we focus only on the computation for one input string , i.e. , on the stages 2 to 4 , the full rate matrix @xmath80 in eq .
( [ eq me generic ] ) decomposes into blocks for each input string @xmath29 , i.e. , it has the form @xmath81 and no transition between different blocks is allowed . here , we labeled the different input strings @xmath82 in some canonical way and note that each block can be infinitely large if the computation does not halt . for each @xmath83
the me describes a simple one - step process and by rearranging the states appropriately we can write each block as a tridiagonal matrix @xmath84 where @xmath47 ( @xmath85 ) denotes the forward ( backward ) rate at step @xmath50 in the me ( [ eq me one step process ] ) .
hence , to conclude , although the state space @xmath61 is extremely large , the rate matrix is also extremely _ sparse _ , i.e. , it contains only a small number of non - zero elements ( in relation to the total number of elements ) .
finally , we would need to associate a consistent energy landscape to our system and the rate of forward and backward transitions would then need to obey local detailed balance , which fixes the temperature of the environment .
we will discuss this issue in the next section .
so far we have shown that a stochastic , logically reversible tm can be modeled by a simple one - step process as given by the me ( [ eq me one step process ] ) . to interpret it thermodynamically we still need to associate a consistent energy landscape to it , which we could control externally via a work source or , alternatively , a bias in chemical potentials as it would be the case for biochemical processes .
for the sake of simplicity , we will choose below a linear energy landscape along the computational path , i.e. , the difference in energy between a logical state and its successor state is taken to be the constant @xmath86 ( i.e. , for @xmath87 the computation proceeds on average in the forward direction along a chain of states with decreasing energies ) .
this choice is in agreement with the one usually appearing in the literature @xcite . before we proceed , however , we discuss and justify this choice in more detail . first of all , in sec .
[ sec effective fpe ] we will actually discuss the thermodynamics of our model on a coarse - grained level of description .
that is to say we will be interested in the regime where the computer was running already for quite a long time such that the variance @xmath88 of the number of computational steps is large compared to unity where we defined @xmath89 . in this picture ,
@xmath86 might denote just an average slope in the energy landscape , i.e. , we explicitly allow for spatial irregularities in the energy landscape as long as they are not too large .
more specifially , if @xmath90 denotes the energy of state @xmath50 according to the me ( [ eq me one step process ] ) , we demand that @xmath91 holds for all @xmath50 and for @xmath92 of the order of the variance @xmath88 such that the energy landscape looks linear at the coarse - grained level .
second , it is worth pointing out that in fact except for the spatially allowed irregularities
no other energy landscape seems to be feasible for a general purpose computer .
the reason for this is twofold : first , we are interested in a steady state regime , i.e. , the average dissipation _ per step _ should be independent of the number of computational steps performed so far .
this demand rules out quadratic or exponential energy landscape .
second , we are also still faced with the halting problem .
this implies that we can not know in advance the number of computational steps we need for one computational cycle .
thus , associating any particularly shaped energy landscapes like a sine or a hill ( as in @xcite ) is unfeasible because we do not know , for instance , how to choose a senseful period for the sine .
hence , we conclude : the only feasible energy landscape with which we can ensure to control the speed and direction of computation independently of the number of computational steps ( which we can not know in advance ) and which is translationally invariant on the state space @xmath78 of the me ( [ eq me one step process ] ) is an on average linear landscape .
having agreed on the ( on average ) linear energy landscape we will choose the transition rates in eq .
( [ eq me one step process ] ) as follows : @xmath93 here , @xmath94 is some rate setting the overall time - scale of our problem and we see that the rates fulfill local detailed balance , i.e. , @xmath95 = -\beta ( e_n - e_{n+1 } ) = -\beta\epsilon$ ] where @xmath96 favors a computation in the forward direction . in the limit where the mean @xmath97 and variance @xmath88 are large compared to one , we can approximate derivatives by @xmath98 then
, the fokker - planck equation ( fpe ) corresponding to the me ( [ eq me one step process ] ) reads @xmath99 p_n(t).\ ] ] this fpe describes the movement of an overdamped brownian particle in a constant force field and with a constant diffusion coefficient with @xmath100 denoting the position of the particle .
it even admits an explicit solution : assuming that the machine has started initially at some fixed state @xmath101 , i.e. , @xmath102 , we obtain @xmath103 ^ 2}{4\gamma\cosh(\beta\epsilon/2)t}\right\}. \end{split}\ ] ] discussing the thermodynamic behaviour of eq .
( [ eq fpe ] ) can be done using standard methods , see e.g. ref .
we first of all compute the mean number of computational steps , which is @xmath104 and hence , the _ speed _ of computation becomes @xmath105 which as expected can be controlled by @xmath86 .
especially , we see that we have @xmath106 for @xmath87 and vice versa .
the variance of the distribution is @xmath107 and grows linearly with time .
based on this we might ask the question when does the computation become approximately deterministic , i.e. , when does the mean dominate the standard deviation ? calculating their ratio yields @xmath108 if we want this quantity to be much larger than one , we obtain a condition for the minimum amount of time we have to wait until our machine computes almost in a deterministic fashion : @xmath109 here , we performed an expansion in @xmath110 at the end . thus , the closer we get to the reversible limit , i.e. , the smaller @xmath111 ( see below ) , the longer we have to wait until the computer starts to work reliably . furthermore , we can explicitly calculate the shannon entropy of our distribution , which is @xmath112 using eq .
( [ eq average n ] ) we can also write the shannon entropy as @xmath113,\ ] ] i.e. , the shannon entropy scales with the average number @xmath114 of computational steps as @xmath115 .
the rate at which entropy is produced is given by the change in shannon entropy plus @xmath2 times the heat flow dissipated into the environment @xcite . since the latter is simply @xmath116 , we can write for the entropy production rate @xmath117 we now note that for @xmath118 the last term vanishes quadratically , i.e. , the heat dissipated can be made arbitrarily small in this limit . the first term , however , is independent of @xmath86 but vanishes for @xmath119 .
hence , we have @xmath120 thus , a brownian computer can work thermodynamically reversibly ( i.e. , with zero entropy production rate ) in the steady state regime if the bias @xmath86 is small enough .
this would confirm the conclusions from ref .
@xcite .
however , we can also confirm norton s perspective on the matter @xcite .
starting initially at @xmath121 we see that the total amount of entropy produced up to time @xmath69 is @xmath122 thus , even for @xmath123 the shannon entropy still grows logarithmically with the number of computational steps because the probability distribution of the machine spreads over the available phase space similarly to the free expansion of a one - molecule gas , which is a thermodynamically irreversible process .
if we think in terms of thermodynamic cycles instead of a thermodynamic machine , which works in the stationary regime , we would have to dissipate an amount of entropy proportional to @xmath124 to reset the brownian computer to its initial zero entropy state .
however , again , compared to the number of computational steps @xmath114 taken , the ratio @xmath125 becomes arbitrarily small for a large number of steps , i.e. , for a long computation .
finally , we remark that a full treatment in terms of the me ( [ eq me one step process ] ) instead of the fpe ( [ eq fpe ] ) would provide very similar results .
in fact , the first two cumulants ( mean and variance ) can be shown to coincide .
let us summarize our main findings and discuss possible interesting open questions based on our findings . we have started with an arbitrary tm as a model for a general purpose computer , which is , however , in general logically irreversible and free from any thermodynamic interpretation .
we then discussed how a computation can be regarded as a thermodynamic process and we decided to investigate the thermodynamics of a logically reversible stochastic computer , which simulates the tm considered at the beginning .
we explained in detail how to obtain a logically reversible computer out of an irreversible one and then used bidirectional probabilistic transition rules instead of unidirectional deterministic rules to model the stochastic or random motion of our computer .
although the problem seems to be very complex , we have seen that the resulting me has the very simple structure of a one - step process , which describes how the computer follows a well - defined path in a large state space and eventually maps the input signals to output signals ( the results of the computation ) .
we have argued that the only feasible energy landscape for such a computer is an approximately linear one , which not only simplified the thermodynamic discussion , but also allowed us to solve a corresponding fpe describing the drift and diffusion of the computer exactly . we have then seen that our stochastic computer can work thermodynamically reversibly , i.e. , in a dissipation - free fashion , in a steady state regime and in this respect feynman was indeed right with his initially quoted statement .
however , just because the entropy production rate can become arbitrary small , this does not imply that the overall integrated entropy production is zero .
especially , if we think in terms of a computational cycle , in which we want to reset the computer to its initial state at the end , there is an unavoidable cost due to the increasing shannon entropy of the probability distribution during the computation .
in fact , this additional cost is _ not _ independent of the number of computational steps but scales logarithmically with it and it seems that this effect has been only recognized by norton so far @xcite . here
, we have verified this result in a conceptually clean and general framework .
furthermore , it is worth emphasizing that our computer works _ error - free _ at a _
finite _ entropy production .
in fact , by construction our model does only allow for temporary errors in the computation ( due to the fact that our stochastic machine can randomly hop back to its previous state ) , but in the long run each temporary error is corrected by the next step in the computational forward direction and there are no other sources of errors allowed .
including errors ( for instance , random bit flips or to avoid the halting problem and an infinitely long computation
one could decide to terminate the computation after a fixed number @xmath126 of steps ) in our scheme and investigating the thermodynamic cost to correct them might be an interesting project for the future .
another interesting question is whether we can sensefully assign a notion of _ efficiency _ to our computer . from a purely physical point of view
the machine we have considered is actually senseless because it describes only a simple ( and never - ending ) relaxation process .
however , the machine is indeed `` working '' , i.e. , doing something `` useful '' for us , because it tells us the answer to many questions . but how can we quantify the usefulness of our machine ? having a rigorous notion of a thermodynamic efficiency for a computer would allow us to study question of , e.g. , efficiency at maximum power , which is an important question for the design of realistic machines , see , e.g. , @xcite . in this context , one can also ask the question whether a logically reversible computer is really desirable or whether a logically irreversible computer might indeed be able to work at a fundamentally better efficiency . at the end
, it seems that biochemical processes in our body work very efficiently , but not necessarily in a logically reversible way . finally , let us say a few words about the relation between the present work and the devices investigated in refs .
@xcite ( also see feynman for a simple example of such a device who called them information - driven engines @xcite ) . indeed , as in our case , these information - driven engines are coupled to an external tape or `` information reservoir '' .
this additional reservoir can then be used to extract work from a single heat bath while simultaneously writing information on the tape ( i.e. , increasing its shannon entropy ) .
this picture , however , does not carry over to our situation .
in fact , the shannon entropies of the incoming and outgoing tapes are _
equal _ because the input tape gets mapped to itself and the output tape is uniquely determined by the input . for a logically irreversible computer this does not need to be true as it can be already seen from the devices in refs .
@xcite where it was also shown that they can be used as an information eraser .
financial support by the dfg ( scha 1646/3 - 1 , sfb 910 , and grk 1558 ) is gratefully acknowledged .
33 r. landauer , _ irreversibility and heat generation in the computing process _ , ibm j. res
. dev . * 5 * , 183 ( 1961 ) . c. h. bennett , _ logical reversibility of computation _ , ibm j. res
* 17 * , 525 ( 1973 ) . c. h. bennett , _ the thermodynamics of computation a review _ , int .
j. theor . phys . * 21 * , 905 ( 1982 ) . c. h. bennett ,
_ notes on the history of reversible computation _ , ibm j. res
. dev . * 32 * , 16 ( 1988 ) .
r. w. keyes and r. landauer , _ minimal energy dissipation in logic _
, ibm j. res .
* 14 * , 152 ( 1970 ) .
k. likharev , _ classical and quantum limitations on energy consumption in computation _ , int
* 21 * , 311 ( 1982 ) . c. h. bennett and r. landauer , _ the fundamental physical limits of computation _ , scientific american * 253 * , 48 ( 1985 ) .
r. p. feynman , _
feynman lectures on computation _ ( addison - wesley publishing company , 1996 ) .
h. s. leff and a. f. rex ( editors ) , _
maxwell s demon 2 : entropy , classical and quantum information , computing _
( iop publishing , bristol , 2003 ) .
j. earman and j. d. norton , _ exorcist xiv : the wrath of maxwell s demon .
part i. from maxwell to szilard _ , stud .
mod . phys . * 29 * , 435 ( 1998 ) .
j. earman and j. d. norton , _ exorcist xiv : the wrath of maxwell s demon .
part ii . from szilard to landauer and beyond _ , stud .
. mod . phys . * 30 * , 1 ( 1999 ) .
o. r. shenker , _
maxwell s demon and baron munchausen : free will as a perpetuum mobile _ , stud .
phys . * 30 * , 347 ( 1999 ) .
maxwell s demon and the thermodynamics of computation _ , stud .
mod . phys . * 32 * , 569 ( 2001 ) . c. h. bennett , _ notes on landauer s principle , reversible computation , and maxwell s demon _ , stud .
* 34 * , 501 ( 2003 ) . o. j. e. maroney , _ the ( absence of a ) relationship between thermodynamic and logical reversibility _ , stud .
phys . * 36 * , 355 ( 2005 ) .
j. d. norton , _ eaters of the lotus : landauer s principle and the return of maxwell s demon _ , stud .
36 * , 375 ( 2005 ) .
j. ladyman , s. presnell , a. j. short , and b. groisman , _ the connection between logical and thermodynamic irreversibility _ , stud .
38 * , 58 ( 2007 ) .
m. hemmo and o. r. shenker , _
maxwell s demon _ ,
j. philos . *
107 * , 389 ( 2010 ) .
j. d. norton , _ waiting for landauer _ , stud . hist .
* 42 * , 184 ( 2011 ) .
l. b. kish and c. g. granqvist , _ energy requirement of control : comments on szilard s engine and maxwell s demon _
. lett . * 98 * , 68001 ( 2012 ) .
j. d. norton , _ brownian computation is thermodynamically irreversible _ , found .
phys . * 43 * , 1384 ( 2013 ) .
j. d. norton , _ all shook up : fluctuations , maxwell s demon and the thermodynamics of computation _
, entropy * 15 * , 4432 ( 2013 ) .
kish , c. g. granqvist , s. p. khatri , and h. wen , _ demons : maxwell s demon , szilard s engine and landauer s erasure - dissipation _ , international journal of modern physics : conference series * 33 * ( 2014 ) .
r. alicki , _ information is not physical _ , arxiv : 1402.2414 .
r. alicki , _ breaking information - thermodynamics link _ , arxiv : 1406.5879 . j. d. norton , _ on brownian computation _ , int .
ser . * 33 * , 1460366 ( 2014 ) .
m. l. minsky , _ computation : finite and infinite machines _
( prentice - hall , london , 1967 ) .
j. schnakenberg , _ network theory of microscopic and macroscopic behavior of master equation systems _ , rev .
* 48 * , 571 ( 1976 ) . n. g. van kampen , _
stochastic processes in physics and chemistry _
( north - holland publishing company , amsterdam , 3rd edition , 2007 ) .
u. seifert , _
stochastic thermodynamics , fluctuation theorems and molecular machines _ , rep .
phys . * 75 * , 126001 ( 2012 ) .
a. e. allahverdyan , d. janzing , and g. mahler , _ thermodynamic efficiency of information and heat flow _ , j. stat
* p09011 * ( 2009 ) .
d. mandal and c. jarzynski , _ work and information processing in a solvable model of maxwell s demon _ , proc .
sci . * 109 * , 11641 ( 2012 ) .
a. c. barato and u. seifert , _ an autonomous and reversible maxwell s demon _ , europhys .
* 101 * , 60001 ( 2013 ) .
d. mandal , h. t. quan , and c. jarzynski , _
maxwell s refrigerator : an exactly solvable model _
lett . * 111 * , 030602 ( 2013 ) .
t. munakata and m. l. rosinberg , _ feedback cooling , measurement errors , and entropy production _ , j. stat .
p06014 * ( 2013 ) .
n. shiraishi , s. ito , k. kawaguchi , and t. sagawa , _ role of measurement - feedback separation in autonomous maxwell s demons _ , new . j. phys .
* 17 * , 045012 ( 2015 ) .
s. deffner and c. jarzynski , _ information processing and the second law of thermodynamics : an inclusive , hamiltonian approach _ ,
phys . rev .
x * 3 * , 041003 ( 2013 ) .
d. hartich , a. c. barato , and u. seifert , _
stochastic thermodynamics of bipartite systems : transfer entropy inequalities and a maxwell s demon interpretation _ , j. stat .
p02016 * ( 2014 ) . j. m. horowitz and m. esposito , _ thermodynamics with continuous information flow _ ,
x * 4 * , 031015 ( 2014 ) .
a. c. barato and u. seifert , _
stochastic thermodynamics with information reservoirs _ , phys .
e * 90 * , 042150 ( 2014 ) .
n. shiraishi and t. sagawa , _ fluctuation theorem for partially masked nonequilibrium dynamics _ phys .
e * 91 * , 012130 ( 2015 ) .
a. e. allahverdyan and q. a. wang , _ adaptive machine and its thermodynamic costs _ , phys .
e * 87 * , 032139 ( 2013 ) .
a. c. barato , d. hartich , and u. seifert , _ information - theoretic versus thermodynamic entropy production in autonomous sensory networks _ phys .
e * 87 * , 042104 ( 2013 ) .
a. c. barato , d. hartich , and u. seifert . ,
_ efficiency of cellular information processing _
, new j. phys . * 16 * , 103024 ( 2014 ) .
p. sartori , l. granger , c. f. lee , and j. m. horowitz , _ thermodynamic costs of information processing in sensory adaptation _ , plos comput .
biol . * 10*(12 ) , e1003974 ( 2014 ) .
p. sartori and s. pigolotti , _ thermodynamics of error correction _
, arxiv : 1504.06407 .
p. strasberg , g. schaller , t. brandes , and m. esposito , _ thermodynamics of a physical model implementing a maxwell demon _ , phys .
* 110 * , 040601 ( 2013 ) .
p. strasberg , g. schaller , t. brandes , and c. jarzynski , _ the second laws for an information driven current through a spin valve _ ,
e * 90 * , 062107 ( 2014 ) .
a. brut , a. arakelyan , a. petrosyan , s. ciliberto , r. dillenschneider , and e. lutz , _ experimental verification of landauer s principle linking information and thermodynamics _ ,
nature * 483 * , 187189 ( 2012 ) .
y. jun , m. gavrilov , and j. bechhoefer , _ high - precision test of landauer s principle in a feedback trap _ , phys .
* 113 * 190601 ( 2014 ) .
j. hong , b. lambson , s. dhuey , and j. bokor , _ experimental verification of landauer s principle in erasure of nanomagnetic memory bits _
, arxiv : 1411.6730 .
j. p. p. silva , r. s. sarthour , a. m. souza , i. s. oliveira , j. goold , k. modi , d. o. soares - pinto , and l. c. cleri , _ experimental demonstration of information to energy conversion in a quantum system at the landauer limit _
, arxiv : 1412.6490 .
a. brut , a. petrosyan , and s. ciliberto , _ information and thermodynamics : experimental verification of landauer s erasure principle _ ,
k. sekimoto , _ stochastic energetics _ ( lect . notes phys .
799 , springer , berlin heidelberg , 2010 ) .
f. l. curzon and b. ahlborn , _ efficiency of a carnot engine at maximum power output _ am .
* 43 * , 22 ( 1975 ) . c. van den broeck , _ thermodynamic efficiency at maximum power _ phys .
lett . * 95 * , 190602 ( 2005 ) . t. schmiedl and u. seifert , _ efficiency at maximum power : an analytically solvable model for stochastic heat engines _
, europhys . lett . *
81 * , 20003 ( 2008 ) .
m. esposito , k. lindenberg , and c. van den broeck , _ universality of efficiency at maximum power _
* 102 * , 130602 ( 2009 ) .
m. esposito , k. lindenberg , and c. van den broeck , _ thermoelectric efficiency at maximum power in a quantum dot _ , europhys .
lett . * 85 * , 60010 ( 2009 ) .
we here provide the detailed rules for the machine behaviour at each stage of the computation . these rules are , of course , not unique . however , because we are not primarily interested in the speed or efficiency of our machine , but only in _ what _ it can do , possibly different implementations are unimportant for the present context .
furthermore , note that at each stage the machine is only manipulating two tapes whereas the other two tapes remain fixed ( see table [ table ] ) .
we will therefore use the notation @xmath127 $ ] where @xmath128 denotes the internal machine state at stage @xmath129 , @xmath130 the symbol @xmath5 printed on square @xmath131 of the first tape of interest and @xmath132 the symbol @xmath69 printed on square @xmath50 of the second tape of interest .
what are the tapes of interest will become clear in the treatment of each stage .
note that the notation is different from the one used by bennett @xcite .
we want to copy the input on the input tape ( first tape of interest ) to the working tape ( second tape of interest ) .
the input is given in the form @xmath133 and we assume that the machine scans initially the symbol at the far right ( i.e. , @xmath134 ) ( see also table [ table ] ) . furthermore , the working tape is by construction initially completely blank ( that this is so can only be seen after the completion of all five stages , of course ) .
the copy stage then proceeds as follows : @xmath135 \rightarrow&~ [ q_1^{(1 ) } , ( s_m)_m , ( s_m)_n ] \\ \rightarrow&~ [ q_0^{(1 ) } , ( s_{m-1})_{m-1 } , b_{n-1 } ] \\ \rightarrow&~ [ q_1^{(1 ) } , ( s_{m-1})_{m-1 } , ( s_{m-1})_{n-1 } ] \\ & ~ \vdots \\ \rightarrow&~ [ q_1^{(1 ) } , ( s_1)_{m-(m-1 ) } , ( s_1)_{n-(m-1 ) } ] \\ \rightarrow&~ [ q_0^{(1 ) } , b_{m - m } , b_{n - m } ] \\ \rightarrow&~ [ \text{r}^{(2)},b_{m - m } , b_{n - m } ] .
\end{split}\ ] ] hence , we see that during the copy operation the machine changes between the two states @xmath136 and @xmath137 where the first is responsible for copying the symbol on the input tape to the working tape and the second is responsible for a shift of both tapes .
this procedure goes on until it hits the first blank symbol to the left of the input string .
the machine then changes to the `` ready '' state r@xmath138 , which is the special initial state for the second stage .
note that due to the fact that the copying procedure is unidirectional , i.e. , the machine moves the tape always in the same direction
each map has a clear logical inverse .
furthermore , the positions @xmath131 and @xmath50 of the squares are in general arbitrary and can be chosen initially at will .
this is the central stage of our computational cycle .
if we would not bother about logical reversibility , this would be the only stage to execute .
we thus have to explicitly think about how to make the map ( [ eq tm standard map ] ) logically reversible .
the idea is the following @xcite : first , because each standard ( i.e. , irreversible ) tm is defined by its @xmath33 many quintuples , we introduce a set @xmath139 with @xmath140 of additional machine states @xmath141 .
then , to each map @xmath142 we can associate a special state @xmath143 , which _ uniquely _ labels each quintuple .
second , we will make use of an additional tape called the history tape , which remembers the past @xmath143 such that the machine is able to uniquely retrace its computational path . stage 2 is thus a bit more complicated .
the corresponding maps are @xmath144&~~~\rightarrow [ \tilde q^{(2)}_{r^{(2)}b } , b_{n - m } , b_{m+1 } ] \\ & ~~~\rightarrow [ q_1 , ( s_1)_{n - m+1 } , ( \tilde q_{\text{r}^{(2)}b})_{m+1 } ] \\ & ~~~ ~~~ ~~ \vdots \\ \text{step } \ell & \left\{\begin{array}{ll } \rightarrow & [ q_\ell^{(2)},(s_\ell)_{n'},(\tilde q_{q_{\ell-1}^{(2)}s_{\ell-1}})_{m+\ell } ] \\ \rightarrow & [ \tilde q^{(2)}_{q_{\ell}^{(2)}s_{\ell}},(s'_\ell)_{n'},b_{m+\ell+1 } ] \\ \rightarrow & [ q'^{(2)}_{\ell},(s_{\ell+1})_{n'+d'},(\tilde q_{q_{\ell}^{(2)}s_{\ell}})_{m+\ell+1 } ]
\\ \end{array}\right . \\
& ~~~ ~~~ ~~ \vdots \\ \text{step } \nu & ~~~\rightarrow [ \tilde q^{(2)}_{q_{\nu}^{(2)}s_{\nu}},(s'_{\nu})_{n''-1},b_{m+\nu+1 } ] \\ & ~~~\rightarrow [ \text{h}^{(2)},b_{n''},(\tilde q_{q_\nu^{(2)}s_\nu})_{m+\nu+1 } ] .
\end{split}\ ] ] the most important step to understand is the one in the middle , which corresponds to the @xmath145th computational step of the ordinary irreversible tm defined by ( [ eq tm standard map ] ) .
initially , the machine is in state @xmath146 and scans the symbol @xmath147 on the square @xmath148 of the working tape .
the history tape contains the state @xmath149 , which uniquely labels the _ previous _ computational step .
then , the machine changes its state to @xmath150 , writes the symbol @xmath151 on the square according to the function @xmath152 and shifts the history tape one square to the left , which contains a blank symbol .
finally , we write @xmath153 to the history tape and change the machine state to @xmath154 .
furthermore , we shift the working tape one square according to @xmath155 such that the machine now scans the new symbol @xmath156
. then , the whole procedure can start again where in order that we are able to apply map ( [ eq tm standard map ] ) we identify @xmath157 because the final state of the machine at the end of step @xmath145 is the initial state for step @xmath158 . the first and last two lines of eq .
( [ eq stage 2 explicit ] ) then simply describe the initial and final steps of the computation .
initially , the machines starts in r@xmath138 and then shifts the working tape to the left such that it reads the first symbol @xmath22 and starts in the state @xmath159 . finally , if the machine halts , it reaches the state h and stops at the first blank to the right of the output of the computation .
we now want to copy the output @xmath160 of the computation from stage 2 from the working tape ( first tape of interest ) to the output tape ( second tape of interest ) .
this goes as follows @xmath161 \rightarrow&~ [ q_0^{(3 ) } , b_{n '' } , b_m ] \\ \rightarrow&~ [ q_1^{(3 ) } , ( s_{m'})_{n''-1 } , b_{m-1 } ] \\ \rightarrow&~ [ q_0^{(3 ) } , ( s_{m'})_{n''-1 } , ( s_{m'})_{m-1 } ] \\ & ~ \vdots \\ \rightarrow&~ [ q_0^{(3 ) } , ( s_1)_{n''-m ' } , ( s_1)_{m - m ' } ] \\ \rightarrow&~ [ q_1^{(3 ) } , b_{n''-m'-1 } , b_{m - m'-1 } ] \end{split}\ ] ] and is very similar to ( [ eq stage 2 explicit ] ) .
finally , however , to prepare the machine for the next stage , we want that it scans the output on the working tape at the very right again ( at the moment it scans the blank on the working tape to the left of the output ) . to accomplish this
we use two more machine states : @xmath162 \\ & \rightarrow [ q_2^{(3 ) } , b_{n''-m'-1 } , b_{m - m'-1 } ] \\ & \rightarrow [ q_3^{(3 ) } , ( s_1)_{n''-m ' } , b_{m - m'-2 } ] \\ & \rightarrow
[ q_3^{(3 ) } , ( s_2)_{n''-m'+1 } , b_{m - m'-3 } ] \\ & ~~~~~ \vdots \\ & \rightarrow [ q_3^{(3 ) } , ( s_{m'})_{n''-1 } , b_{m-2 m ' } ] \\ & \rightarrow [ q_3^{(3 ) } , b_{n '' } , b_{m-2m'-1 } ] . \end{split}\ ] ] here , @xmath163 is an intermediate state and its sole purpose is to indicate that the copying procedure is over and the machine starts now only to shift the working tape without changing it .
the state @xmath164 then actually accomplishes this task by shifting the working tape one step to the left while simultaneously shifting the output tape one step to the right until it reaches the first blank symbol to the right of the output on the working tape , which will indicate the start of stage 4 . in this stage
we will basically apply the inverse of stage 2 such that at the end the working tape contains the input again and the history tape is returned to its initial blank state .
this goes as follows : @xmath165 \\ & \rightarrow [ \text{h}^{(4 ) } , b_{n '' } , ( \tilde q_{q_\nu^{(2)}s_\nu})_{m+\nu } ] \\ & \rightarrow [ \tilde q^{(4)}_{q_\nu^{(2)}s_\nu } , ( s'_\nu)_{n''-1},b_{m+\nu } ] \\ & \rightarrow [ q_\nu^{(4 ) } , ( s'_{\nu-1})_{n''-1},(\tilde q_{q_{\nu-1}^{(2)}s_{\nu-1}})_{m+\nu-1 } ] \\ & ~~~~~ \vdots \\ & \rightarrow [ q'^{(4)}_\ell , ( s_{\ell+1})_{n'+d ' } , ( \tilde q_{q_\ell^{(2)}s_\ell})_{m+\ell+1 } ] \\ & \rightarrow [ \tilde q^{(4)}_{q_\ell^{(2)}s_\ell } , ( s'_\ell)_{n ' } , b_{m+\ell+1 } ] \\ & \rightarrow [ q_\ell^{(4 ) } , ( s_\ell)_{n ' } , ( \tilde q_{q_{\ell-1}^{(2)}s_{\ell-1}})_{m+\ell } ] \\ & ~~~~~ \vdots \\ & \rightarrow [ q_1^{(4 ) } , ( s_1)_{n - m+1 } , ( \tilde q_{\text{r}^{(2)}b})_{m+1 } ] \\ & \rightarrow [ \tilde q^{(4)}_{\text{r}^{(2)}b } , b_{n - m } , b_{m+1 } ] \\ & \rightarrow [ \text{r}^{(4 ) } , b_{n - m } , b_m ] . \end{split}\ ] ] note that we are using the superscript @xmath166 on the _ internal _ machine states to explicitly distinguish them from the states of stage 2 indicating that we are truly in a different stage here .
the last step consists in erasing the content on the working tape such that it is blank again and ready for the next computation .
note that the `` erasure '' of the working tape does not actually erase any information because the working tape contains the same input @xmath133 as the input tape .
as in stage 1 we choose the first tape of interest to be the input tape and the second tape of interest is the working tape .
then , we actually only have to apply the inverse of stage 1 , i.e. , @xmath167 \rightarrow&~ [ q_0^{(5)},b_{m - m } , b_{n - m } ] \\ \rightarrow&~ [ q_1^{(5)},(s_1)_{m-(m-1 ) } , ( s_1)_{n-(m-1 ) } ] \\ \rightarrow&~ [ q_0^{(5)},(s_1)_{m-(m-1 ) } , b_{n-(m-1 ) } ] \\ & ~ \vdots
\\ \rightarrow&~ [ q_1^{(5)},(s_m)_{m } , ( s_m)_{n } ] \\ \rightarrow&~ [ q_0^{(5)},(s_m)_{m } , b_{n } ] \\ \rightarrow&~ [ q_1^{(5)},b_{m+1 } , b_{n+1 } ] .
\end{split}\ ] ] if we would postulate the final transition rule @xmath168 \rightarrow [ q_0^{(1 ) } , ( s_m)_m , b_n]$ ] we would be exactly back at the initial state of stage 1 and the only change would be that we have printed the result of the computation on the output tape .
however , if this were true , we would be doomed to repeat the same computation again , while in fact we want to compute with the _ next _ input @xmath169 on the input string . to achieve this we add the following rules : @xmath170 \rightarrow&~ [ q_2^{(5 ) } , \star_{m+1 } , b_{n+1 } ] \\
\rightarrow&~ [ q_2^{(5 ) } , ( s_m)_{m } , b_{n+2 } ] \\ \rightarrow&~ [ q_2^{(5 ) } , ( s_{m-1})_{m-1 } , b_{n+3 } ] \\ & ~ \vdots \\
\rightarrow&~ [ q_2^{(5 ) } , ( s_{1})_{m - m } , b_{n+m+1 } ] \\ \rightarrow&~ [ q_2^{(5 ) } , b_{m - m-1 } , b_{n+m+2 } ] \\ \rightarrow&~ [ q_3^{(5 ) } , b_{m - m-1 } , b_{n+m+2 } ] \\ & ~ \vdots \\ \rightarrow&~ [ q_3^{(5 ) } , ( s'_{m'})_{m ' } , b_{\hat n } ] \\
\rightarrow&~ [ q_0^{(1 ) } , ( s'_{m'})_{m ' } , b_{\hat n } ] . \end{split}\ ] ] here , we first of all marked the input @xmath79 with a @xmath54 to indicate that we have done already a computation for that input .
we then used an additional state @xmath171 , which simply traverses the input string @xmath79 until it hits a blank symbol .
the machine then changes to the state @xmath172 and goes further to the left until it hits the next _ non - blank _ symbol on the input string .
this symbol then indicates the beginning of the next input @xmath169 such that starting from stage 1 again we can readily execute the next computational cycle .
suppose that the irreversible tm from sec .
[ sec tms ] has @xmath173 many internal states , @xmath174 many different symbols ( including the blank ) on the tape and hence , it has @xmath175 many quintuples . furthermore , suppose it was given an input of length @xmath176 and produced an output of length @xmath177 after @xmath178 computational steps in total
. then , our reversible machine has @xmath179 states from the first , second , ... , fifth stage of the computation ,
i.e. , in total @xmath180 states .
furthermore , it needs @xmath181 many computational steps .
here , the @xmath70 denotes the number of unknown blank symbols separating the current input from the next input on the input tape ( see stage 5 ) .
note that we need @xmath182 such that there is enough space for the symbol @xmath54 and to guarantee that all input strings ( including potentially the symbol @xmath54 ) are separated by at least one blank symbol from eachother . | in analogy to brownian computers we explicitly show how to construct stochastic models , which mimic the behaviour of a general purpose computer ( a turing machine ) .
our models are discrete state systems obeying a markovian master equation , which are logically reversible and have a well - defined and consistent thermodynamic interpretation . the resulting master equation , which describes a simple one - step process on an enormously large state space , allows us to thoroughly investigate the thermodynamics of computation for this situation . especially , in the stationary regime we can well approximate the master equation by a simple fokker - planck equation in one dimension .
we then show that the entropy production rate at steady state can be made arbitrarily small , but the total ( integrated ) entropy production is finite and grows logarithmically with the number of computational steps . |
for the past twenty years , much work has been done to understand the spin structure of the nucleons .
there has been progress in determining the contribution of the lightest quarks to the spin , but there is still uncertain knowledge about the gluon contribution .
transversity studies have contributed additional insight about quark dynamics , but little is known about the the orbital angular momentum of the constituents.@xcite this paper will summarize a project that provides a method of gaining insight into the nature of the orbital angular momentum of the nucleon constituents .
recent experiments @xcite have significantly lowered the measurement errors of the quark longitudinal spin contribution ( @xmath2 ) to the proton .
the compass collaboration analysis quotes a result @xmath3 while the hermes collaboration analysis quotes a result @xmath4 these groups and others @xcite have been working on providing a significant measure of the proton s spin weighted gluon density , @xmath5 where @xmath6 is the bjorken scaling variable and @xmath7 is the @xmath8 evolution variable .
the combination of these measurements is summarized in terms of the @xmath9 sum rule : @xmath10 here @xmath11 and @xmath12 are the projections of the spin carried by all quarks and the gluons on the @xmath13-axis , respectively .
also @xmath1 is the net @xmath13-component of the orbital angular momentum of the constituents .
we do not attempt to separate the flavor components of @xmath1 within the sum rule .
experimental groups at the compass , hermes and rhic collaborations are measuring both the gluon polarization and the asymmetry , @xmath14 to determine the gluon polarization @xcite .
since there is no suitable theoretical model for @xmath15 , we have devised a way to model the asymmetry , @xmath16 to gain insight into the structure of @xmath15 .
this , coupled with the @xmath9 sum rule can then shed light on the nature of the orbital angular momentum of the constituents , @xmath1 . to model @xmath16 , we write the polarized gluon asymmetry using the decomposition @xmath17 where @xmath18 \label{a0def}\ ] ] is a scale invariant calculable reference form @xcite . here
@xmath19 represents the difference between the calculated and gauge - invariant asymmetry .
since @xmath15 is unknown , a useful form is to write equation ( [ adef ] ) as @xmath20 although the quantity @xmath21 is not a physical parameter , it allows the theoretical development of the calculable quantity , @xmath22 .
once an asymmetry is generated from equations ( [ a0def ] ) and ( [ dg ] ) , the gauge - invariant quantity @xmath16 can be compared to data .
thus , each ansatz for @xmath21 gives a corresponding form for @xmath15 and a parametrization for @xmath1 .
these can be compared to existing data to provide a range of suitable models for these contributions . with the definition for the asymmetry in equation ( [ a0def ] ) , the dglap equations
can then be used to evaluate the evolution terms on the right side . @xmath23 .
\label{a00}\ ] ] the polarized gluon distribution in the numerator of equation ( [ a00 ] ) is replaced by @xmath24 . for certain unpolarized distributions ,
there are points at which the denominator vanishes .
to avoid this , we write equation ( [ a00 ] ) as : @xmath25 \\
\nonumber & = & a_0\cdot{{\partial{g}}\over { \partial{t } } } \\ & = & ( 2/\beta_0 ) a_0 \bigl[p_{gq}^{lo}\otimes q+p_{gg}^{lo}\otimes g\bigr ] .
\nonumber \label{adge}\end{aligned}\ ] ] the nlo form is essentially the same as equation ( 9 ) with the splitting functions @xmath26 replaced with their nlo counterparts .
the quark and gluon unpolarized distributions are cteq5 and the polarized quark distributions are a modified ggr set .
@xcite there are constraints on @xmath27 that must be imposed to satisfy the physical behavior of the gluon asymmetry , @xmath28 .
these are : * positivity : @xmath29 for all x , and * endpoint values : @xmath30 and @xmath31 note that the constraint of @xmath32 is built in to satisfy the assumption that the large @xmath6 parton distributions are dominated by the valence up quarks in the proton . the convolutions are dominated by the quark terms , which force the asymmetry to unity as @xmath33 . to investigate the possible asymmetry models
, we use a parameterization for @xmath22 in the form @xmath34 which automatically satisfies the constraints that @xmath30 and @xmath31 .
once a parametrization for @xmath21 is chosen , equation ( 9 ) is used to determine the parameters in equation ( [ a0form ] ) .
the models for @xmath21 that led to asymmetries that satisfied these constraints were all in the range @xmath35 , with positive and negative values included .
larger values of @xmath36 violate one or both of the constraints .
a representative sample of models that satisfy the constraints are listed in table 1 .
note that the integrals for @xmath15 are all positive , ranging from about 0.01 to 0.42 .
the models that gave negative values for these integrals did not agree with the existing asymmetry data , reported at this workshop to be : * @xmath37 at @xmath38 from compass , @xmath39 gev@xmath40 * @xmath41 at @xmath42 from compass , @xmath43 gev@xmath40 * @xmath44 at @xmath45 from hermes , factorization method * @xmath46 at @xmath47 from hermes , approximate method . the models in table 1 that are within one @xmath48 of the preliminary data stated above are in the third , fourth and sixth rows , respectively .
plots of the full asymmetry are shown in figure 1 .
none of the models in table 1 are ruled out by the data since they fall within two @xmath48 of the data for our values of @xmath39 gev@xmath40 .
all of these models except for the fourth row in table 1 ( impulses in figure 1 ) generate total asymmetries @xmath49 that are close to @xmath50 .
ironically , early assumptions of the polarized gluon assumed this functional form as a naive estimate to the asymmetry .
next - to - leading order corrections to these asymmetries tend to bring them less positive , but with the same general shape .
a full set of viable asymmetries will be presented in an upcoming paper . @xcite using the data on @xmath2 in section 1
, the relation between @xmath51 and @xmath52 can be written as : @xmath53 the three models of the asymmetry that agree most closely with existing data give values of @xmath15 in the approximate range of @xmath54 .
thus , the existing data with equation [ dglz ] imply the approximate relation @xmath55 .
thus , the contribution of the orbital motion of the constituents to the proton spin may be comparable to the total quark contribution
. a recent lattice calculation of the contribution of the quark orbital motion to the proton spin ( @xmath56 ) is consistent with zero .
@xcite thus , the gluonic orbital motion appears to provide the majority contribution to @xmath1 in the @xmath9 sum rule .
it is clear that future measurements of @xmath15 and @xmath57 must be made in a wider kinematic range of @xmath6 and @xmath8 with improved precision to better specify the appropriate model of the asymmetry and to extract the @xmath6 and @xmath8 dependence of the orbital angular momentum of the constituents . 99 x. ji , aip conf . proc . * 915 * : 16 , 2007 , x. ji , phys . rev . lett . * 78 * , 610 ( 1997 ) .
see talk by y. bedfer , this proceedings , i. savin , aip conf .
* 915 * : 399 , 2007 , and s. platchkov , nucl . phys . * a790 * : 58 , 2007 .
see talk by s. belostotski , this proceedings and l. de nardo , aip conf .
* 915 * : 404 , 2007 , see talk by g. bunce , this proceedings and r. fatemi , arxiv 0710.3207 ( hep - ex ) .
g. ramsey , proceedings of the 16th international spin physics symposium , trieste , italy , c2005 , world scientific press , p. 310 .
gordon , m. goshtaspbour and g.p .
ramsey , phys . rev . * d58 * , 094017 ( 1997 ) . y. binder , g. ramsey and d. sivers , in preparation . j. w. negele , hep - lat/0509100 . | determination of the orbital angular momentum of the proton is a difficult but important part of understanding fundamental structure .
insight can be gained from suitable models of the gluon asymmetry applied to the @xmath0 sum rule .
we have constrained the models of the asymmetry to gain possible scenarios for the angular momentum of the protons constituents .
results and phenomenology for determining @xmath1 are presented . |
entanglement entropy has been an intensively studied subject in the last few years , and indeed a primary role has been played by the well known formula for its holographic computation . given a fixed time codimension one submanifold @xmath2 , with its complement @xmath3 , and a density matrix @xmath4 , the entanglement entropy is defined to be the von neumann entropy for the reduced density matrix @xmath5 $ ] , where @xmath6 ( resp .
@xmath7 ) is the hilbert space living on @xmath2 ( resp . @xmath3 ): @xmath8 $ ] .
then , if we have a dual static euclidean background , @xcite proposed the formula @xmath9 where @xmath10 is the newton constant in the holographic @xmath11 dimensional space @xmath12 , and @xmath13 is the minimal surface area inside @xmath12 , extending towards @xmath14 where the boundary @xmath15 is located , and such that @xmath16 . not only this remarkable formula provides a relatively simple computational tool for @xmath17 , so far only possible at weak coupling using the replica trick for path integration , it also allowed many new developments in studying the properties of entanglement entropy .
we may mention for example an alternative proof of the c - theorem and possible higher dimensional generalizations @xcite ; numerous achievements in reconstructing bulk and gravity properties from field theory data , for instance @xcite @xcite @xcite and @xcite , in particular enlightening the connection between gravity equations of motion and a sort of thermodynamic first law for entanglement entropy @xcite ( extended to a zeroth and second law in @xcite ) .
further the ryu - takayanagi formula may be used to construct an incredibly simple proof of strong subadditivity @xcite , although its counterpart for time dependent backgrounds looks more tricky and deeply interconnected with the null energy condition @xcite , and for instance @xcite @xcite for the case of vaidya space .
even if many consistency checks have been provided , a proof for this simple formula was so far lacking and has been searched for a few years ; a first proposal by fursaev @xcite was later discovered by @xcite to use gravitational solutions with conical singularities .
later a proof was given at @xmath18 @xmath19 , by @xcite and @xcite , and for generic dimension but @xmath2 restricted to a sphere by @xcite .
only recently the complete generalization was provided by @xcite , through the introduction of the concept of generalized gravitational entropy ( see @xcite for some comments ) .
the proposed proof develops from the holographic correspondence between @xmath20 $ ] , with @xmath21 the non normalized time evolution operator @xmath22 , @xmath23 , and the exponential of a certain gravitational action @xmath24 @xmath25=e^{-s^{grav}_{eucl}(\beta ) } , \ ] ] with the following properties : first the new , regular background @xmath26 , on which the action is evaluated , contains a circle noncontractible on @xmath27 , along which a euclidean time coordinate @xmath28 runs .
second , when @xmath29 , there are two periodicity conditions to be imposed on the fields @xmath30 entering @xmath24 : @xmath31 and @xmath32 .
needs a priory to be defined only on @xmath27
. we will however extend it to the full bulk .
] third , if @xmath0 is analytically continued from integers to reals , so that the two periodic conditions are no longer compatible , only @xmath33 is preserved and the time integral for the action is redefined as @xmath34 . then we can compute the entropy for @xmath35 $ ] as @xmath36=\partial_{\beta}s^{grav}_{eucl}(\beta)|_{\beta=1}-s^{grav}_{eucl}(1)$ ] .
now let us restrict to the boundary @xmath27 , and consider a codimension one submanifold @xmath2 at @xmath37 inside @xmath27 , with @xmath38 a different coordinate chosen to wind around ( not along ) @xmath39 .
locally @xmath38 is the angle obtained by transforming @xmath40 and @xmath41 to polar coordinates @xmath42 , where @xmath43 span the space perpendicular to @xmath39 .
as we wind around multiple times we are naturally implementing the replica trick used for the field theory computation of @xmath17 , that is instead of path integrating over a @xmath0-fold cover of the boundary space with cuts along @xmath2 , we compactify the time direction and go around @xmath0-times , with @xmath2 as a cauchy surface on @xmath27 .
when we look for its holographic counterpart @xmath26 , we are led to the construction described in the above paragraph . and
@xmath43 on @xmath27 . on the right
we show the radial additional direction of @xmath26 and the eventual shrinking of the circle parameterized by @xmath38 , somewhere in the interior ; because of its definition , contributes to @xmath44 come only from there.,scaledwidth=90.0% ] thus the ryu - takayanagi formula can be translated to a purely gravitational computation : @xmath45 one might wonder were does the time dependence on the left hand side of the above ( [ ass2 ] ) come from , if on the right hand side we are considering a static background ?
in polar coordinates @xmath42 , the part of the regular metric at the boundary of @xmath27 spanning the plane perpendicular to @xmath46 , may be chosen to locally look like @xmath47 to avoid the proper length of the @xmath38-circle to shrink at zero at @xmath48 , we need to perform a bulk coordinate transformation such to induce a weyl rescaling on @xmath27 corresponding to a local multiplication by @xmath49 .
if the shape of @xmath46 is generic , this can be done only at the price of introducing a time dependence in the rescaled metric .
the intent of the present paper is to study in detail and explicitly check the assumption , made in @xcite , regarding the analytic continuation of @xmath0 for the gravitational solution @xmath26 , used to compute @xmath24 .
although reasonable , it is in fact not the unique choice that could have been made , and we therefore think it is compelling to better understand it .
one can of course follow a more pragmatic point of view and accept it by the a posteriori justification that it works , but this would then spoil the status of proof of the full construction .
we will work perturbatively in the time dependent deformation of the static bulk metric .
a second goal is then to construct a formalism for comparing minimal area to generalized gravitational entropies at first order , with particular care on the cutoff procedure ( that has so far been quite neglected in the literature ) and on the possible periodicity conditions for the gravitational solution .
these techniques could then be applied to more general theories where the equivalent of a minimal area formula is not known , for example in order to study the dynamics of the entanglement entropy in these cases ; we will discuss it further in the conclusions .
consider a weakly time dependent hamiltonian @xmath50 as we have discussed below equation [ ass2 ] , this would correspond to @xmath2 being a submanifold only marginally different from either half of the ads boundary or a sphere , cases for which the rescaled metric remains @xmath51independent .
the problem is that this change of shape for @xmath46 makes difficult to compute minimal surfaces , even at first order in the deformation .
another possibility for introducing weak time dependence is to consider the perturbed ads metric , with perturbations satisfying the linearized equations of motion ( and being in general time dependent ) , and @xmath2 remaining untouched .
our @xmath26 will then be ads + first order perturbations , the nontrivial @xmath0 dependence contained inside these last ones and fixed by the periodicity conditions along the time direction that we will choose .
we want to compare the generalized gravitational entropy on this background computed using different choices for the analytic continuation in @xmath0 , with the results from minimal area computation on the same background at @xmath52 .
note that , even if for time dependent lorentzian manifolds the recipe is to look for extremal surfaces instead of minimal , @xcite , as we are at first order the shape of the extremal surface remains the same as in the unperturbed ads , and it then coincides with the minimal one , while its area will obviously be modified .
the strategy will be as follows .
first we will consider the analytic continuation in @xmath0 of @xmath20 $ ] from a purely field theoretical point of view , expanding the hamiltonian as in ( [ hami ] ) in order to show that , in principle , different possibilities arise and how to select the correct one .
this will be done in section [ one ] . in section [ two ]
we introduce the bulk gravitational version of the same problem , representing the euclidean time evolution operator holographically as in ( [ ass ] ) , and show what kind of additional complications arise in trying to select the correct choice when @xmath0 becomes real . to warm up and set the technique we will start with simple unperturbed ads in section [ sec1 ] ( that is @xmath53 ) , matching the minimal area with the generalized gravitational entropy result as in ( [ ass2 ] ) .
this will be done by selecting once and for all the simple case where @xmath2 is just half of the boundary ; in particular we will see the nontrivial role played by the uv cutoffs that has not been covered in detail by @xcite .
understood the static case , for which the analytic continuation in @xmath0 is a trivial problem , we will start to tackle the time dependent case by perturbing the ads metric or , equivalently , by switching on @xmath54 in the hamiltonian . in section [ aaa ]
we set some common results and , finally , in sections [ bbb ] and [ ccc ] we consider two different possibilities for the analytic continuation of the gravitational solution .
the first choice will be selected by requiring the gravity solution to preserve periodicity after a time shift of @xmath55 , when @xmath0 becomes real , and thus giving up periodicity for time shifts of @xmath56 .
this is different by what have been used in @xcite ; the lewkowycz - maldacena choice instead ( preserving @xmath56-periodicity and giving up the @xmath57-one ) will be considered in the last section .
both results will be compared to what we obtain using the ryu - takayanagi formula , which is done at @xmath52 and is thus independent by whatever choice we made for the analytic continuation .
finally we present our conclusions .
for the boundary field theory how to define the analytic continuation of @xmath0 in @xmath20 $ ] is in principle clear : first we define the modular hamiltonian as minus the logarithm of @xmath21 : @xmath58 then , simply @xmath59 thus ( with @xmath60 the normalized @xmath21 ) the entropy computed from the replica trick @xmath61}|_{\beta=1}+\log{tr\left[\rho\right]}=-\partial_{\beta}\log{tr \left[e^{-\beta\mathcal{h}}\right]}|_{\beta=1}+\log{tr\left [ e^{-\mathcal{h}}\right]}=\ ] ] @xmath62}{tr\left [ e^{-\mathcal{h}}\right]}+\log{tr\left [ e^{-\mathcal{h}}\right]}\ ] ] correctly coincides with @xmath63=-\frac{tr\left[\rho\log{\rho}\right]}{tr [ \rho]}+\log{tr [ \rho]}=\frac{tr \left[\mathcal{h } e^{-\mathcal{h}}\right]}{tr\left [ e^{-\mathcal{h}}\right]}+\log{tr\left [ e^{-\mathcal{h}}\right ] } .\ ] ] the problem is that , in general , the exact form for the modular hamiltonian is not known .
what is achievable is an order by order expansion formula called magnus expansion : @xmath64\ ] ] @xmath65+[[h(t_1),h(t_2)],h(t_3)]\right)+\dots .\ ] ] we consider the case @xmath66 and @xmath67 time independent .
the goal is to compute @xmath68 $ ] at first order in @xmath69 .
the computation is done in appendix ( [ c ] ) and it gives @xmath70_{order \ ; o(\delta)}=tr \left[\mathcal{t}\left(\int_{0}^{2\pi}dt\;h(t ) e^{-\int_{0}^{2\pi}d\tilde{t}\;h(\tilde{t})}\right)\right]_{order \;o(\delta ) } .\ ] ] the above formula agrees with the entanglement entropy variation computed in @xcite for a generic normalized density matrix @xmath71 , and reproducing the first law of @xcite @xcite , when @xmath21 is taken to be ( [ dens ] ) and @xmath72 with @xmath73 .
now suppose we did nt know about the modular hamiltonian and of ( [ corret ] ) , and we were looking for the analytic continuation of @xmath74 from the very definition of @xmath21 .
one possible guess would be to use the equality valid for @xmath75 and periodic hamiltonian @xmath76 , @xmath77=tr\left[\mathcal{t}e^{-\int_0^{2\beta\pi}h(t)}\right ] \;\;\;\;\ ; \beta\in \mathbb{z},\;\;\;\;\ ; h(t+2\pi)=h(t),\ ] ] and extend it to @xmath1 .
this however fails for real @xmath0 as @xmath78_{\beta=1}= tr\left [ 2\pi \mathcal{t}\left ( h(0 ) e^{-\int_0^{2\pi}h(t)}\right)\right ] \neq tr \left[\mathcal{h } e^{-\mathcal{h}}\right ] , \ ] ] where the last inequality is already evident at first order in @xmath69 .
a second guess might be @xmath79 \;\;\;\;\ ; h(t+2\pi)=h(t ) , \ ] ] that correctly reproduces ( [ ham ] ) at first order in @xmath69 , @xmath80_{\beta=1}= tr\left [ \mathcal{t}\left ( \int_{0}^{2\pi}dt\ ; h(t ) e^{-\int_0^{2\pi}d\tilde{t}\;h(\tilde{t})}\right)\right ] , \ ] ] but fails to equate @xmath74 for integer @xmath0
@xmath81\neq tr\left[\mathcal{t}e^{-\beta\int_0^{2\pi}h(t)}\right ] \;\;\;\;\ ; \beta\in \mathbb{z},\;\;\;\;\ ; h(t+2\pi)=h(t).\ ] ] so far we have considered an hamiltonian that , even in its analytically continued version , remained @xmath56 periodic and @xmath0 independent , but other possible choices could involve different periodicities , as for example with @xmath82 period , and/or nontrivial @xmath0 dependence of the hamiltonian .
the moral then is that the correct way of choosing the analytic continuation is far from trivial and simple guesses may easily turn out to be wrong .
let us now look at what happens when we move to the bulk .
we would like to consider now the holographic counterpart of the previous section . given the classical correspondence @xmath83=e^{-s_{eucl}^{grav}(\beta ) } , \;\;\;\;\ ; \beta\in \mathbb{z}\ ] ] the question is how do we analytically continue the gravitational side in order to obtain the dual of @xmath20 $ ] for @xmath1 ?
this time we do not have an analogue of the boundary modular hamiltonian that we can use ; on the other side , as it is a classical problem , all the troubles given by the time ordering disappear .
we will start by considering what periodicity we can impose on the analytically continued action ; when @xmath1 the double periodicity with periods @xmath56 ( for representing the same operator @xmath21 @xmath0-times ) and @xmath82 ( for closing the trace ) is no longer possible , except for constant functions , so we have to choose between preserving one or the other .
when we keep @xmath84 it is natural to define the analytic continuation by time integrating between zero and @xmath56 and then multiplying by @xmath0 : @xmath85\leftrightarrow e^{-\beta\int_{0}^{2\pi}dt\;l_{eucl}^{grav}(\beta , t ) } , \;\;\;\;\ ; \beta\in \mathbb{r } .\ ] ] if instead we require @xmath86 we are lead to the second proposal : @xmath87\leftrightarrow e^{-\int_{0}^{2 \beta\pi}dt\;l_{eucl}^{grav}(\beta , t ) } , \;\;\;\;\ ; \beta\in \mathbb{r } .\ ] ] note that when @xmath0 is restricted to be an integer ( [ ddue ] ) automatically becomes periodic also in @xmath88 .
instead the opposite does not happen for ( [ unno ] ) , so we need to require as part of the definition , that @xmath89 in this case obeys the double periodicity whenever @xmath0 is restricted to be an integer .
this poses obvious constraints on the action and makes @xmath90 different from the simple rescaling of the integration extrema of @xmath91 .
we note that ( [ ddue ] ) corresponds in fact to the prescription given in @xcite .
we would then expect it to give the correct result and any other choice to fail . despite this , and because the final goal is to learn more about the analytic continuation , we will still pursue the computation for both proposals above .
further let us point out that any relationship between analytic continuations in the bulk , as the above ( [ unno ] ) , ( [ ddue ] ) , and their possible boundary counterparts is in general complicated .
thus correctly guessing what is the dual of ( [ corret ] ) , or ( [ unob ] ) and ( [ dueb ] ) is not a simple task .
one would love to now repeat the analysis done in the previous section , deriving with respect to @xmath0 and trying to relate the result to the boundary stress - energy tensor .
the computation however , although apparently simple , is far from trivial ; for ( [ unno ] ) we have @xmath92 @xmath93 while for ( [ ddue ] ) @xmath94 if the dependence by @xmath0 was only in the metric we could rewrite @xmath95 where @xmath96 is the boundary induced metric ( on the cutoff surfaces we will soon introduce ) , and @xmath97 the boundary induced stress energy momentum tensor .
this because @xmath98 vanishes because of the bulk equations of motion , and the only @xmath0 dependent metric component is @xmath99 , as we will see .
what goes wrong however , apart from the additional terms in ( [ p1 ] ) , is the fact that the very cutoff surface we will use to define our gravitational action , are in general @xmath0 dependent , creating in this way further contributions .
thus we need a more precise computation to check the validity or not of either ( [ ddue ] ) and ( [ unno ] ) at order @xmath100 , in order to correctly reproduce the entanglement entropy through the replica trick .
let us start from the time independent case , where we consider @xmath101 with the submanifold @xmath2 , from now on , being half of the ads boundary .
first we will compute the minimal surface in this case , setting the benchmark against which to check the results obtained from the generalized gravitational entropy computation .
being time independent the various proposals for `` @xmath102 '' are equivalent ; nonetheless we will here set up the computation technique and introduce the cutoff surfaces we will use throwout the paper .
the poincare coordinates in the euclidean signature cover the full ads space , unlikely in the lorentzian case where they cover only half .
thus we can use them even for computing a minimal surface area that cuts in two the full ads , @xmath103 the area of the minimal surface at @xmath104 whose boundary at @xmath105 is @xmath106 , is simply given by the integration of @xmath107 , where @xmath108 are @xmath109 and @xmath110 : @xmath111 where @xmath112 is the uv cutoff and @xmath113 compactify the @xmath109 direction .
we start from the poincare metric in polar coordinates with period @xmath55 : @xmath114 @xmath115 this metric is regular everywhere and obviously periodic in @xmath116 as well .
the first problem is to find coordinates such that the proper size of the euclidean time circle does nt vanish ( as @xmath117 ) at the boundary of ads , but eventually only in the interior .
const , the slightly uprising ones are @xmath118 const and the vertical line is the cutoff in @xmath119 .
the values displayed are totally random and have no meaning at all.,scaledwidth=50.0% ] because of this we change the bulk coordinates such that the weyl rescaled metric at the boundary of ads has a nonvanishing time circle .
one possibility is the following transformation @xmath120 with inverse @xmath121 they produce @xmath122 the gravitational action is @xmath123 with @xmath124 , @xmath125 the vector normal to @xmath126 pointing outside , that is away from the bulk , and with unit norm , and @xmath96 the induced metric @xmath127 on the boundary .
note that @xmath128 is the determinant on the coordinates of the boundary , while the contraction for @xmath129 is other all . for @xmath126
we mean the cutoff surface approaching the boundary .
we will also temporarily need to introduce an additional cutoff on @xmath130 , @xmath131 ; it will be removed at the end of the computation .
const , the almost circular ones are @xmath132 const , while the vertical and horizontal ones are respectively the cutoffs in @xmath110 and @xmath130.,scaledwidth=50.0% ] the point of using different coordinates lies entirely in having a different cutoff surface that approaches the ads boundary , such that the metric there has a nonvanishing circle proper size for the coordinate @xmath38 .
this new cutoff is achieved fixing @xmath133 , with @xmath134 some constant . on the other hand
the computation in the original poincare coordinates had a cutoff in @xmath135 .
the relationship between these two cutoffs is given by the first equation in ( [ inveq ] ) , and it depends on @xmath130 . as we want @xmath134 to be constant we need to fix @xmath130 to some value , effectively changing the cutoff procedure in the computation of the generalized gravitational entropy . because the minimal area surface will be computed in @xmath136 coordinates , we can still use the cutoff @xmath135 for that computation provided that the two cutoffs agree at the value of @xmath130 at which the minimal surface is located , that is @xmath48 regularized as @xmath131 : @xmath137 the cutoff on @xmath130 induces an equivalent cutoff on @xmath21 , which depends on the coordinate @xmath119 as well : @xmath138 we can now compute the action ( [ action ] ) using the normal , outward pointing , unit vector to the boundary ( at @xmath139 ) , which is @xmath140 the result is @xmath141 @xmath142 from this expression it is immediate to compute the result in the limit @xmath143 : @xmath144 that coincides with the result from the minimal surface one might wonder if there is a need for a theta term in the action also for this temporary boundary . if introduced however , it is easy to show that the correction to the above formula vanishes in the @xmath145 limit ] .
having found agreement between the entanglement entropy computed from the ryu - takayanagi formula and the corresponding generalized gravitational entropy when restricted to the static ads case , we now move to study its variation when the ads metric has been perturbed .
as we have already discussed we will first select some analytic continuation , then perturb the metric accordingly , solve its equations of motion and use the result to compute the generalized gravitational entropy .
finally we will compare with what has been obtained using the minimal surface procedure .
but before starting we need some general results , independent of the choice for the analytic continuation . naively one would think that , given a generic metric perturbation @xmath146 , @xmath147 would be zero at first order as the background metric satisfies the equations of motion .
in fact the boundary term of ( [ action ] ) is there exactly to ensure this when a manifold with boundary is considered .
however the caveat is that we work with a cutoff defining a surface on which , as long as the cutoff is not removed , the value of @xmath146 is in general not vanishing ( this will be clear when we will solve the equations of motion for @xmath146 ) . and
the condition for the @xmath129 term in ( [ action ] ) to kill the boundary term is exactly that the perturbation of the metric has to vanish on the boundary .
there are in total four boundary surfaces , two defined by the cutoffs ( [ cutoff1 ] ) and ( [ cutoff2 ] ) , and two by the corresponding infrared versions @xmath148 and @xmath149 , needed to regularize the action when in presence of metric perturbations : @xmath150 @xmath151 it may be argued that @xmath152 and @xmath153 are not on the same level , the first two to eventually be removed at the end of the computation , the seconds to be kept to regularize a truly divergent quantity . sending @xmath145 for example ( or @xmath154 ) ,
removes the corresponding @xmath155 ( @xmath149 ) , while it keeps @xmath153 ) ] .
we will always include a @xmath129 term for the boundaries in @xmath153 , and consider both the case in which a @xmath129 term has been included for those in @xmath155 and @xmath149 , and the one in which it has not
. the final result will be independent by the choice ( for simplicity we temporarily indicate both kind of boundaries as @xmath126 , even though this is not technically correct ) .
from @xcite we have that the variation of the bulk part of the action is @xmath156 with @xmath157 the normal vector to the boundary already encountered and ( calling @xmath158 the ads unperturbed metric ) @xmath159 ( where the last equality comes from symmetric - antisymmetric tensor contraction ) .
moreover @xmath160 with @xmath161 thus it remains @xmath162 note that both terms would vanish if @xmath146 was zero on @xmath126 . without the @xmath129 term instead we would have @xmath163
the final goal is to compute the perturbation to the entanglement entropy from the formula @xmath164 . according to the two choices ( [ unno ] ) and ( [ ddue ] ) we have either ( [ p1 ] ) or ( [ p2 ] ) that we conveniently rewrite @xmath165 @xmath166 as we already discussed , the dependence on @xmath0 of @xmath167 comes both from the metric and the extremes of integration , ( [ cutoff1 ] ) , ( [ cutoff2 ] ) , ( [ cutoff3 ] ) , and ( [ cutoff4 ] ) .
choosing a boundary orientation we can write @xmath168 @xmath169 @xmath170 where @xmath171 and @xmath172 refer to the cutoffs ( [ cutoff2 ] ) and ( [ cutoff4 ] ) evaluated in @xmath173 , and analogously for @xmath174 and @xmath175 .
we notice however that the direct @xmath0 dependence of @xmath155 and @xmath149 in fact cancels with the induced one by fixing @xmath176 , as can be checked straightforwardly .
thus for the two integrals in the first line of ( [ q ] ) , the overall @xmath0 derivative of ( [ pp1 ] ) and ( [ pp2 ] ) passes directly on the integrand .
finally @xmath177 is a short notation for the integrand at fixed time of ( [ m ] ) and @xmath178 for either ( [ m ] ) or ( [ n ] ) .
relative signs appear in ( [ q ] ) as we choose to maintain the same orientation for the perpendicular @xmath125 vector to both ir and uv boundaries , as explained in appendix [ a ] .
then we need to solve the equations for @xmath146 for whatever analytic continuation we have chosen , evaluate the vector @xmath125 and boundary metric @xmath179 , plug the result into ( [ m ] ) ( or ( [ n ] ) ) for computing the above result ( [ q ] ) and then , finally , either ( [ pp1 ] ) or ( [ pp2 ] ) .
the outcome shall be compared with the results obtained from the minimal surface . in both analytic continuations
we will consider the same background metric we already used for pure ads , ( [ f ] ) , dictated by regularity when @xmath28 and , being @xmath51independent , obviously satisfying any @xmath51periodicity conditions we want to ask for any value of @xmath0 .
for convenience we rescale certain perturbations by factors of @xmath180 and by the common factor @xmath181 and write , using the fefferman - graham gauge , @xmath182 @xmath183 where obviously @xmath184 and @xmath185 .
this is the form we will use throwout the paper ( but not in the next section ) .
the difference will come in the explicit shape for the @xmath146 , and the periodicity conditions they will satisfy .
as the minimal area computation is done at @xmath52 , the result should be independent by whichever analytic continuations we are considering when solving for @xmath146 .
however in practice , as the equations of motion will be solved using different choices for the fourier transform , it is more convenient to treat the two cases separately . to compute the perturbation to the minimal area it is better to go back to the poincare coordinates ( [ g ] ) . in the fefferman - graham gauge ( @xmath187 ) @xmath188 @xmath189 the equation for @xmath190
have been solved ( there in the lorentzian signature ) at first order in @xmath69 in @xcite : @xmath191 @xmath192 ( the normalization is to ensure @xmath193 ) .
as the surface is extremal we need only to compute the variation of the area due to the metric variation , and not the change of shape , that is an higher order effect .
then @xmath194 and , using ( [ a ] ) and ( [ b ] ) we obtain @xmath195 as we will always encounter an integral over @xmath109 that , being unbounded , should eventually be regularized , both for minimal surfaces and gravitational action , from now on we will simply compare the corresponding integrand @xmath196 to avoid further issues : @xmath197 then using ( [ c ] ) and ( [ d ] ) , @xmath198 where @xmath199 is the regularized version of the confluent hypergeometric function .
note that we had to introduce a cutoff in @xmath110 also for large values , @xmath200 .
we can at this point , if we wish , compute the space integral over @xmath109 of @xmath201 . as the only @xmath109 dependence is in the exponential factor , the result is simply proportional to a delta function with @xmath202 and @xmath203 support : @xmath204 we want to solve the 15 linearized equations of motion for @xmath146 defined as in ( [ i ] ) , that are listed in appendix [ b ] , while imposing periodicity for @xmath205 for @xmath206 .
the solution looks as follows @xmath207 @xmath208 where the @xmath209 on the right hand side are the same fourier coefficients of ( [ c ] ) whose explicit expression is ( [ d ] ) , and we have allowed an additional @xmath0 dependence inside @xmath210 .
in addition there are five algebraic constraints : , @xmath211 , @xmath212 , @xmath213 and @xmath214 @xmath215 using the above solution , after a change of coordinates to @xmath216 , we want to evaluate ( [ pp1 ] ) .
we begin by computing @xmath217 that is the @xmath0 derivative of ( [ q ] ) .
the explicit expression for the various terms appearing in ( [ q ] ) are given in appendix ( [ a ] ) as a function of the perturbation @xmath146 ; plugging in the above solution we obtain the required result .
we start with the first line of ( [ q ] ) and notice that the two terms just differ by the sign and the replacement @xmath218 .
thus we limit ourselves to the computation of the first using the formula for @xmath177 ( [ sun ] ) .
as we have already seen , the extremes of integration happen to be @xmath0 independent , so that the derivative acts only on the integrand .
we have @xmath219 we are interested in the limit @xmath145 .
as we do not know how to compute the integral , what we can do is to write the integrand as a series in positive and negative powers of @xmath21 and @xmath119 , compute the integral for each single term and analyse finite and divergent contributions .
then , temporarily ignoring the time integral , @xmath220_{\sqrt{\epsilon_r^2+\epsilon_z^2}}^{\sqrt{\epsilon_r^2+\epsilon_z^2}\frac{l_r}{\epsilon_r}}\ ] ] @xmath221_{\sqrt{\epsilon_r^2+\epsilon_z^2}}^{\sqrt{\epsilon_r^2+\epsilon_z^2}\frac{l_r}{\epsilon_r } } .\ ] ] from direct computation we can check that @xmath222 for every @xmath223 and @xmath224 ( no log term ) , and for every @xmath125 and @xmath225 .
then terms not vanishing in the @xmath145 limit , either divergent or @xmath226 independent , can come only from @xmath227 and @xmath228 .
in particular the case with @xmath229 has contribution when @xmath145 from both integration extrema , while for the others values of @xmath223 only at @xmath230 .
we can compute the @xmath21 and time integral of the @xmath229 , @xmath227 term ( zero order in @xmath226 of the integrand and generic function in @xmath21 , as the negative powers are vanishing ) , and obtain @xmath231 @xmath232 when @xmath226 goes to zero the hypergeometric in the first term simply converges to a function of the only @xmath112 .
the other instead is badly divergent in @xmath226 ( and @xmath233 ) , going like @xmath234 .
we disregard this divergence and keep only the finite contribution .
as we will encounter @xmath226 singularities throughout the paper let me briefly comment on it .
one of the properties of the gravitational solution considered in @xcite was to contain only mild singularities whenever @xmath117 , that would automatically drop off once computing the entropy @xmath44 ; more specifically it was supposed to be possible to constrain the solution such that this behaviour would happen .
as here we are presently considering a different periodicity then the one used by lewkowycz and maldacena , one would think that the issue arises precisely because of this and it would disappear once the l - m periodic conditions on the gravitational solution are applied .
nonetheless the same divergences will be present in the next section as well , without any clear possibility to constrain the solution so that they would drop when computing the entropy .
the result is that we will limit ourselves to consider the finite part of the entropy , but a better understanding of this issue is important .
the result of this `` ad hoc '' renormalization , together with the subtraction of the corresponding @xmath235 term , is a first contribution to the entropy : @xmath236 @xmath237 where we wrote @xmath238 then we pass to compute the @xmath227 and @xmath239 coefficients , for each power of @xmath125 and integrated over time .
what happens is that for each one we obtain a divergent term ( negative power in @xmath226 ) , generically @xmath112 dependent , that we throw away , and a finite contribution always @xmath112 independent .
thus when we consider the corresponding term with opposite sign after the @xmath218 substitution , all of them simply vanish .
let us move to the second line of ( [ q ] ) .
now the integration extrema are @xmath0 dependent ; we can however change integration variable from @xmath240 in such a way to remove the @xmath0 dependence from the integration extrema and make them finite in the limit @xmath145 .
so we have @xmath241 that more explicitly can be written as @xmath242 @xmath243 we start from the case with @xmath129 term ; the above integrand is given by ( [ srhosn ] ) , transforming to @xmath244 and plugging in our solution for @xmath146 .
in particular direct computation shows that the terms @xmath233 independent vanish in the limit @xmath145 while the ones containing @xmath233 give rise only to the same badly exponentially divergent contribution already encountered .
as we have already done we disregard them .
an analogous story can be told when we have a @xmath129 term by using ( [ srhocn ] ) . we remain with @xmath245 the computation proceeds more or less in the same way , with the only finite term coming from the @xmath226 independent term in the integrand of ( [ sun ] ) (
@xmath229 , @xmath227 in the formalism used so far ) , and discarding divergences .
the result is @xmath246 @xmath247 as no more contributions are present , we have then achieved an expression for ( [ pp1 ] ) , the finite result given by the sum @xmath248 of ( [ c1 ] ) and ( [ cc1 ] ) .
this is not however the end , as explained in the next section .
our metric perturbation @xmath146 has been constructed in ( [ h ] ) to be periodic after a shift of the time variable @xmath205 for any value of @xmath0 .
however when @xmath29 the present solution does not become automatically periodic for @xmath249 , as we would like . in other words
what we would need is to further constrain the @xmath250 in order to enforce @xmath249 periodicity as long as @xmath29 .
the solution will then retain by construction only the @xmath251 periodicity once @xmath206 .
let us first show what @xmath250 should look like in order to have periodicity in @xmath205 and @xmath249 when @xmath29 . to this end let us transform in polar coordinates the fourier momenta @xmath252 and @xmath253 : @xmath254 so that @xmath255 fourier transforming @xmath256 in this way , will make the resulting function in @xmath42 clearly periodic for @xmath205 .
moreover , if @xmath256 is periodic in @xmath257 , we can rescale the integration variable @xmath258 in order to absorb also the shift @xmath249 , as long as @xmath259 .
that is we need periodicity for @xmath256 also when @xmath260 , which is compatible with @xmath257 only if @xmath29 .
let us implement the solution of @xmath261 and write it as a function of @xmath262 .
the transformation @xmath263 translates in @xmath262 variables to @xmath264 a function of @xmath262 invariant under ( [ trsf ] ) should depend only on @xmath265\equiv k_1 ^ 2+k_{t_e}^2 $ ] and @xmath266\equiv e^{i n\beta \arctan[k_1,k_{t_e}]}$ ] , @xmath267.$ ] we mean a function that gives the angle whose tangent is the ratio @xmath268 , and taking into account the quadrant ( @xmath269 when @xmath270 , @xmath271 when @xmath272 , @xmath273 when @xmath274 and @xmath275 when @xmath276 ) . ]
further we rewrite the expressions in ( [ h ] ) for @xmath277 without the spurious terms in @xmath278 by integrating by parts in the fourier space , @xmath279 as we want invariance under ( [ trsf ] ) we first require @xmath280 and @xmath281 to be functions of only @xmath282 and @xmath283 .
then we constrain @xmath284 and @xmath285 , the only two remaining independent functions according to ( [ h2 ] ) , so that all the expressions on the right hand side of the exponential in ( [ h3 ] ) are themselves functions of only @xmath282 and @xmath283 .
that is ( focusing only on the @xmath262 variables ) @xmath286 for generic @xmath287 , @xmath288 etc ... after a lengthy computation we arrive at the , fortunately simple result @xmath289 note that these two functions are not by themselves invariant under ( [ trsf ] ) but they transform in such a way to make each of the @xmath146 given by ( [ h]),([h2 ] ) , ( [ h3 ] ) and ( [ h5 ] ) ultimately periodic in @xmath249 ( for integer @xmath0 ) .
thus we ended up with only three independent functions : @xmath290 , @xmath291 and @xmath292 .
in particular @xmath293 and @xmath294 now are @xmath295 we have chosen to avoid any direct dependence on @xmath0 inside @xmath296 and @xmath297 , but only through @xmath283 .
this will turn out to be the correct prescription .
we can insert the result in ( [ c1 ] ) and , after a change of integration variables from @xmath262 to @xmath108 , an integration by part in @xmath298 , and a change back to @xmath262 , we obtain the result for @xmath299 to be transformed from ( [ c1 ] ) to : @xmath300 @xmath301 @xmath302 and analogously for @xmath303 : @xmath304 @xmath305 @xmath306 their sum gives @xmath307 @xmath308 and exactly matches ( [ e ] ) . dependence to be through the variables
@xmath108 ] this result is surprising as , having different periodic conditions then in @xcite , we would have expected disagreement .
this may be an effect of being at first order in @xmath69 , somehow mimicking the corresponding boundary situation we encountered when using ( [ dueb ] ) , or it may be the sign that multiple choices for the analytic continuation can be done if the goal is to compute the entropy .
this last possibility does not clash with the construction of lewkowycz and maldacena , in fact it even reinforces it as different choices lead to the same result , but it raises the issue of unicity of the analytic continuation and how to physically single out one from another .
now we want to solve the equations of motion for @xmath146 so that the solution remains explicitly periodic in @xmath249 ( but not in @xmath205 ) after analytic continuation of @xmath0 , with the perturbations conveniently chosen as in ( [ i ] ) . to this end
we fourier transform @xmath146 as follows @xmath311 plugging the above form into the expressions of appendix [ b ] , the resulting equations are quite complicated to solve as they are intrinsically coupled between different perturbations .
however they simplify if we restrict to the case @xmath312 .
as can be seen playing with the equations this implies @xmath313 . to compute the action we do not need to solve all the equations as ( [ sun ] ) ,
( [ srhosn ] ) and ( [ srhocn ] ) contain only @xmath314 and @xmath315 ( equal by assumption ) .
their equation reduces to ( @xmath316 but it is valid also for @xmath317 ) @xmath318 where for simplicity we have dropped the dependence of @xmath319 by @xmath320 and @xmath203 . applying the ansatz @xmath321
we can decouple the equation in @xmath110 by the one in @xmath130 requiring @xmath322 to satisfy @xmath323 then the equation in @xmath110 becomes @xmath324 the solutions are respectively ( with @xmath325 ) @xmath326 at this point both @xmath327 and @xmath328 are free parameters to be correctly normalized , that may depend on @xmath329 and @xmath0 .
we choose them in order to select solutions with certain asymptotic behaviours ; two obvious choices are @xmath330 the first ( plus ) making @xmath322 finite in the limit @xmath117 , the second ( minus ) in the limit @xmath331 . when @xmath332 the asymptotic behaviour in @xmath333 is @xmath334independent , so we fix @xmath335 in order to have @xmath336 . given this solution we want to evaluate ( [ pp2 ] ) .
the procedure is analogous as in the previous sections , only with different perturbations involved and , as we are in the case @xmath312 , expressions ( [ sun ] ) , ( [ srhosn ] ) and ( [ srhocn ] ) further simplify . the computation of ( [ q2 ] ) , both when we do and do not have a @xmath129 term , gives the same vanishing plus exponentially divergent results , when the @xmath226 and @xmath233 cutoffs are removed . thus no finite contribution . also ( [ uno ] ) works similarly and the only finite term that we obtain comes from the order zero in the @xmath226 expansion of the integrand . as in doing the time integral between @xmath337 $ ] @xmath338 gives always zero unless @xmath332 , the choice between the two asymptotic solutions of ( [ p ] ) is irrelevant , with @xmath339 .
then the result is ( using @xmath340 that comes from the first equation of ( [ aleq ] ) ) @xmath341 @xmath342 for comparison with the above result we have to compute the minimal area surface when the perturbation @xmath146 has been fourier transformed as in the above section , with the position of the minimal area surface translated from @xmath343 into @xmath344 : @xmath345 and @xmath346 where we have chosen the first solution of equation ( [ p ] ) and the contribution coming only from the @xmath332 sector , the rest being infinitesimal in the limit @xmath145 .
this coincides with ( [ res ] ) if we impose the simple condition on @xmath347 : @xmath348 if we wish we can again perform the integration over @xmath109 to obtain : @xmath349
what we have done in this paper is to consider two different possible analytic continuations in @xmath0 of the holographic counterpart of @xmath350 $ ] .
the guideline for choosing them has been what periodicity to keep for the gravitational solution , out of the two we should impose when @xmath29 .
one possible choice is the one used by lewkowycz and maldacena , that allows for a proof of the ryu - takayanagi formula , the other is essentially its complement . as the choice done in @xcite is an assumption , the idea was to bring additional evidence from a first order computation ; instead we have found that both solutions work correctly at first order , reproducing the corresponding result obtained from computing the minimal area .
thus it may in principle be possible to consider additional different , more involved gravitational metrics and to expect some of them to provide the correct result at first order .
this is somehow reminiscent of the corresponding boundary problem , where the wrong choice ( [ dueb ] ) could nonetheless reproduce the right first order result .
an obvious possibility for better understanding the issue would be to go to second order in @xmath69 .
this however seems impractical , both because of the complication in solving the einstein s equations and as , at second order , the minimal surface shape should be backreacted .
nonetheless some results appeared recently at second order in @xcite so , perhaps , an attempt to go to second order may be worth . in general so far how to provide a more robust understanding for the holographic dual of @xmath351 $ ] for real @xmath0 is still an open problem . besides the above story
two additional considerations arise : one being the divergences in @xmath226 that we had to throw away when performing the computation .
the second is the choice for the @xmath0 dependence inside @xmath352 that we have made in order to reproduce the correct result from the minimal area . concerning the divergences we do not have much to say ; we solved einstein s equations at first order for the perturbation of a regular background , but the entropy evaluated from the gravitational action presents divergencies , in contrast with the general analysis in @xcite .
further these divergencies appear both as negative powers of @xmath226 , that may be regularized after an appropriate choice of counterterms ( see for example @xcite ) , and as exponentials that are clearly more problematic
. a possibility would be to constrain the solution in order to make the entropy finite , but how to do so is not clear .
besides this fact , the finite part of the computation correctly agrees with the minimal area result , for both analytic continuations , only if we assume a certain @xmath0 dependence of the fourier coefficients , leading respectively to the discussion below equation ( [ h6 ] ) and to equation ( [ hh6 ] ) .
respecting these additional constrains is not a problem in practice , but their very presence is a novelty .
finally we treated in detail the issue of cutoffs and coordinates in ads .
this may seem a technical problem but in fact most of the work is directly connected with it : the @xmath353 to @xmath354 transformation permits to correctly compute the entropy and sets the cutoff in @xmath354 coordinates and their relationship with the ones in @xmath353 ; these then fix the boundaries on which @xmath355 lives ( making it non zero ) , make its @xmath0 dependence non trivial and , by sending @xmath145 , directly implement the restriction of the entropy to the region in ads where the time circle shrinks .
as most of these issues were so far not explicitly covered in the literature , at least in practical examples , we felt compelling a more precise discussion .
the generalized gravitational entropy is not only a useful tool for providing a proof of the ryu - takayanagi formula , but it also allows to extend the holographic description of entanglement entropy beyond the original range of applicability ; for example for studying bulk quantum corrections @xcite and @xcite , or higher derivatives extensions of the gravity theory , for instance @xcite @xcite @xcite @xcite @xcite and @xcite .
this last case case in particular is interesting , as we do not know of any functional such that its minimization would provide a generalization of the ryu - takayanagi formula for a bulk theory with generic higher curvature terms . still using the generalized gravitational entropy it has been possible to construct some formulas for the holographic computation of the entanglement entropy .
the technique developed in the present paper may then be extended to some of these theories , for example to compute holographically the reaction and dynamics of the entanglement entropy to sudden changes of parameters , known as quantum quenches .
i would like to thank aitor lewkowycz , tadashi takayanagi , masaki shigemori and diego trancanelli for suggestions and comments on the paper .
further the result of appendix [ c ] has been taken from shigemori s notes , although the proof here is different , and the first motivation for the present work has sourced from several discussions with him .
this work was founded by fapesp fellowship 2013/10460 - 9 .
we want to prove the equation @xmath358_{order \ ; o(\delta)}=tr \left[\mathcal{t}\left(\int_{0}^{2\pi}dt\;h(t ) e^{-\int_{0}^{2\pi}d\tilde{t}\;h(\tilde{t})}\right)\right]_{order \;o(\delta ) } , \ ] ] where @xmath359 and @xmath360 at order zero in @xmath69 the relationship is trivial as , given ( [ magnus ] ) @xmath361_{order \ ; o(\delta^0)}=tr \left[2\pi h_0 e^{-2\pi h_0}\right]=tr \left[\int_{0}^{2\pi } dt\;e^{-(2\pi - t ) h_0 } h_0 \;e^{-t h_0}\right]\ ] ] and the right hand side is just the @xmath362 term of @xmath363=tr \left[\mathcal{t}\left(\int_{0}^{2\pi}dt\;h(t ) e^{-\int_{0}^{2\pi}d\tilde{t}\;h(\tilde{t})}\right)\right ] .\ ] ] at first order in @xmath69 we should work out two additional terms , respectively when @xmath364 is picked up either from the @xmath356 `` downstairs '' in ( [ ham2 ] ) or from @xmath365 . using ( [ magnus ] ) the first one is @xmath366=\ ] ] @xmath367=tr \left[\delta \int_0^{2\pi}dt\;e^{-(2\pi - t ) h_0}h(t)\;e^{-t h_0}\right ] , \ ] ] and the cyclicity of the trace has been used in the first equality to kill all the terms containing a commutator ( with @xmath368 and @xmath67 obviously commuting ) . again
the final expression is what we would obtain from ( [ exp ] ) at order @xmath100 , picking up @xmath369 from the hamiltonian below .
the final piece is then @xmath370=tr\left [ 2\pi h_0 \mathcal{t}\left ( e^{-\int_0^{2\pi}h(t)}\right)_{o(\delta)}\right]=\ ] ] @xmath371+tr\left[2\pi h_0 \mathcal{t}\left ( e^{-\int_t^{2\pi}d\tilde{t}h(\tilde{t})}\right)_{o(\delta ) } \mathcal{t}\left ( e^{-\int_0^{t}d\tilde{t}h(\tilde{t})}\right)_{o(\delta^0)}\right ] , \ ] ] and again using the cyclicity of the trace and the fact that @xmath67 and @xmath372 commute we can rewrite the above expression as the @xmath100 order of @xmath373 , \ ] ] which is again trivially what we were missing of ( [ exp])@xmath374 .
here we list the 15 equations of motion from the independent components of the einstein tensor ( plus cosmological constant ) , expanded at first order in @xmath69 .
five equations involve only derivatives with respect to @xmath375 and @xmath376 , and so after fourier transforming these coordinates they become algebraic .
the remaining ten instead are differential equations in @xmath110 .
as the fourier transformation involving coordinates @xmath377 will be different depending on the analytic continuation we want to achieve , in this appendix we fourier transform the @xmath146 functions only in @xmath109 : @xmath378 further the form of the equations is simpler if expressed in @xmath379 coordinates . after some rearranging the five `` algebraic '' equations are @xmath380 and the ten `` differential '' : @xmath381
let us explicitly evaluate the integrands of ( [ q ] ) for perturbations @xmath146 defined as in ( [ i ] ) .
this appendix does not distinguish between the two possible periodicities after analytic continuation of @xmath0 .
further we will work in @xmath216 coordinates , as these are the ones to be used in order to naturally implement the cutoff procedure .
we first need the unit normal outward pointing vector @xmath125 , for the four boundaries , at first order in @xmath69 .
the expressions are @xmath382 and @xmath383 note that we have decided for simplicity , to keep the same sign for the vector for both the uv and ir cutoff , and include the relative minus sign directly in the expression ( [ q ] ) ( which is odd in @xmath125 ) . given these expressions the results are @xmath384 @xmath385 and @xmath386 @xmath387 or , without the @xmath129 term , @xmath388 @xmath389 we can simplify these expressions using the equations of motion , essentially subtracting terms proportionals to the first two equations of ( [ aleq ] ) ( rewritten in @xmath390 variables and fourier transformed back into functions of @xmath109 ) , and reduce them to respectively @xmath391 @xmath392 @xmath393 and @xmath394 this last one without the @xmath129 term . | we study the analytic continuation used by lewkowycz and maldacena to prove the ryu - takayanagi formula for entanglement entropy , which is the holographic dual of the trace of the @xmath0-power of the time evolution operator when @xmath1 .
this will be done perturbatively by using a weakly time dependent hamiltonian , corresponding to a small shift of the dual static background .
depending on the periodicity we impose on the gravitational solution , we consider two different possibilities and compare the associated entropies with the results obtained through a minimal area computation . to our surprise
we discover that , at first order , both choices correctly reproduce the associated entanglement entropy .
furthermore we find unexpected divergent contributions that we have to discard in order to fit the minimal area computation , and an additional requirement that needs to be imposed on the @xmath0 dependence on the metric . * a perturbative study on the analytic continuation for generalized gravitational entropy * + andrea prudenziati 2em |
conformal field theories in two dimensions can be used to describe string theory from the worldsheet perspective and statistical systems at a second order phase transition .
the study of these field theories is often manageable because the infinite dimensional symmetry algebras which exist can reduce the field content to a finite number of representations . in such minimal cases , there are ` null vectors ' which should decouple from all correlation functions . as a result
the non - zero correlation functions satisfy differential equations which enable one to solve the theory completely .
each symmetry algebra has its own series of minimal models for which this procedure works .
the simplest case is that of minimal models of the virasoro algebra . in @xcite , belavin
et al showed how to relate correlation functions of arbitrary fields in a virasoro minimal model to those of a particular class of fields , called primary fields , and in turn showed how differential equations for four - point functions of primary fields could be found from the singular vectors that are present in these models .
the virasoro algebra can be extended by extra generators to include supersymmetry , and the simplest of these extensions is the @xmath0 superconformal algebra .
although the @xmath0 superconformal algebra includes the virasoro algebra , there is an infinite set of minimal models of the @xmath0 superconformal algebra which only includes a few minimal models of the virasoro algebra , and the methods of @xcite need to be generalised to find the differential equations satisfied by the correlation functions of primary fields of the @xmath0 superconformal algebra .
the first thing to note is that the superconformal algebra comes in two forms , called the neveu - schwarz ( ns ) and ramond ( r ) algebras , and accordingly has two classes of representations and fields associated to these representations .
the differential equations for correlations functions of four ns type fields were worked out in @xcite , but the extension to correlation functions of four ramond fields has problems @xcite and the solution to these problems presented here requires an extension of ideas in @xcite . in this paper
we perform this extension and find differential equations for correlation functions of four r type fields .
the structure of the paper is as follows : we first introduce the ns and r algebras and describe their representations and singular vectors .
we discuss their three point functions and describe the chiral blocks and how the singular vectors allow them to be found level - by - level .
next we describe the toy model of the ising model and show how one can obtain a differential equation for the four - spin field correlation function , as in @xcite .
next we apply these ideas to the ramond correlation functions .
this lets us write down a matrix - differential equation for the correlation functions .
we check these by comparison with known results and present the exact solution for a variety of cases .
the algebra of chiral superconformal transformations in the plane is generated by two fields , @xmath1 and @xmath2 . according to the choice of boundary conditions on @xmath3
, one may choose the labels @xmath4 on @xmath5 to be integer or half - integer ; the algebra in these two sectors are called the ramond and neveu schwarz algebras respectively . in both sectors
the generators obey @xmath6 & = & \frac c{12 } m(m^2 - 1)\delta_{m+n,0 } + ( m - n)l_{m+n } \;,\\ \{g_m , g_n\ } & = & \frac c3 ( m^2 - \frac 14 ) \delta_{m+n,0 } + 2 l_{m+n } \;,\\ { } ~[l_m , g_n ] & = & ( \frac m2 - n)g_{m+n}\;.\end{aligned}\ ] ] it is usual to adjoin the operator @xmath7 to the superconformal algebra , where @xmath8 is the fermion number operator which anti - commutes with @xmath5 and commutes with @xmath9 .
this allows one to consider states of definite parity and is essential in many constructions . in a superconformal field theory
the physically relevant representations of the superconformal algebra are irreducible highest weight representations .
these are graded by @xmath10 eigenvalue , or level .
see @xcite for more details .
highest weight representations of the neveu - schwarz algebra have a state of least @xmath10 eigenvalue @xmath11 such that @xmath12 if one considers the extended algebra then it is usual to take the highest weight state bosonic ( @xmath13 ) or fermionic ( @xmath14 ) in which case the addition of the fermion number operator does not alter the representation theory .
it is useful to parametrise @xmath15 and @xmath16 as @xmath17 since there is a singular vector of @xmath10 eigenvalue @xmath18 whenever @xmath19 and @xmath20 and @xmath21 , @xmath22 ( see refs .
note that the highest weight state can be either bosonic or fermionic , for example , the ns vacuum state @xmath23 is usually regarded as bosonic and so the ns highest weight state @xmath24 is fermionic .
the highest weight representation @xmath25 is spanned by states of the form @xmath26 and the level of such as state is @xmath27 .
states with even numbers of modes of @xmath28 have integer level and states with odd numbers of modes have half - integer level and hence @xmath29 where @xmath30 is spanned by integer level states and @xmath31 by half - integer level states .
highest weight representations of the extended ramond algebra have , in general , a two - dimensional highest weight space spanned by states @xmath32 of definite fermion parity @xmath33 satisfying @xmath34 where @xmath35 and @xmath36 and @xmath37 can be chosen freely ; representations with different @xmath37 are equivalent .
such a representation has two singular vectors at level @xmath38 whenever @xmath39 , @xmath40 an odd integer , @xmath20 and @xmath41 , @xmath42 we shall mostly be concerned with the constraints arising from the vanishing of the singular vectors at level 1 in the representation @xmath43 which take the form @xmath44
we would like to find the chiral blocks of four ramond fields @xmath45 where the intermediate states lie in a particular ns representation .
we can do this step - by - step by first calculating the operator product expansions ( opes ) @xmath46 and then forming the chiral block by summing over intermediate states . to do this
we need to define the vertex operators @xmath47 of fixed fermion parity and we also need to specify whether the ns representation in the intermediate channel is bosonic or fermionic . for the field operators , we shall use the definition in @xcite : @xmath48 where the combinations @xmath49 and @xmath50 are @xmath51 @xmath52 , @xmath53 and @xmath4 can be integral or half - integral as circumstances dictate ( here @xmath54 play the same role as @xmath55 in and @xmath56 as @xmath37 ) .
it turns out that the four opes in depend on only two quantities .
it is convenient to form the following four linear combinations : @xmath57 we shall first consider the case where the ns highest weight state is bosonic and the opes depend on the two constants @xmath58 . using and the standard result for the virasoro algebra ( see )
, we can calculate the overlaps of @xmath59 and @xmath60 .
we list here the first few , taking @xmath61 for convenience : @xmath62 if we take the ns representation to be fermionic rather than bosonic , then @xmath63 and @xmath64 just swap roles in these equations , so without loss of generality we consider henceforth the intermediate channel to be bosonic .
given these combinations @xmath65 , it is possible to define the chiral blocks and calculate them to be @xmath66 where ` even ' denotes projection of the intermediate states onto @xmath67 and ` odd ' , projection onto @xmath68 , where we again note that we have assumed that @xmath11 is bosonic .
the vanishing of the state imposes constraints on the allowed fusions of the fields @xmath69 and in fact allows one to determine the operator products @xmath70 and @xmath71 recursively .
we take @xmath72 as this simplifies the constraints on @xmath15 , and we shall denote @xmath73 by @xmath63 and @xmath74 by @xmath75 for simplicity .
if we use the equation ( valid for any virasoro primary field of weight @xmath15 ) @xmath76 and the relations , we find that @xmath63 and @xmath75 satisfy @xmath77 where we assume the representation @xmath15 to be bosonic .
taking @xmath15 fermionic simply swaps @xmath63 with @xmath75 . considering the contribution of the highest weight state @xmath11 to @xmath63
, we see that @xmath78 so that @xmath63 is only non - zero if @xmath79 .
the two allowed fusions and the operator products are thus @xmath80 & & \hbox { $ -$ channel:}\qquad ( 1,2 ) \times ( 1,s ) \longrightarrow ( 1,s{-}1 ) \label{eq :- channel } \\ & { { \ensuremath{\cal f}}}^- & = \vec h + \tfrac{st{+}t-2}{st{-}2 } l_{-1}\vec h + \ldots \nonumber\\ & { { \ensuremath{\cal g}}}^- & = -\tfrac{\sqrt{2t}}{2t{-}2 } g_{-1/2}\vec h + \sqrt{\tfrac t2}\tfrac 1{st{-}4}\left ( g_{-3/2 } - \tfrac{2(st{+}t{-}4)}{st{-}2}l_{-1}g_{-1/2}\right)\vec h + \ldots \nonumber\end{aligned}\ ] ] using these expressions it is then possible in principle to calculate the chiral blocks order by order and turn the recursion relations for the operator product expansions into differential equations on the chiral blocks .
rather than carry this out in detail we will instead turn to a method to derive differential equations directly
. we would like to find differential equations for a chiral block of the form @xmath81 coming from the vanishing of , where @xmath82 , @xmath83 and @xmath84 are possibly matrix - valued ramond fields . as a first step , we consider the contribution coming from the mode @xmath85 .
this is easy to calculate using the relations for a virasoro primary field of weight @xmath15 @xmath86 & = & z^{m+1}\phi'(z ) + z^m h(m+1)\phi(z ) \;. \label{eq.l.c4}\end{aligned}\ ] ] the only complication is the need to remove the term in @xmath87 that would normally arise .
this can be done by considering @xmath88 to find @xmath89 f(z ) \;. \label{eq : l-1}\ ] ] the difficulty now is to treat @xmath90 .
we shall adapt some ideas used in the construction of the co - invariant spaces that are used to classify the fusion algebra of virasoro and superconformal algebra representations . in this method
, one finds linear combinations of modes which annihilate the state @xmath91 .
the simplest way to find these is to find polynomials @xmath92 such that @xmath93 has at most a simple pole at @xmath94 , @xmath95 and @xmath96 .
the modes of @xmath97 then take simple values when acting on @xmath91 .
for example , for @xmath98 @xmath99 satisfies @xmath100 as a toy example , we shall first apply this idea to the ising model to reproduce the result of @xcite before applying it to the superconformal algebra and ramond fields .
the ising model can be formulated as the theory of a single free fermion field @xmath101 . as with the superconformal algebra ,
this field can have half - integer ( ns ) or integer ( r ) modes @xmath102 satisfying @xmath103 the ramond algebra has a zero mode @xmath104 satisfying @xmath105 and so there are two inequivalent irreducible highest weight representations of the ramond algebra , with highest weights @xmath106 satisfying @xmath107 a unitary highest weight representation of the ramond algebra will have a highest weight space on which @xmath104 is represented by a matrix with eigenvalues @xmath108 .
we shall denote the ( vector - valued ) highest weight of such a general representation by @xmath109 and the chiral fields corresponding to such a state by @xmath110 .
the energy - momentum tensor can be written in terms of @xmath111 as @xmath112 so that @xmath113 this last equation can be viewed as a null - vector equation , the analogue of and we can use it to find a differential equation for the correlation functions of the form @xmath114 here , each @xmath115 is some vector - value representation as is the function @xmath8 .
we will not actually need to specify the exact form of these representations , as we will see shortly .
we know that the space of virasoro conformal blocks with the correct properties is two dimensional so we have to consider @xmath116 as a vector in some space of solutions and the differential equation we obtain will be a matrix differential equation . according to the idea outlined above
, we want to find combinations of modes of the ramond algebra @xmath117 and @xmath118 so that @xmath119 we will require that these operators square to @xmath120 and mutually anti - commute , that is satisfy the algebra @xmath121 we start by defining operators @xmath122 as integrals , @xmath123 where the contour encloses the origin but not the points @xmath94 or @xmath95 and @xmath124 are functions which remove the unwanted singularities in @xmath125 from the state @xmath126 since the operator product of the field @xmath111 with a ramond field @xmath115 is of the form @xmath127 suitable combinations are @xmath128 it is easy to calculate the anti - commutators of these operators either as contour integrals or directly in terms of the modes to find @xmath129@xmath130 consequently we can define new combinations @xmath131 which satisfy as @xmath132 we can now use the combinations @xmath131 to replace the mode @xmath133 in the singular vector . there remains a large degree of choice in how to do this as we can replace @xmath133 by any of the @xmath131 as follows : @xmath134 without loss of generality , we will now just consider the correlation functions @xmath135 .
the general expression for the null vector relation we can obtain in terms of @xmath131 using is of the form @xmath136 where @xmath137 and @xmath138 are functions of @xmath95 .
acting on this equation on the left by @xmath139 leads to a matrix differential equations for the correlation functions of the form @xmath140 where @xmath141 are matrices representing the action of the zero modes on the fields at @xmath95 , 1 and @xmath96 satisfying the clifford algebra @xmath142 we must now choose a representation of this algebra .
the smallest representations of this algebra are two dimensional and there are two inequivalent choices for which @xmath143 .
it is essential for constructing the correct differential equation for the correlation functions of the spin field that the correct representation is chosen .
to fix the equivalence class of the representation , we note that the operators @xmath131 satisfy further relations , for example @xmath144 either of these is sufficient to fix the class of the representation .
since we are considering @xmath135 , we note that @xmath145 implies we must choose the representation of the clifford algebra for which @xmath146 , ie for which @xmath147 . returning to equation ,
the simplest choices are for two of the functions @xmath148 to be zero , the other non - zero , so that only a single matrix @xmath149 appears in the matrix differential equation . in this case
we can take the matrix to be diagonal ( or alternatively consider the one - dimensional representations of the algebra @xmath150 ) and we obtain a set of first order differential equations for the correlation functions as follows which give the two components of the function @xmath151 as the chiral blocks associated to a particular channel .
we illustrate this below . setting @xmath153
, we find @xmath154 and @xmath155 , so that the differential equation becomes @xmath156 at this stage the only requirement on @xmath157 is that it squares to @xmath94 , so we can consider one - dimensional subspaces of the space of correlation functions on which @xmath157 takes values @xmath33 , leading to the two solutions @xmath158 which are the well known chiral blocks of the virasoro algebra associated to the following choice of channel : @xmath159 the differential equation becomes @xmath161 taking @xmath162 diagonal with values @xmath33 leads to the two solutions @xmath163 which are the chiral blocks of the virasoro algebra associated to the following channel : @xmath164 the differential equation becomes @xmath165 taking @xmath166 diagonal with values @xmath33 leads to the two solutions @xmath167 which are the well - known chiral blocks of the virasoro algebra associated to the channel : @xmath168
taking inspiration from the case of the ising model , we will try to find differential equations for the chiral blocks @xmath169
the expectation is that we can express the singular vector in terms of combinations of the modes @xmath5 which ( inside the correlation function ) lead to a matrix representation of the algebra of the zero modes acting on the primary fields inserted at @xmath94 , @xmath95 and @xmath96 . that is , we will try to find combinations @xmath170 of the modes @xmath5 ( for @xmath171 ) which , when taken inside a four - point function lead to a matrix representation @xmath172 of the zero - mode algebra @xmath173 and which will lead to a matrix differential equation for in the form @xmath174 the first step in repeating the analysis of the ising model for the superconformal algebra is to identify analogues of the combinations @xmath131 . from , the operator product of @xmath3 with a ramond field @xmath175 takes the form @xmath176 where @xmath177 is a matrix representation of the zero mode satisfying @xmath178 , @xmath179 and @xmath180 .
furthermore , @xmath181 consequently , we are motivated to consider the three combinations which remove all the singularities at two of the points @xmath182 and 1 , and turn the leading singularity at the remaining point into a simple pole : @xmath183 these operators , however , do not square to constants as in , nor do they simply anti - commute .
their algebra is more complicated , and only simplifies inside the four - point functions .
we can express their algebra in terms of suitable combinations of modes of the virasoro algebra : @xmath184 these combinations have the following properties when acting on the state @xmath185 @xmath186 @xmath187 in terms of these operators , the @xmath131 satisfy @xmath188 we thus define the combinations @xmath170 as @xmath189 inside the four - point function , the terms in @xmath190 and @xmath191 vanish and @xmath192 , so that the action of the operators @xmath170 inside is given by matrices @xmath172 satisfying the algebra . to use these , we have to express the singular vector in terms of the @xmath131 . if we act with any of the @xmath131 in the highest weight state @xmath193 , the leading term is a multiple of @xmath194 , so that it is not possible to express the singular vector in terms of just one of the @xmath131 .
instead , it is necessary to use all three and one finds that @xmath195 combining with leads to the following matrix differential equation for , the main result of this article : @xmath196 & & ( z-1 ) f ' + \textstyle ( h_{12 } + h_1 + h_z - h_{\infty } ) f(z ) \nonumber \\[2 mm ] & + & \textstyle \sqrt{\frac t2}\left [ \frac{1+z}{2z } \lambda_{12 } - i \left(\frac{1-z}z\right)^{1/2 } \!\!\!\ !
\!\!g_1 - \frac{(1-z)^{1/2}}z g_z + z^{-1/2 } g_\infty \right ] f(z ) = 0 \;.
\\[3 mm ] \end{array } } \label{eq : gde}\ ] ] we now turn to the analysis of this equation and its solutions .
it will also be convenient to parametrise @xmath197 according to as @xmath198 the first thing we can do is to check the indices of the solutions around the points @xmath199 and @xmath200 , and then we can find the explicit solutions in various cases .
finally we compare these to solutions known by other methods .
these are the equations for the leading behaviour of the solution around a singular point .
the equation has singular points @xmath201 and @xmath96 .
we first consider the point @xmath200 around which the solution will have an expansion of the form @xmath202 where @xmath203 are vectors .
note that the expansion will be in half - integer powers of @xmath95 as the differential equation explicitly contains @xmath204 .
around @xmath205 , the leading behaviour is determined by substituting in and examining the coefficient of @xmath206 : @xmath207 ) a_0 = 0 \;. \label{eq : gde2 } \end{aligned}\ ] ] since this equation only involves @xmath208 , we can take @xmath208 to be diagonal with eigenvalues @xmath209 .
this leads to two solutions for @xmath210 , @xmath211 \;.\ ] ] if we use the parametrisation @xmath212 , we find @xmath213 these values are exactly the expected exponents for the chiral blocks shown below : @xmath214 it is easy to check that similar results hold for the other two channels , corresponding to expanding the solution @xmath8 around @xmath96 in powers of @xmath215 and around @xmath94 in powers of @xmath216 . to solve the full equation we must choose a matrix representation for @xmath172 . up to now we have not had to specify the action of the zero modes on the primary fields , and as in the ising model , we do not need to do it now .
there are only two inequivalent representations for which @xmath217 , and the choice of representation is invariant under monodromy around @xmath205 and @xmath61 .
we shall take @xmath172 to be given in terms of the pauli matrices as @xmath218 where @xmath219 .
this choice turns the equation for @xmath8 into a real matrix differential equation and taking @xmath208 diagonal also means that the two components of @xmath8 have expansions in @xmath95 ( and not @xmath204 ) .
the two components of @xmath8 can be identified by comparison with and and are the odd and even chiral blocks ( up to a sign ) @xmath220
in some cases it is possible to find exact solutions to . the simplest case to consider is where all the @xmath197 are equal to @xmath221 . in this case @xmath222 and the fusion rules force @xmath223 . writing the components of @xmath8 as @xmath224 , with the representation , the differential equation becomes @xmath225 the coefficients in the equation as presented have branch cuts at @xmath205 and @xmath61 , and it is convenient to remove these by changing variables to @xmath226 with this change and a redefinition @xmath227 to remove the leading singularity in @xmath95 , the equations simplify dramatically to @xmath228 with solutions @xmath229 g^- = \left(1{-}u^4\right)^{\frac{3}{4}-\frac{3 t}{2 } } \pmatrix { \frac{2t{-}1}{t{-}2 } \cdot u^3 \cdot { f\ ! } \left(\frac{3}{2}{-}t,\frac{3}{2}{-}\frac{3t}{2};2{-}\frac{t}{2};u^4\right ) \cr { f\ ! } \left(\frac{1}{2}{-}t,\frac{3}{2}{-}\frac{3t}{2};1{-}\frac{t}{2};u^4\right ) } \end{aligned}\ ] ] where @xmath230 is the standard hypergeometric function .
for special values of @xmath231 , the blocks with identity intermediate channel simplify further .
for example at @xmath232 , that is for four - point blocks of the field with @xmath233 in the tri - critical ising model , @xmath234 simplifies to @xmath235 as can be seen , the two components are related by @xmath236 which corresponds to @xmath95 encircling the branch point at @xmath61 once .
the blocks can also be found exactly in the more general case @xmath237 , that is @xmath238 in the parametrisation .
for example , the solutions for @xmath8 with @xmath223 are @xmath239 f^- & = & u^{\frac{3}{4}{-}\frac{1}{2 } ( s{+}1 )
t } \left(1{-}u^2\right)^{\frac{1}{4 } \left(2 s{-}1{-}2 ( r{-}1 ) t{-}s^2 t\right ) } \left(1{+}u^2\right)^{\frac{1}{4 } ( 4{-}2 r t{+}s ( s t{-}2 ) ) } \\ & & \pmatrix { \frac{2rt{-}2}{st{-}4 } \cdot u^3 \cdot { f\ ! } \left(\frac{3{-}rt}{2},\frac{6{-}2rt{-}st}{2};\frac{8{-}st}{4};u^4\right ) \cr { f\ ! }
\left(\frac{1{-}rt}{2},\frac{6{-}2rt{-}st}{2};\frac{4{-}st}{4};u^4\right ) } \end{aligned}\ ] ]
we can check the differential equation and its solutions against solutions known by other methods .
principally , there are three values of the central charge in the superconformal minimal series which also appear in list of virasoro minimal models , so that a - series of the superconformal minimal models can be identified with the following invariants of the virasoro minimal models , @xmath240 the representations in the superconformal models can be found as sums of representations in the virasoro minimal models , and the chiral blocks of the superconformal models must be sums of virasoro chiral blocks .
power series expansions of the virasoro chiral blocks can be found easily , either by using one of the recursion relations of zamolodchikov @xcite or by solving the differential equation from the singular vector .
we present two examples here to show how this works . in this model , the tricritical ising model , the ramond representations of the superconformal algebra
are @xmath43 and @xmath242 with conformal dimensions @xmath243 and @xmath244 , and consequently all correlation functions of four ramond fields can be found using the method in this paper . if we consider just one case , the following two blocks can be found by series solution of the differential equation @xmath245 @xmath246 since the ramond representations of the unextended superconformal algebra and the even and odd sectors of the intermediate channel are each irreducible representations of the virasoro algebra , these blocks can also be found using the representation theory of the virasoro algebra . in this case , the irreducible virasoro representation of weight @xmath247 has virasoro kac labels @xmath248 and so the odd and even chiral blocks are two solutions of a fourth order differential equation .
this differential equation can also be solved for a series solution and two of the solutions are exactly those given in and .
this model is related to the @xmath250 invariant of the virasoro minimal model @xmath251 .
the irreducible super virasoro representations of interest split into direct sums of irreducible virasoro representations as follows @xmath252 most of the superconformal chiral blocks are given by sums of virasoro chiral blocks , but in some cases there is only a single virasoro representation contributing to the intermediate channel and so the results of solving and the virasoro null vector equations can be compared directly .
if we denote virasoro representations by @xmath253 then two such cases are @xmath254 @xmath255 this time the blocks and can be calculated either as series solutions of or as the series solutions of eighth order differential equations corresponding to the level eight null vector in the virasoro representation @xmath256 , giving the same answers shown .
they can also be compared with the general series solution for virasoro chiral blocks found by zamolodchikov @xcite .
we have found a matrix differential equation for ramond four - point chiral blocks . this can solved completely in some cases but as yet
the full general solution is now known .
these solutions were known already exactly in some other cases and as series solutions in a few more cases based on relations with virasoro minimal models .
the differential equations presented here reproduce these known results .
exact integral formulae based on free - field constructions are also known @xcite and it remains to check that these satisfy the equations we have found .
recently , recursive formulae generalising zamolodchikov s elliptic recursion formulae in @xcite have been found @xcite and again it remains to check that these give series expansions which satisfy our equations . for the future , these equations and their solutions should enable one to extend calculations that have only fully been worked out in the virasoro case to the superconformal case , such as the construction of the full set of boundary structure constants in @xcite gmtw is very grateful to f. wagner and i. runkel for discussions on many aspects of fusion and superconformal field theory and to ir for detailed comments on the manuscript .
this work was supported by pparc and stfc through rolling grants pp / c507145/1 and st / g000395/1 and by the studentship ppa / s / s/2005/4103 .
99 z. a. qiu , , nucl .
b * 270 * ( 1986 ) 205234 .
g. mussardo , g. sotkov and m. stanishkov , _ ramond sector of the supersymmetric minimal models _ , phys .
b * 195 * , 397 ( 1987 ) .
g. mussardo , g. sotkov and h. stanishkov , _ fine structure of the supersymmetric operator product expansion algebras , _ , nucl .
b * 305 * , 69 ( 1988 ) .
v. g. kac , , proc .
int . congress math .
1978 , helsinki .
b. l. feigin and d. b. fuchs , , repts .
stockholm univ .
( 1986 ) , published in ` _ representations of infinite - dimensional lie groups and lie algebras _ ' , eds .
a. vershik and d. zhelobenko , gordon and breach ( 1989 ) .
a. b. zamolodchikov , _ conformal symmetry in two - dimensions : an explicit recurrence formula for the conformal partial wave amplitude , _ commun .
* 96 * ( 1984 ) 419 .
g mussardo , g m sotkov , and m s stanishkov , _ fine structure of the supersymmetric operator product expansion algebras , _ nucl .
, * b305 * [ fs23 ] ( 1988 ) 69 . | we consider chiral blocks of four ramond fields of the @xmath0 super virasoro algebra where one of the fields is in the ( 1,2 ) representation . we show how the null vector in the ( 1,2 ) representation determines the chiral blocks as series expansions .
we then turn to the ising model to find an algebraic method to determine differential equations for the blocks of four spin fields . extending these ideas to the super virasoro case
, we find a first order differential equation for blocks of four ramond fields .
we are able to find exact solutions in many cases .
we compare our blocks with results known from other methods . |
the generation of nonclassical states of light is one of the primary research areas in quantum optics .
in particular , the preparation of _ entangled states _ of light has attracted a considerable amount of attention recently , since these states have been identified as a key resource for quantum information processing @xcite .
other interesting applications of the entangled states include , for instance , ultra - high precision measurements @xcite and quantum optical lithography @xcite .
the range of interactions between light fields that are experimentally accessible and feasible is rather limited , which restricts the class of quantum states of the optical field that can be generated in the lab .
however , this class can be significantly extended if one considers probabilistic generation schemes , whose success is conditioned on the detection of a particular outcome of a measurement performed on some ancilla system .
schemes for probabilistic preparation of fock states @xcite , arbitrary superpositions of fock states of single - mode field @xcite , and superpositions of classically distinguishable states ( schrdinger - cat - like states ) @xcite , have been found .
several setups for the generation of two - mode @xmath0-photon path - entangled states have been suggested @xcite .
recently , the experimental conditional preparation of a single - photon fock state with negative wigner function has been reported @xcite . however , all the above mentioned schemes are still quite restrictive since they are capable to prepare only single - mode states or only a particular class of two - mode states ( such as the @xmath0-photon states ) . in this paper
, we design a truly _ universal _ scheme that can be used to probabilistically generate an _
arbitrary _ multimode entangled state of light , provided that each mode does not contain more than @xmath0 photons , where @xmath0 is an arbitrary but finite integer .
the resources required for the present scheme comprise the passive linear optical elements ( beam splitters and phase shifters ) , single photon sources , strong coherent laser beams , and photodetectors .
it was shown recently that these resources are sufficient for performing universal quantum computing @xcite .
our proposal provides another illustration of the surprising versatility and power of such an approach , relying only on linear optics and single photons . in this context
it is worth mentioning that , very recently , clausen _ et al .
_ suggested a scheme for approximate conditional implementation of general single- or multi - mode operators acting on states of traveling optical field @xcite .
the scheme described in @xcite is also based on linear optics and may be in principle used for ( approximate ) quantum state preparation .
we emphasize that the approach proposed in the present paper conceptually differs from that of ref .
we are primarily interested in exact state preparation and our setup is specifically tailored for that purpose .
we will first explain all the essential features of the state - preparation procedure on the simplest yet nontrivial example of two - mode entangled state that is formed by a superposition of two product states , @xmath1 where @xmath2 is a normalization factor .
then we will generalize the preparation procedure and design a scheme for the generation of arbitrary multimode entangled states of light .
we note in passing that even the simple entangled states ( [ target ] ) may find applications in the investigations of the foundations of quantum mechanics .
for instance , it was shown recently that the states of the form ( [ target ] ) yield a strong violation of bell - chsh inequalities when alice and bob perform balanced homodyne measurements followed by an appropriate binning @xcite .
the paper is structured as follows . in section
ii , we propose a method to probabilistically generate the two - mode entangled state ( [ target ] ) with the help of passive linear optics , single photon sources , strong coherent laser fields and photodetectors . in section
iii we shall extend our scheme to allow for more than two modes and more than two terms in the superposition .
we will thus design a universal device that enables one to conditionally prepare an arbitrary multimode state of light .
finally , the conclusions are drawn in section iv .
let us first note that , in order to simplify the notation , we shall omit the normalization factors in front of the quantum states in what follows .
since the state @xmath3 is rather complicated , we divide its preparation into several steps .
the main simplification stems from the observation that the entangled state ( [ target ] ) can be prepared by means of entanglement swapping @xcite if we possess the two three - mode states @xmath4 where @xmath5 and @xmath6 denote the state of a single photon in spatial mode @xmath7 ( or @xmath8 ) that is polarized vertically or horizontally , respectively . as shown in fig . 1
, we easily obtain the state ( [ target ] ) from the state @xmath9 if we project the single photons in spatial modes @xmath7 and @xmath8 onto the maximally entangled bell state @xmath10 the generation of state ( [ target ] ) thus boils down to the preparation of the entangled state ( [ targetsp ] ) .
[ state ( [ targetspprime ] ) can be prepared similarly ] .
this latter task will be accomplished with the help of a quantum non - demolition ( qnd ) measurement of a single photon , which creates entanglement .
the proposed setup is depicted in fig .
the input polarization modes @xmath11 and @xmath12 are prepared in the pure single - mode states @xmath13 and @xmath14 that contain no more than @xmath15 photons , @xmath16 where @xmath17 denotes the @xmath18-photon fock state
. the states ( [ fgtildeexpansion ] ) are , of course , closely related to the states @xmath19 and @xmath20 appearing in eq .
( [ targetsp ] ) . the exact relationship between them will be specified later .
note that it was shown by dakna _
_ @xcite that the single - mode states ( [ fgtildeexpansion ] ) can be probabilistically generated with the help of passive linear optics , single photon sources , strong coherent laser pulses and photodetectors with single photon sensitivity . as shown in fig .
2 , the input beam containing the states @xmath13 and @xmath14 in the @xmath21 and @xmath22 polarization modes impinges on a beam splitter @xmath23 with transmittance @xmath24 and reflectance @xmath25 .
the purpose of this beam splitter is to separate a single photon from one of the input states and thus create entanglement .
of course , @xmath23 may split more than one photon or no photon at all .
therefore we must verify that there is exactly a single photon present in the output spatial mode @xmath26 , without disturbing its polarization state .
such qnd measurements of a single photon have been recently thoroughly discussed in ref .
@xcite where several schemes have been proposed and analyzed .
one possible method relies on teleportation @xcite . with the help of the conditional c - not gate for photonic qubits
@xcite we have to first prepare the singlet state @xmath27 in the auxiliary modes @xmath28
. then we teleport the polarization state of a photon in mode @xmath26 onto the photon in mode @xmath29 by performing a bell measurement on modes @xmath26 and @xmath30 .
this measurement should be carried out with two detectors that are capable to resolve the number of photons .
if exactly two photons are detected in coincidence at the two detectors , then we know that there was a single photon in mode @xmath26 and its polarization state has been coherently transfered onto the state of a single photon in mode @xmath29 . kok _ et al . _ also proposed qnd measurement schemes that do not require a - priori entanglement @xcite . for our purposes it suffices to employ the simple interferometric scheme depicted in fig .
3 which performs a partial qnd photon number measurement . the coincident detection of a single photon in detectors @xmath31 and
@xmath32 indicates that there was at least one photon in the input mode .
moreover , if there was _ exactly _ a single photon at the input , then its polarization state is unperturbed by the measurement .
the probability of successful qnd measurement of a single photon provided that there _ is _ a single photon in the spatial mode @xmath26 is equal to @xmath33 . in the preparation scheme shown in fig .
1 the two qnd measurements are followed by the entanglement swapping that succeeds only if each photodetector in fig . 1 detects exactly one photon . in this way we select the events when there is exactly a single photon in each mode @xmath7 and @xmath8 .
the state after the qnd measurement can be written in the form @xmath34 where the non - unitary operators @xmath35 and @xmath36 describe removal of none or a single photon at @xmath37 , respectively , @xmath38 the qnd measurement has thus created the entanglement . still , @xmath39 and @xmath40 are states of two different polarization modes while @xmath41 and @xmath42 in eq .
( [ targetsp ] ) are states of a _ single _ mode .
we achieve this by erasing the information about the polarization .
this is done with the help of a polarizing beam splitter @xmath43 , see fig .
the beam splitter is rotated such that it combines the @xmath21 and @xmath22 polarizations .
if we associate the annihilation operators @xmath44 and @xmath45 with the modes @xmath11 and @xmath12 , then the transformation carried out by @xmath43 is given by @xmath46 where the polarization modes @xmath47 and @xmath48 are _ spatially separated _ at the output of @xmath43 as indicated in fig .
2 . the erasing succeeds if and only if the detector @xmath49 placed on the @xmath48 mode does not detect any photon , which ensures that all photons have been transmitted to the linearly polarized output mode @xmath47 .
since the input states do not contain more than @xmath15 photons each and since a single photon was subtracted on @xmath23 , the state after erasure of the polarization information reads @xmath50 where the output complex amplitudes @xmath51 and @xmath52 are given by the following formula , @xmath53 it follows that we can fix the values of output coefficients @xmath51 by properly choosing the complex amplitudes of the input states .
we have the recurrence equations @xmath54 and a similar formula holds for @xmath55 .
the amplitudes @xmath56 and @xmath55 are calculated as follows .
first , some nonzero values for @xmath57 and @xmath58 are chosen .
then one uses repeatedly eq .
( [ inoutsolution ] ) to calculate @xmath56 and @xmath55 from @xmath59 and @xmath60 , @xmath61 , and the iterations stop at @xmath62 .
we have thus almost obtained the state ( [ targetsp ] ) .
however , while the effective states @xmath41 and @xmath42 in ( [ psierasure ] ) have the correct structure in the subspace spanned by the first @xmath15 fock states @xmath63 , they also contain an unwanted tail in the subspace spanned by the fock states @xmath64 .
the last step in our scheme is to get rid of that tail .
we do so with the help of the quantum scissors @xcite that will project the state in the mode @xmath65 onto the subspace spanned by the first @xmath15 fock states .
this action is formally described by the projector @xmath66 ) in the output modes 1 and 2 of our device .
the quantum scissors have been discussed in detail in several recent papers @xcite .
the schemes proposed there can be implemented with the use of linear optics , single photon sources , and photodetectors .
the setup that we shall consider in the present paper is depicted in fig .
the modes @xmath26 and @xmath30 are prepared in a pure @xmath0-photon two - mode entangled state @xmath67 it was shown recently that these states may be conditionally generated with the help of linear optics , single photon sources and photodetectors @xcite .
we then combine the modes @xmath65 and @xmath26 on a balanced beam splitter and measure the number of photons in output modes @xmath65 and @xmath26 by means of two photodetectors @xmath31 and @xmath32 with single - photon resolution .
the truncation is a conditional operation that succeeds when certain particular measurement outcome has been detected , say , no photons at @xmath31 and @xmath0 photons at @xmath32 , which we denote as @xmath68 .
note , that one could also consider other detection events , but the total number of photons detected by the photodetectors should be equal to @xmath0 .
the quantum scissors is essentially a quantum state teleportation in the subspace of the first @xmath15 fock states , where the state ( [ psiscissors ] ) serves as the quantum channel , and the bell - type measurement is carried out with the use of the beam splitter in fig .
4 that couples the modes @xmath65 and @xmath26 . for our particular choice of the detection event @xmath68 , and assuming a balanced beam splitter
, we find that @xmath69 should hold , where @xmath70 is the probability of the successful quantum - state truncation that projects onto the subspace spanned by @xmath71 .
so far we have considered the generation of a two - mode state that consists of a superposition of two product states . in this section
we shall generalize our scheme in several ways .
first , we will discuss the preparation of a @xmath72-mode entangled state which has a form limited to the superposition of two product states , namely @xmath73 to prepare this state , we first generate @xmath72 states @xmath74 as before , and then project the @xmath72 single photons onto the maximally entangled state @xmath75 this projection can be performed with the greenberger - horne - zeilinger ( ghz ) state analyzer described by pan and zeilinger @xcite .
this analyzer consists of polarizing beam splitters and half - wave plates and allows one to distinguish two ghz states of m qubits .
our second generalization extends the preparation procedure to two - mode states that are superpositions of @xmath76 product states , @xmath77 obviously , a natural strategy is to prepare this state via @xmath76-dimensional entanglement swapping from two states of the form @xmath78 where @xmath79 denotes a state of a single photon in the @xmath80th auxiliary spatial mode . to prepare the states ( [ psiquditsp ] ) , we proceed essentially as in sec .
the extended scheme is depicted in fig .
5 . initially ,
@xmath76 spatial modes are prepared in states @xmath81 .
the array of @xmath76 beam splitters @xmath82 extracts , with a certain probability , a single photon .
we must verify the presence of exactly one single photon in total in the modes @xmath83 while preserving the coherence among these modes by performing a qnd measurement .
this can be accomplished by @xmath76-dimensional teleportation similarly as discussed in section ii .
however , this time we need to teleport a qudit encoded as a state of a single photon in @xmath76 spatial modes .
fortunately , it was shown recently that the probabilistic gates for quantum computing with linear optics @xcite , that were introduced for qubits , can be easily extended to qudits @xcite .
this means that all the necessary manipulations , such as the generation of maximally entangled state of two qudits and the bell measurement , can be probabilistically carried out with the resources that we assume here .
after the qnd measurement , we must erase the spatial information contained in the resulting state @xmath84 where @xmath85 are given by eq .
( [ b01operators ] ) .
this is done by the second array of @xmath86 beam splitters indicated by the shaded zone in fig .
the preparation is finished by the application of the quantum scissors , that truncate the fock state expansion of the output state at @xmath0 photons .
suppose that the output mode after erasing is a balanced linear superposition of all @xmath76 modes @xmath87 the complex amplitudes of the states @xmath88 can then be expressed in terms of the complex amplitudes of the input states @xmath89 as follows @xmath90 where the prime indicates summation over all @xmath91 satisfying the constraint @xmath92 .
note that the formula ( [ finoutglobal ] ) may be easily inverted and we can determine the complex amplitudes @xmath93 of the input states for any given prescribed output states @xmath94 . finally , we point out that those two generalizations can be combined and we can thus generate an arbitrary multimode entangled state of light where each mode contains no more than @xmath0 photons .
any @xmath72-mode state of this kind can be written as a superposition of no more than @xmath95 product states , @xmath96 note that the norm of each product state in the superposition ( [ superposition ] ) may be arbitrary .
we can conditionally generate the state ( [ superposition ] ) if we first prepare @xmath72 entangled states ( @xmath97 ) @xmath98 where @xmath99 denotes a single photon state of the @xmath100-th auxiliary spatial mode . as discussed above , the states ( [ superpositionsp ] )
can be prepared with the help of the scheme depicted in fig .
, we carry out a joint measurement on all auxiliary photons .
the state @xmath101 is obtained if we project the auxiliary modes onto an @xmath72-photon ghz entangled state : @xmath102 if we denote @xmath103 then we can write @xmath104 the above described procedure for the preparation of the generic @xmath72-mode entangled state clearly shows that our approach is indeed general since it allows us , in principle , to generate an arbitrary state .
but the success probability of the scheme will be exponentially small in general .
the preparation procedure will therefore have to be repeated exponentially many times to produce the desired state . in the rest of this section
we shall argue that the number of operations required to produce the desired state can be significantly decreased if a quantum memory is available .
the situation is similar to that analyzed in ref .
@xcite where the operations required to produce `` cat states '' was drastically decreased by using a recursive procedure that uses a quantum memory .
the advantage of the quantum memory is that it allows one to wait until successive steps in the state generation protocol have succeeded before proceeding with the next steps .
for instance in the scheme described in fig . 5 , one can first produce the states @xmath81 and store them in a memory before proceeding with the production of the entangled state ( [ psiquditsp ] ) .
suppose that the state @xmath109 can be prepared with the probability @xmath110 . without quantum memory the probability of simultaneous preparation of the @xmath76 states
@xmath109 is @xmath111 and the number of required operations thus scales as @xmath112 .
if , after a successful preparation , we store each state @xmath13 in a memory , we reduce the number of required operations to @xmath113 .
a more elaborate use of a quantum memory is to build the state using a _ recursive procedure _ based on the schmidt decomposition .
the method is best explained on an explicit example .
suppose we would like to prepare a four - mode state @xmath114 .
the schmidt decomposition of this four - mode state with respect to a bi - partite splitting into the @xmath115 and @xmath116 modes can be written as follows , @xmath117 where @xmath118 and @xmath119 form orthogonal bases in some subspaces of the hilbert space of modes @xmath115 and @xmath116 , respectively .
the number of the terms in the schmidt decomposition ( [ schmidt ] ) is bounded by @xmath120 while , in general , @xmath121 terms were necessary in the general procedure of section iiic .
the decomposition ( [ schmidt ] ) suggests that we could obtain the state @xmath114 via entanglement swapping if we first prepare two entangled states @xmath122 now the entangled states @xmath123 and @xmath124 can be prepared using the general procedure of sec .
however , for states of the specific form ( [ abx ] ) , a simpler procedure can be devised .
one first generates @xmath76 states @xmath125 where the tilde indicates that @xmath126 is related to @xmath127 via relations similar to those described by eqs .
( [ inoutsolution ] ) .
the state @xmath123 is then prepared with the help of the device shown in fig .
5 where the mode @xmath128 of @xmath125 is sent to the @xmath80th input port of the interferometer .
the only modification is that we must also erase the spatial information carried by the @xmath129 modes of the states @xmath125 and perform the quantum state truncation after the erasing .
briefly , all @xmath76 modes @xmath130 are combined on an array of @xmath86 beam splitters and the detectors monitor the first @xmath86 output ports .
if these detectors do not register any photon , then the erasing procedure succeeded and we apply the quantum scissors to the state in the @xmath76th output port . clearly , this recursive procedure can be extended to any number of modes .
the resulting scheme resembles a tree structure involving repeated applications of the entangling scheme shown in fig .
5 and the entanglement swapping that produces an @xmath131-mode state from two @xmath132-mode states entangled with auxiliary single photons .
this scheme relying on quantum memory may possibly lead to a substantial improvement of the generation probability , similarly to the case of the two - mode @xmath0-photon cat states @xcite .
in summary , the scheme proposed in the present paper is universal in that arbitrary multimode entangled states of light can be probabilistically generated .
the resources required are passive linear optics , single photon sources , strong coherent states , and detectors with single - photon resolution .
this result clearly illustrates the versatility and power of the approach relying only on linear optics and single photons .
it is fair to say , though , that the suggested setup is rather complicated and involves several nontrivial operations such as the quantum non - demolition measurement of a single photon , quantum scissors , and transformations of photonic qudits encoded as a state of a single photon in @xmath76 spatial modes .
we have shown that all these operations can be ( probabilistically ) implemented with the resources that we consider here .
the resulting procedure may become very complex for general states .
however , it should be emphasized that building a state like ( [ superposition ] ) must necessarily be complicated because of the large number of parameters [ @xmath133 complex numbers ] that characterize this state and that must be fixed by the preparation scheme .
we have also argued that the use of a quantum memory can decrease the number of operations required to produce the desired state .
we hope that this paper will stimulate the efforts towards experimental demonstrations of the basic building blocks of our scheme , such as the conditional generation of single - mode finite superpositions of fock states and the preparation of two - mode entangled @xmath0-photon states required for the quantum state truncation .
we are grateful to m. duek , r. filip and p. grangier for stimulating discussions .
we acknowledge financial support from the communaut franaise de belgique under grant arc 00/05 - 251 , from the iuap programme of the belgian government under grant v-18 , from the eu under project resq ( ist-2001 - 37559 ) and chic ( ist-2001 - 32150 ) .
j.f . was also partially supported by the grant ln00a015 of the czech ministry of education . n.j .
cerf , j. fiurek , s. iblisdir , and s. massar , quant - ph/0210059 ; to appear in the proceedings of the sixth international conference on quantum communication , measurement and computing ( qcmc02 ) , boston ( 2002 ) . | we propose a universal scheme for the probabilistic generation of an arbitrary multimode entangled state of light with finite expansion in fock basis .
the suggested setup involves passive linear optics , single photon sources , strong coherent laser beams , and photodetectors with single - photon resolution .
the efficiency of this setup may be greatly enhanced if , in addition , a quantum memory is available . |
is the higgs boson recently found by the atlas and cms experiments the only scalar particle , or are there other higgs - like states around the fermi scale ?
this question is of fundamental importance for particle physics , and motivates a detailed study of the phenomenology of additional scalars , as well as the prospects for their discovery at the lhc and future colliders@xcite .
the simplest example of an extended higgs sector is realised adding just a real scalar field , singlet under all the known gauge groups , to the standard model ( sm ) . despite its great simplicity ,
this scenario is of considerable physical relevance , since it can easily arise in many of the most natural extensions of the sm
e.g. the next - to - minimal supersymmetric sm ( nmssm ) , twin higgs , some composite higgs models .
in general , such a singlet will mix with the higgs boson . as a consequence ,
both physical scalar states are coupled to sm particles , hence they can both be produced at colliders and be observed by means of their visible decays . in the following , after briefly reviewing the main properties of a generic singlet - like scalar , i shall present the constraints on the existence of such a particle that arise from both direct searches and higgs couplings precision measurements .
let us call @xmath0 and @xmath1 the two neutral , cp - even propagating degrees of freedom , with masses @xmath2 gev and @xmath3 .
they are related to the higgs and singlet gauge eigenstates via a mixing angle @xmath4 . in a weakly interacting theory ,
the couplings of @xmath0 and @xmath1 are just the ones of a standard higgs boson with the same mass , rescaled by a universal factor of @xmath5 or @xmath6 , respectively . as a consequence ,
their signal strengths @xmath7 are @xmath8 where @xmath9 is the corresponding signal strength of a sm higgs with mass @xmath10 , and @xmath11 is the branching ratio of @xmath1 into two 125 gev higgs bosons .
the phenomenology of the higgs system is therefore completely described by three parameters : @xmath3 , @xmath6 , and @xmath11 .
the second state @xmath1 behaves like a heavy sm higgs boson , with reduced couplings and an additional decay width into @xmath12 .
notice that the mixing angle @xmath4 and @xmath3 are not independent quantities , since the former has to vanish when the mass tends to infinity .
indeed , @xmath13 where @xmath14 is the first diagonal entry of the mass matrix of the scalar system before diagonalisation , which is proportional to the electroweak scale . in the limit of large @xmath3 , the goldstone boson equivalence theorem sets the relations @xmath15 the exact formulae for the @xmath16 and @xmath17 couplings are reported in reference@xcite . 0.1 in .
_ current and expected precisions on higgs couplings@xcite . _ [ cols="^,^,^,^,^,^",options="header " , ] [ table ] the measurement of the higgs signal strengths provides a constraint on the mixing angle @xmath4 through eq .
( [ muh ] ) . at present , a global fit to 8 tev lhc data constrain it to be @xmath18 at 95% c.l.@xcite .
projections for the reach of future hadron and lepton colliders@xcite are listed in table [ table ] .
large modifications to the triple higgs coupling can arise in some regions of the parameter space , even if the deviation in the signal strengths is moderate .
future collider experiments , and even the lhc , could in principle be sensitive to these modifications . more details about higgs couplings can be found in@xcite .
( left ) and @xmath19 ( right ) . in the left panel ,
the @xmath20 exclusion from higgs couplings is also superimposed , assuming a @xmath21 branching ratio into vectors.[direct]_,title="fig:",scaledwidth=50.0% ] ( left ) and @xmath19 ( right ) . in the left panel , the @xmath20 exclusion from higgs couplings is also superimposed , assuming a @xmath21 branching ratio into vectors.[direct]_,title="fig:",scaledwidth=50.0% ]
@xmath14 . @xmath11 has been fixed to 0.25 for simplicity . left : region relevant for the lhc .
right : projections for future colliders .
the notation for the lines is the same as in figure [ direct].[all]_,title="fig:",scaledwidth=45.0%]@xmath14 .
@xmath11 has been fixed to 0.25 for simplicity .
left : region relevant for the lhc .
right : projections for future colliders .
the notation for the lines is the same as in figure [ direct].[all]_,title="fig:",scaledwidth=46.4% ] the main decay channels of a heavy singlet are into a pair of @xmath22 and @xmath23 vector bosons , or into a pair of higgs bosons , if kinematically allowed .
both the atlas and cms collaborations provide a combined limit from all the @xmath24 and @xmath25 channels@xcite , with the strongest bound always coming from searches in the @xmath26 and @xmath27 final states . in the di - higgs channel ,
the main constraint comes from the @xmath28 final state@xcite .
all these searches are already sensitive to cross - sections smaller than the ones for a sm higgs at the same mass , and exceed the reach of higgs coupling measurements for low enough @xmath3 .
projections for future colliders have been obtained in@xcite , rescaling the expected limits from the 8 tev lhc with the parton luminosities of the backgrounds , following the procedure presented in@xcite .
the colliders that have been considered are : the 8 tev , 13 tev , and 14 tev lhc , its high - luminosity upgrade , a possible 33 tev energy upgrade , and a futuristic 100 tev fcc - hh . figure [ direct ] shows the present and extrapolated limits on the @xmath29 and @xmath19 signal strengths , normalised to sm values of the cross - sections . in the left panel
the projections for 125 gev higgs couplings measurements are also shown , in the limit of small @xmath11 .
figure [ all ] again shows a comparison between direct and indirect searches , but this time in the @xmath3@xmath14 plane , and for @xmath30 .
the direct exclusion is dominated by @xmath31 .
the higgs sector of the nmssm@xcite contains the two usual doublets @xmath32 , plus a singlet scalar @xmath33 , coupled through a yukawa interaction @xmath34 in the superpotential .
an extra contribution to the higgs mass is generated at tree - level by @xmath35 , and reduces the size of the radiative correction needed to obtain @xmath36 gev .
at the same time , the fine - tuning of the electroweak scale @xmath37 is reduced . in the decoupling limit for the heavy doublet ,
the cp - even states are the sm higgs and the singlet , and can be matched to the previous scenario via@xcite @xmath38 where @xmath39 is the radiative correction and @xmath40 .
figure [ models ] ( left ) shows the current exclusions and projections from both direct searches and higgs couplings , in the plane @xmath3@xmath41 , for fixed values of @xmath42 and @xmath43 gev . and @xmath43 gev ( left ) , and the twin higgs radial mode ( right ) .
the notation is the same as in figure [ direct ] . in the purple region the width @xmath44.[models]_,title="fig:",scaledwidth=45.0% ] and @xmath43 gev ( left ) , and the twin higgs radial mode ( right ) .
the notation is the same as in figure [ direct ] . in the purple region the width @xmath44.[models]_,title="fig:",scaledwidth=48.2% ] in twin higgs models@xcite , a naturally light higgs is obtained without the presence of coloured particles close to the tev scale .
this is achieved introducing a copy of the sm field content and gauge symmetries , @xmath45 .
the higgs potential has an approximate global @xmath46 symmetry , which is spontaneously broken at a scale @xmath47 , and the higgs @xmath48 is a goldstone boson of this breaking .
quadratic `` divergences '' in the higgs mass cancel between the a and b sectors , while all the new twin particles are sm singlets .
the phenomenology of the `` radial mode '' @xmath49 is described by eq .
( [ mus ] ) , ( [ mushh ] ) .
the mixing angle is proportional to @xmath50 , and one has @xmath51 the only difference with respect to the previous cases is the presence of an invisible width into @xmath52 and @xmath53 bosons .
figure [ models ] ( right ) illustrates the present and future constraints in the plane @xmath54@xmath47 , which are the only two free parameters of the model .
one can see that direct searches for the radial mode are the most powerful probe for a twin higgs scenario , at least for not too large values of @xmath54 and @xmath47 .
searches for scalar singlets at colliders can be an important probe for the extended higgs sectors of many physically motivated models , and complementary to the measurement of higgs couplings . by means of only three parameters that determine the phenomenology in a completely general way
, the reach of future colliders in the relevant @xmath55 and @xmath12 channels has been studied . on the other hand , already
the second run of the lhc can efficiently explore this scenario , and will provide valuable information in the near future .
i thank filippo sala and andrea tesi for the stimulating collaboration on which the results presented here are based .
this research was supported in part by the swiss national science foundation ( snf ) under contract 200021 - 159720 .
99 d. buttazzo , f. sala , a. tesi , jhep * 1511 * 033 ( 2015 ) , [ http://arxiv.org/abs/1505.05488[arxiv:1505.05488 ] ] .
giardino _ et al .
_ , jhep * 1405 * 046 ( 2014 ) , [ http://arxiv.org/abs/1303.3570[arxiv:1303.3570 ] ] .
s. dawson _ et al .
_ , working group report : higgs boson , community summer study 2013 : snowmass on the mississippi ( 2013 ) , [ http://arxiv.org/abs/1310.8361[arxiv:1310.8361 ] ] .
atlas collaboration , [ http://arxiv.org/abs/1507.05930[arxiv:1507.05930 ] ] .
+ cms collaboration , jhep * 1510 * 144 ( 2015 ) , [ http://arxiv.org/abs/1504.00936[arxiv:1504.00936 ] ] .
atlas collaboration , phys .
d * 92 * 032004 ( 2015 ) [ http://arxiv.org/abs/1509.04670[arxiv:1509.04670 ] ] .
+ cms collaboration , phys .
b * 749 * 560 ( 2015 ) , [ http://arxiv.org/abs/1503.04114[arxiv:1503.04114 ] ] .
a. thamm , r. torre , a. wulzer , jhep * 1507 * 100 ( 2015 ) , [ http://arxiv.org/abs/1502.01701[arxiv:1502.01701 ] ] .
see e.g. u. ellwanger , c. hugonie , a. m. teixeira , phys . rept .
* 496 * 1 ( 2010 ) , [ http://arxiv.org/abs/0910.1785[arxiv:0910.1785 ] ] .
r. barbieri _ et al .
_ , phys .
d * 87 * , 115018 ( 2013 ) , [ http://arxiv.org/abs/1304.3670[arxiv:1304.3670 ] ] . z. chacko , h .- s .
goh , r. harnik , phys .
lett . * 96 * , 231802 ( 2006 ) , [ http://arxiv.org/abs/hep-ph/0506256[hep-ph/0506256 ] ] . | a scalar singlet , coupled to the other particles only through its mixing with the higgs boson , appears in several motivated extensions of the standard model .
the prospects for the discovery of a generic singlet at the various stages of the lhc , as well as at future high - energy colliders , are studied , and the reach of direct searches is compared with the precision attainable with higgs couplings measurements .
the results are then applied to the nmssm and twin higgs .
= 11.6pt |
branched cyclic covers of @xmath0 have played a major role in topology , and continue to appear in a wide variety of contexts . for example , branched cyclic covers of @xmath0 branched over two - bridge knots have recently appeared in combinatorial work bounding the matveev complexity of a 3-manifold @xcite , in algebraic and topological work determining relations between @xmath1-spaces , left - orderability , and taut foliations @xcite , and in geometric work giving information on maps of character varieties @xcite .
they provide a wealth of examples , and a useful collection of manifolds on which to study conjectures . given their wide applicability , and their continued relevance , it is useful to have many explicit descriptions of these manifolds . in this paper , we give a new and particularly elegant construction of the branched cyclic covers of two - bridge knots , using the bi - twist construction of @xcite .
while other presentations of these manifolds are known ( see , for example @xcite ) , we feel our descriptions have several advantages , as follows .
first , they follow from a recipe involving exactly the parameters necessary to describe a two - bridge knot , namely , continued fraction parameters .
our descriptions apply uniformly to all two - bridge knots , and all branched cyclic covers of @xmath0 branched over two - bridge knots .
second , they are obtained from a description of a two - bridge knot using a particularly straight - forward bi - twisted face pairing construction , as in @xcite .
bi - twisted face - pairings are known to produce all closed orientable 3-manifolds .
the examples of this paper show , in addition , that bi - twist constructions are often efficient and natural . while a generic face - pairing will yield a pseudomanifold , which , with probability 1 , will not be an actual manifold @xcite , bi - twisted face - pairings avoid this problem . in this paper
, we will review necessary information on bi - twisted face pairings , so no prior specialized knowledge is required to understand our constructions .
third , our description leads to immediate consequences in geometric group theory .
we obtain a simple proof of the fact that the fundamental group of the @xmath2-fold branched cyclic covering of @xmath0 , branched over a two - bridge knot , has a cyclic presentation .
our description also gives immediate presentations of two well known families of groups , the fibonacci and sieradski groups .
these are known to arise as fundamental groups of branched cyclic covers of @xmath0 branched over the figure - eight and trefoil knots , respectively .
these groups have received considerable attention from geometric group theorists ; see , for example @xcite for further references , and section [ sec - history ] for more history .
our methods recover the fact that the first homology groups of sieradski manifolds are periodic .
we also give a proof that their fundamental groups are distinct using milnor s characterization of these spaces .
we consider orders of abelianizations of fibonacci groups , as well .
these orders form an interesting sequence related to the fibonacci sequence , which we shall see .
we will see that the bi - twist description of any two - bridge knot is encoded as the image of the north - south axis in a ball labeled as in figure [ fig - tbf1 ] , along with an associated vector of integer multipliers . for the branched cyclic
cover , the description is encoded by adding additional longitudinal arcs to the sphere .
we now describe the construction briefly , in order to state the main results of the paper .
a more detailed description of the construction , with examples , is given in section [ sec - trefoil ] .
begin with a finite graph @xmath3 in the @xmath4-sphere @xmath5 that is the union of the equator @xmath6 , one longitude @xmath7 from the north pole @xmath8 to the south pole @xmath9 , and @xmath10 latitudinal circles , such that @xmath3 is invariant under reflection @xmath11 in the equator .
then @xmath3 divides @xmath12 into @xmath13 faces that are paired by @xmath14 .
this face - pairing is shown in figure [ fig - tbf1 ] .
as with any face - pairing , the edges fall into edge cycles .
the equator @xmath6 forms one edge cycle @xmath15 since the reflection @xmath14 leaves @xmath6 invariant .
each other edge of the graph is matched with its reflection to form another edge cycle @xmath16 .
we number these edge cycles from @xmath17 through @xmath18 , with even numbers associated with latitudinal edges , as indicated in figure [ fig - tbf2 ] .
now choose nonzero integer _
, denoted @xmath19 , @xmath20 , @xmath21 , @xmath22 , @xmath23 , for the edge cycles @xmath16 .
in the case at hand , restrict the choice of multipliers @xmath24 as follows .
each latitudinal edge cycle @xmath25 is assigned either @xmath26 or @xmath27 as multiplier .
each longitudinal edge cycle @xmath28 may be assigned any integer multiplier @xmath29 whatsoever , including @xmath17 .
the multiplier @xmath30 is usually forbidden , but in this case indicates that the two edges of edge - class @xmath28 must be collapsed to a point before the bi - twist construction is engaged .
finally , for the general bi - twist construction , we obtain a closed manifold @xmath31 by taking the following quotient .
first , subdivide each edge in the edge cycle @xmath16 into @xmath32 sub - edges .
insert an additional edge between each adjacent positive and negative edge , if any . then twist each sub - edge by one sub - edge in a direction indicated by the sign of @xmath24 .
finally , apply the face - pairing map @xmath14 to glue bi - twisted faces .
this is the bi - twist construction . in theorem [ thm - qiss3 ]
, we prove that the bi - twist manifold @xmath33 described above is the @xmath34-sphere @xmath0 .
the image of the north south axis in @xmath0 is a two - bridge knot .
in fact , we prove more . recall that every two - bridge knot is the closure of a rational tangle .
see @xcite for an elementary exposition .
a rational tangle is determined up to isotopy by a single rational number , which we call the rational number invariant of the tangle .
there are two natural ways to close a tangle so that it becomes a knot or link , the numerator closure and the denominator closure .
the full statement of theorem [ thm - qiss3 ] is below .
theorem [ thm - qiss3 ] the bi - twist manifold @xmath33 is the @xmath34-sphere @xmath0 .
the image of the north south axis in @xmath0 is the two - bridge knot which is the numerator closure of the tangle @xmath35 whose rational number invariant @xmath36 is @xmath37 the @xmath4 s in the continued fraction indicate that the tangle is constructed using only full twists instead of the possible mixture of full and half twists . the simplest case , with only equator and longitude , yields the trefoil and figure - eight knots , as we shall see in theorem [ thm : trefoil ] .
simple subdivisions yield their branched cyclic covers , the sieradski @xcite and fibonacci @xcite manifolds .
[ def - normlized ] we say that the multiplier function @xmath38 is _ normalized _ if the following hold : 1 .
@xmath39 , and 2 .
if @xmath40 for some @xmath41 , then @xmath42 . with this definition ,
the previous theorem and well - known results involving two - bridge knots yield the following corollary .
corollary [ cor - denomclosure ] every normalized multiplier function yields a nontrivial two - bridge knot .
conversely , every nontrivial two - bridge knot @xmath43 is realized by either one or two normalized multiplier functions . if @xmath43 is the numerator closure of the tangle @xmath35 , then it has exactly one such realization if and only if @xmath44 notice that the @xmath2-th branched cyclic covering of @xmath0 , branched over @xmath43 , can be obtained by unwinding the description @xmath2 times about the unknotted axis that represents @xmath43 , unwinding the initial face - pairing as in figure [ fig - tbf22 ] .
this has the following application .
theorem [ thm - gn ] the fundamental group of the @xmath2-th branched cyclic covering of @xmath0 , branched over @xmath43 , has a cyclic presentation .
[ prob : knots ] how should one carry out the analogous construction for arbitrary knots ? since the knots in the face - pairing description appear as the unknotted axis in @xmath45 , it is easy to unwind @xmath45 around the axis to obtain face pairings for the branched cyclic coverings of @xmath0 , branched over the trefoil knot and the figure - eight knot . for the trefoil knot ,
the @xmath2-th branched cyclic cover @xmath46 is called the @xmath2-th sieradski manifold . for the figure - eight knot , the @xmath2-th branched cyclic cover @xmath47 is called the @xmath2-th fibonacci manifold .
we will prove : theorem [ thm : fnsnpresentation ] the fundamental group @xmath48 is the @xmath49-th fibonacci group with presentation @xmath50 the fundamental group @xmath51 is the @xmath2th sieradski group with presentation @xmath52 the group presentations are well - known once the manifolds are recognized as branched cyclic covers of @xmath0 , branched over the figure - eight knot and the trefoil knot .
but these group presentations also follow immediately from the description of the bi - twist face - pairings , as we shall see .
the first homology of the sieradski manifolds has an intriguing periodicity property , which is well - known ( see for example rolfsen @xcite ) .
in particular , it is periodic of period @xmath53 .
the following theorem , concerning their fundamental groups , is not as well known ; it is difficult to find in the literature .
we give a proof using milnor s characterization of these spaces .
theorem [ thm : siergps ] no two of the sieradski groups are isomorphic .
hence no two of the branched cyclic covers of @xmath0 , branched over the trefoil knot , are homeomorphic .
this paper is organized as follows . in section [ sec - trefoil ]
, we will give a more careful description of the bi - twisted face - pairing , and work through the description for two examples , which will correspond to the trefoil and figure - eight knots .
in section [ sec - heegaard ] , we recall many of the results in our previous papers @xcite to make explicit the connections between face - pairings , heegaard splittings , and surgery descriptions of 3-manifolds .
we apply these to the examples of bi - twisted face pairings given here , to give surgery descriptions .
we use these descriptions in section [ sec - twobridge ] to prove that our constructions yield two - bridge knots .
the proofs of the main geometric theorems are given in this section .
in section [ sec - cyclicpresentations ] we turn to geometric group theory .
we prove that our presentations easily lead to well - known results on presentations of fundamental groups .
we also give results on fibonacci and sieradski groups in this section . finally , section [ sec - history ] explains some of the history of these problems .
we thank the referees of an earlier version of this paper for numerous helpful comments .
purcell is partially supported by nsf grant dms-1252687 . though leer lambert spent his life as an actuary and a musician and was a loving father of nine girls and one boy , leer had always wanted to earn an advanced degree as a mathematician . with the encouragement of his wife , he earned his ph.d . in mathematics at the age of 68 .
many of the results of this paper appeared in his byu ph.d .
dissertation . at the age of 71 , leer died of bone cancer .
we miss you , leer .
in this section we step through the bi - twisted face - pairing description more carefully .
we believe it will be most useful to work through a pair of examples .
we will see in subsequent sections that these examples lead to fibonacci and sieradski manifolds . as an example , consider the simplest model , shown in figure [ fig - tbf3](a ) .
the graph has three edges and three vertices , and divides the sphere into two singular `` triangles '' , which are then matched by reflection @xmath14 in the equator @xmath6 . [ cols="^,^,^ " , ] we use the same labels @xmath54 as group generators .
the corresponding faces and generators for the southern hemisphere are @xmath55 .
we distinguish three types of faces : those bordering on the equator , which are designated as type 0 , those touching the poles , which are designated as type 2 , and all others , designated type 1 .
we initially assume that @xmath56 so that we do nt have faces that are both type 0 and type 2 .
since edge classes have size 1 or size 2 , the words associated with a face - edge pair have length 1 or length 2 .
figure [ fig - tbf23 ] shows edges of the three types of faces labeled with those face - edge words .
these words are then raised to the appropriate powers and multiplied together to give the word associated with the corresponding face .
we call these words @xmath57 s since they are the relators of the fundamental group . @xmath58^{\ell_0 } [ x(0,j)x(0,j+1)^{-1}]^{m_0 } [ x(0,j)x(1,j)^{-1}]^{\ell_1 } [ x(0,j)x(0,j-1)^{-1}]^{m_0 } \\ r(i , j ) & = [ x(i , j)x(i-1,j)^{-1}]^{\ell_i } [ x(i , j)x(i , j+1)^{-1}]^{m_i } [ x(i , j)x(i+1,j)^{-1}]^{\ell_{i+1 } } [ x(i , j)x(i , j-1)^{-1}]^{m_i } \\ r(k ,
j ) & = [ x(k , j)x(k-1,j)^{-1}]^{\ell_k } [ x(k , j)x(k , j+1)^{-1}]^{m_k } [ x(k , j)x(k , j-1)^{-1}]^{m_k}\end{aligned}\ ] ] we conclude that the fundamental group has a presentation @xmath59 since each of the multipliers @xmath60 is either @xmath26 or @xmath27 , the letter @xmath61 appears at most once in the relator @xmath62 .
similarly , the letter @xmath63 appears at most once in the relator @xmath64 , for @xmath65 , and the letter @xmath66 appears at most once in the relator @xmath67 .
hence , these relators may be solved for @xmath68 iteratively , and then these relators and generators may be removed .
the only generators remaining are the generators @xmath69 , with @xmath70 ; and , with appropriate generator substitutions made , the only remaining relators are the relators @xmath71 .
the presentation @xmath72 is clearly a cyclic presentation .
finally , if @xmath73 , then every face is both type 0 and type 2 . in this case the presentation is @xmath74 , which is cyclic .
we illustrate these group calculations with the fibonacci manifolds and the sieradski manifolds .
we prove the following theorem .
[ thm : fnsnpresentation ] the fundamental group @xmath48 is the @xmath49-th fibonacci group with presentation @xmath50 the fundamental group @xmath51 is the @xmath2-th sieradski group with presentation @xmath52 the faceted 3-ball that serves as the model for the face pairings is the same for both manifolds ; it is as in figure [ fig - tbf22 ] with @xmath73 , so without interior latitudinal circles . for the fibonacci manifolds , we label the faces of the northern hemisphere as @xmath75 , @xmath76 , @xmath21 , @xmath77 .
all subscript calculations are modulo @xmath49 .
we obtain the following cyclic presentation for the fundamental group : @xmath78\cdot [ x(2j)x(2j-2)\inv ] , \ , j=1,2,\ldots,\,n\rangle.\end{aligned}\ ] ] we can then introduce intermediate generators @xmath79 .
the presentation becomes the standard presentation for the @xmath49-th fibonacci group , as desired : @xmath80 for the sieradski manifolds , we label the faces of the northern hemisphere as @xmath81 , @xmath82 , @xmath21 , @xmath83 .
subscript calculations are modulo @xmath2 .
we obtain the following cyclic presentation for the fundamental group : @xmath84 \cdot [ y(j)y(j-1)\inv ] , j=1,\dotsc , n\rangle,\ ] ] or , reversing the order of the subscripts so that @xmath85 , @xmath86 the standard presentation for the @xmath2th sieradski group . in this section
we consider first homology groups of our cyclic branched covers of @xmath0 .
this is a topic which has received and still receives considerable attention .
there are two very different behaviors .
the first homology groups of the @xmath2-fold cyclic covers @xmath87 of @xmath0 branched over a knot @xmath43 are either periodic in @xmath2 or their orders grow exponentially fast .
specifically , gordon @xcite proved that when the roots of the alexander polynomial of @xmath43 are all roots of unity , then @xmath88 is periodic in @xmath2 .
riley @xcite and , independently , gonzalez - acuna and short @xcite proved that if the roots of the alexander polynomial are not all roots of unity , then the finite values of @xmath88 grow exponentially fast in @xmath2 .
silver and williams @xcite extended these results to links and replaced `` finite values '' with `` orders of torsion subgroups '' .
see also le @xcite , bergeron and venkatesh @xcite and brock and dunfield @xcite for more recent results and conjectures on this topic .
we are particularly fascinated by the first homology of the branched cyclic covers of @xmath0 branched over the knots that are two - strand braids .
these knots are the only two - bridge knots that are not hyperbolic .
the northern hemisphere of the model before bi - twisting looks like figure [ fig - tbf22 ] .
we construct the @xmath2-fold branched cyclic cover of @xmath0 , branched over a knot that is a @xmath4-strand braid , by using @xmath89 latitudes and @xmath2 longitudes in the open northern hemisphere , assigning multipliers @xmath27 to the latitudinal edges , and assigning multipliers @xmath26 to all longitudinal edges .
we calculate the fundamental group as in the proof of theorem [ thm - gn ] and transform it into a cyclic presentation as explained there .
we then abelianize , and let @xmath90 denote the exponent sums of the generators in the defining cyclic word @xmath91 .
we very briefly indicate by diagram how these integers may be computed .
every relator corresponds to a diagram as follows .
@xmath92 @xmath93 @xmath94 we begin with the diagram for @xmath71 and use the diagrams for @xmath67 , @xmath95 , @xmath21 to successively transform the entries in rows @xmath96 , @xmath97 , @xmath98 to 0 .
the defining cyclic word is the final result in row 0 .
@xmath99 we find that the polynomial @xmath100 is the cyclotomic polynomial @xmath101 ( if @xmath102 , then the polynomial folds on itself because powers are to be identified modulo @xmath2 . however , once @xmath103 , there is no folding . )
the computation indicated by diagram is a continued fraction algorithm . for the fundamental group of a general two - bridge knot
, the corresponding polynomial may be taken to be the numerator of the continued fraction @xmath104 where @xmath105 and @xmath106 we shall prove that , for a given knot realized as a two - strand braid , the abelianizations of the fundamental group of the @xmath2-fold branched cover are periodic functions of @xmath2 . however , as a warm up , we use row reduction of the presentation matrix to prove the much easier theorem that no two of the fibonacci groups @xmath107 are isomorphic for @xmath108 since no two of the abelianizations have the same order .
this problem appears as an exercise on page @xmath109 of @xcite , where johnson suggests using the two - variable presentation of the group .
we use the @xmath2-variable presentation and note that the fibonacci numbers @xmath110 appear in a very natural way . in this case
we have the behavior of exponential growth of orders .
[ thm : fibonacci - growth ] let @xmath111 , with subscripts calculated modulo @xmath2 . for odd @xmath2 , the order of the abelianization is the sum @xmath112 of two fibonacci numbers . for even @xmath2 ,
the order is @xmath113 .
recall that for even @xmath2 these abelianizations are the first homology groups of the fibonacci manifolds .
this theorem gives successive orders of @xmath114 , @xmath114 , @xmath115 , @xmath116 , @xmath117 , @xmath118 , @xmath119 , @xmath120 , @xmath121 , @xmath122 , @xmath21 for the abelianizations of the fibonacci groups .
it is clear from the definition of the fibonacci numbers that these numbers are strictly increasing after the numbers @xmath123 .
these numbers are also known as the _ associated mersenne numbers _
a001350 in sloane s `` the on - line encyclopedia of integer sequences '' .
the sums @xmath112 are also known as lucas numbers .
the presentation matrix has the following form : @xmath124 the absolute value of the determinant of this matrix is the order of the abelianization of the group unless the determinant is @xmath17 . in that case
, the group is infinite .
the goal is to move the entries in the lower left corner to the right by adding multiples of the upper rows .
these operations do not change the determinant .
we use the upper rows in descending order , with each successive row moving the lower - left @xmath125 matrix one column to the right .
we first trace the evolution of the two entries in the next - to - last row : @xmath126 the reader will easily identify the first in the @xmath96th pair as @xmath127 , and the second as @xmath128 .
since the second of these , namely @xmath129 , coincides with the first pair in the bottom row , we see that the bottom row evolves just one step ahead of the next - to - last row .
thus after @xmath96 moves , the @xmath125 matrix evolves into the matrix @xmath130 which has determinant @xmath131 .
after the appropriate number of moves , this matrix will be added to the matrix @xmath132 from the lower right corner to form the very last lower right corner matrix @xmath133 the matrix then has determinant @xmath134 [ f_k^2 + 2\cdot(-1)^k\cdot f_k + 1 ] & -[f_{k+1}\cdot f_{k-1 } + ( -1)^{k+1 } f_{k+1}]\\ & = ( -1)^{k+1 } + 1 + ( -1)^k[f_k + f_{k+2 } ] .
\end{aligned}\ ] ] the absolute value of this determinant is the order of the abelianization , and since the last value of @xmath96 is @xmath135 , it agrees with the value claimed in the theorem .
for the moment , we fix two integers @xmath136 and @xmath89 , and let @xmath137 , with @xmath138 , denote an abelian group with generators @xmath139 such that @xmath140 and with relators @xmath141 for each @xmath142
. then the group has a circulant relator matrix of the form shown in figure [ fig - tbf28 ] . in the following theorem
we have the behavior of periodic homology groups .
let @xmath143 , @xmath96 , and @xmath137 be as immediately above , so that @xmath138 .
assume that @xmath144 is a cyclotomic polynomial , by which we mean that there is a polynomial @xmath145 such that @xmath146 .
then the groups @xmath137 and @xmath147 are isomorphic .
we manipulate the relator matrix for @xmath148 using integral row and column operations .
see figure [ fig - tbf28 ] .
we use the rows at the top of the matrix to remove entries from the triangle at the lower left corner of the matrix .
let @xmath149 be such an entry in row @xmath150 .
let @xmath151 denote the row whose initial entry on the diagonal is above @xmath149 .
subtract from row @xmath150 the sum @xmath152 $ ] .
the effect is to move entry @xmath149 to the right @xmath153 places .
similarly , we move all entries in the lower left triangle @xmath153 places to the right . because @xmath154 , we may use column operations to make every entry to the right of the first @xmath155 @xmath156 s equal to 0 . the lower right @xmath157 block of the resulting matrix is the relator matrix for @xmath137 .
the theorem follows .
the same calculation can be carried out if the polynomial is any integer multiple @xmath158 of a cyclotomic polynomial @xmath159 , except that the diagonal entries above the periodic box all become @xmath160 s .
thus the abelianization has a periodic component together with an increasing direct sum of @xmath161 s .
it can be shown that these are the only polynomials with these periodicity properties .
[ cor:2strandbraidperiodic ] if @xmath43 is a knot that is a 2-strand braid and @xmath87 is the @xmath2-fold cyclic branched cover of @xmath0 over @xmath43 , then the homology groups @xmath162 are periodic in @xmath2 .
lambert , in his ph.d .
dissertation at brigham young university @xcite , explicitly calculated all of the homology groups of the branched cyclic covers of @xmath0 , branched over knots that are @xmath4-strand braids .
these are the only two - bridge knots that are not hyperbolic .
his tables give an explicit picture of the periodicity we have just proved .
rolfsen @xcite notes that the period for the trefoil is @xmath53 .
we shall also see that as follows .
it suffices to find the appropriate polynomials @xmath163 , and thereby determine the period .
if @xmath164 , as for the trefoil , then the appropriate @xmath163 of smallest degree is @xmath165 so that the period is @xmath166 . with @xmath116 half twists , @xmath167 and @xmath168 and the period is @xmath169 .
each added pair of half twists in the braid adds two terms to @xmath159 , multiplies the negative entries of @xmath163 by @xmath170 , and increases the period by @xmath115 . by gordon @xcite , the homology groups @xmath162 of the cyclic branched covers @xmath87 of the complement of a knot @xmath43 are periodic with period dividing @xmath38 if and only if the first alexander invariant ( the quotient of the first two alexander polynomials ) of @xmath43 is a divisor of the polynomial @xmath171 .
furthermore , if the first alexander invariant is a divisor of @xmath172 and @xmath2 is a positive integer , then @xmath173 , where @xmath174 is the greatest common divisor of @xmath38 and @xmath2 .
since the first alexander invariant of the trefoil knot is @xmath175 , which divides @xmath176 , gordon s theorem shows that the first homology groups of the cyclic branched covers of the trefoil knot are periodic with period @xmath53 and @xmath177 for all @xmath143 .
we may use the calculation of the period of the trefoil in establishing the following theorem .
[ thm : siergps ] no two of the sieradski groups are isomorphic .
hence no two of the branched cyclic covers of @xmath0 , branched over the trefoil knot , are homeomorphic . in @xcite
, milnor defines the brieskorn manifold @xmath178 to be the orientable closed 3-manifold obtained by intersecting the complex algebraic surface given by @xmath179 with the unit sphere given by @xmath180 .
here @xmath181 , @xmath182 , @xmath183 should be integers at least 2 .
theorem 2.1 of @xcite by cavicchioli , hegenbarth , and kim states that @xmath46 is the brieskorn manifold @xmath184 .
this follows from the fact that @xmath46 is the @xmath2-fold cyclic branched cover of @xmath185 branched over the trefoil knot , which is the torus knot of type @xmath186 , and lemma 1.1 of @xcite , which states that the brieskorn manifold @xmath178 is the @xmath183-fold cyclic branched cover of @xmath185 branched over a torus link of type @xmath187 .
the first few @xmath2-fold cyclic covers of @xmath185 branched over the right - hand trefoil knot are discussed in section 10d of rolfsen s book @xcite , which begins on page 304 . here
are the results .
* @xmath188 : the manifold @xmath189 is the 3-sphere @xmath185 , and so @xmath190 .
* @xmath191 : the manifold @xmath192 is the lens space @xmath193 , so @xmath194 . * @xmath195 : the manifold @xmath196 is the spherical 3-manifold with fundamental group @xmath197 the quaternion group of order 8 .
it appears in example 7.2 of @xcite .
this group might be called the binary klein 4-group .
* @xmath198 : the manifold @xmath199 is the spherical 3-manifold with fundamental group @xmath200 the binary tetrahedral group .
* @xmath201 : the manifold @xmath202 is the spherical 3-manifold with fundamental group @xmath203 the binary icosahedral group . in other words , this is the poincar homology sphere . * @xmath204 : the manifold @xmath205 is the heisenberg manifold . here
@xmath206=[y,[x , y]]=1\right\rangle.\ ] ] in @xcite milnor proves that @xmath184 , which we know is homeomorphic to @xmath46 , is an @xmath207-manifold for @xmath208 .
it follows that @xmath209 are distinct and that they are not @xmath207 manifold groups .
because of this and milnor s result that @xmath46 is an @xmath207-manifold for @xmath208 , to prove that the groups @xmath137 are distinct , it suffices to prove that the groups @xmath137 are distinct for @xmath208 .
as stated on page 304 of rolfsen s book @xcite , for every positive integer @xmath2 the first homology group @xmath210 is @xmath211 , 0 , @xmath212 or @xmath213 when @xmath214 .
so to prove that sieradski groups @xmath215 and @xmath137 are distinct , we may assume that @xmath216 .
for the rest of this section suppose that @xmath208 . in @xcite ( see the bottom of page 213 and lemma 3.1 ) milnor proves that @xmath137 is isomorphic to the commutator subgroup of the centrally extended triangle group @xmath217 .
let @xmath218 , a homomorphic image of @xmath219 .
the group @xmath220 is the group of orientation - preserving elements of the @xmath221-triangle group .
let @xmath222 denote the commutator subgroup of @xmath220 .
we see that the quotient group @xmath223 is isomorphic to the group generated by the elements @xmath224 , @xmath225 , and @xmath226 in @xmath227 with relations corresponding to a matrix which row reduces as follows . @xmath228\longrightarrow \left[\begin{matrix}1 & 1 & 1 \\ 0 & -2 & -2 \\ 0 & 3 & 0 \\ 0 & 0 & n
\end{matrix}\right]\longrightarrow \left[\begin{matrix}1 & 1 & 1 \\ 0 & -2 & -2 \\ 0 & 1 & -2 \\ 0 & 0 & n \end{matrix}\right]\longrightarrow \left[\begin{matrix}1 & 0 & 3 \\ 0 & 1 & -2 \\ 0 & 0 & 6 \\ 0 & 0 & n \end{matrix}\right]\ ] ] so @xmath223 is a cyclic group of order @xmath229 .
this computation also shows that @xmath230 if and only if @xmath231 , that @xmath232 if and only if @xmath233 , and that @xmath234 is the smallest power of @xmath235 in @xmath222 .
in particular @xmath234 is a nontrivial elliptic element of @xmath222 .
every element of @xmath222 which commutes with @xmath234 must fix the fixed point of @xmath234 .
it easily follows that the center of @xmath222 is trivial , and in the same way that the center of @xmath220 is trivial .
since the kernel of the homomorphism from @xmath219 to @xmath220 is generated by the central element @xmath236 and the center of @xmath220 is trivial , it follows that the kernel of this homomorphism is the center of @xmath219 .
so @xmath219 modulo its center is isomorphic to @xmath220 .
similarly , @xmath137 modulo its center is isomorphic to @xmath222 .
now suppose that @xmath237
. then @xmath137 modulo its center is isomorphic to @xmath238 .
the largest order of a torsion element in @xmath220 is @xmath2 .
so @xmath215 and @xmath137 are distinct if @xmath239 .
next suppose that @xmath240 . in this case
the largest order of a torsion element in @xmath222 is @xmath241 .
so @xmath215 and @xmath137 are distinct if @xmath242 .
the same argument is valid if @xmath243 .
finally suppose that @xmath244 . in this case neither @xmath245 nor @xmath246 are in @xmath222 . in this case every torsion element in @xmath222 is conjugate to a power of @xmath247 , which has order @xmath248
again @xmath215 and @xmath137 are distinct if @xmath249 .
there is a large literature concerning the fibonacci groups , the sieradski groups , their generalizations , cyclic presentations of groups , the relationship between cyclic presentations and branched cyclic covers of manifolds , two - bridge knots , and their generalizations .
we are incapable of digesting , let alone giving an adequate summary , of this work .
we plead forgiveness forhaving omitted important and beautiful work and for misrepresenting work that we have not adequately studied .
john conway told the first - named author of this paper that he created the fibonacci group @xmath250 , with presentation @xmath251 and asked that his graduate students calculate its structure as an exercise to demonstrate that it is not easy to read the structure of a group from a group presentation .
for example , our straightforward coset enumeration program creates 4 layers and more than 200 vertices before the coset graph collapses to its final 11 elements .
conway presented the calculation as a problem in @xcite .
the definition was immediately generalized to give the group @xmath107 .
coset enumeration showed that @xmath107 is finite for @xmath252 and for @xmath253 .
the cayley graph for group @xmath254 can be constructed systematically and recognized as a @xmath34-dimensional infinite euclidean group .
roger lyndon proved , using small cancellation theory , that @xmath107 is infinite if @xmath255 ( unpublished ) .
a. m. brunner @xcite proved that @xmath256 and @xmath257 are infinite .
george havas , j. s. richardson , and leon s. sterling showed that @xmath258 has a quotient of order @xmath259 , and , finally , m. f. newman @xcite proved that @xmath258 is infinite .
derek f. holt later reported a proof by computer that @xmath258 is automatic , from which it could be seen directly from the word - acceptor that the generators have infinite order .
at the international congress in helsinki ( 1978 ) , bill thurston was advertising the problem ( eventually solved by misha gromov ) of proving that a group of polynomial growth has a nilpotent subgroup of finite index .
the first - named author brought up the example of @xmath254 as such a group .
thurston immediately recognized the group as a branched cyclic cover of @xmath0 , branched over the figure - eight knot . and
before our dinner of reindeer steaks was over , thurston had conjectured that the even - numbered fibonacci groups were probably also branched cyclic covers of @xmath0 , branched over the figure - eight knot .
this conjecture was verified by h. m. hilden , m. t. lozano , and j. m. montesinos - amilibia @xcite and by h. helling , a. c. kim , and j. l. mennicke @xcite .
c. maclachlan @xcite proved that , for odd @xmath2 , the group @xmath107 is not a fundamental group of a hyperbolic @xmath34-orbifold of finite volume .
the sieradski manifolds have a similar rich history , but not one we know as well . as examples
, they were introduced by a. sieradski in 1986 @xcite .
sieradski used the same faceted @xmath34-ball that we employ , though his face pairings were different .
richard m. thomas @xcite shows that the sieradski groups , which he calls @xmath260 , are infinite if and only if @xmath261 and that @xmath262 is metabelian .
a. cavicchioli , f. hegenbarth , and a. c. kim @xcite show that the sieradski manifolds are branched over the trefoil knot .
cyclic presentations are particularly interesting because of their connections with branched cyclic coverings of @xmath34-manifolds
. fundamental results about cyclic presentations appear in the book _ presentations of groups _ by d. l. johnson ( * ? ? ?
* chapter 16 ) .
arye juhsz @xcite considers the question of when cyclically presented groups are finite .
andrzej szczepaski and andrei vesnin @xcite ask which cyclically presented groups can be groups of hyperbolic @xmath34-orbifolds of finite volume and which can not .
alberto cavicchioli and fulvia spaggiari @xcite show that non - isomorphic cyclically presented groups can have the same polynomial .
m. j. dunwoody @xcite managed to enumerate , with parameters , a large class of @xmath34-manifolds admitting heegaard splittings with cyclic symmetry .
the fundamental groups were all cyclically presented .
he observed that the polynomials associated with the cyclic presentations were alexander polynomials of knots and asked whether the spaces were in fact branched cyclic covers of @xmath0 , branched over knots or links .
alberto cavicchioli , friedrich hegenbarth , and ann chi kim @xcite showed that the dunwoody manifolds included branched covers with singularities that were torus knots of specific type .
l. grasselli and m. mulazzani @xcite showed that dunwoody manifolds are cyclic coverings of lens spaces branched over @xmath263-knots .
cavicchiolo , beatrice ruini , and fulvia spaggiari @xcite proved the dunwoody conjecture that the dunwoody manifolds are @xmath2-fold cyclic coverings branched over knots or links .
soo hwan kim and yangkok kim @xcite determined the dunwoody parameters explicity for a family of cyclically presented groups that are the @xmath2-fold cyclic coverings branched over certain torus knots and certain two - bridge knots .
nurullah ankaralioglu and huseyin aydin @xcite identified certain of the dunwoody parameters with generalized sieradski groups .
the first general presentation about the branched cyclic coverings of the two - bridge knots seems to be that of jerome minkus in 1982 @xcite .
a very nice presentation appears in @xcite where cyclic presentations are developed that correspond to cyclically symmetric heegaard decompositions .
the authors alberto cavicchioli , beatrice ruini , and fulvia spaggiari show that the polynomial of the presentation is the alexander polynomial .
they use the very clever and efficient rr descriptions of the heegaard decompositions .
they pass from the heegaard decompositions to face - pairings and determine many of the geometric structures .
michele mulazzani and andrei vesnin @xcite exhibit the many ways cyclic branched coverings can be viewed : polyhedral , heegaard , dehn surgery , colored graph constructions .
significant progress has been made beyond the two - bridge knots . c. maclachlan and a. reid @xcite and a. yu . vesnin and a. ch .
kim @xcite consider @xmath4-fold branched covers over certain @xmath34-braids .
alexander mednykh and andrei vesnin @xcite consider @xmath4-fold branched covers over turk s head links .
alessia cattabriga and michele mulazzani @xcite develop strongly - cyclic branched coverings with cyclic presentations over the class of @xmath263 knots , which includes all of the two - bridge knots as well as many knots in lens spaces .
cristofori , m. mulazzani , and a. vesnin @xcite describe strongly - cyclic branched coverings of knots via @xmath264-decompositions .
every knot admits such a description .
h. m. hilden , m. t. lozano , and j. m. montesinos - amilibia .
the arithmeticity of the figure eight knot orbifolds . in _
topology 90 ( columbus , oh , 1990 ) _ , ohio state univ .
, 1 , pages 133167 , berlin , 1992 . de gruyter . c. maclachlan .
generalisations of fibonacci numbers , groups and manifolds . in _
combinatorial and geometric group theory ( edinburgh , 1993 ) _ , number 204 in london math .
lecture note ser .
, pages 233238 , cambridge , 1995 .
cambridge univ . press .
john milnor . on the @xmath34-dimensional brieskorn manifolds @xmath268 . in _
knots , groups , and @xmath34-manifolds ( papers dedicated to the memory of r. h. fox ) _ , ann . of math . studies .
princeton univ . press , 1975 . | we give uniform , explicit , and simple face - pairing descriptions of all the branched cyclic covers of the 3sphere , branched over two - bridge knots .
our method is to use the bi - twisted face - pairing constructions of cannon , floyd , and parry ; these examples show that the bi - twist construction is often efficient and natural .
finally , we give applications to computations of fundamental groups and homology of these branched cyclic covers . |
quantum system restricted on a bounded domain has become more relevant for theoretical physics . there , the role of boundary conditions are very important not only for the long distance ( infrared ) regime but also for the short distance ( ultraviolet ) regime .
mathematically , the correct framework to treat the boundary conditions in quantum theory is by means of the analysis of von neumann s self - adjoint extension of the hamiltonian operator @xcite . physically speaking
, the variety of boundary conditions provided by the self - adjoint extension of the hamiltonian implies that the very rich structure of point interactions available in quantum theory .
the analysis of self - adjoint extension of the hamiltonian , as the name suggests , is essentially based on the hamiltonian operator approach .
however , in the feynman s path - integral approach , we do not know _ a priori _ how to incorporate the boundary conditions into the integration measure nor into the path - integral weight . as discussed in many textbooks
( see for example @xcite ) , the naive path - integral representation for a system on a bounded domain leads to a wrong boundary behavior and hence requires modification .
the most rigorous way to incorporate the boundary conditions into the feynman kernel is to evaluate it by the operator formalism .
however , the kernel evaluated by the operator formalism becomes the summation over the energy spectrum . in order to switch to the path - integral description
, we have to perform resummation of the energy spectrum to the paths of the space . in general
, this resummation is accomplished by _
trace formulae_. trace formulae provide a direct connection between quantum energy spectrum and classical length spectrum ( periodic orbits ) . however , this connection is in general an asymptotic relation valid for large wave numbers , just as in the case of the gutzwiller trace formula @xcite .
there are only few cases where the trace formulae become identities .
noteworthy among these are the poisson summation formula and the selberg trace formula @xcite , the former is the trace formula for the laplace operator on flat tori and the latter on riemannian manifolds with constant negative curvature .
although in the operator formalism point interactions have been extensively discussed in the literature , the path - integral description of point interactions has not been fully understood yet . in mathematically speaking ,
this is mainly due to the lack of trace formulae suitable for the point interactions . in physically speaking , on the other hand ,
this is mainly due to the lack of our knowledge about the classical trajectories for a particle in the presence of point interactions .
the aim of this paper is to fulfill a gap of the description for boundary conditions between the operator formalism and the path - integral formalism : we would like to propose a physically transparent prescription how to incorporate the boundary conditions obtained in the operator formalism into the path - integral description .
to illustrate our idea in a simple setting , in this paper we will consider a one - particle quantum mechanics on a circle in the presence of a single point interaction . to begin with ,
let us first consider a quantum particle on a circle of circumference @xmath1 in the presence of a @xmath2-interaction described by the hamiltonian @xmath3 , where @xmath4 is the dimensionless coupling constant and prime ( @xmath5 ) indicates the derivative with respect to @xmath6 .
( here as in the following we are using units where @xmath7 . ) it is known that the @xmath2-interaction belongs to the so - called scale - independent subfamily of point interactions @xcite and is verified by the boundary conditions @xmath8\psi(0)$ ] and @xmath9\psi^{\prime}(0)$ ] @xcite , where @xmath10 is the square integrable wave function on an interval @xmath11 . although the feynman kernel of this system has been analyzed in the literature @xcite , the physical interpretation for the weight factors ( see below ) remains open .
we would like to first address this issue .
as ubiquitous in the scale - independent point interactions , in this @xmath2-interaction case the wave numbers are quantized in an integer step so that it is easy to rewrite the feynman kernel @xmath12 evaluated in the operator formalism into the path - integral representation with the help of the poisson summation formula .
the resultant kernel takes the form @xmath13 where @xmath14<\pi$ ] and @xmath15 ( @xmath16 ) sign for @xmath17 ( @xmath18 ) .
@xmath19 is the principal value of the inverse cosine .
notice that the presence of a point interaction breaks the global translational invariance . as a consequence
the kernel is the sum of partial amplitudes for the translational invariant and variant classes of trajectories , which are weighted by the factors @xmath20 and @xmath21 respectively . before going to discuss the physical meaning of the weight factors
, we have to reveal the particle propagations described by . to this end
, it should first be noted that @xmath22 leads to @xmath23 so that eq . becomes the well - known form of the one - particle feynman kernel on a circle with periodic boundary conditions . as discussed in many textbooks
( see for example ref.@xcite ) , in this @xmath22 case the kernel is the sum of partial amplitudes for transitions via classical paths distinguished by the homotopy class of @xmath24 , i.e. , the winding number . for nonzero @xmath25 , however , the classical trajectories of a particle are not so trivial due to the presence of @xmath2-potential , which acts as a point scatterer .
when a particle reaches to the position of the point scatterer , there must be in general two possibilities : reflection or transmission .
thus the paths for a particle interacting @xmath0-times to the point interaction must consist of @xmath26 distinct paths . as an example the classical world lines for @xmath27 and @xmath28 in are depicted in figure [ fig : world_lines ] . [ cols="^,^,^,^ " , ] we emphasize that the eigenvalues of the s - matrix depend only on the three parameters @xmath29 , @xmath30 and @xmath31 , whereas the eigenvectors depend on the full parameters of @xmath32 .
the reason will be explained as follows : since the eigenvalues of the s - matrix on @xmath33 correspond to the energy spectrum of @xmath34 , we here explain why the energy spectrum of @xmath34 does not depend on @xmath35 and @xmath36 .
we first point out that the parity operator @xmath37 is well - defined on @xmath34 .
let us then consider the following ( singular ) unitary transformation : @xmath38 where @xmath39 which acts on the two - component vectors as follows : @xmath40 since the unitary transformation leaves the hamiltonian invariant , the energy spectrum remains the same but the boundary condition is changed as @xmath41 where @xmath42 thus , we found that the unitary transformation has an effect on @xmath43 by a rotation of the angle @xmath44 , i.e. @xmath45 this implies that the energy spectrum should depend on not @xmath43 but their invariant @xmath46 , which is identical to @xmath47 .
thus the spectrum depends only on the three parameters @xmath48 .
the main success of this work is the systematic description for a one - particle feynman kernel on a circle with a point interaction .
the point is that we do not need any knowledge of the spectrum nor the complete set of energy eigenfunctions of the system ( except for the bound states ) .
what we have to know is the classical trajectories of a particle and the one - particle scattering matrix .
we are left with a number of questions , however .
let us close with a few comments on these issues : 1 .
_ more rigorous foundation of partial amplitude unitarity_. + we showed that the particle propagation scattered @xmath0-times from the point interaction should be weighted by the elements of @xmath49 .
as a direct consequence of the unitarity of the s - matrix , these weight factors satisfies the relation @xmath50 , which we proposed to call _ partial amplitude unitarity_. as briefly discussed in section [ sec : examples ] , in the case of reflectionless subfamily of point interactions our partial amplitude unitarity and the laidlaw - dewitt theorem seem to be the same thing .
however , the theorem is essentially based on the homotopy theory so that it could not be applied in general to the other subfamily of point interactions .
we thought that the partial amplitude unitarity would provide a wider notion than the laidraw - dewitt theorem and should be derived from some fundamental properties of the feynman kernel such as the unitarity @xmath51 .
+ another related issue is an algebraic structure for the construction of classical trajectories for a particle on @xmath34 . as mentioned before , the classical trajectories for a particle interacting @xmath0-times to the point interaction consist of @xmath26 distinct paths .
these trajectories are constructed from the more fundamental trajectories depicted in figure [ fig : s - matrix ] by gluing them under the multiplication rule of the matrix @xmath52 .
this fact implies that , in spite of the presence of a point singularity , it might be possible to introduce some kind of notion analogous to the fundamental group to the space @xmath34 .
however , we have no idea to treat these problems
path - integral representation for the kernels_. + the feynman kernels derived in this paper have the forms which will be obtained after performing the path - integration .
it is very interesting to investigate its path - integral representation .
as mentioned in the introduction , boundary conditions are treated unambiguously in operator formalism by von neumann s theory of self - adjoint extension . however , in path - integral formalism we do not know _ a priori _ what kind of trajectories we should integrate over and what kind of `` classical action '' we should adopt as a weight . indeed , even in the system of a free particle in an
infinitely deep well potential , the `` classical action '' we should adopt as a path - integral weight includes the so - called _ topological term _ that localizes at the boundaries @xcite . furthermore , this topological term is proportional to the planck constant so that it is no longer classical . +
to that aim we have to consider the limit @xmath53 of the equation : @xmath54 \langle x|\mathrm{e}^{-ih\delta t}|x_{n-1}\rangle \cdots \langle x_{1}|\mathrm{e}^{-ih\delta t}|x_{0}\rangle,\end{aligned}\ ] ] where @xmath55 .
we already know the exact form of each partial amplitude @xmath56 . in order to evaluate the right hand side we have to tackle with the @xmath57 products of both translational invariant and variant classes of infinite sums .
we would like to report this issue elsewhere .
s.o . would like to thank y. adachi and k. sakamoto for valuable conversations and the latter also for the collaborations on several issues relevant in this paper .
m.s . would like to thank k. takenaga and t. nagasawa for useful discussions .
this work is supported in part by the grant - in - aid for scientific research ( no.18540275 ( m.s . ) ) by the japanese ministry of education , science , sports and culture .
15 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , _ _ ( ) . , _ _ ( ) . , _ _ ( ) . , _ _ , * * ( ) . , _ _ , * * ( ) .
, , _ _ , * * ( ) . , _ _ , * * ( ) . , , _ _ , * * ( ) . , _ _ , * * ( ) , http://arxiv.org/abs/quant-ph/9910062[arxiv:quant-ph/9910062 ] . , , _
_ , * * ( ) , http://arxiv.org/abs/quant-ph/0307002[arxiv:quant-ph/0307002 ] .
, , _ _ , * * ( ) , http://arxiv.org/abs/quant-ph/0105066[arxiv:quant-ph/0105066 ] . , _
_ , * * ( ) , http://arxiv.org/abs/cond-mat/0501110[arxiv:cond-mat/0501110 ] . , _
_ , * * ( ) .
, , _ _ , * * ( ) , http://arxiv.org/abs/0712.4353[arxiv:0712.4353[quant-ph ] ] . , _
_ , * * ( ) .
, , _ _ , * * ( ) , http://arxiv.org/abs/quant-ph/0111057[arxiv:quant-ph/0111057 ] . | we study a particle propagation on a circle in the presence of a point interaction .
we show that the one - particle feynman kernel can be written into the sum of reflected and transmitted trajectories which are weighted by the elements of the @xmath0-th power of the scattering matrix evaluated on a line with a point interaction . as a by - product we find three - parameter family of trace formulae as a generalization of the poisson summation formula . |
the internet and the world wide web ( www ) , the electricity power grid , networks of flight connections , of social contacts , and neuronal networks of the brain are few of the many examples of networks that surround us and that may be usefully described as graphs @xcite .
graph theory is rooted in the 18th century , beginning with the work of euler .
early efforts focused on properties of special ( and usually small ) graphs . in the 1960s ,
paul erds and alfrd rnyi @xcite initiated the study of _ random graphs _ , known also as er graphs .
the unlimited size and randomness of er graphs made them natural contenders for models of large networks encountered in everyday life . in 1967
the psychologist stanley milgram asked himself how many acquaintances , on average , connect between a person on the east coast and another on the west coast of the usa .
following some research he finally concluded that the number of acquaintances is surprisingly small , just about six @xcite .
this celebrated `` six degrees of separation '' achieved widespread popularity .
it was later realized that an average short path , typically of order @xmath0 ( @xmath1 being the number of nodes , or the _ size _ of the network ) , connects between randomly selected nodes in most naturally occurring networks .
recognizing milgram s contribution the effect came to be known as the _ small world _
property of networks .
er graphs possess the small world property .
however , an important ingredient was still missing . in networks of social contact
the people that are connected to a certain individual ( such as family , acquaintances at work , and friends ) are highly likely to be connected among themselves .
this high degree of _ clustering _ is absent in er graphs .
_ small world networks
_ , introduced by watts and strogatz @xcite , exhibit high degrees of clustering as well as the small world property .
they mark an important stage in the modeling of everyday networks and are studied in their own right .
it was recently realized that in addition to a high clustering index and the small world property real life networks exhibit yet a third important characteristic : a _ scale - free _ degree distribution .
the _ degree _ of node @xmath2 is the number of links , @xmath3 , connected to the node .
the likely degree of the various nodes in most observed networks follows a power - law distribution : @xmath4 where @xmath5 is an appropriate normalization factor , @xmath6 is the minimal degree of any given node , and the cutoff degree @xmath7 depends on the size of the network , @xmath8 @xcite .
the term _ scale - free _ refers to the fact that the moments @xmath9 for @xmath10 do not exist(diverge ) , in the limit of @xmath11 .
scale - free networks possess the small world property , and it is possible to construct them with a high degree of clustering .
many important examples follow this pattern , including the internet @xcite and the www ( both for incoming and outgoing links ) @xcite , social networks , and virtually any large network arising in some natural context .
see @xcite , for excellent reviews , and @xcite , for a timely anthology .
an oft - neglected aspect in the modeling of everyday - life networks is the fact that they are embedded in physical ( euclidean ) space and possess a _ geography _
, in addition to their _
topology_. the spatial location of nodes and the length of the connecting links is never a consideration in the models of networks discussed above .
nevertheless , routers of the internet and social networks lay on the two - dimensional surface of the globe ; neuronal networks in brains occupy three - dimensional space , etc . the likelihood of connections between the nodes in such networks is certainly affected by their geographical proximity , and one expects nontrivial consequences arising from this interplay between geography and topology @xcite
yet , most studies so far have focused on topological networks , where the nodes and links exist in some abstract space , devoid of metric . in this paper
we consider the embedding of networks in euclidean space . in section [ general ] ,
we discuss general aspects of embedding and introduce a specific algorithm for the embedding of random networks of arbitrary degree distribution .
this algorithm favors connections to nearest nodes and economizes on the total length of the links .
we show that infinite networks can be thus embedded if and only if the degree ditribution has compact support , that is , provided that there exists a sharp upper cutoff @xmath12 for the degree of any given node .
if the distribution has no compact support , then embedding introduces an artificial cutoff and is only possible in a restricted sense . in section
[ scale - free ] we apply the embedding algorithm to the widespread case of scale - free networks and study key structural properties of the resulting lattices .
we conclude with a discussion of open work and a comparison to an interesting related algorithm , proposed by warren , sander , and sokolov @xcite , in section [ discussion ] .
our general problem is that of embedding a given infinite graph in euclidean space .
suppose first that the graph contains no cycles ( loops ) , i.e. , it is a tree .
let the volume of a node be @xmath13 . the total volume of nodes within chemical shell @xmath14 ( nodes up to @xmath14 links away from a given node ) increases as @xmath15 , where @xmath16 is the average degree of the nodes . on the other hand , the volume enclosed within a euclidean distance @xmath17 scales as @xmath18 ( in @xmath19-dimensions ) .
unless the links become progressively long , the exponential growth in @xmath14 can not be sustained within the much slower algebraic growth in @xmath17 .
if the network is to be statistically homogeneous , a lengthening of successive links is unacceptable .
we conclude that infinite trees , such as the cayley tree ( fig [ cayley - hex]a),are not embeddable in euclidean space @xcite . allowing for loops ,
the volume constraint is reduced to the extent that embedding might be possible while retaining homogeneity , and without a change in the degree distribution .
an example is shown in fig [ cayley - hex]b . ) without loops .
it can not be embedded in euclidean space , unless one allows for loops : the hexagonal lattice ( b ) is one example of embedding achieved in this way .
notice that the degree distribution remains unchanged.,title="fig:",scaledwidth=50.0% ] ) without loops .
it can not be embedded in euclidean space , unless one allows for loops : the hexagonal lattice ( b ) is one example of embedding achieved in this way .
notice that the degree distribution remains unchanged.,title="fig:",scaledwidth=50.0% ] in order to make further headway we restrict ourselves to _ random _ graphs , of arbitrary degree distribution @xmath20 , and the following embedding algorithm @xcite .
to each site @xmath2 of a @xmath19-dimensional lattice assign a random connectivity @xmath3 , taken from the degree distribution @xmath20 .
select site @xmath21 at random and connect it to its closest neighbors , until its preassigned connectivity @xmath22 is realized , or until all sites within distance @xmath23 have been explored . a link to site @xmath14
is allowed provided that : ( a ) the site is not saturated ( its connectivity has not yet reached the preassigned @xmath24 ) , and ( b ) its distance from site @xmath21 is smaller than @xmath25 .
repeat for all sites .
this algorithm makes sense in the context of social networks , where connections are typically confined to one s immediate neighborhood .
the implied economy in the physical length of links might render the algorithm useful for the modeling of other networks of interest .
suppose that one attempts to embed an infinite network , of degree distribution @xmath20 , in an infinite lattice , by the above algorithm .
nodes with a connectivity larger than a certain cutoff @xmath26 can not be realized , because of the possible saturation of surrounding sites .
consider the number of links @xmath27 entering a generic node from a surrounding neighborhood of radius @xmath17 .
sites at distance @xmath28 are linked to the node with probability @xmath29 thus , @xmath30 where @xmath31 is the surface area of the @xmath19-dimensional unit sphere . reversing the order of integration and carrying out the spatial integral
, we obtain @xmath32 where @xmath33 is the volume of the @xmath19-dimensional unit sphere . finally ,
taking the limit @xmath34 , we derive the cutoff @xmath35 we distinguish between two cases for the degree distribution @xmath20 : with and without compact support .
the distribution @xmath20 has compact support if there exists a @xmath12 such that @xmath36 for all @xmath37 .
all finite networks have a distribution with compact support , however infinite networks might too have compact support ( an example is the cayley tree of fig .
[ cayley - hex]a ) . if this is the case , then @xmath38 is finite , and one can always select @xmath39 large enough so that @xmath40 . in other words , _ if there is compact support the network is embeddable_. if the network is infinite and without compact support , then if @xmath41 , for any finite @xmath39 the cutoff @xmath42 is finite and the tail of the distribution for @xmath43 is chopped off .
( infinite networks with diverging @xmath44 are pathological we are not aware of any important practical application though they might be embeddable . )
the cutoff connectivity @xmath42 implies a cutoff length @xmath45 the embedded network displays the original ( chopped ) distribution up to lengthscale @xmath46 and repeats , statistically , at lengthscales larger than @xmath46 .
indeed , @xmath46 is analogous to the correlation length in percolation theory , above criticality , where the infinite cluster is _ fractal _ for @xmath47 and is homogeneous for @xmath48 @xcite .
we emphasize that the cutoff @xmath42 is of consequence even if the distribution @xmath20 is narrow , that is , even if @xmath49 .
indeed , even in such a case the lengthscale @xmath46 is finite ( and controlled mainly by @xmath39 ) .
in summary , finite networks ( all practical situations ) are always embeddable by ourproposed algorithm .
infinite networks are strictly embeddable only if their degree distribution has compact support .
otherwise , they are embeddable in a restricted manner , with a cutoff @xmath42 imposed by the embedding , and statistical repetition at lengthscales @xmath50 .
we now apply the embedding algorithm to the widespread case of scale - free networks , eq .
( [ pscale - free ] ) .
generally , we consider embedding in @xmath19-dimensional lattices of size @xmath51 , though in the numerical simulations shown below we limit ourselves to two - dimensional square lattices with periodic boundary conditions . because the lattice has a finite number of sites , @xmath52 , the degree distribution ( [ pscale - free ] ) has compact support @xcite : @xmath53 this , in conjunction with ( [ r - k ] ) , implies a natural cutoff length @xmath54 the interplay between the three lengthscales , @xmath51 , @xmath46 , @xmath55 , determines the nature of the network .
the embedding cutoff @xmath42 is imposed only if @xmath56 , and there is statistical repetition for @xmath48 .
otherwise ( @xmath57 ) , the natural cutoff of @xmath7 is attained .
as long as @xmath58 , the finite size of the lattice imposes no serious restrictions .
otherwise ( @xmath59 ) , finite - size effects become important . in short , there exist @xmath60 different regimes , characterized by the relative ordering of @xmath51 , @xmath46 , and @xmath55 : regime a : @xmath61 .
natural cutoff @xmath7 is attained , and no finite - size effects .
statistical repetition occurs at lengthscales @xmath62 , rather than @xmath46 .
regime b : @xmath63 .
same as regime a , but statistical repetition occurs at lengthscales @xmath48 .
regime c : @xmath64 .
cutoff @xmath42 is imposed ; nofinite - size effects .
regime d : @xmath65 .
same as regime c. regime e : @xmath66 .
cutoff @xmath42 is imposed ; strong finite - size effects prevent statistical repetition .
regime f : @xmath67 .
natural cutoff @xmath7 is attained ; strong finite - size effects .
the various regimes are demarcated by the lines @xmath68 as shown in fig .
[ regimes ] . and
@xmath69 , embedded in @xmath70-sites square lattices .
( b ) chemical shells , for the same parameters as in ( a ) , but for lattices of size @xmath71 .
notice the statistical repetition at lengthscales greater than@xmath46 , for @xmath72 . ] in fig .
[ pinta]a , we present typical networks that result from our embedding method , for @xmath72 and @xmath73 .
long range links become more noticeable as @xmath74 decreases . in fig .
[ pinta]b , we show the same networks as in fig .
[ pinta]a , where successive chemical shells are shaded differently .
chemical shell @xmath14 consists of all sites that can be reached by a minimal number of @xmath14 connecting links relative to a given site ( the central site , in the figure ) . for our choice of parameters
, @xmath73 falls in the region of @xmath75 , while for @xmath72 , @xmath56 . in the latter case
, we clearly see the statistical repetition beyond the lengthscale @xmath46 ( fig .
[ pinta]b , @xmath72 ) .
the degree distribution of the embedded networks is illustrated in fig .
[ distrib ] . in fig .
[ distrib]a , various values of @xmath39 were selected such that @xmath56 , and the distribution terminates at the imposed cutoff @xmath42 .
the scale - free distribution is altered slightly , for @xmath76 , due to saturation effects , but the overall trend is highly consistent with the original power - law .
the data - collapse shown in the inset confirms the power - law , as well as the scaling of the cutoff @xmath77 . in fig .
[ distrib]b , @xmath78 was fixed and values of @xmath74 were selected such that @xmath75 .
the natural cutoff @xmath7 is now attained , and the data collapse at the inset confirms the power - law distribution as well as the known relation @xmath79 .
we now turn to the relation between the euclidean metric and chemical length in our embedded scale - free networks .
the chemical distance @xmath14 between two sites is the minimal number of links @xmath14 connecting the two .
thus , if the euclidean distance between the sites is @xmath17 , then @xmath80 defines the minimal ( chemical ) length exponent @xmath81 . in order to compute @xmath81
we regard the chemical shells as being roughly smooth , at least in the regime @xmath75 , as suggested by fig .
[ pinta]b(@xmath73 ) .
let the width of shell @xmath14 be @xmath82 , then @xmath83 since @xmath84 .
the number of sites in shell @xmath14 , @xmath85 , is , on the one hand , @xmath86 . on the other hand , since the maximal connectivity in shell @xmath14 is @xmath87 , the thickness of shell @xmath88 is @xmath89 . assuming ( for large @xmath14 ) that @xmath90 , we obtain @xmath91}\;.\ ] ] using this expression in ( [ lint ] ) yields @xmath92 thus , for @xmath93 , the dimension @xmath94 .
this result is opposite to all known disordered media , where @xmath95 .
( a particularly simple example is provided by a polymer chain , or its self - avoiding - walk model , where the end - to - end distance is well - approximated by the flory relation : @xmath96 , i.e.,@xmath97 . )
we have also computed the fractal dimension of the newtorks and found that the interior of the @xmath14-clusters is compact , with @xmath98 @xcite .
[ pinta ] suggests that the hull of the clusters is fractal . )
this resultwas confirmed by simulations .
it then follows that the fractal dimension of the network in chemical space is also anomalous : @xmath99 ( for @xmath93 ) . in fig .
[ d_min]a , we plot @xmath81 as measured from simulations , and compared with the analytical result of eq .
( [ dmin ] ) .
the scaling suggested in fig .
[ d_min]b , @xmath100 , is valid only for @xmath75 . for @xmath101
, we expect that the network is scale - free up to lengthscale @xmath46 and the analogous scaling would be @xmath102 , where @xmath103 .
it remains an open challenge to conceive of a general scaling relation that would encompass all of the regimes a f.
in summary , we propose an algorithm for the embedding of networks in euclidean space .
finite networks can always be embedded in this way .
infinite networks are strictly embeddable only if the degree distribution has compact support . if not , an artificial cutoff @xmath42 results from our embedding technique , and the networkrepeats itself , in a statistical sense , at lengthscales greater than @xmath104 .
we have applied our embedding algorithm to scale - free graphs , and studied the resulting networks . concurrently with our work ( and independently ) , warren , sander and sokolov @xcite proposed an embedding algorithm very similar to ours .
the main difference in their approach is that connecting to saturated sites is allowed , and a node is connected to as many of its closest neighbors as necessary , until its target connectivity is fulfilled . in this case , our computation of @xmath42 in section [ general ] is still valid , only that now it is a _ lower _
cutoff for the connectivity of the sites . on the other hand ,
there is no restriction , in this version of embedding , on the upper cutoff of the distribution .
we anticipate that @xmath46 here plays the role of a ` geometrical ' lengthscale , similar to the lattice spacing for a critical percolation cluster grown on a lattice .
( the cluster is self - similar only at lengthscales larger than the lattice spacing . )
embedding scale - free networks in euclidean space results in some dramatic changes from the original graphs .
consider , for example , the small world property , common to off - lattice scale - free networks grown by any of the known techniques .
the small world property disappears at lengthscales greater than @xmath46 : two sites separated by a physical distance @xmath105 , @xmath106 , would be connected by at least @xmath107 intervening links , due to the statistical repetition .
the number of sites withinradius @xmath17 is @xmath108 , so @xmath109 .
not surprisingly , this is the same scaling as for euclidean @xmath19-dimensional lattices . for distances @xmath47 , wehave @xmath110 , which represents small - world behavior , since it improves on the euclidean @xmath109 , but a far cry from the off - lattice @xmath111 .
another example of a dramatic change is the qualitative difference between percolation inoff - lattice and embedded networks . indeed , warren et al .
, @xcite find that the percolation transition takes place in embedded lattices even for @xmath112 , in contrast to off - lattice scale - free networks @xcite . in view of the fact that many everyday - life networks are embedded in physical space ,
we ought to reconsider how the topological properties of scale - free graphs properties that we normally attribute to the physical networks are affected by the embedding .
we are grateful to nsf grant phy-0140094 ( dba ) for partial support of this research .
b. bollobs , _ random graphs _ ( academic press , london , 1985 ) .
m. molloy , b. reed : random structures and algorithms * 6 * , 161 ( 1995 ) .
r. albert and a .-
barabsi , _ rev . of mod .
phys . _ * 74 * , 47 ( 2002 ) .
s. n. dorogovtsev and j. f. f. mendes , _ adv . in phys .
_ , * 51 * ( 4 ) , ( 2002 ) . s. bornholdt and h.g .
schuster , eds .
, _ handbook of graphs and networks _ , ( wiley - vch , berlin , 2003 ) .
p. erds and a. rnyi,_publicationes mathematicae _ * 6 * , 290 ( 1959 ) .
p. erds and a. and rnyi,_publications of the mathematical institute of the hungarian academy of sciences _ * 5 * , 17 ( 1960 ) .
p. erds and a. rnyi_acta mathematica scientia hungary _ * 12*,261 ( 1961 ) .
s. milgram , _ psychology today _ * 2 * , 60 ( 1967 ) .
d. j. watts , _ small worlds_.(princeton university press , princeton , 1999 ) .
d. j. watts and s. h. strogatz,_nature _ * 393 * , 440 ( 1998 ) .
r. cohen , k. erez , d. ben - avraham , and s. havlin , _ phys .
lett . _ * 85 * , 4626 ( 2000 ) .
s. n. dorogovtsev , and j. f. f. mendes , _ phys .
e _ * 63 * , 062101 ( 2001 ) .
a. l. barabsi and r. albert , _ science _ ,
* 286 * , 509 ( 1999 ) .
m. faloutsos , p. faloutsos , and c. faloutsos , _ comput .
_ * 29 * , 251 ( 1999 ) .
a. -l .
barabsi , r. albert , and h. jeong , _ physica a _ , * 281 * , 2115 ( 2000 ) .
a. broder , r. kumar , f. maghoul , p. raghavan , s. rajagopalan , r. stata , a. tomkins and j. wiener , _ comput .
netw . _ * 33 * , 309 ( 2000 ) . s .-
yook , h. jeong , a .-
barabsi , _ proc .
sci . _ * 99 * , 13382 ( 2002 ) . c. p. warren , l. m. sander and i. m. sokolov , _ phys .
e _ * 66 * , 56105 ( 2002 ) .
an amusing example is provided by starburst dendrimers ( sbd ) star - shaped polymers that grow in the patern of the cayley tree offig .
[ cayley - hex]a .
the sbds typically can not grow beyond chemical shell@xmath113 , due to lack of space .
see , for example , d. ben - avraham , l. s. schulman , s. h. bossmann , c. turro , and n. j. turro,_j .
b _ * 102 * , 5088 ( 1998 )
. a. f. rozenfed , r. cohen , d. ben - avraham and s. havlin,_phys .
. lett . _ * 89 * , 218701 ( 2002 ) .
d. stauffer and a. aharony , _ introduction to percolation theory _ , 2nd edition ( taylor and francis , london , 1991 ) .
a. bunde , and s. havlin ( editors ) , _ fractals and disordered system _ ( springer , new york , 1996 ) .
d. ben - avraham and s. havlin , _ diffusion and reactions in fractals and disordered systems _ ( cambridge university press , 2000 ) . | a method for embedding graphs in euclidean space is suggested .
the method connects nodes to their geographically closest neighbors and economizes on the total physical length of links .
the topological and geometrical properties of scale - free networks embedded by the suggested algorithm are studied both analytically and through simulations .
our findings indicate dramatic changes in the embedded networks , in comparison to their off - lattice counterparts , and call into question the applicability of off - lattice scale - free models to realistic , everyday - life networks . , , , , internet , scale - free , networks , embedding , lattice 89.75.hc , 05.50.+q , 89.75.da |
there has been enormous progress over the past decade in discovering galaxies and qsos at increasingly high redshifts .
we are now probing far enough back in time that the universe at these early epochs was fundamentally different from its predominantly ionized state today .
observations of @xmath24 qsos ( becker et al .
2001 , fan et al .
2002 ) show near - complete absorption of flux at wavelengths short - ward of lyman-@xmath3 ( gunn & peterson 1965 ) , indicating that the universe is optically thick to this line , and that the neutral fraction of hydrogen is much greater than at lower redshifts .
initial cosmic microwave background results from the _ wmap _ satellite indicated that the universe was completely neutral at redshifts of @xmath23 ( kogut et al .
there is an ongoing debate as to what reionized the universe at @xmath25 : is it agn or ionizing photons ( produced in hot , massive , short - lived stars ) escaping from star forming galaxies ?
along with the escape fraction for these ionizing photons , it is crucial to know the global star formation rate at this epoch .
here we detail our work in identifying star - forming galaxies within the first billion years of the big bang through deep imaging with the _ hubble space telescope ( hst)_. our selection method is based on the lyman break technique , and we have shown this is effective in isolating @xmath1 by obtaining spectroscopic confirmation with the 10-m keck telescopes .
our analysis of the _ hubble ultra deep field _ ( the deepest images ever obtained ) enables us to explore the faint end of the luminosity function ( the bulk of galaxies with low star formation rates ) .
our discovery of this @xmath0-drop galaxy population is used to infer the global star formation rate density at this epoch ( @xmath1 ) .
we consider the implications for reionization of the uv flux from these galaxies .
we also use infrared data from the _ spitzer space telescope _ to determine the spectral energy distributions ( seds ) of our @xmath0-drops , and constrain the previous star formation histories of these sources . throughout
we adopt the standard `` concordance '' cosmology of @xmath26 , @xmath27 , and use @xmath28 .
all magnitudes are on the @xmath29 system .
in order to select @xmath1 galaxies , we use the `` lyman break technique '' pioneered at @xmath30 using ground - based telescopes by steidel and co - workers ( steidel , pettini & hamilton 1995 ; steidel et al .
1996 ) and using _ hst _ by madau et al .
( 1996 ) . at @xmath31
, we can efficiently use only two filters , above ( @xmath2-band 9000 ) and below ( @xmath0-band 8000 ) the continuum break at lyman-@xmath3 line ( @xmath32 ) since the integrated optical depth of the lyman-@xmath3 forest is @xmath33 ( see figure [ fig : filters ] ) and there is essentially no flux at shorter wavelengths .
the key issue is to work at a sufficiently - high signal - to - noise ratio that lyman break `` drop - out '' galaxies can be safely identified through detection in a single redder band ( i.e. , @xmath2-band ) .
this approach has been demonstrated to be effective by the sdss collaboration in the detection of @xmath1 qsos using the @xmath0- and @xmath2-bands alone ( fan et al .
2002 ) , using wide - area but shallow ground - based imaging ( sensitive to bright but rare quasars ) .
our group has analysed deep imaging from the advanced camera for surveys ( acs ) on _ hst _ with the same sharp - sided sdss f775w ( @xmath0 ) and f850lp ( @xmath2 ) filters to locate the``@xmath6-drops '' which are candidate @xmath1 galaxies ( fainter but more numerous than the qsos ) . in figures [ fig : tracks]&[fig : izzcolmag ] we illustrate how a colour cut of @xmath34 can be effective in selecting sources with @xmath35 .
[ fig : tracks ] we have analysed the acs images from the _ hst _ treasury `` great observatory origins deep survey '' ( goods ; giavalisco & dickinson 2003 ) to discover this @xmath0-drop population of @xmath1 galaxies . in stanway , bunker & mcmahon ( 2003 ) we searched 150arcmin@xmath36 of the goods - south field ( the _ chandra _ deep field - south ) to select @xmath37 objects , of which six were probable @xmath38 galaxies brighter than @xmath39 .
subsequent papers by the goods team ( dickinson _ et al . _ 2004 ; giavalisco _ et al .
_ 2004 ) reproduced many of these @xmath0-drop candidates . to address potential cosmic variance issues
, we performed a similar analysis in the goods - north field , which yielded a consistent estimate of the surface density of @xmath17 star forming sources ( stanway et al .
. the effectiveness of the lyman - break technique has been demonstrated at lower redshifts through spectroscopic confirmation of hundreds of @xmath13-band dropouts at @xmath12 and @xmath40-band drop - outs at @xmath41 .
however , it is important to demonstrate that the @xmath0-drop selection is similarly isolating galaxies at @xmath1 , if we are to use the surface density of @xmath0-drops to draw global inferences about galaxies within the first billion years .
a known possible contaminant is the extremely red object ( ero ) population of evolved galaxies at @xmath42 which can produce large @xmath43 colours : deep near - infrared imaging ( e.g. , from the _ hst_/nicmos survey of the udf ) should identify eros , and we consider this possible contamination stanway , mcmahon & bunker ( 2005 ) .
low - mass galactic stars ( m / l / t dwarfs ) are another interloper population . we have obtained deep spectroscopy with keck / deimos of several of the brightest @xmath0-drops identified by stanway , bunker & mcmahon ( 2003 ) in goods - south ; in the case of sbm03#3 ( @xmath44 ) and , an @xmath0-drop in the chandra deep field south photometrically - selected from hst / acs images to lie at @xmath1 . for two of these objects
, we discovered single emission lines ( fig .
3 ) at 8245 & 8305 , consistent with lyman-@xmath3 emission from galaxies at @xmath45 ( sbm03#3 , bunker et al . 2003 ) and @xmath46 ( sbm03#1 , stanway et al .
the spectrally - resolved profiles of the emission lines are asymmetric ( as high-@xmath47 lyman-@xmath3 tends to be ) with a p - cygni - like profile and a sharp cut - off on the blue wing ( fig . 2 ) .
the line fluxes are @xmath48 .
the equivalent widths are @xmath49 using the @xmath2-band photometry from hst / acs , which is within the range seen in high-@xmath47 star - forming galaxies at @xmath50 .
the velocity widths of the lyman-@xmath3 are comparatively narrow ( @xmath51 ) and we do not detect the high - ionization line n v@xmath5212140 , which strongly support the view that this line emission is powered by star formation rather than an agn .
our spectroscopic redshifts for these objects confirms the validity of the @xmath0-drop selection technique to select star - forming galaxies at @xmath1 .
subsequently , a programme of spectroscopy with gmos `` nod&shuffle '' on gemini ( glare , stanway et al .
2004b ) has confirmed more galaxies in goods - south .
slitless spectroscopy of the udf with _
hst_/acs ( grapes , malhotra et al .
2005 ) has also confirmed the spectral breaks of the @xmath0-drop galaxies , although the spectral resolution of @xmath53 is insufficient to see the lyman-@xmath3 emission lines .
[ fig : madau_plot ]
we base our measurement of the star formation rate on the intensity of the rest - frame uv continuum emission , redshifted into the @xmath2-band at @xmath1 ; this is dominated by the light from the most massive stars ( @xmath54 which are the hottest and bluest stars ) , which are the shortest lived , and hence a tracer of the instantaneous star formation rate . to go from the number of massive stars formed to the total star formation rate
requires an assumption about the stellar initial mass function ( imf ) . in the absence of dust obscuration ,
the relation between the flux density in the rest - uv around @xmath55 and the star formation rate ( @xmath56 in @xmath57 ) is given by @xmath58 from madau , pozzetti & dickinson ( 1998 ) for a salpeter ( 1955 ) imf with @xmath59 .
our relatively bright magnitude cut of @xmath4 for the goodsv0.5 data ( stanway , bunker & mcmahon 2003 ; stanway et al .
2004a ) corresponds to an unobscured star formation rate of @xmath60 at @xmath61 ( the luminosity - weighted average redshift ) , equivalent to @xmath62 for the @xmath13-band dropout population at @xmath12 .
we recover the @xmath1 rest - frame uv luminosity function from the observed number counts of @xmath0-drops to faint magnitudes in the udf .
although our colour cut selects galaxies with redshifts in the range @xmath63 , an increasing fraction of the @xmath2-band flux is attenuated by the redshifted lyman-@xmath3 forest . at higher redshifts
we probe increasingly shortward of @xmath64 ( where the luminosity function is calculated ) so the @xmath65-corrections become significant beyond @xmath66 .
we now compare the comoving star formation rate deduced for @xmath31 galaxies based on our candidate @xmath0-dropout source counts with predictions based on a range of rest - frame uv luminosity functions .
for convenience we assume that there is no evolution over the sampled redshift range , @xmath67 , spanned by our @xmath0-drop selection ( equivalent to a range between @xmath68gyr after the big bang ) .
we take as a starting point the luminosity function for the well - studied lyman - break @xmath13-dropout population , reported in steidel et al .
( 1999 ) , which has a characteristic rest - uv luminosity @xmath69 and a characteristic comoving number density at @xmath12 is @xmath70 in our cosmology .
the faint end slope of the schechter function at @xmath12 is relatively steep ( @xmath71 ) compared with @xmath72 to @xmath73 for lower - redshift galaxy samples ( e.g. , blanton et al .
2003 ) .
a tantalizing result from our work so far is that at @xmath74 there are far fewer uv - luminous star forming galaxies than would have been predicted if there was no evolution , based on a comparison to the well - studied @xmath75 lyman break population ( steidel et al.1999 ) .
in fact , the volume averaged ( comoving ) star formation rate in galaxies with @xmath76 is @xmath77 _ less _ at @xmath1 than at @xmath12 .
so it would appear that at this crucial epoch , where abundant sources of uv photons are needed to reionize the universe , there is in fact a deficit of luminous star forming galaxies .
we have conclusively shown that the bright end of luminosity function has evolved greatly from @xmath74 to @xmath30 .
other groups have claimed less dramatic evolution or even no evolution in the total volume - averaged star formation rate , based on the same fields ( giavalisco et al .
2004 ; dickinson et al .
2004 ) and similar _
hst_/acs data sets ( bouwens et al . 2003
; yan , windhorst & cohen 2003 ) , but these groups work closer to the detection limit of the images and introduce large completeness corrections for the faint source counts . to do a complete inventory of the uv light from star formation we must address the contribution of low - luminosity star - forming galaxies to the ionizing flux .
the public availability of the hubble ultra deep field ( udf ; beckwith , somerville & stiavelli ) can address this puzzle by pushing down the luminosity function at @xmath1 to well below the equivalent of @xmath78 for the @xmath12 population .
the hubble ultra deep field ( udf ) was a cycle 12 stsci director s discretionary time programme executed over september 2003 january 200 , comprising 400 orbits in 4 broad - band filters ( including f775w @xmath0-band for 144 orbits ; f850lp @xmath2-band for 144 orbits ) .
the udf lies within the chandra deep field south ( cdf - s ) .
as the udf represents the deepest set of images yet taken , significantly deeper than the @xmath79-band exposures of the hubble deep fields ( williams et al .
1996 ) , and adds the longer - wavelength @xmath2-band , it is uniquely suited to the goals of our program .
we performed the first analysis of @xmath0-drops in the udf ( bunker et al .
2004 ) , presenting details of 54 candidate star forming galaxies at @xmath1 in a preprint the day after the release of the udf data .
our analysis was subsequently independently repeated by yan & windhorst ( 2004 ) , with agreement at the 98% level ( see bunker & stanway 2004 for a comparison ) .
we take our magnitude limit as @xmath80 ( a @xmath81 cut).we measure a star formation density of @xmath82 at @xmath1 from galaxies in the udf with sfrs@xmath83 ( equivalent to @xmath11 at @xmath12 ) . at the relatively bright cut of @xmath4
used in stanway , bunker & mcmahon ( 2003 ) from the goodsv0.5 survey , the udf data is 98% complete for sources as extended as @xmath84arcsec .
interestingly , we detect no extended ( low surface brightness ) @xmath0-drops to this magnitude limit in addition to the targetted @xmath0-drop sbm03#1 in the deeper udf data .
this supports our assertion that the @xmath0-drop population is predominantly compact and there can not be a large completeness correction arising from extended objects ( c.f .
lanzetta et al . 2002 ) .
the acs imaging is of course picking out hii star forming regions , and these uv - bright knots of star formation are typically compact ( @xmath85kpc , @xmath86arcsec at @xmath1 ) even within large galaxies at low redshift .
it is interesting that the level of stellar contamination in the udf @xmath0-drops is only 2% , compared with about one in three at the bright end ( @xmath4 ; stanway , bunker & mcmahon 2003 ; stanway et al.2004 )
. this may be because we are seeing through the galactic disk at these faint limiting magnitudes to a regime where there are no stars ( see also pirzkal et al .
we compare our @xmath0-drop luminosity function with the work of madau , haardt & rees ( 1999 ) for the density of star formation . we have updated their equation 27 for the more recent concordance cosmology estimate of the baryon density of spergel et al .
( 2003 ) , and for the predicted mean redshift of our sample ( @xmath87 ) : @xmath88 this relation is based on the same salpeter initial mass function as we have used in deriving our volume - averaged star formation rate .
@xmath89 is the concentration factor of neutral hydrogen , @xmath90 .
simulations suggest @xmath91 ( gnedin & ostriker 1997 ) . the escape fraction of ionizing photons ( @xmath92 ) for high - redshift galaxies is highly uncertain ( e.g. , steidel , pettini & adelberger 2001 ) , but even if we take @xmath93 ( no absorption by h i ) this estimate of the star formation density required is a factor of @xmath94 higher than our measured star formation density of @xmath95 at @xmath1 from galaxies in the udf with sfrs@xmath83 . for faint end slopes of @xmath96 galaxies with @xmath97 account for @xmath98% of the total luminosity
, so would fall short of the required density of lyman continuum photons required to reionize the universe .
if the faint - end slope is as steep as @xmath99 then there would just be enough uv lyman continuum photons generated in star forming galaxies at @xmath1 ( assuming a salpeter imf ) , but the required escape fraction for complete reionization would still have to be implausibly high ( @xmath100 , whereas all high-@xmath47 measurements to date indicate @xmath101 : fernnadez - soto , lanzetta & chen 2003 ; steidel , adelberger & pettini 2001 ) .
agn are also under - abundant at these epochs ( e.g. , dijkestra , haiman & loeb 2004 ) .
if star forming galaxies at redshifts close to @xmath102 were responsible for the bulk of reionization , then a very different initial mass function would be required , or the calculations of the clumping factor of neutral gas would have to be significantly over - estimated ( see also stiavelli , fall & panagia 2004 ) .
alternatively another low - luminosity population ( e.g. , forming globular clusters ; ricotti 2002 ) could be invoked to provide some of the shortfall in ionizing photons .
it is also plausible that the bulk of reionization occurred at redshifts well beyond @xmath102 : the wmap polarization data indicate @xmath103 ( kogut et al . 2003 ) , and it is possible that the gunn - peterson troughs seen in @xmath104 qsos ( becker et al .
2001 ; fan et al .
2002 ) mark the very last period of a neutral igm .
we have recently discovered evidence of an early epoch of star formation ( eyles et al .
some of the spectroscopically - confirmed @xmath0-drops at @xmath1 show evidence of another spectral break in the near - infrared ( comparing _ spitzer _ images with irac at @xmath19 m with _ hst _ nicmos & acs images at @xmath105 m ) .
this age - sensitive balmer/4000break indicates the ir light is dominated by stars older than 100myr ( and most probably 400myr old ) .
this pushes the formation epoch for these galaxies to @xmath106 .
our recent results also indicate that there were at least some galaxies already assembled with stellar masses @xmath21 ( equivalent to @xmath22 today ) within the first billion years
. the early formation of such systems may have played a key role in reionizing the universe at @xmath23 . | we have determined the abundance of @xmath0-band drop - outs in the _
hst_/acs goods surveys and the hubble ultra deep field ( udf ) .
the majority of these sources are likely to be @xmath1 galaxies whose flux decrement between the f775w @xmath0-band and f850lp @xmath2-band arises from lyman - alpha absorption . we have shown with keck / deimos and gemini / gmos spectroscopy that this technique does indeed select high redshift galaxies , and we discovered lyman-@xmath3 emission in the expected redshift range for about a third of the galaxies with @xmath4 in the 150arcmin@xmath5 of the goods - south field .
the @xmath6-drop number counts in the goods - north field are consistent , so cosmic variance is possibly not the dominant uncertainty .
the increased depth of udf enables us to reach a @xmath7 limiting magnitude of @xmath8 ( equivalent to @xmath9 at @xmath10 , or @xmath11 for the @xmath12 @xmath13-drop population ) .
the star formation rate at @xmath1 was approximately @xmath14 _ less _ than at @xmath12 .
this declining comoving star formation rate ( @xmath15 at @xmath1 at @xmath16 for a salpeter imf ) poses an interesting challenge for models which suggest that @xmath16 star forming galaxies at @xmath17 reionized the universe .
the short - fall in ionizing photons might be alleviated by galaxies fainter than our limit , or a radically different imf .
alternatively , the bulk of reionization might have occurred at @xmath18 .
we have recently discovered evidence of an early epoch of star formation in some of the @xmath0-drops at @xmath1 . _
spitzer _ images with irac at @xmath19 m show evidence of the age - sensitive balmer/4000 , dominated by stars older than 100myr ( and most probably 400myr old ) .
this pushes the formation epoch for these galaxies to @xmath20 .
there are at least some galaxies already assembled with stellar masses @xmath21 ( equivalent to @xmath22 today ) within the first billion years .
the early formation of such systems may have played a key role in reionizing the universe at @xmath23 . |
the electron spin has recently made its appearance in organic electronics , generating a new field which takes the suggestive name of molecular - spintronics @xcite .
the main idea is to investigate how spin - currents propagate through organic molecules possibly affecting their internal state .
example of such spin - molecular devices are spin - valves made from carbon nanotubes @xcite , organic insulators @xcite or conducting polymers @xcite , fempto - second spectroscopy of molecular bridges connecting semiconductor nanodots @xcite and three terminal devices using magnetic molecules as active elements @xcite .
organic molecules present genuine advantages over conventional inorganic metals and semiconductors .
typically the spin - orbit interaction , which scales with the atomic number @xmath3 as @xmath4 , is weak in molecules .
for instance the spin - orbit splitting of the valence band of carbon in its diamond form is only 13 mev @xcite , to be compared with 340 mev of gaas .
this yields rather long spin - diffusion lengths ( @xmath5 ) with conservative estimates of 130 nm in multiwall carbon nanotubes @xcite , 200 nm in polymers @xcite , and 5 nm in alq@xmath6 monolayers @xcite .
it is important to note that these values are usually extracted from the magnetoresistance ( mr ) of spin - valves using the jullier s formula @xcite and therefore also include possible spin - flip at the surface of the magnetic electrodes .
this means that the intrinsic spin - diffusion length is such organics may be considerably longer .
furthermore one should remember that at least in polymers and alq@xmath6 the resistivity is rather large and therefore the spin - relaxation times associated to the values of @xmath5 given above are quite long . in organic molecules also the hyperfine interaction , another possible source of spin - decoherence , is weak .
most of the molecules used for spin - transport are @xmath7-conjugate and the conduction is mostly through molecular states with amplitude over the carbon atoms .
carbon , in its most abundant isotopic form , @xmath8c , has no nuclear spin , and therefore is not hyperfine active .
moreover the @xmath7-states are usually delocalized and the interaction is in general weak even over those nuclei with non - vanishing nuclear spin .
in addition to the long spin - diffusion length organic materials offer a range of new physical properties , difficult to find and tune in conventional inorganic materials .
for instance the electrical conductivity of a polymer can be changed by over twelve orders of magnitude with the incorporation of various counterions @xcite .
more interestingly the current - carrying elementary excitations are not band - like but usually are in the form of electronic - vibronic correlated states , with peculiar spin characteristics @xcite .
furthermore magnetic molecules with collective high - spin ground states are available @xcite . in these
the transport can be dominated by long - living metastable states , giving rise to spin - dependent non - linearities in the @xmath0-@xmath1 characteristics @xcite .
finally a less explored avenue is that of exploiting the many alternatives for bonding a molecule to a magnetic surface .
the electronic structure of the anchoring group can in fact change quite drastically the magneto - transport response of a molecule @xcite paving the way for a chemical design of magnetic devices .
this is precisely the problem we address in the present work . in particular
we investigate how spin - polarized surface states , almost always present in the case of organic spin - valves , can affect the transport .
all the calculations presented in this work have been performed with the code _ smeagol _ @xcite .
_ smeagol _ combines the non - equilibrium green s function ( negf ) method @xcite with the kohn - sham @xcite form of density functional theory ( dft ) @xcite . in its present implementation _ smeagol _ uses the code siesta @xcite as the main dft platform .
this has the advantage of expanding the kohn - sham problem over an efficient linear combination of localized numerical orbitals @xcite , making large systems tractable without the need of massive computational resources .
in addition core electrons are removed from the self - consistent calculation and described by standard norm - conserving scalar relativistic troullier - martins pseudopotentials .
the central quantity that _ smeagol _ evaluates is the non - equilibrium green s function @xmath9 for the portion of the device where the potential drops .
usually this is called `` extended molecule '' ( em ) and comprises the molecule itself and part of the current / voltage probes .
such a green s function simply reads @xmath10^{-1}\ : , \label{negfmx}\ ] ] where @xmath11 is the hamiltonian of the em , @xmath12 is the energy and @xmath13 ( @xmath14 ) is the self - energy for the left - hand side ( right - hand side ) electrode .
these latter ones are non - hermitian energy - dependent matrices describing the interaction of the em with the current / voltage electrodes and effectively allow us to map a hermitian problem for an open non - periodic system onto a non - hermitian problem for a finite system .
they can be written as @xmath15 with @xmath16 the coupling matrix between the leads @xmath17 ( @xmath18l , r ) and the em .
the @xmath19 are surface green s functions for the leads , i.e. the green s function for a semi - infinite lead evaluated at the termination plane .
they are constructed using a semi - analytical expression @xcite , after removing the singularities of the coupling matrices with a renormalization scheme @xcite .
importantly when @xmath11 is a function of the charge density @xmath20 , as in dft , then a simple prescription for a self - consistent calculation can be designed .
in fact , a close formula for the non - equilibrium charge density in terms of the green s function @xmath9 is available @xcite , and can be used in conjunction with equation ( [ negfmx ] ) for deriving the electronic structure of the em in presence of the leads .
external bias can be added in the self - consistent procedure by imposing the appropriate boundary conditions to the hartree potential in the leads .
this simply produces a rigid shift of the whole spectra in the current / voltage electrodes , since they are good metals and therefore local charge neutral .
finally the two probe current can be extracted by using a landauer - like formula @xmath21[f(e-\mu_\mathrm{l})-f(e-\mu_\mathrm{r})]\ ; , \label{currnegf}\ ] ] where @xmath22 , @xmath23 $ ] , @xmath24 is the fermi function and @xmath25 with @xmath26 the fermi energy of the leads and @xmath1 the external bias .
note that ultimately the current is nothing but the integral of the bias - dependent transmission coefficient @xmath27 evaluated over the bias window .
in our simulations the em includes three ni planes at each side of the molecule and the cross section is large enough to prevent intermolecular interaction .
we use periodic boundary conditions in the direction orthogonal to the transport with a 9 k - points in the two - dimensional brillouin zone .
the real - space grid cut - off is fixed at 450 ry @xcite and multiple-@xmath28 basis sets are constructed in the following way : double-@xmath28 for the @xmath29 shell of h , c , s , te and ni , and for the ni @xmath2 shell , double-@xmath28 plus polarization for the @xmath30 shells of c , s and te , and single-@xmath28 for ni @xmath30 orbitals .
finally the charge density is obtained by integrating the green s function over 96 imaginary and 300 real energy points @xcite .
consider a typical spin - valve constructed from a strong ferromagnet and a non - magnetic molecule . in this set - up
the transport is determined by the alignment of the molecular levels with the fermi energy ( @xmath26 ) of the electrodes @xcite .
let us consider the situation in which the transmission through the molecular level is small , but the anchoring groups form a spin - polarized localized state with the @xmath2-orbitals of the magnetic electrodes .
the spin - splitting of such state will be comparable with the band - exchange of the ferromagnet , although its actual form and position will depend on the details of the anchoring structure .
cartoon representing the alignment of the spin - polarized surface states associated to the two magnetic electrodes forming a spin - valve : ( a ) parallel alignment and forward bias , ( b ) antiparallel alignment and forward bias , ( c ) antiparallel alignment and reverse bias .
panels to the left are for @xmath1=0 and the bias increases from left to right .
the red boxes mark the three resonant conditions.,scaledwidth=45.0% ] in this condition of strong coupling the position of such surface state ( ss ) is strongly pinned at the chemical potential of the relevant electrode and shifts rigidly with the bias . then , the alignment of the ss associated to the left electrode of the device with that of the right determines the current .
in particular for identical electrodes and anchoring groups there are three different resonant conditions ( see figure [ fig1 ] ) depending on the bias and the mutual alignment of the magnetization vectors in the contacts . in the case of parallel alignment ( pa )
the state associated to the left - hand side electrode is lined up with that of the right only at zero bias and for both spin directions ( left panel of figure [ fig1]a ) .
if it is also located at @xmath26 , then the current will be large around @xmath1=0 .
a finite bias reduces the overlap between the states , regardless of the bias polarity ( in the picture we report forward bias only ) .
this means that the polarization of the currents decreases with bias and approximately vanishes for a bias voltage as large as the state width .
the other two resonant conditions are found for the antiparallel alignment ( aa ) of the electrodes .
in fact the sss on the two sides of the device overlap for a bias of the order of the exchange splitting of the state itself . for forward bias
such overlap is between the majority component of the ss to the left with the minority of that to the right ( last panel to the right of figure [ fig1]b ) .
the opposite is for reverse bias ( last panel to the right of figure [ fig1]c ) .
this gives the interesting result that for the aa both the spin - currents are highly asymmetric as a function of bias , although the total current is not .
thus , if we define the mr ratio as @xmath31 , we will expect a minimum in @xmath32 for biases of the order of the ss exchange splitting .
this occurs in correspondence of a large current in the aa configuration .
a completely different situation is encountered when the attachment of the molecule to a surface is highly asymmetric . in this case different resonance conditions will be present depending on the details of the structure and both the spin - current and the mr might have non - trivial bias dependence @xcite .
similar arguments have been also used for explaining rectification in non - magnetic molecular junctions using electrodes with @xmath2 states close to @xmath26 @xcite .
we start our analysis by investigating a benzene molecule chemically attached to the hollow site of two ni [ 001 ] surfaces . as anchoring groups we use s on one side and te on the other .
these both form a stable chemical bond with ni , although the two bond lengths are quite different ( 1.28 for s and 1.77 for te ) reflecting the rather different atomic radii of the anchoring atoms . with this configuration
we expect an asymmetric @xmath0-@xmath1 characteristics , although we do not expect any rectification , since the coupling with the electrodes is strong . in figure [ fig2 ] we present the @xmath0-@xmath1 characteristics for both the pa and aa , the mr ratio as a function of the bias voltage and the transmission coefficients as a function of energy for zero bias in the pa configuration .
transport properties of a s - benzene - te molecule sandwiched between ni [ 001 ] surfaces . in the large panels the @xmath0-@xmath1 characteristic for both the pa ( a ) and aa ( b ) . in the left inset
the mr ratio as a function of bias , and in the right one the transmission coefficient as a function of energy for the pa and zero bias.,scaledwidth=45.0% ] from the transmission coefficient it is clear that most of the current through the device is carried by one broad transmission peak located just below the fermi level ( @xmath26 ) for the majority spins and just above for the minority .
these are associated with a hybrid ss formed from ni-@xmath2 orbitals with the @xmath30-shells of both s and te .
the resulting transport properties are those expected from the simple discussion of the previous section . in the pa
the two spin - currents are rather similar since no resonant condition is ever met except for zero bias .
note also that both spin - currents almost saturate at around @xmath330.5 volt , which approximately corresponds to the width of the ss .
finally it is clear that the spin polarization of the current , defined as @xmath34 ( @xmath35 is the spin @xmath36 current ) , decreases as the bias increases and approximately vanishes for @xmath371 volt . in the aa the two spin - currents
individually present a rather asymmetric bias dependence reflecting the two resonant conditions .
however the total current is almost completely symmetric and the tiny asymmetry is associated with the slightly different bonding of s and te with the ni hollow site . importantly also in this case the current saturates at around @xmath330.5 volt , when the resonant condition is not met any longer .
this gives a mr ratio with a minimum and negative sign at around @xmath330.5 volt .
interestingly the spin - resolved mr ratio @xmath38 ( note that @xmath39 ) is rather asymmetric with @xmath1 , reflecting the asymmetry of the spin - currents .
note finally that the mr ratio changes sign when the bias is increased from zero , since we gradually move from the resonant condition for the pa to those for the aa .
similar results have been recently reported for different anchoring structures @xcite and molecules @xcite .
the junction described in the previous section presents rather large currents and the typical transmission coefficient approaches @xmath40 .
this is not the ideal situation for studying resonances , which in contrast dominate the transport in the case of overall small transmission .
for this reason we move to investigating a stm - type geometry . here
we consider the same benzene molecule attached to the hollow site of a ni [ 001 ] surface via s and terminated with a thiol group .
this latter is then in loose contact with a single - atom ni stm tip positioned on top of s ( with a ni - s separation of 4 ) and shifted perpendicularly to the benzene plane by @xmath411.2 ( see inset of figure [ fig3 ] ) . in this configuration
we probe the @xmath42 orbital of s ( i.e. the one perpendicular to the benzene plane ) , which is involved in the transport .
@xmath0-@xmath1 characteristic for a benzene - thiol - thiolate molecule attached to the hollow site of ni [ 001 ] and probed with a ni stm tip . in the insets
the geometrical configuration of the device and @xmath32 as a function of bias.,scaledwidth=45.0% ] in figure [ fig3 ] we present both the @xmath0-@xmath1 characteristics and the mr for such a device . at variance with the previous case the total currents for
the pa and aa are rather similar for forward bias , but differ substantially for the negative voltages . in this case
the pa current drastically increases for @xmath1 between -0.1 volt and -0.5 volt and then saturates . in
turns this produces a large positive peak in the mr reaching 70% for @xmath1 around -0.5 volt .
away from this peak the mr ratio is a smooth function of the bias and generally it is quite small . in order to have a better understanding of this second situation in figure [ fig4 ] we show the orbital resolved density of states for our device , where in particular we project over the ni [ 001 ] surface , the benzene molecule and the @xmath2 orbitals of the tip . from the picture
we can identify the two spin - polarized states mainly responsible for the transport .
the first is the tip @xmath2-orbitals band - edge ( red circles in figure [ fig4 ] ) with majority and minority components approximately 200 mev below the corresponding @xmath2 band - edge of the substrate .
orbital resolved density of state for benzene - thilolate - thiol molecule attached to ni [ 001 ] and probed by a ni stm tip .
the tip and surface states are indicated respectively with red circles and blue squares.,scaledwidth=45.0% ] the second is a hybrid ss with spin - polarized components ( blue squares ) approximately 200 mev above the corresponding @xmath2 band - edge of the substrate .
it is then clear that a resonant condition is found whenever the tip @xmath2 band - edge is aligned with ss of the substrate .
however , in contrast to the cartoon of figure [ fig1 ] , the resonant states of the left - hand side electrode ( associated to the ni [ 001 ] substrate ) and of right - hand side electrode ( associated to the tip ) are in different positions .
the alignment that we obtain from figure [ fig4 ] immediately suggests that , regardless of the magnetization alignment of the electrodes , there can not be any resonant condition for positive bias ( current going from the substrate to the tip ) , since the tip state is lower than the ss of the substrate .
in contrast a resonant condition can be matched at negative bias for pa and both spins . in figure [ fig5 ]
we demonstrate this resonant condition and present the transmission coefficient as a function of energy in the pa for various biases .
we note that for zero bias @xmath43 displays two 0.5 ev wide peaks around @xmath26 for both spin directions .
these directly reflect the tip @xmath2 orbital band - edge and the spin - polarized ss . as the negative bias increases the tip and substrate state move to resonance as demonstrated by the large transmission peak at @xmath37 -0.4 volt .
this occurs for both majority and minority spins , however only the minority contribute to the current since only the minority peak falls within the bias window .
transmission coefficient as a function of energy at different bias for benzene - thilolate - thiol molecule attached to ni [ 001 ] and probed by a ni stm tip : parallel alignment only .
note the sharp increase of the transmission for v@xmath44 -0.4 volt .
the solid ( dashed ) line represent respectively majority and minority spins .
the vertical red lines indicate the bias window.,scaledwidth=45.0% ] finally as the bias keeps increasing the resonance is destroyed and the enlargement of the bias window is compensated by a drastic reduction of the transmission peaks .
this causes the current to saturate .
note that for positive bias no resonant behavior is observed within the bias window at any of the voltages investigated , although some resonances are observed approximately 1 ev below @xmath26 .
a similar non - resonant situation is encountered for the aa case and it is not presented here .
in conclusion we have investigated resonant transport in organic spin - valves . in the case of a single resonant state
we predict a maximum in the current for the antiparallel alignment of the electrodes at a bias comparable with the exchange - splitting of the resonant state .
in contrast in the case of highly asymmetric anchoring structures multiple resonant states may appear and different resonant conditions can be found . in particular
we investigate the almost symmetric case of benzene molecule chemically bonded to ni [ 001 ] surface via s and te , and that of a stm - type geometry .
importantly the mr ratio is a non - trivial function of bias and can assume both positive and negative values .
therefore such a strong variation of @xmath32 with bias seems to be a distinctive feature of organic spin - valves .
this work is supported by science foundation of ireland under the grants sfi02/in1/i175 and sfi05/rfp / phy0062 .
computational resources have been provided by the hea iitac project managed by the trinity center for high performance computing and by ichec . | we investigate theoretically the effects of surface states over the magnetoresistance of ni - based organic spin - valves .
in particular we perform _ ab initio _ electronic transport calculations for a benzene - thiolate molecule chemically attached to a ni [ 001 ] surface and contacted either by te to another ni [ 001 ] surface , or terminated by a thiol group and probed by a ni stm tip . in the case of s- and te - bonded molecules we find a large asymmetry in the spin - currents as a function of the bias , although the @xmath0-@xmath1 is rather symmetric .
this leads to a smooth although not monotonic dependence of the magnetoresistance over the bias .
in contrast , in the case of a stm - type geometry we demonstrate that the spin - current and the magnetoresistance can be drastically changed with bias .
this is the result of a resonance between a spin - polarized surface state of the substrate and the @xmath2-shell band - edge of the tip . |
surface plasmon resonance ( spr ) is a well accepted direct detection technique for monitoring biological processes @xcite . while ellipsometry is another well known method for analyzing thin film properties @xcite , its use in liquid medium for monitoring biochemical reactions
is made difficult by the varying environment through which the probing light beam has to propagate . in the kretschmann configuration , the laser generating the spr evanescent wave is only propagating through the substrate , leading to a better control over the influence of the various buffer solutions used during a protein adsorption experiment .
the use of various acoustic wave devices for monitoring bound mass changes in liquid media is well known @xcite .
love mode devices , based on the propagation of a guided shear acoustic wave , present sensitivity improvements over the more usual quartz crystal microbalance @xcite as well as a compatibility with measurements in liquids .
we take advantage of the unique geometrical setup of the surface acoustic wave ( saw ) device which leaves the backside of the quartz wafer free of electrodes to inject a laser in order to generate an evanescent surface plasmon on the gold coated sensing area .
such a setup enables simultaneous estimates of the bound mass and dielectric surface properties changes during electrochemical and biochemical reactions occurring on the sensing electrode @xcite .
estimates of water content in protein layers is an important topic in the development of biosensors , since the detection sensitivity is directly related to the number of active sites on the surface of the sensor to which the receptor molecules are bound .
when physical measurements of this bound layer are performed , such as via spr or acoustic measurements , a large water content will lead to overestimates of the potential detection limit of the biosensor due to an overestimate of number of molecules bound to the surface . from a more fundamental point of view , determining water level content should provide more accurate physical parameters of the protein layer itself such as density and optical index by allowing the subtraction of the influence of water once it has been identified .
we use a modified commercial spr instrument ( ibis ii , ibis technologies bv , the netherlands ) to detect the surface plasmon resonance angle after replacing the gold coated glass slide by a love mode saw device ( fig .
[ fig1 ] ) .
the excitation laser in this instrument has a wavelength of 670 nm . all reflected intensity _ vs. _ angle curves ( 6 @xmath3 angle span recorded on a 200 pixels ccd array )
were recorded and later fitted by a polynomial to extract with high accuracy the position of the dip .
the acoustic wave device is made of a 500 @xmath4 m thick st - cut quartz wafer on which 200 nm thick sputtered @xmath5 interdigitated electrodes are patterned .
the surface is coated by a 1.13 @xmath4 m pecvd silicon dioxide layer acting as a guiding layer , and the [email protected] mm@xmath7 sensing area is coated with 10 nm @xmath8 and 50 nm @xmath9 .
this area acts both as a working electrode for electrochemistry or a grounded electrode during biochemical experiments , as well as a supporting layer for the surface plasmon resonance generation .
the influence of the saw device substrate over the detection of the spr is limited to interference patterns due to the birefringence of the quartz , and the optical index mismatch between the quartz and the deposited silicon dioxide layer .
the former effect is reduced by orienting the optical axis of quartz so that it is in the plane defined by the normal to the sensing surface and the wavevector of the laser , thus minimizing the optical index difference between the ordinary and the extraordinary axis .
the remaining interference patterns lead to fringes with a low enough contrast that the surface plasmon peak is easily identified and tracked during the experiments .
the phase of the saw delay line is monitored at a fixed frequency using a network analyzer hp4396a at 123.200 mhz .
the phase shift is converted to a frequency shift thanks to the linear phase to frequency relationship recorded in the bode plot .
the observed phase shift leads to a frequency shift ( as would be observed in a phase locked loop configuration ) which in turn can be converted to an adsorbed mass change through the sensitivity of the device .
we first electrodeposited copper on the surface in order to calibrate the mass sensitivity of the acoustic wave device in liquid medium ( fig .
[ fig2 ] ) @xcite . at the same time
, the spr displayed resonance angle shifts due to the varying potential @xcite .
when the voltage applied by the potentiostat is above 0.2 v with respect to the @xmath10 pseudo reference electrode , the spr angle slightly shifts due to the electroreflectance effect described by ktz _
_ @xcite . below 0.2
v , under potential deposition starts depositing a mono layer of @xmath10 atoms on the @xmath9 surface as visible both in the phase shift of the saw device and as a reversal of the trend of the angle shift of the spr .
below 0 v , a rough @xmath10 layer is deposited , leading to a loss of the spr peak and a large phase shift of the saw device due to the large added mass .
the rough , discontinuous film of copper clusters a few tens to hundreds of nanometers high was observed by in - situ afm imaging ( data not shown @xcite ) .
simulating the influence of sub - wavelength metallic structures on the surface plasmon resonance is a difficult task @xcite which we have not attempted to solve .
simultaneously , the saw device displays a frequency drop @xmath11 due to the added mass @xmath12 to the sensing electrode , leading with the combination of the current measurement from the potentiostat to a direct estimate of the mass sensitivity @xmath13 where @xmath14 cm@xmath7 is the area of the sensing working electrode and @xmath16 mhz the base frequency of the saw device .
the electrodeposited mass is deduced from the current measurement @xmath17 with a sampling period @xmath18 : @xmath19 where @xmath20 c is the faraday constant , @xmath21 the molar weight of the electrodeposited metal and @xmath22 the number of electrons exchanged for each metallic atom reduced .
the resulting sensitivity for the love mode device under consideration here is 150@xmath215 cm@xmath7/g .
once the mass sensitivity of the saw is calibrated , we adsorbed on the same device a crystalline monolayer of s - layer proteins ( 100 @xmath4g / ml of the protein sbpa of _ bacillus sphaericus _ ccm 2177 in 0.5 mm tris , 10 mm @xmath23 , ph=9 buffer ) @xcite while monitoring simultaneously with the saw device the mass variation and with spr the surface dielectric change .
the adsorption and desorption steps , the latter being possible with the use of 2 % @xmath24 , was repeated several times ( fig .
[ fig3 ] ) .
a property of this protein which makes it suitable for calibration of a new instrument is that it only forms a monolayer and will not stack to multiple layers even at high concentrations .
the parameters required for analyzing the saw data are the layer thickness @xmath25 and the layer density @xmath26 . the parameters required for analyzing the spr data are the common layer thickness @xmath25 and the optical index @xmath27 of the adsorbed layer .
we thus have three free parameters and only two measurements , namely the adsorbed mass per unit area and an spr resonance angle shift which is a function of both @xmath27 and @xmath25 .
the two options are to add one more measurement parameter , such as an spr dip position as a function of wavelength which would lead to a unique identification of @xmath25 and @xmath27 @xcite , or to reduce the number of variables by adding the assumption that the layer is made of a homogeneous mixture of a proportion @xmath28 of protein and @xmath29 of water .
the layer density @xmath26 is then @xmath30 where @xmath31 is assumed to be equal in the @xmath32 g/@xmath33 to 1.4 g/@xmath33 range @xcite and @xmath34 g/@xmath33 is the density of water .
similarly , we use then the optical index of the layer @xmath35 with the optical index of the protein layer assumed to be in the @xmath36 to @xmath37 range @xcite and that of water @xmath38 .
the remaining unknown parameters are then the layer thickness @xmath25 and the water content proportion @xmath29 . by simulating a stack of planar multilayers ( glass @xmath39 , 1200 nm silicon dioxide @xmath40 , 2 nm titanium @xmath41 @xcite , 50 nm gold @xmath42 @xcite , proteins , water ) for the angle shift as a function of water content and layer thickness
, one obtains a set of pairs of values for these two parameters compatible with the observed angle shift ( fig .
[ fig4 ] , bottom ) .
then , by calculating the mass per unit area @xmath43 , a comparison with the experimentally observed adsorbed mass as seen from the phase shift of the saw device leads to a unique set of parameters both compatible with the optical index change and the adsorbed mass ( fig .
[ fig4 ] , top ) . in the case under consideration here , an observed angle shift of 380 to 400 m@xmath3 and an adsorbed mass of 540 to 580 ng / cm@xmath7 is only compatible with @xmath44 and @xmath45 nm . the thickness result is compatible with atomic force microscope ( afm ) measurements in liquid @xcite while the mass per unit area is compatible with morphological data obtained by electronic microscopy @xcite . while the main source of uncertainty is due to the wide possible values of the density and optical index reported in the literature , the experimental results display good reproducibility .
the water content is lower than that observed by other authors for different kinds of proteins @xcite but compatible with the 30 to 78% value cited in ref .
the dense packing of protein is explained by the regular arrangement of identical s - layer subunits in the p4 lattice @xcite .
we have shown here how the combination of love mode saw device with spr provides advantageous combinations of information on the bound mass as well as on the dielectric changes when monitoring protein adsorption . the resulting protein layer thickness and protein content percentage , respectively @xmath45 nm and @xmath46 % , is in agreement with independent afm estimates .
the saw transducers were fabricated within the framework of a belgian phd scholarship program ( fonds pour la formation la recherche dans lindustrie et dans lagriculture fria ) .
we wish to thank r. giust ( lopmd , besanon , france ) for kindly providing spr simulation routines .
b. liedberg , c. nylander , and i. lundstrm , sensors and actuators * 4 * , 299 ( 1983 ) b. ivarsson , m. malmqvist , _ surface plasmon resonance : development and use of biacore instruments for biomolecular interaction analysis _ , in biomolecular sensors , ed .
e. gizeli and c.r .
lowe ( taylor & francis , london ) 2002 f. hk , b. kasemo , t. nylander , c. fant , k. sott , and h. elwing , anal . chem . *
73 * 5796 ( 2001 ) e. gizeli , _ acoustic transducers _ , in biomolecular sensors , ed .
e. gizeli and c.r .
lowe ( taylor & francis , london ) 2002 l.e .
bailey , d. kambhampati , k.k .
kanazawa , w. knoll , and c.w .
franck , langmuir * 18 * 479 ( 2002 ) a. laschitsch , b. menges , and d. johansmann , appl .
phys . lett . *
77 * 2252 ( 2000 ) j .- m .
friedt , l. francis , k .- h .
choi , and a. campitelli , j. vac .
a ( to be published ) r. ktz , d.m .
kolb , and j.k .
sass , surf .
sci . * 64 * 96 ( 1977 ) t.w .
ebbesen , h.j .
lezec , h.f .
ghaemi , t. thio , and p.a .
wolff , nature * 391 * 667 ( 1998 ) a.j .
haes , and r.p .
van duyne , j. am .
124 * 10596 ( 2002 ) m. sra , and u.b .
sleytr , micron * 27 * 141 - 156 ( 1996 ) e.s .
gyrvary , a. oriordan , a.j .
quinn , g. redmon , d. pum , and u.b .
sleytr , nanoletters * 3 * 315 ( 2003 ) b.p .
nelson , a.g .
frutos , j.m .
brockman , and r.m .
corn , anal . chem .
* 71 * 3928 ( 1999 ) f. caruso , d.n .
furlong , k. ariga , i. ichinose , and t. kunitake , langmuir * 14 * 4559 ( 1998 ) f. hk , j. vrs , m. rodahl , r. kurrat , p. bni , j.j .
ramsden , m. textor , n.d .
spencer , p. tengvall , j. gold , and b. kasemo , colloids and surfaces b * 24 * 155 ( 2002 ) r.j .
marsch , r.a.l .
jones , and m. sferrazza , _ adsorption and displacement of a globular protein on hydrophilic and hydrophobic surfaces _ colloids and surfaces b * 23 * 31 ( 2002 ) e. palik , _ handbook of optical constants of solids _ , academic press ( 1997 ) e. stenberg , b. persson , h. roos , and c. urbaniczky , journal of colloid and interface science * 143 * 513 ( 1991 ) m. weygand , b. wetzer , d. pum , u.b .
sleytr , n. cuvillier , k. kjaer , p.b . howes and m. lsche ,
j. * 76 * 458 ( 1999 ) m. weygand , m. schalke , p.b .
howes , k. kjaer , j. friedmann , b. wetzer , d. pum , u.b .
sleytr , and m. lsche , j. mater . chem . * 10 * 141 ( 2000 ) | we present results from an instrument combining surface acoustic wave ( saw ) propagation and surface plasmon resonance ( spr ) measurements .
the objective is to use two independent methods , the former based on adsorbed mass change measurements and the latter on surface dielectric properties variations , to identify physical properties of protein layers , and more specifically their water content .
we display mass sensitivity calibration curves using electrodeposition of copper leading to a sensitivity in liquid of 150@xmath0 @xmath1 for the love mode device used here , and the application to monitoring biological processes .
the extraction of protein layer thickness and protein to water content ratio is also presented for s - layer proteins under investigation .
we obtain respectively [email protected] nm and 75@xmath215% . |
the class of hyperluminous infrared galaxies , whose bolometric luminosities exceed @xmath9 , have been found through far - infrared and submillimetre surveys by iras and scuba ( e.g. , rowan - robinson 2000 ) , and more are expected from the spitzer space telescope .
some of them may be massive galaxies in formation at high redshift and , given the suggested link between galaxy spheroids and central black holes in them ( e.g. , tremaine et al 2002 ) , they might represent an important galaxy evolution phase where a quasar nucleus is forming amid heavy dust obscuration ( e.g. , sanders et al 1988 ) .
a population of mid - infrared - selected hyperluminous objects at @xmath10 , which appear to be powered mainly by obscured active galactic nuclei ( agn ) , are emerging from spitzer surveys ( houck et al 2005 ) . as well as the origin of the enormous luminosity being of a great interest , examining the role of luminous counterparts of nearby compton - thick seyfert agn to the x - ray background ( xrb ) and
their detectability is also important , since the higher energy part of the xrb is not yet fully resolved ( worsley et al 2005 ) .
iras f15307 + 3252 is one of the few classical iras - selected hyperluminous infrared galaxies at redshift of @xmath11 ( cutri et al 1994 ) . with the currently popular cosmology with @xmath12 , @xmath13 , and @xmath14 , the luminosity distance is 6.1 gpc and and the bolometric luminosity is estimated to be @xmath15 .
the large bolometric luminosity was once suspected to be the result of gravitational lensing ( liu , graham & wright 1996 ) as in the case of iras f10214 + 4724 at @xmath16 ( e.g. , broadhurst & lehar 1995 ) .
however farrah et al ( 2002 ) have ruled this out based on hst imaging .
therefore , iras f15307 + 3252 indeed belongs among the most luminous objects in the universe .
the uv / optical emission - line spectrum of iras f15307 + 3252 is of seyfert 2 type ( cutri et al 1994 ; soifer et al 1994 ; liu et al 1996 ; evans et al 1998 ) .
optical spectropolarimetry has revealed broad mgii @xmath17 emission and a rest - frame uv spectrum which is very similar in shape and intensity to that of typical quasars , indicating the presence of a dust - shrouded quasar nucleus ( hines et al 1995 ) . despite the optical evidence for hidden active nuclei ,
hyperluminous infrared galaxies are generally very faint in x - ray ( wilman et al 1998 ; 2003 ; see franceschini et al 2000 ; iwasawa , ettori & fabian 2000 for iras p09104 + 4101 ; iwasawa 2001 ; alexander et al 2005 for iras f10214 + 4724 ) . iras f15307 + 3252
is no exception and previous x - ray observations with rosat and asca failed to detect any x - rays ( fabian et al 1996 ; ogasaka et al 1997 ) .
we report here the detection of faint x - ray emission from this hyperluminous infrared galaxy and its properties obtained from xmm - newton observations , and discuss their implications .
iras f15307 + 3252 was observed with the epic cameras of xmm - newton on three occasions ( table 1 ) . both epic pn and mos cameras operated in the full - window mode with the medium filter in all the three observations .
a significant fraction of the datasets are affected by high radiation background . a larger fraction of the observed intervals are affected in the pn data than in the mos data .
time intervals of quiescent background ( count rate less than 0.3 for each of the mos and less than 1.21.5 for the pn ) were selected , based on the single event light curves at energies above 10 kev taken from the whole field of each detector .
the useful exposures from the three observations are 6 , 10 , and 5 ks ( 21 ks in total ) for the pn camera , and 10 , 12 , and 11 ks ( 33 ks in total ) for the mos cameras .
single and double events recorded by the detectors were used for the data analysis presented here .
the data reduction was performed by the standard xmm - newton software package sas 6.1 and lheasoft 5.3.1 .
.xmm - newton observations of iras f15307 + 3252 .
useful exposure times in unit of ks ( and whole duration of exposure including the high background intervals in parenthesis ) for the respective epic cameras . [
cols="<,^,^ " , ]
iras f15307 + 3252 is detected with xmm - newton but is very faint , particularly at energies above 2 kev .
with so few detected counts , the individual observations do not warrant an independent analysis .
we integrated all three datasets to examine source detection first in different energy bands .
compton - thick obscuration can be a plausible explanation for the x - ray faintness of iras f15307 + 3252 .
since a strong fe k@xmath18 feature at rest - frame 67 kev would characterise the x - ray spectrum in such a case , we designed a photometric study by dividing the whole bandpass into the following four bands : 0.52 kev , 23 kev , 34 kev , and 510 kev , so that fe k@xmath18 feature falls in the 34 kev band by the galaxy redshift ( @xmath11 ) .
x - ray images in the four bands were constructed by adding all the pn and mos data from the three observations , which are shown in fig . 1 . iras f15307 + 3252
is detected at more than @xmath19 in all but the 23 kev band images ( see table 1 ) .
the source is strongly detected at lower energies below 2 kev for which further details are presented below .
this component does not seem to stretch well above 2 kev , as suggested by the lack of detection in the 23 kev band .
note , however , a remarkable recovery of the source in the 34 kev band .
this indicates a strong excess in the 34 kev range , where fe k@xmath18 emission is expected .
the 23 kev drop - out and the marginal detection in the 510 kev band suggest a rather hard underlying continuum in the 210 kev ( the rest frame 420 kev ) range . these are characteristics of the x - ray spectrum of a heavily absorbed active nucleus .
the soft x - ray emission below 2 kev found at the position of iras f15307 + 3252 is not point - like .
the radial profile of the soft band source is compared with that of the nuclear source in ngc4395 , which was also observed with the xmm - newton in the full - window mode ( fig .
since the x - ray emission from ngc4395 has been found to be point - like as viewed by the chandra acis - s ( moran et al 2005 ) , it represents the point spread function ( psf ) . although adding different detectors and observations together might have resulted in a blurred psf at some degree , a comparison of the radial profile with other point sources in the field confirms the extended nature of the soft x - ray emission from iras f15307 + 3252 . the angular scale at @xmath11 with the adopted cosmology is @xmath20 kpc arcsec@xmath21 .
detection of a source size broader than the psf ( fwhm @xmath225 - 6 arcsec ) means that the source is extending at least @xmath23 kpc .
fitting a lorentzian function to the respective radial profiles of iras f15307 + 3252 and ngc4395 implies that the fwhm of the intrinsic source extension is about @xmath24 arcsec , corresponding to 85 kpc in radius .
this is larger than the effective radius ( @xmath25 kpc , farrah et al 2002 ) of the giant elliptical host of iras f15307 + 3252 , of which the surface brightness distribution is described well with the de vaucouleur s profile ( see also liu et al 1996 ) .
no galaxy cluster has so far been reported for the region around iras f15307 + 3252 . however , iras f15307 + 3252 is a luminous ( @xmath26 , farrah et al 2002 ) giant elliptical , inside which some substructures have been found , indicating galaxy mergers or interaction ( fig .
3 , also soifer et al 1994 ; liu et al 1996 ; farrah et al 2002 ) .
the galaxy is @xmath27 mag more luminous than the virgo brightest galaxy m87 , and should be in a relatively massive dark matter halo .
3 shows the hst / acs image of the region around iras f15307 + 3252 , retrieved from the hst archive at stsci .
the data were taken on 2002 august 18 in a 2,120-s exposure , using the f814w filter ( approximately @xmath28 band ) . while there is no clear evidence for a galaxy concentration centred on the iras galaxy as such expected in a rich cluster ,
a chain of several small galaxies 1115 arcsec ( @xmath29 kpc ) to the south is seen , which lies within the x - ray extension , indicating a moderate galaxy over - density .
the hst image also shows that both iras f15307 + 3252 and the several nearby compact knots ( which farrah et al 2002 ascribe to the remnants of merging companions ) are embedded in an envelope of diffuse emission out to a radius of @xmath303 arcsec .
there is considerable structure within this emission , including a sharp conical - shaped cavity to the se with its apex at the agn .
further arc - shaped structures concentric to the active nucleus and perpendicular to the cavity are also apparent , which could be due to [ oii]@xmath31 located near the edge of the filter bandpass , representing ionization cones .
fitting jointly the pn and mos spectra ( fig .
4 ) gives a temperature of @xmath32 kev ( hereafter , the errors for spectral parameters are of the 90 per cent confidence region for one parameter of interest ) .
the 0.42.8 kev band was fitted with half solar metallicity and galactic absorption of @xmath33(dickey & lockman 1990 ) being assumed .
the observed 0.52 kev flux is @xmath34 .
the absorption - corrected bolometric luminosity is estimated to be @xmath35 .
the temperature and bolometric luminosity are on the @xmath5-@xmath6 relation for galaxy clusters , groups and elliptical galaxies ( e.g. , fukazawa et al 2004 ) , and iras f15307 + 3252 lies in the region for a poor cluster ( it is comparable to the virgo cluster in temperature and luminosity ) .
we tentatively identify the extended soft x - ray emission with hot gas associated with a relatively poor cluster around the iras galaxy .
whilst soft x - ray emission in some hyperluminous infrared galaxies has been ascribed to star formation ( wilman et al 2003 ; alexander et al 2005 ) , the large source size , high temperature and large soft x - ray to infrared luminosity ratio suggest that a starburst is unlikely to be a major component of the soft x - ray emission in iras f15307 + 3252 ( see also section 4.2 ) .
the cluster emission has a relatively short radiative cooling time of @xmath36 gyr .
this would lead to a cooling flow of @xmath37 yr@xmath21 if there is no heating to balance cooling ( there is a 6 mjy first radio source at 1.4 ghz ; becker , white & helfand 1995 ) .
the photometric study ( section 3.1 ) demonstrated that x - ray emission of iras f15307 + 3252 shows a strong excess in the 34 kev range , which is probably due to a redshifted fe k@xmath18 line .
there is a very faint underlying continuum , which extends to higher energies .
the faint 510 kev band is too noisy for a spectral study and only fit for a photometric measurement .
the spectral data in the 25 kev band can be fitted with a narrow gaussian with a flat power - law ( @xmath38 ) .
the centroid of the gaussian is @xmath39 kev in the rest frame .
the fit shows a line width of 0.1 kev in gaussian dispersion , but the data also agree with no broadening . assuming this feature is all due to line emission yields a line flux of @xmath40 .
since the continuum is too faint to obtain a reasonable constraint , the equivalent width ( ew ) of the line is accordingly uncertain , but it is likely to be larger than 2 kev .
the observed 25 kev flux is @xmath41 ( corresponding to a 410 kev rest frame luminosity of @xmath42 ) , most of which originates from the line feature .
extrapolating the 25 kev continuum model with @xmath43 predicts source counts which are comparable to the detected counts in the 510 kev image at the given exposure times .
the observed 510 kev flux derived from the extrapolation is @xmath44 , and the rest - frame 1020 kev luminosity estimated from the flux is @xmath45 .
the detection of the prominent fe k@xmath18 line strongly indicates the presence of a compton - thick agn in iras f15307 + 3252 .
the large ew of the fe k@xmath1 and the deficit of x - ray emission just below the line imply the lack of the direct continuum emission in the fe k band , which means the absorbing column density to be @xmath46 or larger .
thus , together with the line centroid at 6.4 kev , the hard x - ray emission from iras f15307 + 3252 is dominated by reflection .
using the observed line luminosity , we made a rough estimate of the minimum rest - frame 210 kev luminosity which is required to produce the observed fe k emission .
assumptions made for this calculation are as following : fe k@xmath18 emission is produced in cold matter with solar abundance ( @xmath47 , morrison & mccammon 1983 ) , irradiated by a power - law source with a typical quasar photon - index of @xmath48 ; the continuum photons above the k - shell ionization threshold energy ( 7.11 kev ) are converted into fe k@xmath18 at 6.4 kev with the efficiency calculated by basko ( 1978 ) and george & fabian ( 1991 ) , as such that would produce @xmath49 ev , if an observer see both the illuminating source and reflection from the cold slab behind it ; and the solid angle of the cold matter viewed by the illuminating source is @xmath50 . with these assumptions ,
the ratio @xmath51 is found to be about 3 per cent . since
not all the reflecting surfaces are likely visible to an observer , the minimum 210 kev luminosity to produce the observed fe k line is estimated to be @xmath52 .
the ratio of bolometric to the 210 kev band luminosity , @xmath53 , is typically 3050 for quasars
( elvis et al 1994 ) . with this bolometric correction , the bolometric luminosity of the agn in iras f15307 + 3252 is found to be @xmath54 , where @xmath55 .
therefore a large fraction of the bolometric luminosity ( @xmath56 ) is likely to be powered by the hidden agn , consistent with previous estimates ( e.g. , yun & scoville 1998 ; aussel et al 1998 ; verma et al 2001 ; peeters , spoon & tielens 2004 ) .
the reflected x - ray emission originates from optically thick matter and is of a distinct origin from the scattered uv / optical light revealed by spectropolarimetry ( hines et al 1995 ) . the large polarization in the uv / optical bands ( 1620 per cent ) is presumably due to dust scattering .
the black hole masss hosted in iras f15307 + 3252 is estimated to be @xmath57 , using the empirical relation based on the virial theorem for the blr obtained by mclure & jarvis ( 2002 ) with the mgii line width of fwhm@xmath58 and the 3000 luminosity ( @xmath59 ) of the intrinsic uv spectrum derived from the spectropolarimetry ( hines et al 1995 ) .
the bolometric luminosity of this object is then about half the eddington luminosity for the estimated black hole mass .
iras f15307 + 3252 shares many properties with another hyperluminous infrared galaxy iras 09104 + 4109 . in their spectral energy distribution ,
the infrared to optical and 60 @xmath60 m to 1.5 ghz luminosity ratios for both galaxies are around 300 and 50 , respectively ( cutri et al 1994 ) .
both contain a hidden quasar revealed by optical spectropolarimetry ( hines et al 1999 ) , which appears to dominate the energetics of the infrared emission , and their x - ray spectra show characteristics of compton - thick agn ( franceschini et al 2000 ; iwasawa et al 2001 )
. the black hole mass of iras 09104 + 4109 estimated by the same technique as described above is @xmath61(v.d .
ivanov , priv comm ) , similar to that of iras f15307 + 3252 .
these suggest that both galaxies have a well - grown black hole emitting at a high efficiency , although their radiation is strongly attenuated by dust shrouds in the optical wavelength and by cold gas with a large thomson depth ( @xmath62 ) in the line - of - sight in x - ray , respectively .
non - detection of co ( evans et al 1998 ; yun & scoville 1998 ) in either galaxies limits the material needed for a vigorous star formation .
the deficit of cold ( @xmath63 k ) dust in both galaxies inferred by the non - detection with scuba ( deane & trentham 2001 ; which places the limit of the contribution of cold dust to the bolometric luminosity less than 0.3 per cent ) can be a natural consequence of the absence of a strong starburst .
perhaps in these two galaxies , a mature black hole is still efficiently accreting material from small radii , while star formation on a large scale has already been terminated . as hines et al ( 1995 ) suggested ,
the agns in these galaxies are viewed from an unfavourable direction but otherwise are indistinguishable from normal quasars .
one remaining similarity between the two objects is their environment .
iras 09104 + 4109 is a giant elliptical located at the centre of a rich cluster at @xmath7 ( kleinmann et al 1988 ) , which is one of the most luminous x - ray clusters with @xmath64(fabian & crawford 1995 ; iwasawa et al 2001 ) . despite the much lower luminosity and higher redshift , iras f15307 + 3252 appears to be in a cluster as the extended x - ray emission indicates .
if this is the case , galaxy mergers in an overdensity region may be a necessary condition for making a luminous quasar .
this contrasts with nearby lower luminosity infrared galaxies which are not usually found in a rich environment ( sanders & mirabel 1996 ) .
on the other hand , quasars ( wold et al 2000 , 2001 ) and possibly hyperluminous infrared galaxies ( farrah et al 2004 ) tend to reside in a moderately dense environment , and which is in agreement with galaxy and quasar formation model based on a hierarchical assembly scenario ( e.g. , enoki , nagashima & gouda 2003 ) .
this has implications for a search for type ii quasars in x - ray .
in an unobscured quasar , the nuclear emission dominates the total x - ray emission even if there is some emission from a cluster with which the quasar is associated .
however , in an obscured quasar , the nuclear emission is suppressed and soft x - ray emission from a putative cluster could be a main x - ray source rather than the quasar itself . whether faint reflected light of a hidden nucleus is detectable depends on its luminosity .
since such a component was only barely detected in iras f15307 + 3252 , one of the most luminous galaxies , with a 30 ks xmm - newton exposure , there are clear difficulties in detecting compton - thick agn at redshift of @xmath65 .
this paper is based on observations obtained with xmm - newton , an esa science mission with instruments and contributions directly funded by esa member states and nasa .
acf and csc thank the royal society for support .
ki thanks pparc for support . | we report the detection of x - ray emission from the hyperluminous infrared galaxy iras f15307 + 3252 at @xmath0 and its properties obtained from xmm - newton observations .
although the x - ray emission is very faint and the data are noisy , a prominent line - like feature in the observed 34 kev range is inferred from both photometric and spectroscopic techniques .
it indicates an x - ray spectrum dominated by 6.4 kev fe k@xmath1 emission and the presence of a compton - thick agn .
our estimate of the luminosity of the illuminating source ( @xmath2 ) , required to produce the observed fe k@xmath1 emission in reflection from cold matter , means that the hidden quasar nucleus accounts for a significant fraction of the large bolometric luminosity . the soft x - ray emission below 2 kev is found to be spatially extended and probably of a separate origin .
the temperature and bolometric luminosity ( @xmath3 kev and @xmath4 ) obtained from a thermal spectrum place this x - ray source on the @xmath5-@xmath6 relation of galaxy clusters .
the possible association with a galaxy cluster can be added to the list of remarkable similarities between iras f15307 + 3252 and another hyperluminous infrared galaxy iras 09104 + 4109 ( @xmath7 ) , both of which have bolometric luminosities dominated by hidden quasar nuclei .
our result on iras f15307 + 3252 illustrates how difficult it is to detect compton thick type ii quasars at @xmath8 , particularly if their bolometric outputs do not rival the hyperluminous population .
galaxies : individual : iras f15307 + 3252 x - rays : galaxies x - rays : galaxies : clusters infrared : galaxies |
mcneil s nebula was recently illuminated by the outburst of v1647 ori ( mcneil 2004 ) , a young star that is embedded in the lynds 1630 dark cloud and coincides with the 850@xmath4 continuum source oribsmm55 ( mitchell et al .
v1647 ori has a flat sed in the mid - infrared , and is thus a class i young stellar object ( yso ; andrews et al .
v1647 ori underwent a similar outburst as recently as 1966 ( aspin et al .
2006 ) , indicating that v1647 ori is also an exor .
such pre - main sequence stars undergo eruptive events that dramatically increase their luminosity for periods of months to years ( hartmann 1998 ) .
the outbursts are thought to be triggered by a rapid increase in the stellar accretion rate ( hartmann & kenyon 1996 ) .
the eruption in november 2003 of v1647 ori lasted two years . during the outburst ,
the star brightened by a factor of 50 in x - rays ( kastner et al .
2006 ) , a factor of 250 in the red ( 6 mag in the @xmath5 band ; fedele et al .
2007a ; briceo et al .
2004 ) , a factor of 15 in the near - ir(@xmath63 mag in the @xmath7 , @xmath8 , and @xmath9 bands ; reipurth & aspin 2004 ) , and a factor of @xmath615 at wavelengths from @xmath10 to @xmath11 ( muzerolle et al . 2005 ) . from the overall brightening of the source , muzerolle et al .
( 2005 ) concluded that the bolometric luminosity increased by a factor of 15 to 44@xmath12 ( see also andrews et al . 2004 ) and that the stellar accretion rate increased from @xmath13 yr@xmath0 to @xmath14 yr@xmath0 .
similarly , gibb et al .
( 2006 ) inferred a stellar accretion rate of @xmath15 yr@xmath0 from the luminosity of the br@xmath16 emission one year later .
this is somewhat larger than the typical accretion rate of a young low mass star ( @xmath17 yr@xmath0 ; bouvier et al .
2007 ) , yet lower than is expected for a star of the fuor type ( @xmath18 yr@xmath0 ; hartman & kenyon 1996 ) . during the onset of the outburst of v1647 ori ,
observations of atomic lines with p cygni profiles provided evidence for a hot ( @xmath1910,000 k ) , high velocity ( @xmath20 km s@xmath0 ) wind ( briceo et al . 2004 ; reipurth et al . 2004 ; vacca et al . 2004 ; walter et al .
2004 ; ojha et al .
2006 ; fedele et al .
2007a ) with a mass - loss rate of @xmath21=4@xmath2210@xmath23 m@xmath24 yr@xmath0(vacca
et al . 2004 ) .
this mass loss rate is much lower than that of the typical fuor ( hartmann & kenyon 1996 ) and comparable to that of a classical t tauri star ( ctts ; hartigan et al . 1995 ) .
the absorption component of the p cygni profile of several lines ( e.g. pa@xmath25 ) disappeared within a few months following the peak of the outburst in early 2004 ( gibb et al .
2006 ) . however , p cygni structure in the h@xmath26 profile indicated that a weaker wind continued throughout the outburst phase ( ojha et al .
2006 ; fedele et al .
2007a ) .
in contrast to the hydrogen and helium lines , the fundamental near - infrared ro - vibrational emission lines of co , observed 2004 february 27 , were broad , centrally peaked , and compatible in their intensity with an excitation temperature of 2500 k ( rettig et al . 2005 ) .
the width of the lines was shown to be consistent with keplerian orbital motion of the gas within the inner disk surrounding the central star , similar to the broad emission line profiles that are observed around cttss and herbig ae / be stars ( haebes ; najita et al .
2003 ; blake & boogert 2004 ) .
a later observation , obtained on 2004 july 30 , showed that the co lines remained broad but the temperature of the gas decreased to 1700 k ( gibb et al .
neither observation showed any indication of co in an outflow , as a blue shifted absorption component was not detected .
we report followup observations of co from v1647 ori , which were acquired 2006 february , 2006 december , and 2007 february . by the time of our initial 2006 observation , v1647 ori had returned to quiescence and
the absorption component in the h@xmath26 line profile had disappeared ( fedele et al .
2007a and references therein ) .
presumably the accretion rate had fallen by two orders of magnitude to its pre - outburst accretion rate as the star faded to its pre - outburst brightness ( see muzerolle et al . 2005 ) . as suggested by the work of najita et al .
( 2003 ) , only a continued decrease in co line intensity from the warm gas in the disk was therefore expected .
however , the first post - outburst observation revealed the striking metamorphosis of these lines from centrally peaked emission features to emission lines with blue - shifted absorption .
subsequently , by late 2006 and early 2007 , the co emission lines returned to their original centrally peaked structure , indicating that the production of the outflow diminished within one year of the end of the outburst . in this paper
we discuss three scenarios that can give rise to such a phenomenon .
v1647 ori was observed with nirspec at the w. m. keck observatory on 2006 february 17 using one @xmath27-band setting ( table 1 ; figs . 1 and 2 ) .
nirspec is a high - resolution ( @xmath28 ) , near - infrared ( 15 @xmath4 ) , cross - dispersed spectrometer ( mclean et al . 1998 ) .
the nirspec data were reduced and calibrated using standard reduction techniques , which are described in disanti et al .
( 2001 ) , brittain et al .
( 2003 ) and brittain ( 2004 ) .
additional observations of v1647 ori were taken with the phoenix spectrometer on gemini south on 2006 december 01 , 2007 february 01 , and 2007 february 06 , using one setting centered at 2011 cm@xmath0 ( table 1 ) .
phoenix is a high - resolution ( @xmath30 ) , near - infrared ( 15@xmath4 ) spectrometer covering 1550 km s@xmath0 per observation ( hinkle et al .
1998 ; hinkle et al .
2000 ; hinkle et al . 2003 ) .
the same reduction algorithms used on the nirspec data were applied to the phoenix data .
spectra of the p30 and p31 co lines from the four epochs spanning 20042007 are presented in figure 1 in order to illustrate how the co profiles have evolved over time .
gaps in the spectra represent frequencies with less than 50% atmospheric transmittance .
the sample error bars presented for each spectrum represent the standard deviation of the continuum . because the individual phoenix spectra have lower s / n , we plot just the average of those datasets , weighting each observation by the standard deviation of the continuum .
the equivalent widths of the emission and absorption lines measured from the new datasets are presented in table 2 .
upper limits on the absorption lines in the phoenix spectrum were calculated by adopting the width of the absorption lines observed in 2006 february , setting the depth of the line equal to the standard deviation of the continuum , and assuming a gaussian line profile .
the barycentric velocity of l1630 is + 26 km s@xmath0 ( lada , bally , & stark 1991 ; gibb 1999 ; mitchell et al . 2001 ) .
since the velocity of young embedded stars is typically within a few km s@xmath0 of the surrounding cloud , we adopt this as the barycentric velocity of v1647 ori .
the emission lines , corrected for the motion of the earth , are centered on the barycentric velocity of the star under the assumption that it has the same velocity as l1630 ( fig .
1 ) . rettig et al .
( 2005 ) noted that the unblended co emission lines observed during the early phase of the outburst of v1647 ori were symmetric and centrally peaked , hence they concluded that the co emission originated in the inner disk , as in cttss and haebes .
the co emission lines from these sources are generally centrally peaked , although a few of them exhibit double - peaked structure due to rotational broadening .
the broadening of the emission lines , the absence of blue - shifted absorption , and the excitation temperature indicate that the emitting gas around cttss and haebes lies within a circumstellar disk ( najita et al .
our 2006 february observation of v1647 ori , however , reveals the surprising appearance of a blue - shifted absorption feature superimposed on the co emission lines .
this absorption feature was blue - shifted from line center by 30 km s@xmath0 and had a width of 20 km s@xmath0 .
roughly one year later , in 2006 december and 2007 february , the strength of the absorption decreased by at least a factor of 9 in such prominent lines as p30 ( table 2 ) and was no longer observed ( fig .
an excitation plot for the co emission lines observed in 2006 february ( after the star had returned to its preoutburst brightness ) implies that the effective rotational temperature of the emitting gas was 1400@xmath31200 k. to estimate the temperature of the absorption lines , we fit the observations with a synthetic spectrum , which we calculate by assuming the lines have a gaussian profile , the @xmath32co/@xmath33co ratio is 60 , and the gas is characterized by a single temperature .
the free parameters in our model are the intrinsic line width , @xmath34 , the co column density , n(co ) , and the rotational temperature of the gas , @xmath35 .
we present the predicted equivalent widths of @xmath32co absorption lines in table 2 and three models in figure 2 .
the parameters for each model are as follows .
red spectrum : @xmath36k , @xmath37 km s@xmath0 , n(co)@xmath3810@xmath39 @xmath2 ; green spectrum : @xmath40k , @xmath41 km s@xmath0 , n(co)=@xmath42 @xmath2 ; and blue spectrum , @xmath43k , @xmath44 km s@xmath0 , n(co)=5@xmath2210@xmath39 @xmath2 .
the three sets of parameters do an equally good job of fitting the low-@xmath7 @xmath32co lines
. if the gas is 1000k or hotter , the high-@xmath7 lines are too deep .
if the gas is 600k or cooler , the low-@xmath45co lines from 21272151 cm@xmath0 ( r9r15 ) become too strong .
our 1@xmath46 upper limit on the observed equivalent widths of @xmath33co lines not blended with other features is 0.006 cm@xmath0 . for gas at 600k , 700k and 1000k , the predicted strengths of these lines are @xmath60.018 cm@xmath0 , @xmath60.011 cm@xmath0 and @xmath60.006 cm@xmath0 , respectively .
the model with @xmath40k ( green ) provides the best overall fit , so we conclude that @xmath47 k , n(co)@xmath48 @xmath2 , and @xmath49 km s@xmath0 .
during quiescence , v1647 ori is a class i yso ( andrews et al .
however , its mass - loss rate during outburst was similar to a strongly accreting ctts ( vacca et al .
while outflows from cttss and haebes are common , fundamental ro - vibrational co emission lines with a blue - shifted absorption component have not been observed around any of the more than 300 such sources observed to date ( najita et al .
2000 , 2003 ; blake & boogert 2004 ; rettig et al .
2006 ; brittain et al . 2007 ; j.
brown private communication ) .
further , class i ysos such as gss 30 irs 1 , hl tau and rno 91 do not show ro - vibrational co emission lines with blue - shifted absorption components ( pontoppidan et al .
2002 , brittain et al .
2005 , and rettig et al .
2006 , respectively ) .
in contrast to these systems , the fuor v1057 cyg has shown blue - shifted overtone ro - vibrational co absorption lines ( hartmann et al .
2004 ) . while the similarity of the unusual co outflows is intriguing , there are important differences between fuors such as v1057 cyg and exors such as v1647 ori .
first , fuors have greater accretion rates ( @xmath610@xmath50 m@xmath24 yr@xmath0 ) and more extreme winds than exors ( hartmann & kenyon 1996 ) .
secondly , co is always and _ only _ detected in absorption toward fuors , and is thought to originate in the accretion disk ( calvet et al .
1993 ; calvet et al .
1991 ) . despite the significant differences between v1057 cyg and v1647 ori ,
both stars have undergone major eruptions that have generated outflows in co , suggesting that a different mass - loss mechanism may be at work in these systems than in other young stars . given the unusual nature of the outflowing co from v1647 ori , it is of interest to consider the mechanisms that could give rise to this phenomenon .
one possibility is that the absorbing co condensed out of the hot outflowing wind that was observed during the outburst . there were two outflow components noted in the h@xmath26 absorption feature : a variable component at 400 km s@xmath0 and a steady component at 150 km s@xmath0 ( fedele et al .
if the lower - velocity gas decelerated at a constant rate to @xmath51 km s@xmath0 ( the velocity of the outflowing co ) , by the time of our second co observation in 2004 july the wind would have expanded to 10 au .
there was no evidence of blue - shifted co absorption on this date ( fig .
1 ) . by 2006 february ,
when the absorption was observed , the moderate velocity wind would have expanded to nearly 40 au .
however , it seems unlikely that the wind could still retain a kinetic temperature of 700k at a distance of 40au from the star .
furthermore , it is not clear why co would condense out of this outflow to reveal warm , blue shifted absorption but not out of any of the other outflows with similar or even greater mass - loss rates . thus we conclude that this scenario is unlikely .
a second possibility is that the co absorption formed in a shell of material that was swept up by the atomic wind . in this case
the absorption did not appear until the column density of material was sufficient to produce measurable absorption , and the heating was the result of the interaction of the wind with the shell .
when the mass loss decreased at the end of the outburst phase , this heating was eventually shut down . if the co / h@xmath52 ratio in such a shell was similar to that of a dense molecular cloud , 1.5 @xmath2210@xmath50 , then the column density of gas was 2@xmath2210@xmath53 @xmath2 . adopting a normal interstellar extinction to gas ratio , i.e. , @xmath54 mag @xmath55 atom@xmath0 ( bohlin et al .
1978 ) , we find that the observed column density of co corresponded to @xmath620 mags of visible extinction .
this is a lower limit , as it is possible that selective dissociation of gas in the nebula could drive down the relative abundance of co. however , the extinction measured on the line of sight toward v1647 ori appears to have remained relatively unchanged over the entire course of the outburst at @xmath611 mags ( vacca et al . 2004 ; gibb et al . 2006 ) , of which 6.5 mags is due to the nebula ( fedele et al .
there is no evidence for an additional 1020 mags of extinction toward v1647 ori , and so we conclude that this scenario is also unlikely .
a final scenario we consider is that the co outflow was launched in response to the reorganization of the stellar magnetic field following the sharp drop in the accretion rate .
pre - main sequence stars tend to have kilogauss magnetic fields which mediate stellar accretion and outflows ( e.g. johns - krull et al .
this stellar magnetic field truncates the accretion disk where the ram pressure from accretion balances the magnetic pressure from the magnetosphere ( camenzind 1990 ) .
consequently , the truncation radius of the disk , @xmath56 , is inversely and nonlinearly proportional to the accretion rate , @xmath57 , and given by @xmath58 ( e.g. bouvier et al .
thus the two order of magnitude change in the stellar accretion rate experienced by v1647 ori would have resulted in the truncation radius being shifted by nearly a factor of four .
two years into the outburst , v1647 ori rapidly faded by a factor of 40 in the r@xmath59-band in just 180 days to return to its pre - outburst level ( fedele et al .
the mid - infrared flux also returned to its pre - outburst level , indicating that the drop in the optical / near - infrared lightcurve was due to the intrinsic fading of the source ( fedele et al .
this sharp drop in the luminosity of the source indicates that the accretion rate fell rapidly as the star returned to quiescence . in response to the drop in ram - pressure from the accretion flow ,
the truncation radius was pushed back by the magnetosphere .
evidence from simulations suggests that disk - magnetosphere systems tend to form outflows when the system is undergoing the greatest amount of dynamical rearrangement ( e.g. , fig .
7 in balsara 2004 ) .
it is possible that the realignment of the magnetic field as it pushed out against the circumstellar disk resulted in a warm , shortlived outflow , an outflow that is not observed toward any other cttss or haebes .
while virtually all accreting low - mass stars drive outflows , the co outflow from v1647 ori is highly unusual .
indeed , the transformation of centrally peaked ro - vibrational co emission lines to co emission lines with blue - shifted absorption is unique .
we suggest that the mechanism responsible for producing this outflow is distinct from the one that drives the outflow from typical cttss .
the coincidence between the rapid fading of v1647 ori and the subsequent observation of the co outflow hints at a connection .
better sampling of the fundamental ro - vibrational co spectrum of exors as they brighten and fade is crucial for determining whether this coincidence is significant .
the rapid and dramatic change in the accretion rate that characterizes the exor phenomenon provides an important opportunity to study the interplay between stellar accretion , the inner disk , and outflows .
this insight is key to reaching a satisfactory theoretical understanding of these events .
the data presented herein were obtained ( in part ) at the w.m .
keck observatory , which is operated as a scientific partnership among the california institute of technology , the university of california and nasa .
the observatory was made possible by the generous financial support of the w.m . keck foundation .
also based in part on observations obtained at the gemini observatory , which is operated by aura , under a cooperative agreement with the nsf on behalf of the gemini partnership .
the phoenix infrared spectrograph was developed and is operated by noao .
the phoenix spectra were obtained as part of program gs-2006a - dd-1 and gs-2006b - dd-1 .
s.d.b . performed this work ( in part ) under contract with jpl funded by nasa through the michelson fellowship program .
andrews , s. m. , rothberg , b. , & simon , t. 2004 , , 610 , l45 aspin , c. , barbieri , c. , boschi , f. , di mille , f. , rampazzi , f. , reipurth , b. , & tsvetkov , m. 2006 , , 132 , 1298 balsara , d. s. 2004 , , 151 , 149 blake , g. a. , & boogert , a. c. a. 2004 , , 606 , l73 bohlin , r. c. , savage , b. d. , & drake , j. f. 1978 , , 224 , 132 bouvier , j. , alencar , s. h. p. , harries , t. j. , johns - krull , c. m. , & romanova , m. m. 2007 , protostars and planets v , 479 briceo , c. , et al .
2004 , , 606 , l123 brittain , s. simon , t. , najita , j. r. , & rettig , t .w .
2007 , , 659 , 685 brittain , s. d. , rettig , t. w. , simon , t. , & kulesa , c. 2005 , , 626 , 283 brittain , s. d. 2004 , ph.d .
thesis , brittain , s. d. , rettig , t. w. , simon , t. , kulesa , c. , disanti , m. a. , & dello russo , n. 2003 , , 588 , 535 brown , j. m. , boogert , a. c. a. , salyk , c. , & blake , g. a. 2005 , protostars and planets v , 8513 2006 feb 17 & keck 2/nirspec & 21512119 , & 480 & 29 + & & 20181987 & & 28 + 2006 dec 01 & gemini s./phoenix & 20162006 & 960 & 6 + 2007 feb 01 & gemini s./phoenix & 20162006 & 4800 & 11 + 2007 feb 06 & gemini s./phoenix & 20162006 & 3360 & 8 + + r1 & 2150.86 & 4.41 @xmath31 0.60 & 4.52 & 4.69 & 4.81 & ( 3.4 ) & + r0 & 2147.08 & 3.49 @xmath31 0.60 & 3.30 & 3.98 & 4.29 & ( 3.4 ) & + p1 & 2139.43 & 3.40 @xmath31 1.00 & 3.27 & 3.96 & 4.26 & ( 1.7 ) & + p2 & 2135.55 & 5.17 @xmath31 0.80 & 4.47 & 4.64 & 4.77 & ( 1.7 ) & + p3 & 2131.63 & 4.33 @xmath31 0.60 & 5.05 & 4.97 & 5.02 & ( 2.2 ) & + p4 & 2127.68 & 5.03 @xmath31 0.60 & 5.39 & 5.18 & 5.17 & [email protected] & + p5 & 2123.70 & 6.21@xmath31 0.60 & 5.62 & 5.31 & 5.28 & [email protected] & + p6 & 2119.68 & 5.09@xmath31 1.00 & 5.79 & 5.40 & 5.35 & [email protected] & + p30 & 2013.35 & 3.57 @xmath31 0.36 & 4.45 & 3.42 & 3.25 & [email protected] & + p31 & 2008.53 & 2.42 @xmath31 0.24 & 4.23 & 3.15 & 2.97 & [email protected] & + p32 & 2003.67 & 1.44 @xmath31 0.94 & 4.00 & 2.85 & 2.66 & [email protected] & + p33 & 1998.78 & 2.86 @xmath31 0.64 & 3.75 & 2.53 & 2.32 & [email protected] & + | we present new observations of the fundamental ro - vibrational co spectrum of v1647 ori , the young star whose recent outburst illuminated mcneil s nebula .
previous spectra , acquired during outburst in 2004 february and july , had shown the co emission lines to be broad and centrally peaked similar to the co spectrum of a typical classical t tauri star . in this paper , we present co spectra acquired shortly after the luminosity of the source returned to its pre - outburst level ( 2006 february ) and roughly one year later ( 2006 december and 2007 february ) .
the spectrum taken in 2006 february revealed blue - shifted co absorption lines superimposed on the previously observed co emission lines .
the projected velocity , column density , and temperature of this outflowing gas was 30 km s@xmath0 , @xmath1 @xmath2 , and 700@xmath3 k , respectively .
the absorption lines were not observed in the 2006 december and 2007 february data , and so their strengths must have decreased in the interim by a factor of 9 or more .
we discuss three mechanisms that could give rise to this unusual outflow . |
the weak - localization approach ( wla ) has been a very popular tool to estimate whether electronic states in a weakly disordered system tend to localize or to delocalize on large scales .
moreover , it enables us to calculate the magnetoresistance in the presence of a weak magnetic field and weak scattering . a central result of
the wla is that on large scales there might be diffusion due to one or more massless modes .
this has been studied in great detail for conventional metals @xcite and more recently for graphene @xcite and for the surface of topological insulators @xcite , using a one - band projection for the two - band system .
t he existence of a diffusive mode , which is a necessary ( but not a sufficient ) condition for metallic behavior , has been debated for the one - band projected graphene model . it was found that either a single diffusive channel exists @xcite or no diffusion @xcite in the presence of generic disorder . on the other hand , the weak - scattering approach ( wsa ) , where transport properties are studied within the expansion in powers of @xmath0 ( @xmath1 is the scattering rate and @xmath2 is the bandwidth ) @xcite , a non - abelian chiral symmetry was identified , describing diffusion in two - band systems due to spontaneous symmetry breaking @xcite .
this is also the origin of a massless fermion mode found for 2d dirac fermions with a random gap in ref .
@xcite . in the wsa disorder fluctuations of the two - band model
are approximated by gaussian fluctuations around a saddle - point of the original model , expressed in terms of a functional integral @xcite .
the saddle point is equivalent to the self - consistent born approximation ( scba ) of the one - particle green s function , while the gaussian fluctuations are equivalent to the wla .
the latter consists of one - particle and two - particle diagrams which are partially summed up in terms of geometric series ( cf .
[ sect : wla ] ) . within the wsa
it is also possible to analyze the fluctuations with respect to the non - abelian chiral symmetry .
the projection onto these fluctuations generates a nonlinear field which allows us to go beyond the gaussian approximation within the expansion in powers of @xmath0 .
this idea is analogous to the nonlinear sigma model , derived originally for one - band hamiltonians by schfer and wegner @xcite .
the difference between the one - band and the two - band hamiltonians is that the former can be formulated either by a symmetric replica space or a supersymmetric fermion - boson @xcite , whereas the latter can be expressed in terms of a non - symmetric fermion - boson theory @xcite .
therefore , in the derivation of a nonlinear sigma model it is crucial to take the two - band structure into account .
a projection onto a single band would destroy the relevant symmetries of the system .
in more physical terms , the two - band structure is essential for supporting diffusion in a two - dimensional system , since it allows for klein tunneling .
the latter enables a particle in a potential barrier to transmute to a hole , for which the potential barrier is not an obstacle .
our aim is to establish a direct connection between the wla and the gaussian fluctuations around the saddle point for 2d dirac fermions with a random gap , and to provide a general discussion about the existence of diffusive modes due to ladder and maximally crossed contributions in two - band systems .
finally , these results will be employed to calculate the conductivity corrections , and the resulting conductivities will be compared with experimental measurements in graphene .
the results can also be applied to other 2d two - band systems such as the surface of topological insulators @xcite .
the paper is organized as follows . in sect .
[ sect : model ] we introduce a general description for the two - band hamiltonian and various types of random scattering .
the main ideas of the wla are discussed in sect .
[ sect : wla ] , which includes the self - consistent born approximation for the average one - particles green s function , the ladder and the maximally crossed contribution of the average two - particle green s function . in sect .
[ sect : special ] we study the long - range behavior of the average two - particle green s function for a one - band hamiltonian ( sect . [ sect : scalar ] ) and for the two - band hamiltonian ( sect .
[ sect : spinor_gf ] ) .
these results are used to calculate the conductivity ( sect .
[ sect : correction ] ) . and
finally , in sect .
[ sect : discussion ] we discuss the connection of the wla with the wsa , the robustness of the diffusion pole structure with respect to a one - band projection of the two - band hamiltonian and the symmetry properties of the inter - node scattering .
quasiparticles in a system with two bands are described by a spinor wavefunction .
the corresponding hamiltonian can be expanded in terms of pauli matrices @xmath3 . here
we will consider either a gapless hamiltonian h_0=h_1_1+h_2_2 or a gapped hamiltonian h_m = h_1_1+h_2_2+m_3 .
[ gap_ham ] the gapless hamiltonian changes its sign under a chiral transformation _ 3 h_0_3=-h_0 , [ ham0 ] which implies the continuous abelian chiral symmetry @xmath4 the situation is more subtle for @xmath5 because its transformation properties depends on the properties of @xmath6 .
we distinguish here two cases , namely @xmath7 ( dirac fermions , @xmath8 is the transposition , acting on real space ) , where _ 1 h_m^t_1=-h_m [ ham1 ] and @xmath9 ( e.g. , bilayer graphene ) , where _ 2 h_m^t_2=-h_m .
[ ham2 ] the transformation properties of the hamiltonians imply a relation between the wavefunctions in the upper and in the lower band .
in particular , eq . ( [ ham0 ] ) implies that @xmath10 , eq .
( [ ham1 ] ) implies that @xmath11 and eq .
( [ ham2 ] ) that @xmath12 .
disorder is included by an additional random term @xmath13 in the hamiltonians @xmath14 and @xmath5 . in the following
we will consider the case of @xmath15 ( scalar potential ) , @xmath16 ( random gap ) and independent random diagonal elements @xmath17 , @xmath18 , assuming that the matrix elements @xmath19 have mean zero and variance @xmath20 .
starting point of the wla is that the green s function @xmath21 of the hamiltonian @xmath14 is perturbed by @xmath22 and creates the green s function @xmath23 .
the latter relation is equivalent to the matrix identity @xmath24 which connects the unperturbed green s function @xmath25 with the perturbed green s function @xmath26 .
this equation can be iterated to give a geometric series g = g_0-g_0vg = g_0-g_0vg_0+g_0vg_0vg = ... = g_0_l0(-vg_0)^l . [ dyson2 ]
now we assume that @xmath22 is a random quantity with mean @xmath27 such that the averaged dyson equation becomes g = g_0+g_0vg_0vg .
when we assume that the correlation between the green s function and @xmath22 is weak , the factorization of the average @xmath28 is possible , which creates a linear equation for @xmath29 : gg_0+g_0vg_0vg , [ dyson3 ] whose solution reads @xmath30 .
this result is known as the born approximation with the self - energy @xmath31 . the self - consistent born approximation ( scba )
is provided by the replacement v_ig_0,ij v_jv_i v_jg_ij = : _ ij [ scba ] on the right - hand side of eq .
( [ dyson3 ] ) : g=(g_0 ^ -1-)^-1 . the two - particle green s function @xmath32 , created from the two one - particle green s functions @xmath33 ( e.g. , the adavanced and the retarded green s function ) , reads with the help of the corresponding dyson equations ( [ dyson2 ] ) ( using the summation convention in this section ) g^+_ijg^-_kl = t^-1_ik , mng^+_0,mjg^-_0,nl with t_ik , mn= _ im_kn - g^+_0,imv_mm g^-_0,knv_nn .
[ bse01 ] on the left - hand side is the two - particle green s function , while the right - hand side depends only on products of @xmath25 .
this equation is known as the bethe - salpeter equation .
now we can perform the average with respect to the random scatterers @xmath34 , assuming that this is a gaussian variable with zero mean . here
it is convenient to define @xmath35 such that t_ik , mn= _
im_kn -^+_im^-_kn .
the expansion of @xmath36 leads to a geometric series in @xmath37 . averaging this series with respect to a gaussian distribution
can be achieved by what is known as wick s theorem : the average of the product of random variable @xmath38 is expressed as a sum over all possible products of pairs @xmath39 .
this series includes a ladder contribution and a maximally crossed contribution @xcite ( cf .
[ fig : diagrams ] ) as special cases .
we also include the contribution from the iterated dyson equation of sect .
[ sect : dyson ] in terms of the scba and obtain eventually g^+_ijg^-_kl [ t^-1_ik , mn_l+ t^-1_ik , mn_m-_im_kn -t_ik , mn ] |g^+_mj|g^-_nl , [ bse02 ] where the last two terms on the right - hand side are introduced to avoid overcounting in the geometric series . and
maximally crossed contributions of @xmath40 , respectively.,title="fig:",width=188 ] and maximally crossed contributions of @xmath40 , respectively.,title="fig:",width=188 ] [ fig : diagrams ] beginning with the ladder contribution , we obtain t^-1_ik , mn_l = ( * 1 * -t)^-1_ik , mn with t_m_1n_1,m_2n_2 = ^+_m_1m_1v_m_1m_2|g^-_n_1n_1v_n_1n_2 . [ series2 ] then the maximally crossed contribution is created by re - arranging the order of factors in the geometric series which after averaging results in t^-1_ik , mn_m = ( * 1 * -)^-1_in , mk with _
m_1m_1,m_2m_2 = ^+_m_1n_1v_n_1m_2 contribution to the maximally crossed contribution is achieved simply through changing @xmath41 by the transposition @xmath42 . in the following
we will discuss the two - particle green s functions of ( [ bse02])([series4 ] ) and the scba for the two - band hamiltonians of sect . [ sect : model ] with an additional random term @xmath13 , which is either a random scalar potential , a random gap or independent random diagonal elements .
for simplicity it is assumed that the random terms are spatially uncorrelated .
the general expressions in ( [ series2 ] ) and in ( [ series4 ] ) shall now be applied to specific cases . for a diagonal random matrix @xmath43
we obtain t_m_1n_1,m_2n_2 = before we start to discuss the two - band hamiltonians , the simpler one - band hamiltonian with fourier components @xmath44 is used to explain briefly the main ideas of the wla . for this case
we consider the following coordinates in real space : @xmath45 , @xmath46 . then the ladder and the maximally crossed contributions read for uncorrelated disorder @xmath47 t_r_1r_2,r_3r_4=g|g^+_r_1r_3|g^-_r_2r_3_r_3,r_4 , _
r_1r_2,r_3r_4=g|g^+_r_1r_3|g^-_r_4r_3_r_2,r_3 , and in the geometric series @xmath48 ( @xmath49 ) only t_r_2 r_3:=t_r_2r_2,r_3r_3=g|g^+_r_2r_3|g^-_r_2r_3 , _
r_2 r_3:=_r_2r_3,r_3r_2=g|g^+_r_2r_3|g^-_r_2r_3 [ t - matrix2 ] contributes .
( [ bse02 ] ) reads g^+_rrg^-_rr _ r[(*1 * -t)^-1_rr+(*1 * -)^-1_rr-_rr
-t_rr]|g^+_rr|g^-_rr .
moreover , with ( [ t - matrix2 ] ) the extra factor ^+_rr|g^-_rr=t_rr=_rr , can be replaced by @xmath50 and @xmath51 , respectively , to provide g^+_rrg^-_rr ( * 1 * -t)^-1_rr+(*1 * -)^-1_rr-(t^2_rr+t_rr+2_rr ) . the green s function is symmetric in the absence of a magnetic field , such that the fourier components of the hamiltonian @xmath44 satisfy the relation @xmath52 .
then there is no difference between ladder and maximally crossed contributions .
we consider @xmath53 for the long - range behavior and obtain from the expansion in powers of the momentum @xmath54
t_q~g_r self - energy @xmath55 : @xmath56 , @xmath57 with @xmath58 .
this gives @xmath59 , such that @xmath60 becomes the diffusion propagator ~ .
this result is remarkable because it implies that @xmath60 diverges like @xmath61 for @xmath62 and @xmath63 , reflecting the well - known massless two - particle mode for diffusion @xcite . for the spinor hamiltonian @xmath5 of eq .
( [ gap_ham ] ) we introduce the coordinate @xmath64 and the pauli matrix index @xmath65 and adopt the same procedure as for the scalar case . with the correlation @xmath66 and with @xmath56 , @xmath57 for @xmath67 , where @xmath68 is the renormalized fermi energy ( i.e. , the bare fermi energy which is shifted by the real part of the self - energy @xmath69 ) and @xmath1 is the scattering rate ( i.e. , the imaginary part of the self - energy ) .
now we use the fourier representation of @xmath5 and get with @xmath70 for the green s function g_0,k(z ) = ( z_0+h_1_1+h_2_2+m_3 ) , [ 1pgf0 ] and t_q;ab , cd = g_cd_k g_0;k , ac g_0;q+k , bd^ * , _
q;ab , cd = g_cb_k g_0;k , acg_0;q+k , db^ * [ matrices1 ] . for random scalar potential @xmath71 ( random gap @xmath72 ) the prefactors read @xmath73 ( @xmath74 ) , and for independent random diagonal elements @xmath75 , @xmath76
. then we get from eq .
( [ matrices1 ] ) for @xmath77 the matrices t_0= , _
0= , [ 2pgf_mb ] where @xmath78 for a random gap , @xmath79 for random scalar potential , @xmath80 for independent random diagonal elements . for dirac fermions we have @xmath81 and
the matrix elements are the following expressions : @xmath82 i=_k , = g_k . using the scba
we obtain the integral _
k=- , which implies @xmath83 .
thus , besides two independent diagonal matrix elements , we get two eigenvalues for the non - diagonal @xmath84 submatrices of @xmath85 and @xmath86 , namely _
l^=1- , _ m^=1-[eigenvals ] with the parameters @xmath87 , @xmath88 and @xmath89 , @xmath90 . for @xmath80 the eigenvalues @xmath91
are always massive : _ m^=1= and for @xmath92 we consider two special cases , the behavior at the dirac node and the gapless case : _ ( i ) at the dirac node @xmath93 _ : for the hamiltonian @xmath5 of eq .
( [ gap_ham ] ) we get the parameter @xmath94 , and the eigenvalues of eq .
( [ eigenvals ] ) now read _
l^=1- gi(^2+m^2 ) ] , _ m^=1-s(^2-m^2)gi .
[ eigenv1 ] thus , the ladder contribution @xmath95 is always massless , in contrast to @xmath96 and the maximally crossed contributions , which are all massive . _
( ii ) gapless spectrum @xmath97 _ : for the hamiltonian @xmath14 we have @xmath98 , and the eigenvalues read _
l^=1-g|z|^2i(1-g|z|^2i ) , _ m^=1-sg|z|^2i(1-g|z|^2i ) , [ eigenv2 ] such that there is a massless mode @xmath99 for any @xmath100 , like for the dirac node , and an additional massless mode from the maximally crossed contribution @xmath101 for @xmath79 .
these results are summarized in table [ m_eigenvalues ] .
the findings of the previous section can be used to evaluate the correlation function as g_rr,abg^*_rr,cd ( * 1 * -t)^-1_rr;ac , bd+(*1 * -)^-1_rr;ad , bc -(t^2_rr;ac , bd + t_rr;ac , bd+2_rr_ac_bd ) . [
2pgf2 ] for the long - range behavior with @xmath102 it is sufficient to consider the two - particle propagators @xmath103 and @xmath104 for @xmath62 .
then we can focus on the massless ( diffusion ) modes as the most important contributions to get ~ , [ diff_prop ] where the second expression is obtained by the analytic continuation @xmath105 .
the diffusion coefficients @xmath106 are obtained from the @xmath54 expansion of the eigenvalues of ( [ matrices1 ] ) as @xmath107+o(g^2)\ ] ] and additionally for @xmath79 @xmath108+o(g^2 ) \ .\ ] ] both coefficients agree , at least up to terms of @xmath109 , and give us d_t = d_=(1+)+o(g^2 ) ( =/ ) .
[ diff_const ] .vanishing eigenvalues of @xmath110 and @xmath111 ( cf .
( [ 2pgf_mb ] ) , ( [ eigenv1 ] ) , ( [ eigenv2 ] ) ) . [ cols="^,^,^,^",options="header " , ]
next , the results of the wla will be used to evaluate the quantum corrections to the boltzmann - drude conductivity .
the latter is usually calculated in terms of the current - current correlation function for the real part of the conductivity @xcite _ ~ ( j_gj_g^ ) .
the advantage of using this expression is that the current - current correlation function @xmath112 is closely related to the action of the corresponding nonlinear sigma model @xmath113 @xcite , since the current operator @xmath114 of a conventional one - band model is proportional to the momentum operator @xmath115 .
therefore , the renormalization of the parameter @xmath50 is similar to the renormalization of the current - current correlation function .
this relation , however , breaks down for dirac fermions , where @xmath114 is proportional to the pauli matrix @xmath116 . for this reason it is not obvious that the renormalization of the nonlinear sigma model is linked to the renormalization of the current - current correlation function . in this case
it is better to use an alternative kubo formula , which is based on the density - density correlation function @xcite : _
= -^2_r r_^2tr_2g_r0g^_0r .
[ conductivity ] this expression is closely related to diffusion and the einstein relation @xcite .
the classical approximation assumes a weak correlation between the two green s functions @xmath26 , @xmath117 such that we can factorize the expectation value as @xmath118 and obtain the boltzmann - drude conductivity as = -^2_r r_^2tr_2g_r0g^_0r .
furthermore , the average one - particles green s functions are evaluated within the scba as @xmath119 . for the gapless case @xmath97 and for the parameter @xmath120 we can write _
r r_^2tr_2g_r0g^_0r_r r_^2tr_2|g_r0|g^_0r .
[ 2pgf3 ] in particular , for @xmath121 the boltzmann - drude conductivity reads = .
this expression is obviously not the boltzmann - drude conductivity of a conventional metal with one - band hamiltonian . at the dirac node @xmath93
it has a frequency independent conductivity @xmath122 and decreases monotonically from @xmath123 to zero as we move the fermi energy @xmath68 away from the dirac node .
this indicates a cross - over from the optical conductivity of the two - band model at the dirac node to the boltzmann - drude behavior of a conventional metal , where the optical conductivity is much lower than in the two - band case . here
it should be noticed that there are additional corrections for @xmath121 , which increase the optical conductivity to @xmath124 .
these are not taken into account here , since they disappear in the dc limit @xcite . for @xmath125 ,
on the other hand , the boltzmann approximation is invalid , which is reflected by a negative boltzmann - drude conductivity that vanishes like @xmath126 .
for this regime we must consider the corrections @xmath127 which can be evaluated from eq .
( [ 2pgf2 ] ) as _ 0 ^ 2_r(r_-r_)^2g_rr,abg^*_rr,ab _ 0 ^ 2_r(r_-r_)^2 [ ( * 1 * -t)^-1_rr;aa , bb+(*1 * -)^-1_rr;ab , ba ] . [ 2pgf2a ] with the help of the propagator in eq .
( [ diff_prop ] ) and the diffusion coefficient ( [ diff_const ] ) the conductivity corrections then read = ^2|_q=0 = 2d = ( 1+)+o(g^2 ) ~(1+^2/3 ) .
[ ddc ] there is an additional factor 2 for @xmath79 due to the extra massless mode from @xmath101 in that case .
this result , which is depicted in fig .
[ fig : conductivity ] , agrees with previous calculations based on the wsa @xcite as well as with the v - shape conductivity with respect to @xmath128 in graphene @xcite . from the expression in eq .
( [ ddc]).,width=377 ]
our main result is that the ladder and the maximally crossed contributions are quite different for the spinor hamiltonians , in contrast to their agreement for the scalar hamiltonian of sect .
[ sect : scalar ] .
this situation is reminiscent of the bose - fermi functional representation of the average two - particle green s function in the case of dirac fermions with a random gap , where it was observed that the bosonic and the fermionic propagators are distinct @xcite . similar to the expressions in eq .
( [ 2pgf_mb ] ) , the inverse bosonic two - particle propagator is also a @xmath129 matrix for @xmath130 @xmath131 while its fermionic counterpart reads @xmath132 a straightforward calculation shows that only the fermionic propagator has a massless mode , very similar to @xmath133 with @xmath78 of the wla .
therefore , the comparison of the wsa with the wla sheds some light on the role of bose - fermi ( or super- ) symmetry breaking , which can be associated with the difference between ladder diagrams and maximally crossed diagrams in the wla . in particular , the massless mode of the wla agree with the massless fermion mode of the wsa , where the latter is caused by a spontaneous breaking of a non - abelian chiral symmetry @xcite . it remains to discuss whether or not the structure of the diffusion poles is robust with respect to approximations . in the one - band projection of the one - particle green s function
the pole of the second band has been ignored @xcite , since it is energetically too far away from the fermi surface .
for @xmath134 we have g_0;k , ab = u_1a^*u_1b+u_2a^*u_2b u_1a^*u_1b with u= and @xmath135 , which provides the expression g_0(_0-_1k_1/k-_2 k_2/k ) .
for the projected green s function the matrices ( [ 2pgf_mb ] ) then read t_0= , _ 0= , i=_k , where we get @xmath136 from the scba .
thus the eigenvalues of @xmath110 are @xmath137 and the eigenvalues of @xmath111 are @xmath138 . like in the case of the two - band green s function of sect .
[ sect : spinor_gf ] there is one massless mode for @xmath110 for any value of @xmath100 and one massless mode for @xmath111 if @xmath79 .
this indicates that the result of the one - band projection preserves the structure of two - band result of table [ m_eigenvalues ] in terms of the number of massless modes .
the agreement of the results from the one - band projected green s function and the two - band green s function reflects the fact that the type of diffusive modes is not sensitive to the scattering to the second band .
finally , we briefly discuss the effect of inter - node scattering for the hamiltonian @xmath14 . for this purpose
we introduce the extended hamiltonian _ 0= , which describes inter - node scattering by the random scattering terms @xmath71 .
now we could evaluate the inverse two - particle green s functions within the wla and study the vanishing eigenvalues , which would be associated with diffusive behavior .
alternatively , we can also start from the symmetry argument and analyze the underlying chiral symmetry whose spontaneous breaking would create a massless mode , analogously to the treatment of random scattering terms in ref .
@xcite . following this concept
, we first realize that the hamiltonian @xmath139 changes its sign under the transformation in eq .
( [ ham0 ] ) _ 0s|h_0s=-|h_0 , s= . according to the general procedure of ref .
@xcite , this leads for the hamiltonian @xmath140 to the non - abelian chiral symmetry e^sh_0e^s = h_0 , s= for independent continuous parameters @xmath141 , @xmath142 , since @xmath143 and @xmath144 anticommute .
this symmetry is spontaneously broken due to the scattering rate @xmath1 , causing the appearance of a massless mode . here
it should be noticed though that the appearance of a massless mode is only a necessary but not a sufficient condition for a diffusive behavior because the interaction of the nonlinear symmetry fields generated by @xmath141 , @xmath142 , could lead to anderson localization @xcite .
the calculation of the ladder and maximally crossed contributions of the average two - particle green s function has revealed a characteristic diffusion pole structure . depending on the type of randomness
, the ladder contributions always have one diffusion pole , provided the fermi energy is at the dirac node or away from the dirac node but in the absence of a gap .
moreover , for a random scalar potential there is an additional diffusion pole from the maximally crossed contributions .
all these results require the existence of a nonzero scattering rate , obtained as a solution of the self - consistent born relation ( scba ) .
no diffusion pole have been found away from the dirac node in the presence of a one - particle gap .
the diffusion pole structure of the wla is identical to that of the wsa , at least for the case of a random gap .
this enabled us to identify the origin of the diffusion poles with massless modes which are created by spontaneously broken symmetries .
these are chiral symmetries , associated with the symmetry of the two bands .
such a symmetry exists also in the presence of inter - node scattering .
further work is necessary though , to compare the relation between the wla and the wsa for other types of disorder scattering .
the diffusion pole structure of the two - particle green s function is preserved when we employ a one - band projection of the one - particle green s function by removing one pole of the latter .
although this projection changes the form of the diffusion coefficient ( cf .
@xcite ) , it may serve as a good approximation that reduces the computational effort significantly . as already mentioned in the introduction ,
the existence of a diffusion pole is a necessary condition but does not guarantee a metallic behavior .
higher order terms beyond the ladder and maximally crossed contribution can destroy diffusion and eventually lead to anderson localization .
this was studied recently in terms of a strong scattering expansion @xcite , which revealed an exponential decay when the scattering rate @xmath1 is larger than the band width @xmath2 . | we study a weakly disordered 2d electron gas with two bands and a spectral node within the weak - localization approach and compare its results with those of gaussian fluctuations around the self - consistent born approximation .
the appearance of diffusive modes depends on the type of disorder .
in particular , we find for a random gap a diffusive mode only from ladder contributions , whereas for a random scalar potential the diffusive mode is created by ladder and by maximally crossed contributions .
the ladder ( maximally crossed ) contributions correspond to fermionic ( bosonic ) gaussian fluctuations .
we calculate the conductivity corrections from the density density kubo formula and find a good agreement with the experimentally observed v - shape conductivity of graphene . |
liquids cooled toward their glass transition exhibit remarkable dynamical behavior @xcite .
the initial slowing down of many liquids at temperatures well above their glass transition temperature @xmath0 can be described to great extent by the mode coupling theory @xcite , which predicts diverging relaxation times at a dynamical critical temperature @xmath1 despite the absence of a diverging or even growing static correlation length @xcite . at the same time , experiments in the temperature range @xmath2 , and simulations in the range @xmath3 , have shown that it is possible to select subsets of particles in the liquid ( or monomers in the case of polymers ) that move differently from the bulk on time scales less than the structural relaxation time @xcite .
the question then arises as to whether glass - forming liquids exhibit _ spatially heterogeneous _ dynamics , and if so , is there a growing length scale associated with this dynamical heterogeneity ? we have employed three complementary computational approaches to address this question : 1 .
_ subset approach .
_ in this approach @xcite , we monitor the displacement of each particle in a time window @xmath4 , rank the displacements from largest to smallest , and examine spatial correlations within subsets of particles exhibiting either extremely large or extremely small displacements . in this way
, we have demonstrated that particles of similar mobility form clusters that grow with decreasing temperature @xmath5 .
moreover , particles exhibiting the largest displacements within a time window in the late-@xmath6/early-@xmath7 relaxation regime ( `` mobile '' particles ) form low - dimensional clusters that grow and percolate at the mode coupling transition temperature @xmath1 .
2 . _ displacement - displacement correlation function approach .
_ in this approach @xcite , we monitor the displacement of each particle in a time window @xmath4 as above , and calculate a bulk , equilibrium correlation function that measures the correlations between the local fluctuations in the particle displacements . because this approach does not rely on defining subsets via arbitrary thresholds or cutoffs , it may prove more amenable to future analytical treatment .
first passage time approach .
_ in this approach @xcite , we monitor the first passage time @xmath8 for each particle to move a distance @xmath9 , and then for a given @xmath9 , examine spatial correlations between the different @xmath8 values either by grouping particles into subsets of similar @xmath8 , or by calculating a @xmath8-@xmath8 correlation function using all the particles .
this approach shows that particles with similar first passage times to travel any @xmath9 are spatially correlated .
it also allows us to identify particular values of @xmath9 for which the first passage times are most correlated , and thereby extract detailed information on , e.g. , the cage size . in this paper
, we present results from the first two approaches outlined above , and show that as an equilibrium lennard - jones liquid approaches @xmath1 , correlations between local fluctuations in particle displacements develop , grow , become increasingly long - ranged , and appear to diverge at @xmath1 .
similar results for a simulated polymer blend are presented elsewhere @xcite .
consider a liquid confined to a volume @xmath10 , consisting of @xmath11 identical particles , each with no internal degrees of freedom .
let the position of each particle @xmath12 be denoted @xmath13 .
the motion of a particle @xmath12 may be described by its displacement @xmath14 over some interval of time @xmath4 , starting from a reference time @xmath15 .
the question of whether particle motions in a liquid are correlated can be addressed by modifying the usual definition of the density - density correlation function @xmath16 ( which measures spatial correlations of fluctuations in local density away from the average value @xcite ) so that the contribution of each particle @xmath12 to @xmath16 is weighted by @xmath17 .
thus , we define a `` displacement - displacement '' correlation function @xcite , @xmath18 \ , \bigl [ u({{\mbox{\boldmath$r$\unboldmath}}'},t,\delta t ) - \langle u \rangle \bigr ] \bigl\rangle,\ ] ] where @xmath19 and where @xmath20 denotes an ensemble average .
@xmath21 measures the correlations of fluctuations of local displacements away from their average value .
since we are considering an equilibrium liquid , and because @xmath22 is defined as an ensemble average , @xmath22 does not depend on the choice of the reference time @xmath15 .
for the same reasons , @xmath23 only depends on @xmath4 for a homogeneous , equilibrium liquid .
@xmath21 can be separated into self and distinct parts , so that we can identify a spatial correlation function @xmath24 analogous to the static pair correlation function @xmath25 conventionally used to characterize the structure of a liquid @xcite : @xmath26,\ ] ] where @xmath27 here we have defined the `` total displacement '' @xmath28 as @xmath29 , and its ensemble average @xmath30 .
we have also defined @xmath31 , and @xmath32 . in a constant-@xmath11 ensemble , @xmath33 and @xmath34
are readily evaluated as @xmath35 and @xmath36 .
note that @xmath37 is different from the distinct part of the time - dependent van hove correlation function @xmath38 , because it correlates information between pairs of particles using information about the position of each particle at _ two _ different times .
thus @xmath37 is in some sense a four - point correlation function , whereas @xmath38 is a two - point correlation function .
a `` structure factor '' for the particle displacements can be defined as @xmath39 \biggr \rangle.\end{aligned}\ ] ] the fluctuations of @xmath28 are related to the volume integral of @xmath21 : @xmath40 ^ 2 \bigr \rangle = \int d{{\mbox{\boldmath$r$\unboldmath } } } \ , g_u({{\mbox{\boldmath$r$\unboldmath}}},\delta t ) \equiv \bigl \langle u \bigr \rangle \bigl \langle u \bigr \rangle kt \kappa_u , \label{flucuu}\ ] ] where @xmath41 is boltzmann s constant .
recall that the volume integral of the density - density correlation function @xmath16 is proportional to the isothermal compressibility @xmath42 , which diverges at a conventional critical point because @xmath16 becomes long - ranged .
likewise , the fluctuations in @xmath28 will provide information regarding the range of @xmath21 .
thus , in analogy with conventional critical phenomena , we have defined an isothermal `` susceptibility '' @xmath43 . for a fixed choice of @xmath4 ,
note that if the displacement were always the same for every particle , then @xmath24 and @xmath25 would be identical for all @xmath44 ( and @xmath45 and @xmath46 would be identical for all @xmath47 ) .
hence , it is deviations of @xmath24 ( @xmath45 ) from @xmath25 ( @xmath46 ) that will inform us of correlations of fluctuations of local displacements away from the average value , that are in excess of those that would be expected based on a knowledge of @xmath25 or @xmath46 alone .
we have measured the spatial correlations in particle displacements using data obtained @xcite from a molecular dynamics simulation of a model lennard - jones glass - former .
the system is a three - dimensional binary mixture ( 80:20 ) of 8000 particles interacting via lennard - jones interaction parameters @xcite .
we analyze data from seven @xmath48 state points that lie on a line in the @xmath49 plane , approaching the mode - coupling dynamical critical temperature @xmath50 at a pressure @xmath51 and density @xmath52 @xcite ( all values are quoted in reduced units @xcite ) . the highest and lowest temperature state points simulated are @xmath53 and @xmath54 , respectively . following equilibration at each state point , the particle displacements are monitored in the nve ensemble for up to @xmath55 lennard - jones time units ( 25.4 ns in argon units ) for the coldest @xmath5 .
complete simulation details may be found in @xcite . for all seven state points ,
a `` plateau '' exists in both the mean square displacement and the intermediate scattering function @xmath56 as a function of @xmath15 @xcite .
the plateau separates an early time ballistic regime from a late time diffusive regime , and indicates `` caging '' of the particles typical of low @xmath5 , high density liquids .
the same model liquid , simulated along a different path toward the same mode coupling critical point , was found to be well described by the ideal mode coupling theory @xcite .
for the present ( larger ) simulation , the @xmath7-relaxation time @xmath57 that describes the long - time decay of @xmath56 at the value of @xmath58 corresponding to the first peak in the static structure factor @xmath46 increases by 2.4 orders of magnitude , and follows a power law @xmath59 , with @xmath60 and @xmath61 .
the diffusion coefficient @xmath62 is found to behave as @xmath63 .
the simulated liquid states analyzed here therefore exhibit the complex bulk relaxation behavior characteristic of a supercooled liquid approaching its glass transition .
throughout the following subsection , all quantities are calculated using all 8000 particles in the system .
results do not change substantially if the minority particles are not included in the analysis . in the cluster analysis in subsection _ [ seccluster ]
_ , only the majority particles are included in the analysis .
however , our results do not change quantitatively when the minority particles are included .
recall that in the mode coupling analysis of this model liquid by kob and andersen , the same mode coupling transition temperature was found for the majority particles as for the minority particles @xcite .
@xmath64 is shown for different @xmath5 in fig .
[ figsuu ] , calculated for a value of @xmath4 chosen on the order of @xmath57 for each @xmath5 . for intermediate and large @xmath58 , @xmath65 coincides with the static structure factor @xmath66\bigr \rangle$ ] ( see inset ) .
however , for @xmath67 a peak in @xmath65 develops and grows with decreasing @xmath5 , suggesting the presence of increasing long range correlations in @xmath21 .
no such growing peak at @xmath68 appears in the static structure factor @xmath69 ( cf .
inset ) , indicating the absence of long range correlations in the particle positions .
to indeed , a comparison of @xmath37 and @xmath70 shows that at every @xmath5 simulated , the particle displacements are more correlated than their positions . at the lowest @xmath5
, this excess correlation persists for values of @xmath71 up to 6 interparticle distances @xcite . in fig.[figsuu
] @xmath4 is chosen on the timescale of the @xmath7-relaxation time .
the significance of the time interval @xmath4 can be seen by examining the `` susceptibility '' @xmath72 .
here we have calculated @xmath73 directly from the fluctuations in the total displacement @xmath28 .
[ figchidt ] demonstrates four important points : ( i ) as @xmath74 , @xmath75 .
note that in the limit @xmath76 , @xmath77 is equal to the magnitude of the instantaneous velocity .
thus correlations can not be observed by looking at a `` snapshot '' of the system , that is , by measuring correlations in the instantaneous velocity @xcite .
( ii ) initially , as @xmath4 increases , @xmath43 increases .
thus the spatial correlation of particle displacements develops over time .
( iii ) there is a time window @xmath78 during which @xmath43 is maximum .
both the maximum value of @xmath43 and corresponding time window @xmath78 increase with decreasing @xmath5 .
( iv ) as @xmath79 , @xmath43 decreases .
thus in the diffusive regime , where the particles act like brownian particles , there are no correlations in the particle motions . to fig .
[ figdt ] shows that the time @xmath78 at which the fluctuations in the particle displacements are a maximum follows a power law with t : an excellent fit to the form @xmath80 is obtained when @xmath81 , and yields @xmath82 , where the highest value of @xmath83 is obtained with the lower bound on @xmath1 .
our estimated value for @xmath83 differs from that describing the divergence at @xmath1 of the structural relaxation time @xmath57 ( @xmath84 ) , but ( within our numerical uncertainty ) can not be distinguished from the exponent governing the apparent vanishing of the diffusion coefficient @xmath62 ( @xmath85 ) . to in fig [ figchit ]
we show the @xmath5-dependence of @xmath86 .
we find that @xmath87 grows monotonically with decreasing @xmath5 , indicating that the range of the correlation measured by @xmath88 is growing with decreasing @xmath5 .
the data can be fitted extremely well with a power law @xmath89 with @xmath90 , and yields @xmath91 .
( as shown in the figure , a larger ( closer to curie - weiss - type ) exponent can be obtained by fitting the data with a power law that diverges at a slightly lower value of @xmath1 . )
thus the previously determined mode coupling critical temperature coincides , within our numerical error , with the temperature where @xmath92 becomes long ranged . to the displacement - displacement correlation function measures the tendency for particles of similar mobility to be spatially correlated ( cf .
[ figextreme ] ) . to to extract a correlation length from the data , we have attempted to fit @xmath93 using an orstein - zernike form , @xmath94 .
this form assumes that @xmath95 is asymptotically proportional to @xmath96 , where @xmath97 is the correlation length @xcite .
we find that this form fits well at the highest @xmath5 but fails to fit the data on approaching @xmath1 .
the data can instead be fitted at all @xmath5 using the more general form @xmath98 .
the best fit gives @xmath99 and 3.54 for our highest and lowest @xmath5 , respectively .
however , because @xmath100 depends on @xmath5 , @xmath97 does not show a definite trend with temperature , giving values of roughly three particle diameters .
larger simulations may be required to accurately determine the correct functional form for @xmath101 at small @xmath58 .
nevertheless , the data show unambiguously that as @xmath102 , spatial correlations between the displacements of particles arise and grow and become long - ranged at @xmath1 . to obtain further evidence for a growing dynamical length scale ,
we turn now to the subset method described in the introduction @xcite .
examination of the distribution of particle displacements ( the self part of the usual time - dependent van hove correlation function @xcite ) immediately shows the problems inherent in defining subsets of particles of extremely low mobility . over the time range coincident with the plateau in the mean square displacement ,
this function exhibits a narrow peak containing most of the particles , and a long tail to large distances containing a small percentage of the particles .
the non - gaussian parameter @xmath103 has been used to identify a time when this tail is most pronounced @xcite . at that time , approximately 95% of the particles are contained in the peak , and about 5% are contained in the tail , regardless of temperature . because of this tail , the most mobile particles in the liquid `` distinguish '' themselves from the bulk , while it is more difficult to define a threshold with which to identify particles exhibiting extremely small displacements ( `` immobile '' particles ) . by examining the _ maximum _ displacement achieved by a particle in a time window @xmath4 , a somewhat better definition of immobility can be achieved .
it was shown in ref .
@xcite that particles exhibiting the smallest displacement defined in this way are spatially correlated , and form relatively compact clusters . using this definition of maximum displacement , which for the most mobile particles gives the same results as the simple definition of displacement used above , we find that highly mobile particles also cluster .
these clusters are very ramified , and are composed of smaller `` strings '' of particles that follow one another ( cf .
[ figdino ] ) . to in this subsection , we are interested in the growth of clusters of highly mobile particles with decreasing temperature . to this end
, we have constructed clusters of nearest neighbor particles whose maximum displacement in a time window @xmath104 falls within the top 3% , 5% and 7% of all displacements @xcite .
the distribution of clusters of size @xmath105 constructed using the top 5% of the particles is shown in the inset of fig .
[ figperc ] for four different @xmath5 . also shown in fig .
[ figperc ] is the @xmath5-dependence of the mean cluster size @xmath106 for each subset @xcite .
we find that for each subset @xmath107 with @xmath108 , and @xmath109 , all close to @xmath1 .
( note that as the number of particles contained in the subset increases , @xmath110 _
decreases_. ) for the subset containing 5% , the data fall on a straight line when plotted log - log against @xmath111 .
thus , we conclude that @xmath1 appears to coincide with a percolation transition of the most mobile particles in the liquid . to the presence of clusters of mobile particles whose size grows with decreasing temperature contributes to the growing range of the displacement - displacement correlation function
. thus we can use the mean cluster size to give a rough estimate of the length scale on which particle motions are correlated . in our coldest simulation ,
the mean cluster size is approximately @xmath112 particles .
since the clusters have a fractal dimension of approximately 1.75 @xcite , that implies a mean radius of gyration larger than 3 particle diameters .
the largest cluster observed at @xmath113 contains over 100 particles , and thus has a radius of gyration that exceeds the size of our simulation box at that @xmath5 .
it is important to note that while the mean size of clusters of mobile particles appears to diverge at @xmath1 , the smaller , cooperatively rearranging `` strings '' that make up these clusters also grow with decreasing @xmath5 , _ but are still finite at @xmath1 _ @xcite .
in the temperature regime studied here , the distribution of string lengths is exponential , and the mean string length appears to diverge at a temperature substantially lower than @xmath1 .
thus while the mode coupling transition may coincide with a percolation transition of clusters of highly mobile particles , our data is compatible with the idea of adam and gibbs that the ideal glass transition at @xmath114 is associated with the growth and possible divergence of the minimum size of a cooperatively rearranging region ( `` strings '' in the present simulation ) .
this idea will be further explored elsewhere @xcite .
in this paper , we have defined a bulk correlation function that quantifies the spatial correlation of single - particle displacements in a liquid , and we have examined spatial correlations within subsets of highly mobile particles . while the first approach has the advantage of allowing a direct calculation of spatial correlations in local particle motions without having to define arbitrary subsets , the second approach has the advantage of permitting a geometrical analysis of clusters of highly mobile and highly immobile particles . using both approaches ,
we have shown in computer simulations of an equilibrium lennard - jones liquid that the displacements of particles are spatially correlated @xcite over a range and time scale that both grow with decreasing @xmath5 as the mode coupling transition is approached .
while mode coupling theory currently makes no predictions concerning a growing dynamical correlation length , calculation of the vector analog of the displacement - displacement correlation function should be tractable within the mode - coupling framework .
finally , the displacement - displacement correlation function has allowed us to identify a fluctuating dynamical variable u whose fluctuations become longer ranged and appear to diverge at @xmath1 . in this
regard , @xmath28 is behaving much like a static order parameter on approaching a second - order static critical point , suggesting the possibility that we can obtain insights into the nature of glass - forming liquids using an extension to dynamically - defined quantities of the framework of ordinary critical phenomena .
999 r. bhmer , r.v . chamberlain , g. diezemann , b. geil , a. heuer , g. hinze , s.c .
kuebler , r. richert , b. schiener , h. sillescu , h.w .
spiess , u. tracht and m. wilhelm 1998 _ j. non - cryst .
solids _ * 235 - 237 * 1 , and the many references therein .
the lennard - jones interaction parameters @xmath115 and @xmath116 are given by : @xmath117 , @xmath118 , @xmath119 , @xmath120 , @xmath121 , @xmath122 .
lengths are defined in units of @xmath123 , temperature @xmath5 in units of @xmath124 , and time @xmath15 in units of @xmath125 .
both types of particles are taken to have the same mass .
indirect evidence for a growing length from studies of glassforming liquids or polymers in confined geometries is reported in , e.g. m. arndt , et al .
lett . * 79 * , 2077 ( 1997 ) ; b. jerome and j. commandeur , nature * 387 * , 589 ( 1997 ) ; and p. ray and k. binder , europhysics lett . * 27 * , 53 ( 1994 ) .
a `` hydrodynamic '' correlation length was calculated for a lj liquid by r. mountain in _ supercooled liquids : advances and novel applications _ , acs symposium series * 676 * , 122 ( 1987 ) .
other recent computational investigations of spatially heterogeneous dynamics include y. yamamoto and a. onuki 1998 _ phys .
e _ * 58 * 3515 ; d. perera and p. harrowell 1998 _ j. non - cryst .
_ 235 - 237 314 ; b. doliwa and a. heuer 1998 _ phys .
* 80 * 4915 ; ref . | we examine the phenomenon of dynamical heterogeneity in computer simulations of an equilibrium , glass - forming liquid .
we describe several approaches to quantify the spatial correlation of single - particle motion , and show that spatial correlations of particle displacements become increasingly long - range as the temperature decreases toward the mode coupling critical temperature . |
there is strong cosmological and astrophysical evidence for the existence of dark matter ( dm ) in the universe .
the observations indicate that dm , about 26% of the total mass - energy of the universe , is non - baryonic , non - relativistic and not subject to electromagnetic interactions @xcite . in the framework of the weakly interacting massive particles ( wimps ) paradigm ,
the visible baryonic part of a galaxy is embedded in the dm halo . in the most common scenario
, wimps can scatter elastically with matter and become trapped in massive astrophysical objects such as the sun@xcite .
there , dm particles could self - annihilate , reaching equilibrium between capture and annihilation rates over the age of the solar system .
the standard scenario assumes that the products of dm annihilation are standard model ( sm ) particles , which interact with the interior of the sun and are largely absorbed @xcite .
however , during this process , high - energy neutrinos may be produced , which can escape and be observed by neutrino detectors , such as antares .
limits on wimp dm annihilation in the sun have been reported already in antares @xcite , and in other neutrino telescopes : baksan @xcite , super - kamiokande @xcite and icecube @xcite .
an alternative hypothesis is based on the idea that dm is `` secluded '' from sm particles and that the annihilation is only possible through a metastable mediator ( @xmath0 ) , which subsequently decays into sm states @xcite .
these models retain the thermal relic wimp dm scenario while at the same time explain the positron - electron ratio observed by pamela @xcite , fermi @xcite , and measured recently by ams - ii with improved accuracy @xcite . in the secluded dark matter ( sdm ) scenario , the presence of a mediator dramatically changes the annihilation signature of dm captured in the sun . if the mediators live long enough to escape the sun before decaying , they can produce fluxes of charged particles , @xmath1-rays or neutrinos @xcite that could reach the earth and be detected . in many of the secluded dark matter models ,
@xmath0 can decay into leptons near the earth .
the signature of leptons arising from @xmath0 decays may differ substantially from other dm models . assuming that the dm mass ( @xmath2 1 tev ) is much greater than the @xmath0 mass ( @xmath2 1 gev ) the leptons are boosted . if these leptons are muons , which is the less constrained case to explain the positron - electron ratio @xcite , the signature in the vicinity of the detector would be two almost parallel muon tracks . in ref .
@xcite this possibility is discussed and the expected sensitivity for the icecube neutrino telescope is calculated .
why a neutrino telescope generally interprets the dimuon signature as a single muon , the different energy deposition can help to discriminate this case from a muon induced by a neutrino interaction @xcite . on the other hand , even for short - lived mediators that decay before reaching the earth , neutrinos from the products of mediator decays could be detected in neutrino telescopes as well . finally , another possibility is that mediators may decay directly into neutrinos @xcite . in this case , the neutrino signal could be enhanced significantly compared to the standard scenario even for quite short - lived mediators .
the mediators will be able to escape the dense core of the sun where high - energy neutrinos can interact with nuclei and be absorbed .
the fact that the solar density decreases exponentially with radius facilitates neutrinos injected by mediators at larger radii to propagate out of the sun . in this work an indirect search for sdm using the 2007 - 2012 data recorded by the antares neutrino telescope is reported .
the analysis treats the different mediator decay products : * direct detection of dimuons * neutrinos from decays of dimuons produced by mediators that decay before reaching the earth * neutrinos produced by mediators that decay directly to neutrinos and antineutrinos .
the analysis procedure is similar to the previous antares search for dm annihilation in the sun @xcite , but optimising the search for the expected signal in the case of sdm . in the following
, neutrino will mean neutrino plus anti - neutrino , unless explicitly stated otherwise .
the antares neutrino telescope @xcite is located in the mediterranean sea , at a depth of 2.5 km , about 40 km offshore from toulon ( france ) .
it is presently the largest neutrino telescope in the northern hemisphere and consists of 885 optical modules ( oms ) arranged in a three - dimensional array .
the operation principle is based on the detection of the cherenkov light induced by relativistic charged particles produced in interactions of high - energy neutrinos in the surroundings of the detector .
the oms are installed along 12 lines anchored to the sea floor and kept vertical by a submerged buoy .
the length of the lines is 450 m and the distance between the lines is 6075 m. the oms are grouped in triplets in order to reduce the effect of optical background produced by potassium-40 decays and bioluminescence .
a line comprises 25 triplets separated by a vertical distance of 14.5 m. the position @xcite and time @xcite information of the photons detected by the oms are used to determine the muon direction .
the reconstruction algorithm is based on the minimisation of a @xmath3like quality parameter of the track reconstruction , q , which uses the difference between the expected and measured times of the detected photons , taking into account the effect of light absorption and scattering in the water @xcite .
@xcite shows the distribution of reconstruted events as a function of q. the installation of the detector was completed in 2008 , although during 2007 five lines were already installed and data taking begun . in this
analysis data recorded between the @xmath4 of january 2007 and the @xmath5 of october 2012 are used , corresponding to a total livetime of 1321 days , without taking into account the visibility of the sun . during this time , the detector consisted of 5 lines for most of 2007 and of successively 10 and 12 lines from 2008 to 2012 .
dimuon and neutrino - induced candidates are events reconstructed as upgoing .
the duty cycle of this search for events from the sun corresponds to 50% of the detector livetime .
two main sources of background are present in the antares data : 1 ) downgoing atmospheric muons resulting from the interaction of cosmic rays in the atmosphere .
these background events are strongly reduced by the deep sea location and by the reconstruction algorithms that are optimised for upgoing events .
cuts on the quality of the tracks are also applied to reject downgoing muons mis - reconstructed as upgoing .
2 ) atmospheric neutrinos produced by cosmic rays .
these neutrinos can traverse the earth , so they can be detected as upgoing tracks and can not be rejected on an event - by - event basis .
both kinds of background have been simulated and good agreement with data has been found @xcite .
nevertheless , the background estimation is performed using scrambled data , by randomising the time of selected events , to reduce the effect of systematic uncertainties ( efficiency of the detector , assumed atmospheric fluxes , etc . ) .
to evaluate the sensitivity of antares to the signal from sdm models for the case a ) in which dimuons are detected directly , a new tool for dimuon signal generation ( dimugen ) has been developed @xcite .
dimugen generates and propagates dimuons produced by the decay of mediators resulting from dm annihilation . for this analysis ,
the mediator arrives from the sun s position during the period under study .
different dm masses in the range between 30 gev to 10 tev have been simulated , using in most cases a typical mass of 1 gev for the mediator @xmath0 .
once the muons are generated in the vicinity of the detector , simulations of the travel and interactions of muons are performed , as well as the detection of the cherenkov light by the optical modules . triggering and reconstruction algorithms
are also included in the process in order to evaluate the global efficiency for the detection of dimuons as a function of the quality parameter , q , and the half - cone angle from the sun direction , @xmath6 . to evaluate the antares sensitivity for the cases where the neutrino is the final decay product that arrives at earth ,
the antares effective areas for neutrinos as a function of the q and @xmath6 according to neutrino simulations have been used . for this , it is necessary to know the energy spectra of neutrinos arriving at the detector . in case
b ) the neutrino spectra have been obtained from michel s spectra of neutrinos from muon decay and taking into account the boost @xcite . for scenario c ) and assuming long mediator lifetimes with respect to the time required to exit from the sun s core , the neutrino spectra are almost flat in the energy range under study @xcite .
for these cases the assumption that after oscillations all neutrino flavours arrive at earth with the same proportion has been made . for case
b ) , this results in a factor of 2/3 for the muon neutrinos with respect to the parent muons , since a muon neutrino and an electron neutrino result from a muon decay , and these neutrinos are spread to all flavours .
there is not any change in the neutrino flavour composition of case c ) due to oscillations since it is assumed that all kinds of neutrinos result from the mediator decays .
in order to avoid any bias in the event selection , a blinding policy has been followed using as observables the angular separation of the track with respect to the sun s direction and the track quality parameter @xmath7 .
the values of the cuts have been chosen before looking at the region where the signal is expected .
the best sensitivities for dimuon ( or neutrino ) fluxes and cross sections are extracted using the model rejection factor ( mrf ) method @xcite .
it consists of finding the set of cuts which provide , on average , the best flux upper limit taking into account the observed background and the efficiency for a possible signal flux .
the procedure to obtain the sensitivity for a neutrino flux coming from dm annihilations in the sun is described in ref .
the detection of a couple of muons arising from the decay of the mediators is performed using the same reconstruction technique described in ref .
@xcite and used for single muons .
a couple of parallel muons with relative separation smaller than the distance of strings in the detector yields a number of cherenkov photons practically indistinguishable from that of a single muon . for this reason , the case
a ) is treated in the same way as cases b ) and c ) .
then , the mrf is used to determine the best value for the cut on the half - cone angle around the sun ( @xmath8 ) and the track quality cut parameters ( @xmath9 ) for the different cases ( a , b , c ) and the different dm masses studied . since in most cases
the difference in flux sensitivities between different optimisations is not large , it was decided to limit the optimisations to four different sets of cuts , see table [ tab:1 ] , that were representative of all possible situations , within a few percent sensitivity difference from the optimal one @xcite .
there are three optimisation sets corresponding roughly to lower , intermediate and larger dm masses for the dimuon detection case . for the neutrino detection cases ,
the latter set is also used for larger dm masses and another additional optimisation is used for lower and intermediate dm masses .
sensitivities to the particle flux at earth,@xmath10 , obtained for the different cases studied are shown in figure [ fig:1 ] .
sensitivity , expected 90% confidence limit , for dm annihilation in the sun for three sdm scenarios with the data recorded by antares between 2007 and 2012.,width=453 ]
after the optimisation of the flux sensitivities using the mrf with scrambled data , the data coming from the sun direction was unblinded .
as an example , figure [ fig:2 ] shows the distribution of events detected for @xmath11 as a function of the angular distance from the sun .
good agreement between data and the expected background obtained from scrambled data is observed .
the green line indicates the angular cut selected for this analysis .
table 1 summarises the number of events observed and expected from background for the different sets of cuts .
since no significant excess is observed , the 90% confidence level ( cl ) upper limit values in the feldman - cousins approach @xcite are used to constrain the model .
differential distribution of the angular separation of the event tracks with respect to the sun s direction with @xmath12 for data ( black ) and expected background ( red curve ) .
the statistical uncertainties are shown for the data.,width=377 ] .[tab:1 ] number of observed events , n@xmath13 , and expected background , n@xmath14 , for the sets of cuts ( track quality parameter , q@xmath15 , and half - cone angle around the sun , @xmath8 ) used for the different sdm scenarios and dm mass .
the resulting 90% confidence level upper limit is also shown ( @xmath16 ) .
[ cols="^,^,^,^,^,^,^",options="header " , ] taking into account the upper limits obtained in table [ tab:1 ] and the corresponding effective areas , the upper limits on the dimuon and neutrino fluxes at earth , @xmath10 , are derived , which almost coincide with the sensitivities shown in figure [ fig:1 ] .
following the arguments given in ref .
@xcite , the dimuon ( or neutrino ) flux at earth can be translated into dm annihilation rate in the sun for scenarios a ) and b ) . assuming a 100% branching ratio for the @xmath17 decay channel of the mediator , and taking into account the solid angle factor and the decay probabilities , the following relationships between the annihilation rate , @xmath18 , and the dimuon and muonic neutrino fluxes , respectively @xmath19 and @xmath20 @xcite ,
are obtained : * @xmath21 * @xmath22 where @xmath23 is a characteristic distance related to the detector size , @xmath24 is the distance between the sun and the earth ; @xmath25 is the radius of the sun ; and @xmath26 is the mediator s decay length , @xmath27 , i.e. the product of the mediator s lifetime , @xmath28 , the speed of light , @xmath29 , and the relativistic boost factor @xmath30
. for the case in which mediators decay directly into neutrinos , only the situation in which the mediator lifetime is long enough has been considered , so that the absorption of neutrinos in the sun becomes negligible .
if the lifetime of the mediator is small , the final energy spectrum of neutrinos would be quite similar to the case of typical dm searches @xcite . for long - lived mediators , ( @xmath31 km ) , in this scenario , the relationship between @xmath18 and @xmath20 is @xcite : * @xmath32 constraints on the annihilation rates as a function of mediator lifetime and dark matter mass have been obtained . for example , figure [ fig:3 ] shows the antares exclusion limits for the sdm scenarios for dm masses of 0.5 ( left ) and 5 ( right ) tev using the typical @xmath0 mass of 1 gev .
blue lines indicate the exclusion region in the di - muon case , either by direct detection ( dotted - dashed line ) or through detection of neutrinos ( solid line ) . for large decay lengths @xmath26 ( @xmath33 ) , i.e. long mediator lifetime ,
the direct detection of dimuons is more efficient than neutrino detection for small dm masses , whereas the opposite holds for larger masses .
the transition is around 0.8 tev in dm mass .
naturally , for small @xmath26 ( @xmath34 ) neutrino detection is much more efficient .
green lines indicate the exclusion regions of the parameter space for the scenario of sdm with mediator decaying into neutrinos .
more stringent constraints are obtained in this scenario mainly due to the harder neutrino energy spectrum .
antares exclusion limits for the sdm cases studied by products of dm annihilation in the sun through mediators decaying into : dimuons ( dashed - dotted blue ) , neutrinos from dimuons ( solid blue ) , directly into neutrinos ( green ) as a function of the annihilation rate ( @xmath18 ) and the decay length ( @xmath35 ) for 0.5 and 5 tev dm masses .
the shadowed regions are excluded for these models.,title="fig:",width=287 ] antares exclusion limits for the sdm cases studied by products of dm annihilation in the sun through mediators decaying into : dimuons ( dashed - dotted blue ) , neutrinos from dimuons ( solid blue ) , directly into neutrinos ( green ) as a function of the annihilation rate ( @xmath18 ) and the decay length ( @xmath35 ) for 0.5 and 5 tev dm masses .
the shadowed regions are excluded for these models.,title="fig:",width=287 ] limits on the dm - nucleon interaction can also be derived for these cases .
assuming that , as expected , there is equilibrium of the dm population in the sun , i.e. , the annihilation balances the dm capture , @xmath36 , and that according to @xcite the capture is approximately @xmath37 where , @xmath38 and @xmath39 are the spin - dependent ( sd ) and spin - independent ( si ) cross sections , respectively , and @xmath40 is the dm mass .
the limits on the sd and si wimp - proton scattering cross - sections are derived for the case in which one of them is dominant .
the sensitivity in terms of annihilation rates depends on the lifetime of the mediator @xmath0 . to assess the potential to constrain these models , lifetime values for which the sensitivities are optimal
have been assumed . for the dimuon case ,
the lifetime has to be long enough to assure that the mediator reaches the vicinity of the earth , so mediators with a decay length of about the sun - earth distance are shown . in
both neutrino cases the lifetime of the mediator for best sensitivity has to be long enough to ensure that the mediator escapes the sun , but not so long that it decays before reaching the earth .
figure [ fig:4 ] shows the antares nucleon wimp cross section limits for the sdm scenario with the products of dm annihilation in the sun through mediators decaying into : di - muons ( blue ) and directly into neutrinos ( green ) for the selected mediator s lifetimes . for
sufficiently long - lived , but unstable mediators , the limits imposed to these models are much more restrictive than those derived in direct detection searches for the case of spin - dependent interaction . in the case of spin - independent interactions ,
the direct detection search is more competitive for low and intermediate masses , but the sdm search becomes more competitive for larger masses ( @xmath41 tev ) .
antares 90% cl upper limits on wimp - nucleon cross section as a function of wimp mass .
the left panel refers to spin - dependent and the right one to spin - independent wimp interactions .
two favourable mediator lifetimes are considered .
the current bounds from pico @xcite , lux @xcite and xenon @xcite are also shown.,title="fig:",width=287 ] antares 90% cl upper limits on wimp - nucleon cross section as a function of wimp mass .
the left panel refers to spin - dependent and the right one to spin - independent wimp interactions .
two favourable mediator lifetimes are considered .
the current bounds from pico @xcite , lux @xcite and xenon @xcite are also shown.,title="fig:",width=287 ] the limits derived here are the first experimental limits on sdm models established by a neutrino telescope . compared to other indirect detection methods , such as those using gamma - rays , the limits derived here are in general competitive for large dm masses and favourable mediator lifetimes ( @xmath42 m ) . however , the comparison is not straightforward , since the results are usually given in terms of the @xmath43 parameter , and several astrophysical assumptions have to be made .
therefore , the different indirect searches can be considered complementary .
an analysis of 2007 - 2012 antares data for the search of secluded dark matter annihilation from the sun has been presented . lately
, this kind of models has increased in popularity , since sdm could help to interpret the energy spectrum of the positron / electron ratio measured recently by pamela , fermi - lat and ams - ii .
assuming that dm can be annihilated through a mediator that has a long lifetime , three different situations have been considered : a ) detection of dimuons that result from the mediator decay , or neutrino detection from : b ) mediator that decays into a dimuon and , in turn , into neutrinos , and c ) a mediator that decays directly into neutrinos .
no significant excess over background has been observed for these searches , and limits to these models have been derived .
this is the first time that these models are constrained with dedicated searches in neutrino telescopes .
the limits are the most restrictive available for a wide range of dm masses , and mediator lifetimes , and in particular for the case of spin - dependent dm - proton interaction .
the authors acknowledge the financial support of the funding agencies : centre national de la recherche scientifique ( cnrs ) , commissariat lnergie atomique et aux nergies alternatives ( cea ) , commission europenne ( feder fund and marie curie program ) , institut universitaire de france ( iuf ) , idex program and univearths labex program at sorbonne paris cit ( anr-10-labx-0023 and anr-11-idex-0005 - 02 ) , rgion le - de - france ( dim - acav ) , rgion alsace ( contrat cper ) , rgion provence - alpes - cte dazur , dpartement du var and ville de la seyne - sur - mer , france ; bundesministerium fr bildung und forschung ( bmbf ) , germany ; istituto nazionale di fisica nucleare ( infn ) , italy ; stichting voor fundamenteel onderzoek der materie ( fom ) , nederlandse organisatie voor wetenschappelijk onderzoek ( nwo ) , the netherlands ; council of the president of the russian federation for young scientists and leading scientific schools supporting grants , russia ; national authority for scientific research ( ancs ) , romania ; ministerio de economa y competitividad ( mineco ) , prometeo and grisola programs of generalitat valenciana and multidark , spain ; agence de loriental and cnrst , morocco .
we also acknowledge the technical support of ifremer , aim and foselev marine for the sea operation and the cc - in2p3 for the computing facilities .
l. accardo , et al . , _ high statistics measurement of the positron fraction in primary cosmic rays of 0.5 - 500 gev with the alpha magnetic spectrometer on the international space station _ , _ phys .
_ * 113 * ( 2014 ) 121101 .
m. aguilar et al . , _ precision measurement of the ( @xmath44 ) flux in primary cosmic rays from 0.5 gev to 1 tev with the alpha magnetic spectrometer on the international space station _ , _ phys .
* 113 * ( 2014 ) 221102 .
s. adrin - martnez , _ design and development of an acoustic calibrator for deep - sea neutrino telescopes and first search for secluded dark matter with antares _ , ph .
d. thesis , universitat politcnica de valncia ( 2015 ) . | a search for secluded dark matter annihilation in the sun using 2007 - 2012 data of the antares neutrino telescope is presented .
three different cases are considered : a ) detection of dimuons that result from the decay of the mediator , or neutrino detection from : b ) mediator that decays into a dimuon and , in turn , into neutrinos , and c ) mediator that decays directly into neutrinos . as no significant excess over background is observed , constraints are derived on the dark matter mass and the lifetime of the mediator . |
it would be very interesting if @xmath13 and @xmath0 violation were a consequence of another fundamental symmetry , such as a symmetry between electric and magnetic charges . in ref .
@xcite we considered the possibility that dyons ( electrically charged magnetic monopoles ) could induce parity and time invariance violating electric dipole moments ( edms ) of quantum systems .
the point is that the polarization of the dyon vacuum by the coulomb field may produce not only the usual correction to the electric field , @xmath14 , but also a radial magnetic field @xmath15 , where @xmath11 is the electric and @xmath12 the magnetic charge of the dyons ( the corresponding antidyons would have charges @xmath16 and @xmath17 ) .
the interaction of electrons ( or quarks ) with this magnetic field could generate atomic ( and neutron ) electric dipole moments .
[ in addition , electron and neutron edms @xmath18 could appear due to higher order diagrams ( @xmath19 is the electron or quark mass ; @xmath10 is the dyon mass ) . ]
the edms were found to be proportional to @xmath20 .
if there were a second dyon - antidyon pair , with charges @xmath11 and @xmath17 , and @xmath16 and @xmath12 , then they would also make a contribution , which would cancel out the original contribution ( assuming this second pair had the same mass as the first ) . therefore to have time invariance violation we can only have one of the dyon - antidyon pairs .
our consideration in @xcite gave quite a strong limit on the dyon mass : @xmath21 .
however , this consideration was only based on `` heuristic '' arguments ; it did not prove that the effect exists , only that it is conceivable . in ref .
@xcite a different mechanism was considered , by which dyons may induce an electron edm .
firstly , they calculated the lowest order @xmath0- and @xmath1-odd correction to the effective lagrangian of qed due to dyon vacuum polarization .
this involves finding the change in the energy density of the vacuum dyons when external electric and magnetic fields are applied .
the correction @xmath22 was found to be @xmath23 for ( spin @xmath24 ) dyons with masses @xmath10 , electric charges @xmath11 , and magnetic charges @xmath12 . under time reversal ( @xmath25 ) and parity ( @xmath26 ) transformations
@xmath27 changes sign , while @xmath28 is unchanged , and so @xmath22 is @xmath0- and @xmath1-odd . in ref .
@xcite they used this effective lagrangian to estimate the induced electron edm .
unfortunately , the integrals involved in such calculations are quadratically divergent .
therefore , the calculation can not be done within the limits of the applicability of the lagrangian ( [ epp2b ] ) .
the result depends strongly on the effective cut - off parameter . in this paper
we present further arguments that dyon vacuum polarization may produce @xmath0 and @xmath1 violating effects . as in ref .
@xcite , our calculation is based on the effective lagrangian @xmath22 ( [ epp2b ] ) . however , the integrals involved in the calculation of the effects considered here ( mixed polarizablity , @xmath0- and @xmath1-odd scalar - pseudoscalar interaction between electrons and quarks , and atomic and neutron edms ) are only logarithmically divergent .
therefore , we stay within the limit of the applicability of the effective lagrangian ( [ epp2b ] ) . in @xcite
the photon - photon scattering tensor ( see , e.g. , @xcite , 124 and @xcite , 54.4 for the definition of this ) for low - energy photon - photon scattering due to dyons was derived from @xmath22 ( this is the basis for our calculation of the dyon effects ) .
photon - photon scattering due to dyon vacuum polarization is shown in fig .
the result for this @xmath0- and @xmath1-odd photon - photon scattering tensor is @xmath29 \label{eppst}\end{aligned}\ ] ] in sec .
[ smixpol ] we find the dyon induced @xmath0- and @xmath1-odd mixed polarizability of a particle , as well as a nucleus .
the @xmath0- and @xmath1-odd mixed polarizability , @xmath2 , can be defined by the equation @xmath30 , where @xmath4 is the change in energy of a system when electric ( @xmath5 ) and magnetic ( @xmath6 ) fields are applied . in sec .
[ stpoint ] we find the @xmath0- and @xmath1-odd interaction between two particles that could be induced by dyons and in sec .
[ sdexp ] we use experimental bounds on edms to give limits on dyon properties . finally , in sec .
[ skmeson ] we estimate the contribution of dyons to @xmath13 violation in k - meson decays .
note that it is possible that dyons do not exist as free particles but only as bound dyon - antidyon pairs that can not be separated ( for example , nambu showed that in the standard electroweak model the classical solution is a monopole - antimonopole pair connected by a @xmath31 field string @xcite ) . in this case
dyons with a small mass are not ruled out by experimental searches for new particles ( a bound state of dyons may be hard to produce and identify ) .
the effects considered in this paper apply equally well for free dyons and bound dyons .
we will now find the matrix element corresponding to fig .
[ f2 ] using the photon - photon scattering tensor ( [ eppst ] ) .
note that because of the way in which the photon - photon scattering tensor was symmetrized in @xcite we are actually calculating the matrix element for all of the diagrams involving different arrangements of the photons ( e.g. the diagram similar to fig .
[ f2 ] , but with the lower two photons crossing each other ) put together .
this only covers diagrams with two photons free and two attached to a particle ( or , when two particles are involved , as in fig .
[ f3 ] , two photons attached to each particle ) .
other diagrams ( such as that considered in @xcite ) do not involve logarithmically divergent integrals our aim is not to do a calculation including every type of diagram with dyons , rather it is to show that the effect exists . using the feynman rules we get ( we use @xmath32 and the feynman rules and notation of ref .
@xcite ) @xmath33 where @xmath34 and @xmath35 are the fermion and photon propagators , @xmath36 is the charge and @xmath37 the mass of particle @xmath38 , and we use the notation @xmath39 .
we define @xmath40 as @xmath41 , and so we have @xmath42 . substituting in the propagators and simplifying gives @xmath43 } m^{\mu \nu \lambda \omega } ( k_1,k_2,k_3,k_4 ) \ , d^4 k_2 .
\label{ea2}\end{aligned}\ ] ] we now use the following identity , which can be obtained in a similar way to eq . ( 24 ) of ref .
@xcite : @xmath44 this gives @xmath45 ( k_2 ^ 2 + i \epsilon ) [ ( k_2 - q)^2 + i \epsilon ] } \frac{\partial}{\partial k_{3 \lambda } } \frac{\partial}{\partial k_{4 \omega } } m^{\mu \nu \rho \sigma } \
, d^4 k_2 \right ] \gamma_\nu u(p_a ) , \label{ea4}\end{aligned}\ ] ] where @xmath46 .
\label{ea5}\end{aligned}\ ] ] using this equation ( and keeping in mind that @xmath47 ) , we can see that the integral in eq .
( [ ea4 ] ) will be made up of terms that contain the form @xmath48 [ ( k_2 - p_a)^2 - m_a^2 + i \epsilon ] } \equiv w^{\alpha \beta } , \label{ea6}\ ] ] which we define as @xmath49 for convenience ( @xmath50 and @xmath2 can stand for any indices ) . using this notation
we can write @xmath51 @xmath49 can be worked out using the feynman parameterization technique .
we use the identity ( see , e.g. , @xcite ) @xmath52 ^ 3 } , \label{ea7}\ ] ] with @xmath53 , @xmath54 , and @xmath55 ( we now omit the @xmath56 s ) .
using this , and completing the square in the denominator gives @xmath57 ^ 3 } \
, d^4 k_2 , \label{ea8}\ ] ] where @xmath58 and @xmath59 we now do a change of variable : @xmath60 .
the numerator in the integral in eq .
( [ ea8 ] ) changes as follows : @xmath61 k_2^\alpha k_2^\beta + z ( p_a^\alpha k_2^\beta \hat{k_2 } + p_a^\beta k_2^\alpha \hat{k_2 } ) \nonumber \\ & & \mbox { } + x ( q^\alpha k_2^\beta \hat{k_2 } + q^\beta k_2^\alpha \hat{k_2 } ) - q^\alpha k_2^\beta \hat{k_2 } , \label{ea11}\end{aligned}\ ] ] where we threw away the terms that are odd in @xmath62 , as they will vanish on integration , as well as the terms independent of @xmath62 , as their contribution will be much smaller than the @xmath63 terms , which are logarithmically divergent when integrated . in the integral
we can replace @xmath64 with @xmath65 ( see , e.g , @xcite , 127 ) .
so @xmath66 g^{\alpha \beta } + z ( p_a^\alpha \gamma^\beta + p_a^\beta \gamma^\alpha ) + x ( q^\alpha \gamma^\beta + q^\beta \gamma^\alpha ) - q^\alpha \gamma^\beta \ } \label{ea12}\end{aligned}\ ] ] consider the integral @xmath67 .
for @xmath68 the integrand is proportional to @xmath69 ( since @xmath70 ) and hence the integral will be a logarithm .
we take a lower `` cut - off '' for this integral of @xmath71 , since near this point the @xmath72 behavior of the integrand begins to break down . for our upper `` cut - off ''
we take the dyon mass @xmath10 , because , as stated in @xcite , eq . ( [ eppst ] ) is only valid when the photon momenta are small compared to @xmath10 .
hence we have @xmath73 .
( the factor of @xmath74 appears as @xmath62 is a @xmath75-vector in minkowski space . )
using this and integrating over @xmath76 and @xmath77 gives @xmath78 \label{ea13}\ ] ] we now put this result for @xmath49 into eq .
( [ ea14 ] ) . evaluating
this is a long and involved process and so we will not give the details here .
to give an idea of how the calculation is done , the calculation for one of the terms the term @xmath79 in @xmath49 , where @xmath80 is done in the appendix .
we denote the contribution of this term to @xmath81 by @xmath82 .
the result is [ see eq .
( [ ea35 ] ) ] @xmath83 .
the @xmath84 and @xmath85 terms in @xmath49 give similar structures .
the contribution of the @xmath86 and @xmath87 terms to @xmath81 vanishes this is essentially due to the fact that @xmath88 , which is just the charge conservation condition and can be proved from the dirac equation ( using @xmath41 ) .
note that all of these terms are symmetric in @xmath50 and @xmath2 . the remaining , nonsymmetric term in @xmath49
, @xmath89 gives a different structure : @xmath90 / ( 144 \pi^2 m^4 ) \ln ( m / m_a ) \overline{u } ( p_a^\prime ) \hat{q } \gamma_5 u(p_a ) f_{\alpha \beta } ( k_3 ) f^{\alpha \beta } ( k_4)$ ] .
using all the terms of @xmath49 gives the result @xmath91 .
\label{ea17}\end{aligned}\ ] ] writing @xmath92 and @xmath93 in terms of the components of the electric and magnetic fields gives @xmath94 , \label{ea18}\ ] ] @xmath95 .
\label{ea19}\ ] ] this allows us to rewrite eq .
( [ ea17 ] ) as @xmath96 \nonumber \\ & & \mbox { } + i ( 1/2 ) \overline{u } ( p_a^\prime ) \hat{q } \gamma_5 u(p_a ) [ { \bf b } ( k_3 ) \cdot { \bf b } ( k_4 ) - { \bf e } ( k_3 ) \cdot { \bf e } ( k_4 ) ] \}. \label{ea51}\end{aligned}\ ] ] the first term of this equation describes a @xmath0- and @xmath1-odd mixed polarizability effect . writing this term as a potential , @xmath97 and setting @xmath98 ( as the @xmath0- and @xmath1-odd mixed polarizability involves homogeneous electric and magnetic fields )
, we get @xmath99 if we define the @xmath0- and @xmath1-odd mixed polarizability by the equation @xmath100 , we then have @xmath101 the above holds for a particle of charge @xmath36 . to find the dyon induced @xmath0- and @xmath1-odd mixed polarizability for a nucleus with @xmath102 protons and @xmath103 neutrons
we must sum over all the quarks , giving @xmath104 where @xmath105 is the constituent quark mass .
we used @xmath106 , where @xmath103 is a proton or neutron and @xmath40 is either @xmath107 or @xmath108 ( see , e.g , @xcite ) .
the second term in eq .
( [ ea51 ] ) describes the interaction of the spin of a particle with the gradient of the electromagnetic field s lagrangian density [ @xmath109 in the non - relativistic limit ] .
in this section we will find the @xmath0- and @xmath1-odd interaction between two particles that could be induced by dyons .
we will find the matrix element for the diagram shown in fig .
[ f3 ] , starting from the matrix element for fig .
[ f2 ] found above ( [ ea17 ] ) . taking the first term of eq .
( [ ea17 ] ) we will call it @xmath110 and using eq .
( [ ea16 ] ) we have @xmath111 to convert from fig . [ f2 ] to fig .
[ f3 ] we take @xmath112 and @xmath113 ( corresponding to the two free ends ) and replace them by @xmath114 . carrying out this transformation on eq .
( [ ea40 ] ) ( and using @xmath115 ) gives @xmath116 } \right \ } \gamma_\sigma u(p_b ) \nonumber \\ & = & -i e_a^2 e_b^2 \frac{q g ( q^2 - g^2)}{36 \pi^4 m^4 } \ln \left ( \frac{m}{m_a } \right ) m_a \overline{u } ( p_a^\prime ) u(p_a ) \varepsilon^{\sigma \beta \omega \alpha } \overline{u } ( p_b^\prime ) \gamma_\beta w_{\alpha \omega}^\prime \gamma_\sigma u(p_b ) , \label{ea41}\end{aligned}\ ] ] ( we introduce the dash to distinguish it from @xmath81 for the previous diagram ) where @xmath117 is like @xmath118 [ defined by eq .
( [ ea6 ] ) ] , but with the sign of @xmath40 changed , and so we can modify the result ( [ ea13 ] ) for @xmath117 . because of the @xmath119 only the nonsymmetric part of @xmath117 will contribute .
so we have @xmath120 now @xmath121 and so we have @xmath122 a similar process can be carried out for the second term in eq .
( [ ea17 ] ) . the total result for the matrix element of the @xmath0- and @xmath1-odd interaction between particles @xmath38 and
@xmath123 is @xmath124 \label{ea44}\end{aligned}\ ] ]
in this section we will find limits on the properties of dyons from experimental bounds on the electric dipole moments ( edms ) of atoms and the neutron .
first we will consider the atomic edm that can be produced by the dyon induced @xmath0- and @xmath1-odd electron - nucleon interaction . to find this interaction we let particle @xmath38
be a quark ( @xmath40 ) and particle @xmath123 be an electron ( @xmath125 ) , and sum over the quarks of the nucleon . to begin with we
will consider the first term of eq .
( [ ea44 ] ) the quark scalar - electron pseudoscalar interaction .
it follows from the dirac equation and @xmath126 that @xmath127 .
we rewrite the interaction as a hamiltonian density : @xmath128 where the wave function , @xmath129 , is related to the amplitude @xmath130 by @xmath131 ( @xmath132 is the energy of the particle ) .
summing over the quarks in the proton and neutron gives @xmath133 where @xmath134 denotes the nucleon wave functions and @xmath103 ( nucleon ) is either @xmath135 or @xmath136 .
the hamiltonian density for the other term the quark pseudoscalar - electron scalar interaction is @xmath137 here we have a derivative of a quark axial current ( @xmath138 ) , but we need to express our result in terms of nucleon wave functions .
we make an order of magnitude estimate using pcac ( partial conservation of axial current ) @xcite : @xmath139 .
it does not matter exactly what value we choose for the coefficient on the right hand side of this equation , as due to the fact that the main dependence of the interaction on the dyon mass goes as @xmath140 , the limit on the dyon mass that we will obtain will not be very sensitive to this choice .
the result for the nucleon pseudoscalar - electron scalar interaction is @xmath141 the interactions in eqs .
( [ e17b2 ] ) and ( [ en17a2 ] ) can produce atomic edms .
note that since the pseudoscalar @xmath142 is proportional to the spin of the particle , interaction ( [ e17b2 ] ) only has an effect in atoms with unpaired electrons , while interaction ( [ en17a2 ] ) requires an unpaired nucleon .
the size of such scalar - pseudoscalar and pseudoscalar - scalar interactions can be denoted by the constants @xmath143 and @xmath144 ( see , e.g. , @xcite ) : @xmath145 @xmath146 thus , these constants can be written in terms of dyon parameters as @xmath147 to order of magnitude accuracy , where we have defined @xmath148 for convenience as @xmath149 from experimental limits on atomic edms @xcite and calculations of the edms produced by the interactions ( [ eee1a ] ) and ( [ eee1b ] ) @xcite , limits on @xmath143 and @xmath144 can be deduced , and hence limits on the dyon parameter @xmath148 .
see table [ t1 ] .
note that in eqs .
( [ eea2p][eea2r ] ) , in addition to the unknown parameter @xmath148 there is @xmath10 , which is another ( independent ) unknown parameter , as @xmath11 and @xmath12 are unknown .
however , since @xmath10 only appears in logarithms , which are not very sensitive , we can take @xmath150 , assuming that @xmath151 and @xmath152 this estimate of @xmath12 is based on the dirac condition @xcite : @xmath153 , where @xmath136 is an integer .
the best bound on @xmath148 is @xmath154 .
dyons could also induce an electron - electron scalar - pseudoscalar interaction , which could create an atomic edm in atoms with unpaired electrons .
however , this effect is negligible compared to the effect of the nucleon scalar - electron pseudoscalar interaction . since the wave function of the outer , unpaired electron is enhanced at small distances , the main contribution to the interaction is from @xmath155 , where @xmath38 is the bohr radius . in the case of the electron - electron interaction
only the innermost electrons have sizeable wave functions in this region , while for the nucleon - electron interaction all of the nucleons are able to contribute , so the electron - nucleon interaction is larger by a factor @xmath156 .
in addition , the wave function of the outer electron becomes very large near the nuclear surface and this enhances the nucleon - electron interaction by a factor @xmath157 .
@xcite for a similar result concerning the relative contributions of weak electron - nucleon and electron - electron interactions .
now we consider the @xmath0- and @xmath1-odd quark - quark interaction .
this corresponds to eq .
( [ ea44 ] ) , with particles @xmath38 and @xmath123 both quarks .
the interaction between quarks @xmath158 and @xmath159 is @xmath160 ( of course there is also the scalar - pseudoscalar interaction ) .
here we used @xmath161 , as we did for the electron pseudoscalar in sec .
[ sdexp ] , but note that @xmath162 is the current quark mass : @xmath163 .
the @xmath164 comes from @xmath165 ( as in the appendix ) and is the constituent quark mass ( @xmath166 ) as it takes into account the effects of the gluon cloud .
it is not used in @xmath161 as this involves the difference of two dirac equations and so the effects of the gluon cloud cancel out .
the general quark - quark pseudoscalar - scalar interaction can be written in terms of the dimensionless constant @xmath167 as ( see , e.g. , ref .
@xcite ) @xmath168 where @xmath169 the calculation done in ref .
@xcite allows us to convert the experimental limit on the atomic edm of @xmath170hg , @xmath171 @xcite , to a limit on @xmath167 : @xmath172 .
the quark - quark pseudoscalar - scalar interaction can induce a neutron edm .
the limit on the neutron edm , @xmath173 @xcite implies the weaker limit @xmath174 @xcite .
using the stronger limit , and taking @xmath175 and @xmath176 as @xmath177 , gives @xmath178 .
it would be good if it were possible to make a comprehensive theory of @xmath13- and @xmath0-violation based on dyons that could explain the experimentally observed @xmath13-violation in k - meson decays @xcite .
unfortunately , no such theory exists at the moment , but we are able to make some plausible estimates of dyon induced @xmath13-violation in k - meson decays .
however , we find that the effect induced by dyons is not large enough to explain the observed level of @xmath13-violation . because of electroweak unification ,
since dyons may induce @xmath0- and @xmath1-odd photon - photon scattering processes , we would expect them to also induce w - boson scattering processes .
the w - boson scattering tensor should be similar to as in eq .
( [ eppst ] ) .
@xmath13-violation in k - meson decays could be caused by an interaction described by the following lagrangian @xcite : @xmath179 @xmath13 violation would occur if @xmath180 has a non - zero imaginary part ( the real part of @xmath180 is responsible for the mass difference between @xmath181 and @xmath182 ) . according to @xcite
the observed @xmath13 violation implies that @xmath183 and from @xcite we have @xmath184 where @xmath185 is the cabibbo angle , @xmath186 is the fermi weak interaction constant ( @xmath187 , where @xmath188 is the proton mass ) , and @xmath189 is the mass of the charm quark . a diagram through which dyons could induce a cp - violating interaction in k - mesons is shown in fig .
the intermediate quarks may be either up or charm quarks .
note that such a diagram requires two different kinds of dyons , differing in their electric charges by @xmath125 , to satisfy the conservation of charge at the @xmath190-dyon vertices .
( however , even if this is not true the effect may still appear at higher orders . )
an interaction corresponding to this diagram is @xmath191 where @xmath192 where @xmath193 and @xmath194 are the masses of the strange and down quarks , @xmath195 is the mass of the w - boson , and @xmath196 and @xmath197 are some effective weak electric and weak magnetic charges for the dyons .
this comes from integrating over the region @xmath198 , which assumes that the dyon mass is much greater than the w boson mass .
if in fact the dyon mass were less than the w boson mass then the relevant region of integration would be @xmath199 and the result is @xmath200 we can compare the size of the cp violation induced by this interaction to the experimentally observed cp violation in k - meson decays by looking at the ratio of the constant @xmath201 to @xmath202 [ although @xmath201 and @xmath202 describe two different types of cp violating interactions , we assume that the same order of magnitude of either would induce the experimentally observed cp violation ] . if we first consider the case of the dyon mass being larger than the w boson mass we have [ using eqs .
( [ elcp2 ] ) , ( [ elcpa8 ] ) , and ( [ ew12b ] ) ] @xmath203 .
this can not be larger than about @xmath204 .
if we consider the case of the dyon mass being less than the w boson mass then we get the result @xmath205 , which has a maximum value of about @xmath204 ( here we actually assumed that the dyon mass was larger than the c quark mass ; in the case that it is not we get a result of similar magnitude ) .
these results suggest that this effect can not produce the experimentally observed level of cp - violation in k - meson decays .
we can also consider the case when the intermediate quark is a top quark , rather than up or charm quarks .
if we assume that @xmath206 and integrate over the region @xmath207 we get the result @xmath208 where @xmath209 and @xmath210 are cabibbo - kobayashi - maskawa mixing matrix elements .
the top quark mass is @xmath211 and , according to @xcite , @xmath212 and @xmath213 lie in the ranges 0.0040.013 and 0.0350.042 , respectively
. therefore @xmath214 has a maximum value of about @xmath215 .
again , this is too small to give the observed effect . if we assume that @xmath216 and integrate over the region @xmath217 we get @xmath218 for which @xmath214 has a maximum value of about @xmath204 .
the case of the dyon mass being larger than the top quark mass gives a result of similar magnitude , with the maximum value of @xmath214 being @xmath204 .
dyons could also induce a cp violating interaction through the diagram shown in fig .
however , once again it seems that it can not produce an effect of the magnitude of the experimentally observed effect .
finally , the possibility is not excluded that dyons may generate an electroweak @xmath219-term ( @xmath220 , where @xmath221 $ ] ) which may give an effective @xmath13-violating interaction for four w - bosons that does not contain the fourth power of the dyon mass in the denominator , and in that case the effect could be of considerable size even if the dyon mass is large .
this work was supported by the australian research council .
in this appendix we calculate the contribution of the term @xmath222 in @xmath49 ( [ ea13 ] ) to @xmath81 .
we denote this contribution by @xmath82 .
the contribution of the 6th and 12th terms of the equation to @xmath82 is equal to ( with @xmath226/(360 \pi^4 m^4)$ ] ) @xmath227 using @xmath228 and the dirac equation : @xmath229 .
now , from the fact that @xmath230 and the definition of the dual tensor : @xmath231 , we have @xmath232 therefore @xmath233 the remaining terms ( terms 25 and 811 ) can be written as follows ( note that we swap the dummy indices @xmath234 and @xmath235 around in terms 4 , 5 , 10 , and 11 ) : @xmath236 now we have @xmath237 g^{\omega \nu } \varepsilon^{\sigma \lambda \mu \rho } = -i f_{\sigma \omega } ( k_4 ) g^{\omega \nu } \varepsilon^{\sigma \lambda \mu \rho } = -i { f_\sigma}^\nu ( k_4 ) \varepsilon^{\sigma \lambda \mu \rho}$ ] , which corresponds to the 1st and 3rd terms in eq .
( [ ea31 ] ) .
doing a similar thing for the 2nd and 4th terms as well gives @xmath238 \overline{u } ( p_a^\prime ) [ \gamma_\mu \hat{p}_a \gamma_\nu + \gamma_\nu \hat{p}_a \gamma_\mu ] u(p_a ) \nonumber \\ & = & -c_1 c_2 [ \widetilde{f}^{\sigma \mu } ( k_3 ) { f_\sigma}^\nu ( k_4 ) + { f_\sigma}^\nu ( k_3 ) \widetilde{f}^{\sigma \mu } ( k_4 ) ] \overline{u } ( p_a^\prime ) [ \gamma_\mu \hat{p}_a \gamma_\nu + \gamma_\nu \hat{p}_a \gamma_\mu ] u(p_a ) \label{ea32}\end{aligned}\ ] ] using the expressions for the @xmath239 and @xmath240 matrices in terms of the electric and magnetic field components , the following identity can be derived : @xmath241 g^{\mu \nu } \nonumber \\ & = & \frac{1}{2 } g^{\mu \nu } \widetilde{f}^{\alpha \beta } ( k_3 ) f_{\alpha \beta } ( k_4 ) \label{ea33}\end{aligned}\ ] ] therefore we have @xmath242 so , from eqs .
( [ ea30 ] ) and ( [ ea34 ] ) , the contribution of the term @xmath222 ( in @xmath49 ) to @xmath81 is @xmath243 99 v.v .
flambaum and d.w .
murray , phys .
a * 55 * , 1736 ( 1997 ) .
kovalevich , p. osland , ya .
m. shnir , and e.a .
tolkachev , phys .
d * 55 * , 5807 ( 1997 ) .
berestetski , e.m .
lifshitz and l.p .
pitaevski , _ relativistic quantum theory _
( pergamon , oxford , 1971 ) .
a.i . akhiezer and v.b .
berestetski , _ quantum electrodynamics _
( interscience , new york , 1965 ) . y. nambu , nucl . phys .
b * 130 * , 505 ( 1977 ) .
f. mandl and g. shaw , _ quantum field theory _
( wiley , chichester , 1993 ) .
i.b . khriplovich and s.k .
lamoreaux , _ cp violation without strangeness _
( springer - verlag , berlin , 1997 ) .
treiman , r. jackiw , and d.j .
gross , _ lectures on current algebra and its applications _ ( princeton university press , princeton , 1972 ) .
khriplovich , _ parity nonconservation in atomic phenomena _
( gordon and breach , philadelphia , 1991 ) . e.d .
commins , s.b .
ross , d. demille and b.c .
regan , phys .
a * 50 * , 2960 ( 1994 ) .
jacobs , w.m .
klipstein , s.k .
lamoreaux , b.r .
heckel and e.n .
fortson , phys .
a * 52 * , 3521 ( 1995 ) . a .-
mrtensson - pendrill , phys .
lett . * 54 * , 1153 ( 1985 ) .
dirac , proc .
( london ) * a133 * , 60 ( 1931 ) .
o.p . sushkov and v.v .
flambaum , yad .
* 27 * , 1308 ( 1978 ) [ sov .
* 27 * ( 1978 ) ] . v.m . khatsymovsky and i.b .
khriplovich , phys .
b * 296 * , 219 ( 1992 ) .
i.s . altarev _
et al . _ , yad . fiz . *
59 * , 1204 ( 1996 ) [ phys . at
. nucl . * 59 * , 1152 ( 1996 ) ] .
i.s . altarev _
et al . _ , phys .
b * 276 * , 242 ( 1992 ) .
et al . _ ,
b * 234 * , 191 ( 1990 ) .
christensen , j.w .
cronin , v.l .
fitch , and r. turlay , phys .
* 13 * , 138 ( 1964 ) .
okun , _ leptons and quarks _
( north holland , amsterdam , 1982 ) . c. caso
_ et al . _ ,
j. c * 3 * , 1 ( 1998 ) .
.table showing the experimentally determined upper bounds on various atomic edms ( with references ) and the consequent upper bounds on @xmath244 and @xmath245 and lower bounds on @xmath148 .
all bounds are given at the @xmath246 standard deviation limit .
[ cols="^,^,^,^,^",options="header " , ] | we consider a mechanism by which dyons ( electrically charged magnetic monopoles ) can produce both a @xmath0- and @xmath1-odd ( i.e. time reversal invariance and parity violating ) mixed polarizability @xmath2 [ defined by @xmath3 , where @xmath4 is the energy change when electric ( @xmath5 ) and magnetic ( @xmath6 ) fields are applied to a system ] and a @xmath0- and @xmath1-odd interaction between two particles : @xmath7 , where the @xmath8 are electron and quark spinors .
the latter can create atomic and neutron electric dipole moments ( edms ) . from experimental bounds on these
we find limits on the properties of dyons .
our best limit , using the experimental limit for the edm of the tl atom , is @xmath9 , where @xmath10 is the dyon mass and @xmath11 is the electric and @xmath12 the magnetic charge of the dyons .
the contribution of dyons to @xmath13 violation in k - meson decays is also estimated . |
the coefficients in the zero mode hamiltonian ( [ hzero ] ) are @xmath148\big],\nonumber\\ b&=&\frac{1}{2n_0 ^ 2}\left[\varepsilon(1+m^2 ) -8c_2{\cal o}_{\phi\theta}m^2\right]\nonumber\\ c&=&\frac{1}{3n_0 ^ 2}(\varepsilon-4c_2{\cal o}_{\phi\theta}),\nonumber\\ \alpha&=&\frac{\sqrt{2}}{6n_0 ^ 2}(1 + 3m^2 ) ( \varepsilon-4c_2{\cal o}_{\phi\theta}),\nonumber\\ \beta&=&-\frac{2m}{\sqrt{3}n_0 ^ 2}(\varepsilon -4c_2{\cal o}_{\phi\theta}),\nonumber\\ \gamma&=&-\frac{2m}{n_0 ^ 2}\sqrt{\frac{2}{3}}\left [ \varepsilon - c_2{\cal o}_{\phi\theta}(1 + 3m^2)\right],\nonumber\\ \eta&=&-3c_2n_0 ^ 2{\cal o}_{\phi\phi}/\hbar^2.\nonumber\end{aligned}\ ] ] | we discuss the quantum phases and their diffusion in a spinor-1 atomic bose - einstein condensate . for ferromagnetic interactions ,
we obtain the exact ground state distribution of the phase fluctuations corresponding to the total atom number ( @xmath0 ) , the magnetization ( @xmath1 ) , and the alignment ( or hypercharge ) ( @xmath2 ) of the system .
the mean field ground state is shown to be stable against these fluctuations , which dynamically recover the two continuous symmetries associated with the conservation of @xmath0 and @xmath1 as in current experiments .
since the observation of bose - einstein condensation of trapped atomic clouds @xcite , the coherence properties of the condensate has become the focus of many theoretical studies @xcite . within the mean - field theory ,
it is commonly assumed that the condensate can be described by a @xmath3 symmetry breaking field @xcite , equivalent to a coherent state assumption of the ground state .
( see refs .
@xcite for discussions of @xmath3-symmetric approaches ) .
although quite successful in providing theoretical understanding to many experimental observations , such a coherent state assumption is not necessarily consistent with real experimental situations , where the fluctuations of the atom numbers are difficult to control @xcite .
a coherent state leads to a poisson distribution of atoms . for a ground state with average of @xmath0 atoms ,
the associated number fluctuations are of the order @xmath4 . as was initially pointed out in refs .
@xcite , this number fluctuation of a coherent state condensate leads to the diffusion " ( or spreading ) of its initial phase . in a scalar condensate
, this diffusion , a dynamic attempt to restore the @xmath3 symmetry of the interacting atomic system , can be studied in terms of a zero mode , or the goldstone mode of the condensate @xcite .
more physically meaningful discussions in terms of the relative phase of two condensates were studied soon afterwards @xcite . experimentally , starting with the remarkable direct observation of the first order coherence in an interference experiment @xcite , direct correlations between number and phase fluctuations were observed with a condensate in a periodic potential @xcite , and more recently , in the remarkable mott insulating state @xcite obtained by loading a superfluid condensate into an optical lattice @xcite . the emergence of spinor-1 condensates @xcite ( of atoms with hyperfine quantum number @xmath5 ) has created new opportunities to understand quantum coherence and the associated number / phase dynamics in a three component condensate @xcite . in this paper , we investigate the quantum phase dynamics of a spinor-1 condensate due to atom number fluctuations .
we will focus on ferromagnetic interactions , when the condensate wave functions for the three spin components share the same spatial mode @xcite ; in this case , the phase fluctuations translate into fluctuations of the direction of the macroscopic condensate spin .
we consider a system of @xmath0 spin-1 bosonic atoms interacting via only s - wave scattering @xcite . a weak magnetic field @xmath6 ( as always exists in an experiment )
fixes the quantization z - axis such that quadratic zeeman effect can be neglected . in the second - quantized form ,
the hamiltonian is @xcite @xmath7\psi_i(\vec r)\nonumber\\ & & + \frac{c_0}{2}\sum_{i , j}\int d\,\vec r\psi_i^\dag(\vec r)\psi_j^\dag(\vec r ) \psi_i(\vec r)\psi_j(\vec r)\nonumber\\ & & + \frac{c_2}{2}\sum_{i , j , k , l}\int d\,\vec r\psi_i^\dag(\vec r ) \psi_j^\dag(\vec r)\vec{\mathbf f}_{ik } \cdot\vec{\mathbf f}_{jl}\psi_l(\vec r)\psi_k(\vec r),\hskip 18pt \label{ham}\end{aligned}\ ] ] where @xmath8 ( @xmath9 ) denotes the annihilation operator for the @xmath10-th component of a spinor-1 field .
the trapping potential @xmath11 is assumed harmonic and spin - independent .
the larmor precessing frequency is @xmath12 with @xmath13 the magnetic dipole moment for state @xmath14 .
the pseudo potential coefficients are @xmath15 and @xmath16 , with @xmath17 ( @xmath18 ) the s - wave scattering length for two spin-1 atoms in the combined symmetric channel of total spin @xmath19 ( @xmath20 ) .
@xmath21 is the mass of atom , and @xmath22 is the spin 1 operator @xcite .
hamiltonian ( [ ham ] ) is invariant under u(1 ) gauge transformation @xmath23 and so(3 ) spin rotations @xmath24 ( for @xmath25 ) @xcite .
a non - zero @xmath6 or the conservation of magnetization @xmath26 reduces the so(3 ) to its subgroup so(2 ) generated by @xmath27 .
following the bogoluibov theory we assume there exist three ` large ' condensate components @xmath28 around which we study the small quantum fluctuations ( off - condensate excitations ) via @xmath29 where @xmath30 is the number of condensed atoms in component @xmath10 . at near zero temperatures , higher than quadratic terms of @xmath31 in the hamiltonian ( [ ham ] ) are neglected .
the assumed mean field ground state @xmath32 breaks both continuous symmetries u(1 ) and so(2 ) , thus we expect to observe multiple zero energy goldstone modes @xcite . in the first order
, we obtain the usual coupled gross - pitaevskii equation ( gpe ) for the condensate modes @xmath33 .
the quantum fluctuations @xmath34 obey the usual bogoluibov - de gennes equations ( bdges ) @xcite .
the condensate number fluctuations , can be studied through the number fluctuation operators @xcite @xmath35\nonumber\\ & \approx & { 1\over \sqrt{n}}\int d\vec r [ \psi_j^\dag(\vec r ) \psi_j(\vec
r)-n\phi_j^*(\vec r)\phi_j(\vec r)].\end{aligned}\ ] ] using the gpes and the bdges @xcite their equations of motion are @xcite , @xmath36 from which we find that both @xmath37 and @xmath38 are constants of motion , an obvious outcome since the hamiltonian ( [ ham ] ) commutes with operators of both the total number of atoms and magnetization .
we define the phase operator @xmath39,\end{aligned}\ ] ] with @xmath40 the associated phase mode functions .
denote @xmath41 , the canonical quantization condition @xmath42=i\hbar$ ] is satisfied if the constraint @xmath43 is enforced , where @xmath44 $ ] .
as for a scalar or binary condensate @xcite @xmath45 we note that @xmath46=i\hbar\delta_{jk}j_j$ ] , @xmath47=\delta_{jk}\phi_j(\vec r)$ ] , and @xmath48=-i\hbar\delta_{jk}\theta_j(\vec r)$ ] .
these lead to @xmath49 @xcite , which helps to define the complete dynamic equations for the number and phase fluctuations @xcite .
including both the initial and time phases , the ground state wave functions can be generally expressed as @xmath50 with a common chemical potential @xmath51 , a lagrange multiplier enforcing the conservation of atom numbers .
minimization of the total energy eq .
( [ ham ] ) leads to @xmath52 @xcite .
it was further proven in ref .
@xcite that the steady state solution takes the same spatial mode for each of its three spin components ( for ferromagnetic interactions ) , i.e. @xmath53 where the real - valued mode function @xmath54 is normalized to unity , and governed by an equivalent scalar gpe @xmath55\phi(\vec r ) = \mu\phi(\vec r ) , \label{gpe1}\end{aligned}\ ] ] of a scattering length @xmath18 ( note @xmath56 ) .
define @xmath57 and @xmath58 as the relative atom numbers and relative magnetization , @xmath59 and @xmath60 in the ferromagnetic ground state @xcite .
similarly , the phase functions share the same spatial mode and can be generally expressed as @xmath61 with @xmath62 governed by the following equation @xmath63\theta(\vec r ) = n\tilde{u}\phi(\vec r ) .
\label{eqphase}\end{aligned}\ ] ] the normalization @xmath64 determines the goldstone parameter @xmath65 , which in the thomas - fermi ( tf ) limit is @xmath66(a_{\rm ho}/15 a_2)^{3/5}n^{-3/5}$ ] , with @xmath67 the ground state size @xmath68 of a harmonic trap with a frequency @xmath69 . equations ( [ gpe1 ] ) and ( [ eqphase ] ) are identical to their counterparts for a scalar condensate @xcite . in the ferromagnetic ground state , individual atomic spins align along the same direction , thus they collide only in the symmetric total spin @xmath70 channel . to conserve the magnetization with respect to the @xmath6-field direction , the condensate spin direction is simply tilted at an angle @xmath71 .
if this direction is taken as the quantization axis , then the spinor condensate behaves essentially as a scalar one @xcite when the @xmath6-field is zero . with the mode functions for atom number and phase fluctuations
, we find the zero mode dynamics can be expressed in terms of an associated hamiltonian . after laborious calculation
we arrive at @xcite @xmath72 with @xmath73 , @xmath74 , redefined as canonically conjugated variables @xmath75 and @xmath76 .
@xmath77 & @xmath78 are two hermitian and positive - definite matrices involving only parameters of the system .
thus @xmath79 , the assumed ground state , a mean field symmetry breaking state with coherent condensate amplitudes @xmath32 is stable .
the associated quantum fluctuations of atom numbers can be studied with the linearization approximation eq .
( [ bog ] ) .
matrices @xmath77 & @xmath78 are simplified with @xmath80 and @xmath81 .
we also find @xmath82 with @xmath83\theta(\vec r)$ ] a non - negative quantity .
we note that matrix @xmath84 is diagonalized by an orthogonal transformation @xmath85 , @xmath86 , and @xmath87 @xcite .
the corresponding @xmath88 , @xmath89 , and @xmath90 are respectively fluctuations of the total number of atoms , the magnetization , and the alignment . with these new collective operators ,
the zero mode hamiltonian becomes @xmath91 where all the coefficients are listed in the appendix .
replacing @xmath92 by @xmath93 , hamiltonian ( [ hzero ] ) leads to the ground state distribution of fluctuations @xmath94 with a width @xmath95 and the zero point energy @xmath96^{1/2}$ ] .
we find @xmath97 and @xmath98 in the tf limit ( and also taking the small @xmath99 ) .
figure [ fig2 ] shows selected results for the @xmath0-dependence of @xmath100 and @xmath101 . while @xmath100 increases monotonically with @xmath0 , the ground state width @xmath101 for @xmath102 peaks at some intermediate values of atom numbers and eventually saturates in the limit of large @xmath0 ( or strong interactions ) . and @xmath103 for a @xmath104rb condensate in a cylindrically symmetric trap with @xmath105 ( hz ) and @xmath106 .
@xmath107 ( solid line ) , @xmath108 ( dashed line ) , and @xmath109 ( dash - dotted line ) . , width=312 ] the above ground state distribution of @xmath110 can easily be understood . for a condensate with fixed total number of atoms and magnetization , the conservations of @xmath0 and @xmath1 require @xmath111 , which lead to @xmath112 , i.e. , completely diffused phases
; the ferromagnetic interaction , nevertheless , prepares a correlated ground state such that both distributions for @xmath90 and @xmath102 take a gaussian form with @xmath113 and @xmath114 .
such a distribution in @xmath90 and @xmath102 is in fact the minimal uncertainty coherent state consistent with the symmetry breaking ground state .
however , it is experimentally difficult to produce such a condensate having fixed atom numbers and phase fluctuations .
for any initial state , the dynamic solution @xmath115 as governed by the hamiltonian eq .
( [ hzero ] ) for @xmath116 $ ] can be used .
this solution has a simple structure @xcite ; in addition to oscillating terms of the forms @xmath117 and @xmath118 , the phase fluctuations of @xmath119 and @xmath120 also contain diffusion terms proportional to @xmath121 , which indicates the linearization approximation of eq .
( [ bog ] ) is valid only for a finite duration . finally , let s consider the diffusion of the direction of the condensate spin @xmath122 . because all three spin components share the same spatial mode function @xcite , individual spins of bose condensed atoms are parallel for ferromagnetic interactions , i.e. they act as a macroscopic magnetic dipole pointing along the same direction ( independent of the spatial coordinates ) : @xmath123 while undergoing precessing with respect to the z - axis . @xmath124 .
as the phase dynamics attempts to restore the u(1 ) and so(2 ) symmetries of the system , their respective initial values become irrelevant .
the fluctuation becomes @xmath125 $ ] with a zero average .
using @xmath126 , @xmath127 , and @xmath128 to denote the initial variances of @xmath0 , @xmath1 , and @xmath2 , we find in spherical coordinates @xmath129 ( note @xmath130 ) , @xmath131 and @xmath132 . both are fixed constants @xmath133 ^ 2\right\rangle&= & 3\sigma_n^2,\\ \left\langle\left[\delta f_\theta(t)\right]^2\right\rangle&=&\frac{1}{1-m^2 } \left ( 3m^2\sigma_n^2 + 2\sigma_m^2\right),\end{aligned}\ ] ] ( due to conservations of @xmath0 and @xmath1 ) when respective fluctuations in @xmath0 , @xmath1 , and @xmath2 are uncorrelated . in the azimuthal @xmath134 direction
a simple phase diffusion results @xmath135 with @xmath136/2\hbar,$ ] and a diffusion rate @xmath137 , proportional to @xmath138 , ( which can be explained in terms of the single axis twisting of @xmath139 in the isospin subspace of a spior-1 condensate @xcite ) .
the @xmath0- and @xmath140-dependence of the diffusion parameter @xmath141 are shown in fig .
[ fig3 ] for a @xmath104rb condensate . in the tf limit and
applying the results of ref .
@xcite , we find that @xmath142 , generally a very small quantity .
@xmath143 is the tf radius .
typically , the dephasing rate is a fraction of @xmath144 for a condensate of @xmath145 atoms and @xmath146 , i.e. its macroscopic spin direction is lost in a few cycles of trap oscillation .
figure [ fig3 ] also indicates that the faster a ferromagnetic condensate diffuses its pointing direction along the precession direction , the larger and more tightly confined a condensate is .
but for @xmath141.,width=312 ] when @xmath25 , zero mode dynamics becomes much simpler if the condensate spin direction is taken as z - axis .
although there still exist more than one goldstone mode in this case , the mean field ground state corresponds to all atoms in state @xmath147 @xcite .
thus to leading order , the only relevant fluctuation is that of the total number of atoms ( corresponding to the u(1 ) symmetry ) , and the phase diffusion dynamics becomes the same as in a scalar condensate @xcite .
our theory as developed in this paper does reduce to this simple limit . in conclusion ,
we have studied in detail quantum phase diffusions of a spinor-1 condensate with ferromagnetic interactions .
the condensate ground state is very simple , all three spin components have the same spatial mode function and their associated phase functions are also identical .
we have constructed the zero mode hamiltonian for the condensate number and phase fluctuations , and solved for the ground state distribution of these fluctuations when both @xmath0 and @xmath1 are conserved .
furthermore , we have obtained analytically the dynamic number and phase fluctuations relating to both the quantum phase diffusion and the initial distribution of these fluctuations .
we have identified a quantum phase diffusion coefficient for the pointing direction of the condensate spin and recovered its small - time quadratic t - dependent spreading .
this work is supported by the nsf grant no .
phys-0140073 and by a grant from nsa , arda , and darpa under aro contract no .
daad19 - 01 - 1 - 0667 . |
recently , much attention has been drawn to the problem of recovering a target matrix from a small set of linear measurements .
the estimated matrix is a superposition of low - complexity structure .
it can be found in many different fields , such as medical imaging @xcite , seismology @xcite , information retrieval @xcite and machine learning @xcite .
this problem regained great attention after the publication of the pioneering works of e.j .
cands et al @xcite .
according to paper @xcite , we can build the data model as follows : there exists a large - scale data matrix @xmath0 , where @xmath1 has low - rank , and @xmath2 is sparse component . the main question is how to recover a low - rank matrix @xmath3 and sparse matrix @xmath2 from a small set of linear measurements . in the paper @xcite , e.j .
cands et.al proved that most low - rank matrices and the sparse components can be recovered , provided that the rank of the low - rank component is not too large , and the sparse component is reasonably sparse ; and more importantly they proved that it can be done by solving a simple convex optimization problem , i.e. most matrices of low - rank and the sparse components can be perfectly recovered by solving the optimization problem @xmath4 provided that the rank of the matrix @xmath5 and the cardinality of the sparse component @xmath6 obey @xmath7 where @xmath8 and @xmath9 are positive numerical constants . in practice
, it is necessary to develop efficient and effective tools to process , analyze , and extract useful information from such high dimensional data ( in application , dimensional of data is always very high ) .
because strongly convex optimizations have many advantages , such as optimal solution is unique , many scholars suggest solving their strongly convex approximations , see , e.g. , @xcite , instead of directly solving the original convex optimizations . pertaining to problem ( 1 )
, the authors gave the suitable sufficient conditions under which the strongly convex models lead to the exact low - rank and sparse matrix recovery .
some suggestions were given on how to choose suitable parameters in practical algorithms in the paper @xcite .
however , the results of paper @xcite are limited in a special case , i.e. @xmath10 . in this paper
, we extend this result to the principal component pursuit with reduced linear measurements , i.e. @xmath11 is a @xmath12-dimensional random subspace .
it s easy to note that results of paper @xcite is only a special case of ours . in this subsection
, we will interpret an important strongly convex programming which will be addressed in this paper and list its existence and uniqueness theorems . in the paper @xcite ,
the authors have studied principal component pursuit with reduced linear measurements and given sufficient conditions under which @xmath3 and @xmath2 can be perfectly recovered .
@xmath13 in this paper , we address a strongly convex programming .
we prove it can guarantee exact low - rank matrix recovery .
the proposed optimization is @xmath14 where @xmath15 is some positive penalty parameter and @xmath16 is the orthogonal projection onto the linear subspace @xmath17 .
we also assume @xmath11 is a random subspace(the same assumption considered in paper @xcite ) . when @xmath18 in ( 2 ) , existence and uniqueness theorems is provided in the paper @xcite , as we list them below . in the end , how to choose suitable parameters in the optimization model ( 2 ) is discussed .
@xcite [ t1 ] _ fix any @xmath19 , and let @xmath11 be a @xmath12-dimensional random subspace of @xmath20 ; @xmath3 obeys incoherence condition with parameter @xmath21 , and @xmath22 . then with high probability , the solution of problem(2 ) with @xmath23 is exact , i.e. @xmath24 and @xmath25 , provided that _
@xmath26 _ where , @xmath27 , @xmath28 and @xmath29 are positive numerical constants and @xmath30 . _ we provide a brief summary of the notations which are used throughout the paper .
we denote the operator norm of matrix by @xmath31 , the frobenius norm by @xmath32 , the nuclear norm by @xmath33 , and the dual norm of @xmath34 by @xmath35 . the euclidean inner product between two matrices is defined by the formula @xmath36 .
note that @xmath37 .
the cauchy - schwarz inequality gives @xmath38 , and it is well known that we also have @xmath39 , e.g.@xcite @xcite .
linear transformations which act on the space of matrices are denoted by @xmath40 .
it s easy to see that the operator of @xmath41 is high dimension matrix in substance .
the operator norm of the operator is denoted by @xmath42 .
it should be noted that @xmath43 . the rest of the paper is organized as follows . in section 2
, we list many important lemmas and prove a key lemma on which our main result depends .
suggestions then is given in section 3 , which will guide us to choose suitable parameters in practical algorithms .
conclusion and further works are discussed in section 4 .
in this section , we first list some useful lemmas which will be used throughout this paper and then prove a main lemma . although the main lemma is similar to the corresponding one in the paper @xcite
, the construction of @xmath44 is different .
that leads to our necessary additional work .
[ @xcite , lemma 1][l1 ] _ suppose that @xmath45 .
let @xmath46 so that @xmath47 .
assume that @xmath48 and @xmath49 .
then , @xmath50 is the unique optimal solution to ( 2 ) if there exists a pair @xmath51 satisfying _ @xmath52 _ with @xmath53 , and @xmath54_. [ @xcite , lemma 3][l2 ] _ assume that @xmath55 for some small @xmath56 and the other conditions of theorem 1.2 hold true .
then , the matrix @xmath57 obeys , with high probability . + _
@xmath58 + @xmath59 + @xmath60 [ @xcite , lemma 4][l3 ] _ in addition to the assumptions in the previous lemma , assume that the signs of the non - zero entries of @xmath2 are i.i.d .
then , the matrix @xmath61 obeys , with high probability , + _ @xmath62 + @xmath63 the construction of @xmath57 and @xmath61 can be found in the paper @xcite .
the authors also introduce a new scheme to construct @xmath44 for the principal component pursuit .
however , the matrix @xmath44 constructed in the paper @xcite do not satisfy the requirement of our problem , so we have to modify this construction .
we first give explicit construction of @xmath44 , and then , prove the modification of @xmath44 satisfies the corresponding property .
_ construction of @xmath44 with least modification .
_ we define @xmath44 by the following least squares problem : @xmath64 where @xmath65 .
this construction of @xmath44 do nt satisfy theorem 2.6 only , but also has below lemma .
[ l4 ] _ assume @xmath66 , and that @xmath55 for some small @xmath56 and the assumptions of theorem 1.2 hold true
. then , the matrix @xmath44 obeys , with high probability .
+ _ @xmath67 + @xmath68 in proof of lemma 2.4 , we have to use two important lemmas which are listed below .
[ @xcite lemma 11 ] [ l5 ] let @xmath69 and @xmath70 be any three linear subspaces in @xmath20 satisfying @xmath71 , and @xmath72 and @xmath73 .
we define @xmath74 . then , we have @xmath75 [ @xcite lemma 7 ] assume that @xmath76 .
let @xmath11 be a linear subspace distributed according to the random subspace model .
then , with high probability , we have @xmath77 * a , bounding the behavior of @xmath78*. for convenience , let @xmath79 . + according to triangle inequality , we have @xmath80 in the last equality , we have used @xmath81 .
note that @xmath82 according to the derivation in the paper @xcite , with high probability , we can obtain @xmath83 putting those all together , we get @xmath84 combining with @xmath66 , we can obtain @xmath85 because @xmath44 is the optimum solution of least squares problem , we can use the convergent neumann series expansion .
it s easy to note that @xmath86 according to triangle inequality , we have @xmath87 * b , estimating the first inequality of lemma 2.4*. in order to bound @xmath88 , we have to bound the behavior of @xmath89 .
therefore , we have @xmath90 according to lemma 2.5 , we have , for any @xmath91 , with high probability , @xmath92 according to the paper@xcite , we have @xmath93 with high probability . next
, we will bound @xmath94 . according to the paper @xcite
, @xmath95 has the same distribution as @xmath96 , where @xmath97 is a random gaussian matrix with i.i.d .
entries @xmath98 .
therefore , we can obtain @xmath99 together with lemma 2.6 , we can obtain @xmath100\le e^{-\frac{n^2}{32}}\nonumber\end{aligned}\ ] ] it s easy to note that any entries of @xmath101 have the same distribution as @xmath102 , where @xmath103 are independent identically distributed .
it is obvious to see that @xmath104 and @xmath105 therefore , @xmath102 is distributed according to @xmath106 , where @xmath79 . for simplicity , we define @xmath107 . using the jesen inequality , we have @xmath108\le ( \mathbb e[\|z\|_2 ^ 2])^{1/2}=\sqrt{\frac{p\xi}{n^2}}\nonumber\end{aligned}\ ] ] according to the proposition 2.18 in @xcite
, we can obtain @xmath109 + t\sqrt{\frac{\xi}{n^2}}\right ] \le e^{-t^2/2}\nonumber\end{aligned}\ ] ] setting @xmath110 , after a simple calculation , we can obtain @xmath111 with high probability . for sufficiently large @xmath112 , the first inequality of lemma 2.4 is established . +
* c , estimating the second inequality of lemma 2.4 * , note that @xmath113 similar to the paper @xcite , after a simple calculation , we can obtain @xmath114 where @xmath115 is some constant .
note that for sufficiently large @xmath112 , the second inequality of lemma 2.4 is established .
in this section , we shall provide sufficient conditions under which @xmath117 is the unique solution of the strongly convex programming ( 2 ) with high probability .
afterwards , an explicit lower bound of @xmath116 will be given as well , which will guide us to choose suitable parameters in practical algorithms .
[ result1 ] _ suppose that @xmath45 .
let @xmath46 so that @xmath47 .
assume that @xmath48 and @xmath49 .
if there exists a pair @xmath51 and a matrix @xmath118 satisfying _
@xmath119 _ with _ @xmath120 _ where @xmath121 , @xmath122 are positive parameters satisfying _ @xmath123
_ then @xmath50 is the unique solution of the strongly convex programming ( 2)_. for any feasible perturbation @xmath124 , it s easy to note that @xmath125 .
according to the definition of @xmath126 , we have @xmath127 , therefore @xmath128 . for simplicity ,
let @xmath129 , we can obtain @xmath130 in the second inequality above , we have used the facts @xmath131 in the third inequality above , we have used @xmath125 .
+ we will bound @xmath132 . according to the definition of @xmath126 ,
we have @xmath133 therefore @xmath134 putting those all together , we get @xmath135 this , together with ( 6 ) , implies that @xmath136 is a solution to ( 2 ) .
the uniqueness follows from the strong convexity of the objective in ( 2 )
. we will provide the criterion of the value of @xmath116 .
[ result1 ] _ let @xmath137 , @xmath138 , and @xmath139 .
@xmath140 then , under the other assumptions of theorem 1.1 , @xmath50 is the unique solution to the strongly convex programming ( 2 ) with high probability . in order to check the conditions in theorem 3.1 , we will prove there exists a matrix @xmath141 obeying @xmath142 note that @xmath143 with @xmath57 , @xmath61 and @xmath44 have analytical form constructed in the paper @xcite .
we will check above conditions hold true one by one . for simplicity of proof
, we denote @xmath144 without loss of generality , let @xmath145 . with the help of the construction @xcite of @xmath57 , @xmath61 and @xmath44 ,
it is easy to check the first and second conditions hold true . with respect to the third condition , according to the paper @xcite , we have @xmath146 and @xmath147 . according to the modification of @xmath44 constructed in lemma 2.4
, we have @xmath148 .
it s easy to check that @xmath149 , which implies that the third condition holds true .
consequently , we will provide the last two conditions also hold true under some suitable assumptions . pertaining to the fourth inequality
, we have @xmath150 for the last inequality , noting that @xmath151 and @xmath152 as shown in @xcite , we can obtain @xmath153 in order to satisfy the condition ( 8) , we choose a @xmath116 obeying @xmath154 therefore @xmath155 combining ( 9 ) with ( 6 ) , we can obtain @xmath156 therefore @xmath157 together with ( 10 ) and ( 11 ) , the theorem 3.2 is established . in order to simplify the formula ( 7 ) ,
we suppose @xmath158 and @xmath159 , which satisfy the conditions above . therefore @xmath160 however , note that the exact lower bound is very hard to get , because we only have the information about the given data matrix @xmath161 . noting that @xmath162 and according to the paper @xcite , we have @xmath163 therefore , we can choose @xmath164 it s obvious that @xmath165 .
therefore , we can obtain the result as follows .
[ tau2 ] _ assuming _
@xmath166 _ and the other assumptions of theorem 1.1 , @xmath50 is the unique solution to the strongly convex programming ( 2 ) with high probability . _
in this paper , we have studied strongly convex programming for principal component pursuit with reduced linear measurements .
we first provide sufficient conditions under which the strongly convex models lead to the exact low rank and sparse components recovery ; second , we give the criterion of the choice of the value of @xmath116 , which gives very useful advice on how to set the suitable parameters in designing efficient algorithms .
especially , it is easy to note that the main results of paper @xcite is only the special case of ours . in some sense
, we extend the result of choosing suitable parameters to the general problem .
we would like to thank the reviewers very much for their valuable comments and suggestions .
this research was supported by the national natural science foundation of china ( nsfc ) under grant 61172140 , and 985 key projects for excellent teaching team supporting ( postgraduate ) under grant a1098522 - 02 .
yipeng liu is supported by fwo phd / postdoc grant : g0108.11(compressed sensing ) . | in this paper , we address strongly convex programming for principal component pursuit with reduced linear measurements , which decomposes a superposition of a low - rank matrix and a sparse matrix from a small set of linear measurements .
we first provide sufficient conditions under which the strongly convex models lead to the exact low - rank and sparse matrix recovery ; second , we also give suggestions on how to choose suitable parameters in practical algorithms .
qingshan you and qun wan yipeng liu ( communicated by the associate editor name ) |
at the time of discovery , supernova ( sn ) 2006gy was the most luminous sn ever found @xcite
. generated a great deal of interest ; in addition to being @xmath3100 times more luminous than a typical type ii ( hydrogen - rich , core - collapse ) sn at peak , it exhibited a long rise time ( @xmath370 day ) and slow decline , leading to speculation that it may have been the first observed example of a pair - instability sn ( @xcite ) or a pulsational pair - instability sn @xcite .
was classified as a type iin sn ( see @xcite for a definition of the type iin subclass and @xcite for a review of its spectral properties ) based on the relatively narrow emission features present in the early - time sn spectrum @xcite . some type iin supernovae ( sne iin ) are known to be overluminous relative to their typical sn ii counterparts , probably due to the collision of fast - moving sn ejecta and a dense , possibly clumpy , circumstellar medium ( csm ; @xcite ) . in a companion paper @xcite , a detailed spectroscopic comparison of is made to other sne iin .
is unique within the sn iin subclass , however , because typical interaction models can not explain its early - time behavior , suggesting the need for alternative models for this particular object @xcite .
pair - instability sne @xcite are expected to occur in very massive , low - metallicity stars , such as those that may have been present in the metal - free environment of the very early universe ( e.g. , @xcite ) .
the detection of a pair - instability sn in the comparatively local universe , then , could potentially reveal a great deal about the first generation of stars .
the light curves of pair - instability sne are expected to exhibit a relatively slow rise , followed by a broad turnover after the peak , and a peak luminosity that is considerably larger than that of typical sne @xcite .
qualitatively each of these characteristics matches those observed for .
other more luminous sne have been announced since the discovery of : sne 2005ap @xcite and 2008es @xcite .
this suggests that while these events are rare , there may be a new subclass of very luminous supernovae ( vlsne ) .
the peak luminosity and photometric evolution of both sne 2005ap and 2008es are difficult to explain via the pair - instability model @xcite , implying that peak luminosities @xmath5 few @xmath8 erg s@xmath9 are possible without a pair - instability explosion .
there is a wide diversity of alternative models that have been developed to explain the early - time observations of .
@xcite argue that the peak luminosity and light - curve evolution can be explained via the thermalization of shock energy deposited into a massive ( @xmath310 ) , optically thick shell .
similar explanations were offered for sne 2005ap and 2008es @xcite . based on a model of the light curve near peak ,
@xcite suggest that the combination of ejecta colliding with dense clumps in the csm and @xmath313 of @xmath4ni are responsible for the early - time luminosity of .
@xcite are unable to match the light curve with a standard pair - instability sn model ; however , they can reasonably reproduce the ( @xmath7400 day ) light curve of using a pair - instability model where they artificially reduce the ejecta mass , such that the radioactive heating is less than 100% efficient .
@xcite use a model of a pulsational pair instability within a massive star to explain the light curve of .
some of the models make predictions for the late - time behavior of .
those with a large yield of @xmath4ni @xcite would expect a decline in bolometric luminosity at the rate of @xmath4co deay , 0.98 mag ( 100 day)@xmath9 , or faster if the radioactive - decay energy is not converted to optical emission with 100% efficiency .
the shell - shock model @xcite predicts a rapid decline after @xmath3200 day , though the authors note that this decline may be offset by the production of a large amount of @xmath4ni or continued csm interaction . the pulsational pair - instability model @xcite predicts another sn explosion at the location of about 9 years after the initial outburst from .
more than a year after the explosion , @xcite detected in the near - infrared ( nir ) at a luminosity comparable to that of the peak luminosity of most sne ii .
this property had not been predicted by any of the models . when coupled with the lack of a detection in the radio and x - rays , this led @xcite to conclude that the luminosity could not be powered by the continued interaction of the sn ejecta with csm . @xcite
draw similar conclusions based on their detection of weak h@xmath10 emission at a comparable epoch .
@xcite conclude that only two possibilities are able to explain the late - time observations of : ( i ) the explosion produced @xmath52.5 of @xmath4ni , possibly as a pair - production sn , which was heating dust and consequently generating the large nir excess , or ( ii ) a massive ( @xmath3510 ) , dusty shell , located @xmath31 light year from the site of the sn , was being heated by the radiation produced at peak , and re - radiated that energy as a nir echo @xcite .
for the first case , if the luminosity were powered by radioactive decay , then future observations should indicate a continued decline at the rate of @xmath4co decay .
a dust echo , on the other hand , would result in a nir light curve that stays roughly constant for @xmath312 years before exhibiting a rapid decline
. evolved massive stars , such as red super giants and luminous blue variables ( lbvs ) , are often observed with massive dust shells , so if these stars explode as type iin sne one might expect a late - time ir echo .
many sne iin have been observed to exhibit a late - time nir excess @xcite which in some cases lasted @xmath0 1 year .
this excess has been attributed to nir echoes @xcite , though we note that the formation of new dust has been argued specifically for sn 1998s @xcite and sn 1995n @xcite .
dusty regions near the sn should also lead to uv / optical scattered light echoes @xcite ; thus , dust is capable of providing significant optical and nir emission at late times .
the optical decline of type iin sne at late times is very heterogeneous @xcite , and therefore caution must be applied when determining the source of any late - time emission . in this paper
we present new nir observations of , taken around the peak of optical emission , as well as optical and nir observations obtained more than two years after exploded . in @xmath11 2
we describe the observations and data reduction .
we discuss the results in @xmath11 3 , and in @xmath11 4 we offer some conclusions . throughout this paper
we assume that the distance to ngc 1260 ( the host galaxy of ) is 73.1 mpc , and that the total color excess toward is @xmath12 mag @xcite , of which @xmath13 mag is in ngc 1260 and galactic extinction accounts for @xmath14 mag .
all spectral energy distributions ( seds ) have been corrected for this color excess assuming @xmath15 using the reddening law of @xcite .
nir observations of were obtained simultaneously in with the peters automated infrared imaging telescope ( pairitel ; @xcite ) starting on 2006 october 13 ( 54 day after explosion ) .
pairitel is a 1.3-m robotic telescope , located on mt .
hopkins , az , which obtained observations for 30 min to 1 hr at each epoch through normal queue - scheduled operations for the next 24 months . in total
there were 124 epochs obtained .
all images were processed via an automated pipeline @xcite .
analysis of has proved challenging because the sn is located very close to the nucleus of ngc 1260 ( separation @xmath16 1@xmath17 ; @xcite ) .
pairitel has a large native scale of 2@xmath17 pixel@xmath9 , precluding spatial resolution of the sn from the nucleus .
this necessitates image differencing to obtain the light curve of the sn .
image subtraction was performed with hotpants , and the flux in each difference pair was determined via aperture photometry at the location of the sn .
an example subtraction , which clearly shows flux from after the reference image has been subtracted , is shown in figure [ ptel - subtract ] . despite observations extending more than two years past the date of discovery ,
has not faded beyond the point of detectability with pairitel .
consequently , all images of contain some flux from the sn .
we thus adopted the `` nn2 method '' of @xcite to determine the relative nir flux changes of the sn .
the nn2 method treats all images equally and does not require a template image with no light from the source of interest , .
it uses the subtraction of all @xmath18 possible pairs of images to mitigate against possible errors associated with the use of a single reference template image .
the downside to the nn2 method is that it only produces the differential flux between each of the @xmath19 epochs of observations . to convert these flux differences into magnitudes requires an absolute calibration , which must be obtained independently of the results from the nn2 method .
uncertainties in the individual subtractions were estimated by measuring the scatter in fake sne inserted at locations having a surface brightness similar to that at .
early - time nir observations of ( defined here as those made before passed behind the sun during 2007 ) show a remarkably flat light curve , as seen in figure [ early - lc ] . to transform the relative flux differences from the nn2 method to an absolute scale
, we subtracted the archival two micron all sky survey ( 2mass ; @xcite ) image from each of the images obtained on 2006 nov . 13.24 , 14.25 , 15.38 , and 16.37 to determine the magnitudes of the sn on these dates .
the mean sn flux was then used to transform the relative flux from the nn2 method to the absolute flux of the 2mass system @xcite .
pairitel uses the old 2mass camera and telescope ; hence , we do not expect any large systematic effects in the 2mass subtractions .
the final calibrated photometry is summarized in table [ ptel - data ] . on 2006 nov .
01 , @xcite obtained palomar hale ( 5 m ) ao images in the @xmath20 and @xmath21 bands , clearly resolving from the host - galaxy nucleus , and measured its flux . in the @xmath21 band our calibration and the @xcite measurement
agree to within 1@xmath22 , while the agreement in the @xmath20 band is somewhat worse ( @xmath32@xmath22 ) .
@xcite had only a single 2mass star within the field of view of their ao images , whereas @xmath0100 2mass stars were used to calibrate the pairitel images ; when coupled with the difficulty associated with photometry of ao images , this may explain the differences between the two measurements .
we note that were we to adopt the @xcite measurements as our calibration , there would be an overall systematic shift of our @xmath20 and @xmath21 light curves to brighter values , which in turn would lead to worse agreement between the nir data and early - time optical spectra ( see figure [ early - sed ] ) .
this suggests that our calibration method is sufficient .
we note that the uncertainties in our photometry are dominated by the uncertainty in the calibration , which is @xmath30.03 mag in @xmath20 , @xmath30.06 mag in @xmath23 , and @xmath30.04 mag in @xmath21 .
this uncertainty is the same for all epochs , so a change in the calibration would lead to a systematic shift of the entire light curve .
rccc @xmath24 & @xmath20 mag & @xmath23 mag & @xmath21 mag + ( mjd ) & ( vega ) & ( vega ) & ( vega ) + 54022.24 & 13.33 @xmath25 0.04 & 13.05 @xmath25 0.06 & 12.81 @xmath25 0.05 + 54023.24 & 13.33 @xmath25 0.04 & 13.08 @xmath25 0.09 & 12.83 @xmath25 0.05 + 54025.26 & 13.29 @xmath25 0.04 & 13.03 @xmath25 0.06 & 12.84 @xmath25 0.06 + 54028.17 & 13.32 @xmath25 0.04 & 13.03 @xmath25 0.06 & 12.78 @xmath25 0.04 + 54029.19 & 13.26 @xmath25 0.04 & 13.03 @xmath25 0.06 & 12.81 @xmath25 0.05 + 54030.41 & 13.27 @xmath25 0.03 & 13.02 @xmath25 0.06 & 12.81 @xmath25 0.06 + 54031.43 & 13.27 @xmath25 0.04 & 13.01 @xmath25 0.06 & 12.79 @xmath25 0.04 + 54032.41 & 13.27 @xmath25 0.04 & 13.01 @xmath25 0.06 & 12.81 @xmath25 0.05 + 54036.21 & 13.25 @xmath25 0.04 & 13.02 @xmath25 0.06 & 12.78 @xmath25 0.04 + 54037.20 & 13.21 @xmath25 0.05 & 13.00 @xmath25 0.06 & 12.77 @xmath25 0.04 + 54038.24 & 13.25 @xmath25 0.03 & 13.02 @xmath25 0.06 & 12.76 @xmath25 0.04 + 54040.24 & 13.23 @xmath25 0.04 & 13.01 @xmath25 0.06 & 12.77 @xmath25 0.04 + 54040.69 & 13.25 @xmath25 0.04 & 13.01 @xmath25 0.06 & 12.77 @xmath25 0.05 + 54041.71 & 13.29 @xmath25 0.04 & 13.01 @xmath25 0.06 & 12.78 @xmath25 0.04 + 54042.71 & 13.22 @xmath25 0.05 & 13.03 @xmath25 0.08 & 12.79 @xmath25 0.09 + 54044.66 & 13.26 @xmath25 0.04 & 13.00 @xmath25 0.06 & 12.74 @xmath25 0.04 + 54045.66 & 13.28 @xmath25 0.05 & 13.00 @xmath25 0.06 & 12.79 @xmath25 0.05 + 54046.65 & 13.33 @xmath25 0.04 & 13.02 @xmath25 0.07 & 12.80 @xmath25 0.05 + 54047.64 & 13.26 @xmath25 0.03 & 12.99 @xmath25 0.06 & 12.76 @xmath25 0.05 + 54048.62 & 13.29 @xmath25 0.04 & 13.02 @xmath25 0.06 & 12.74 @xmath25 0.05 + 54049.62 & 13.27 @xmath25 0.04 & 13.01 @xmath25 0.06 & 12.78 @xmath25 0.05 + 54050.66 & 13.25 @xmath25 0.03 & 13.00 @xmath25 0.06 & 12.78 @xmath25 0.04 + 54051.70 & 13.25 @xmath25 0.04 & 13.00 @xmath25 0.06 & 12.76 @xmath25 0.04 + 54052.64 & 13.24 @xmath25 0.04 & 13.00 @xmath25 0.06 & 12.77 @xmath25 0.04 + 54053.66 & 13.25 @xmath25 0.03 & 13.01 @xmath25 0.06 & 12.76 @xmath25 0.04 + 54054.78 & 13.27 @xmath25 0.04 & 13.01 @xmath25 0.06 & 12.76 @xmath25 0.04 + 54055.78 & 13.28 @xmath25 0.04 & 13.01 @xmath25 0.06 & 12.77 @xmath25 0.04 + 54058.57 & 13.27 @xmath25 0.03 & 13.00 @xmath25 0.06 & 12.75 @xmath25 0.04 + 54059.59 & 13.24 @xmath25 0.04 & 13.01 @xmath25 0.06 & 12.74 @xmath25 0.05 + 54060.61 & 13.27 @xmath25 0.03 & 13.01 @xmath25 0.06 & 12.79 @xmath25 0.04 + 54061.60 & 13.24 @xmath25 0.04 & 13.01 @xmath25 0.06 & 12.75 @xmath25 0.05 + 54063.63 & 13.28 @xmath25 0.04 & 13.00 @xmath25 0.06 & 12.74 @xmath25 0.05 + 54064.63 & 13.30 @xmath25 0.04 & 13.01 @xmath25 0.06 & 12.76 @xmath25 0.04 + 54065.63 & 13.24 @xmath25 0.07 & 13.08 @xmath25 0.11 & 12.83 @xmath25 0.19 + 54066.65 & 13.31 @xmath25 0.04 & 13.01 @xmath25 0.06 & 12.76 @xmath25 0.05 + 54069.55 & 13.27 @xmath25 0.04 & 13.04 @xmath25 0.06 & 12.82 @xmath25 0.05 + 54070.96 & 13.31 @xmath25 0.04 & 13.02 @xmath25 0.06 & 12.77 @xmath25 0.04 + 54071.98 & 13.31 @xmath25 0.05 & 13.02 @xmath25 0.06 & 12.83 @xmath25 0.05 + 54072.97 & 13.28 @xmath25 0.04 & 13.01 @xmath25 0.06 & 12.77 @xmath25 0.04 + 54074.92 & 13.31 @xmath25 0.04 & 13.03 @xmath25 0.06 & 12.82 @xmath25 0.05 + 54075.96 & 13.31 @xmath25 0.04 & 13.01 @xmath25 0.06 & 12.77 @xmath25 0.05 + 54076.95 & 13.37 @xmath25 0.04 & 13.02 @xmath25 0.06 & 12.79 @xmath25 0.04 + 54077.95 & 13.37 @xmath25 0.04 & 13.02 @xmath25 0.06 & 12.80 @xmath25 0.04 + 54080.01 & 13.34 @xmath25 0.05 & 13.03 @xmath25 0.06 & 12.80 @xmath25 0.05 + 54081.01 & 13.41 @xmath25 0.05 & 13.06 @xmath25 0.06 & 12.80 @xmath25 0.04 + 54082.01 & 13.40 @xmath25 0.04 & 12.98 @xmath25 0.06 & 12.82 @xmath25 0.05 + 54082.99 & 13.32 @xmath25 0.07 & 13.07 @xmath25 0.07 & 12.79 @xmath25 0.05 + 54083.99 & 13.36 @xmath25 0.09 & 13.00 @xmath25 0.06 & 12.75 @xmath25 0.05 + 54084.97 & 13.39 @xmath25 0.05 & 13.05 @xmath25 0.06 & 12.84 @xmath25 0.08 + 54085.96 & 13.40 @xmath25 0.05 & 13.08 @xmath25 0.07 & 12.84 @xmath25 0.05 + 54086.96 & 13.48 @xmath25 0.06 & 13.06 @xmath25 0.07 & 12.88 @xmath25 0.06 + 54090.91 & 13.42 @xmath25 0.04 & 13.11 @xmath25 0.07 & 12.88 @xmath25 0.05 + 54091.91 & 13.59 @xmath25 0.09 & 13.11 @xmath25 0.07 & 12.93 @xmath25 0.05 + 54093.12 & 13.51 @xmath25 0.11 & 13.06 @xmath25 0.06 & 12.90 @xmath25 0.07 + 54094.95 & 13.68 @xmath25 0.11 & 13.12 @xmath25 0.07 & 12.90 @xmath25 0.05 + 54095.95 & 13.66 @xmath25 0.11 & 13.13 @xmath25 0.07 & 13.00 @xmath25 0.16 + 54100.97 & 13.79 @xmath25 0.08 & 13.15 @xmath25 0.07 & 13.03 @xmath25 0.13 + 54101.42 & 13.81 @xmath25 0.07 & 13.16 @xmath25 0.07 & 13.16 @xmath25 0.14 + [ ptel - data ] as shown by @xcite , and subsequently confirmed by @xcite , the nir evolution of is very slow at late times . given the relatively small change in flux , and the reduced signal - to - noise ratio following the fading of the sn , we were unable to recover reliable flux measurements from pairitel data taken after 2007 sep .
furthermore , unlike the case at early times , the sn had faded below the 2mass detection limit , which means that the late - time subtractions relative to the 2mass template image do not yield meaningful results despite the fact that at @xmath26 the sn is well above the pairitel detection limit . in principle , if deep pairitel images are obtained after fades well beyond the detection limit , it should be possible to recover the late - time nir light curve using template images that contain no light from the sn .
we also observed with the near - infrared camera 2 ( nirc2 ) using the laser guide star ( lgs ) ao system @xcite on the 10-m keck ii telescope in hawaii on 2008 aug .
we have re - reduced the lgs ao images presented by @xcite and derive revised values for the @xmath27-band magnitudes of , as summarized in table [ tbl - keckao].-band .
the uncertainty associated with the transformation between @xmath27-band and @xmath21-band is small compared to the absolute calibration uncertainty and therefore for the late - time ao images @xmath28 . ]
we also present the first measurement of the @xmath23-band flux from the ao images taken on 2007 dec .
2 . to obtain this @xmath23-band measurement , despite a lack of 2mass stars in the field , we measured the @xmath29 color of relative to the @xmath29 color of the host galaxy , and calibrated this against the @xmath30 color of the galaxy in the archival 2mass images .
the large uncertainty for this measurement reflects the accuracy with which we can determine the color of the galaxy from the 2mass images .
lccc date & epoch & filter & mag + ( ut ) & ( day ) & & ( vega ) + 2007 sep .
29 & 401 & @xmath27 & 14.91 @xmath25 0.17 + 2007 dec .
02 & 469 & @xmath23 & 16.8 @xmath25 0.3 + 2007 dec .
02 & 469 & @xmath27 & 15.02 @xmath25 0.17 + 2008 aug .
25 & 736 & @xmath27 & 15.59 @xmath25 0.21 + [ tbl - keckao ] as part of a _ hubble space telescope ( hst ) _ snapshot survey ( go-10877 ; pi li ) , was observed with the wide field planetary camera 2 ( wfpc2 ; @xcite ) using the @xmath31 , @xmath32 , @xmath33 , and @xmath34 filters on 2008 nov .
the data were reduced in the standard fashion using * multidrizzle * @xcite .
ngc 1260 and sn 2006gy were located at the center of the pc chip of wfpc2 , which has a native pixel scale of 0.0455@xmath17 [email protected] follow the recipe of @xcite to do charge transfer efficiency correction and photometric reduction of the wfpc2 images .
our measurements of the magnitudes are summarized in table [ tbl - hst ] .
a false - color image of this detection is shown in figure [ false - color ] .
lcccc filter & @xmath35 & @xmath36 & mag & flux + & ( ) & ( ) & ( vega ) & ( @xmath37jy ) + @xmath31 & 4519 & 957 & 22.16 @xmath25 0.03 & 5.97 @xmath25 0.18 + @xmath32 & 5397 & 1226 & 21.51 @xmath25 0.02 & 9.31 @xmath25 0.19 + @xmath33 & 6697 & 866 & 21.07 @xmath25 0.03 & 11.29 @xmath25 0.34 + @xmath34 & 7924 & 1500 & 20.91 @xmath25 0.04 & 10.80 @xmath25 0.43 +
while the early evolution of is remarkably flat in the nir ( see figure [ early - lc ] ) , we find that these measurements are consistent with radiation from a cooling blackbody . from blackbody
fits to optical spectra , @xcite find that the temperature of monotonically cools after day @xmath350 , roughly two weeks before optical maximum , until day @xmath3165 , where the temperature levels off at @xmath36300 k. in the top panel of figure [ early - sed ] we show the evolution of the sed of for three epochs ( day 59 , 109 , and 135 ) during our early - time nir observations . for each epoch
we show the single - component blackbody spectrum , after adopting the temperature from @xcite and normalizing the spectra to the photometric measurements from @xcite .
their photometric observations come from a series of unfiltered observations taken with the katzman automatic imaging telescope ( kait ; @xcite ) , which are best matched by the @xmath38 band @xcite
. however , the scatter in the transformation between kait unfiltered and @xmath38 can be quite large @xcite , so we adopt a 0.2 mag uncertainty for the calibration of the blackbody spectra .
we also show the nir flux during these epochs , determined from the absolute calibration of the 2mass system @xcite .
this shows excellent agreement with an extrapolation of the early - time spectra of @xcite . between day @xmath355135 roughly 24% of the bolometric luminosity of
was emitted in the nir . in the lower panel of figure [ early - sed ]
we show the fractional excess of the photometric observations relative to the single - component blackbodies ( note that by definition the @xmath38-band excess is set to zero ) . for clarity
we do not show the uncertainties associated with the excess , but after accounting for the large uncertainty in the @xmath38-band calibration , each point is within @xmath31@xmath22 of zero excess .
nevertheless , there is an apparent trend that the excess is growing in the @xmath23 and @xmath21 bands as a function of time .
the trend may be indicative of radiation from warm dust ( see below ) , though we note that this effect would be small .
if this trend toward a nir excess is created by the same source as the late - time nir excess , we would expect a @xmath21-band excess of @xmath30.17 mag , comparable to the uncertainty in our blackbody calibration . with data obtained more than a year after explosion
, @xcite discovered a significant nir excess from .
our nir observations taken 736 day after explosion show that the @xmath27-band flux from has only faded by a factor of @xmath32 over the course of the previous year .
we can not determine the total ir luminosity at late times , because our nir data do not cover the peak of the sed .
furthermore , as first noted by @xcite , the very red @xmath39 color at late times indicates that the ir emission likely peaks in the mid - ir .
we can , however , place lower limits on the nir luminosity by assuming that the sed peaks in the @xmath27 band . following this assumption
, the nir excess constitutes a slowly varying luminosity of @xmath40 for @xmath0 1 yr .
the most likely explanation for this large luminosity in the nir is warm dust .
the observed @xmath29 color from day 436 corresponds to a temperature of @xmath31000 k , assuming a single - temperature blackbody .
if the dust is radiating as a single blackbody with @xmath41 k , this would increase the above luminosity to @xmath42 .
we note that our nir observations are only sensitive to the warmest dust ; it is possible that cooler dust with @xmath43 k k. ] could dominate the ir emission , in which case the luminosity could be significantly higher than the values quoted above .
the location of this dust , and whether it is newly formed or pre - existing at the time of the sn explosion , remain to be determined .
the dust - cooling time is short , meaning that a prolonged heat source is needed to explain the extended excess .
we consider four possibilities for heating the dust : ( i ) radioactive heating from @xmath4co decay , ( ii ) collisional excitation of pre - existing dust , ( iii ) heating via radiation from circumstellar interaction , and ( iv ) a late - time ir dust echo , where pre - existing dust is heated by the radiation produced while the sn was near its optical peak .
@xcite noted that the observed @xmath27-band decay between 2007 september and 2007 december could be explained with radioactive heating from a minimum of 2.5 of @xmath4ni , if a sufficient amount of dust formed in order to move the luminosity into the nir ( though they favored another interpretation ; see below ) .
we show the bolometric evolution of through the first @xmath3800 day post - explosion , including both optical and nir detections , as well as the expected light curve from 2.5 of @xmath4ni , in figure [ bolo - lc ] .
the early - time measurements ( @xmath16 250 day ) come from @xcite .
the optical measurement near day 400 comes from photometric measurements by @xcite , while the optical luminosity on day 825 comes from a direct integration of our _ hst _ observations ( see @xmath11 3.3.1 ) .
the three late - time nir luminosities represent lower limits based on the measured @xmath27-band flux ( see above , @xmath11 3.2 ) .
the decay rate of @xmath4co , 0.98 mag ( 100 day)@xmath9 , is much faster than the observed decay of the @xmath27 flux from , @xmath30.2 mag ( 100 day)@xmath9 .
the late - time nir excess declines at a rate that is too slow to be explained by @xmath4co heating alone .
the pair - instability model of @xcite , which provided good agreement with the early light curve of after an artificial reduction of the total ejecta mass from the evolutionary calculation , was able to reproduce the late - time nir luminosity observed by @xcite .
this model required less than 100% efficiency in the conversion of gamma - rays ( from radioactive decay ) to optical / nir emission , which means that the light curve should decay _
faster _ than 0.98 mag ( 100 day)@xmath9 . the possibility of a pair - instability sn was first invoked to explain the large peak luminosity of @xcite .
this scenario would have required the production of @xmath5 10 @xmath4ni , which in turn would produce a large late - time luminosity that decays at the rate of @xmath4co .
the late - time nir luminosity is not accounted for in either the general pair - instability models or the artificial model of @xcite ; it therefore provides a serious challenge to the pair - instability sn hypothesis .
another possibility is that the dust existed prior to the sn explosion , at large distances from the explosion site , and was heated via collisions with the expanding material in the expanding blast wave .
the intense ultraviolet / optical output from a sn at its peak vaporizes any dust in the vicinity of the sn @xcite .
this radiation near peak creates a dust - free cavity into which the sn ejecta may expand at early times ; however , the ejecta blast wave will eventually reach the edge of the dust - free cavity , at which point collisional excitations may generate nir emission .
the large peak luminosity of , @xmath44 @xcite , provides significant challenges to this collisional heating scenario .
@xcite shows that the radius of the dust - free cavity can be determined from the peak luminosity and the dust vaporization temperature , assuming the grain emissivity @xmath45 is proportional to @xmath46 : @xmath47^{0.5 } , \label{rad - eqn}\ ] ] where @xmath48 is the radius of the dust - free cavity , @xmath49 is the mean grain emissivity , @xmath50 is the peak luminosity , and @xmath51 is the dust vaporization temperature .
following @xcite , if we assume @xmath52 and that @xmath53 @xmath54 , we find that @xmath55 cm , for a vaporization temperature @xmath56 k. if the absorption coefficient is instead proportional to @xmath57 , then the cavity becomes even larger .
based on the observed width of the h@xmath10 line from , @xcite estimate the speed of the blast wave to be 4000 km s@xmath9 , which would mean that this material would take @xmath370 yr to reach the edge of the dust - free cavity .
even if there are ejecta traveling at more typical sn velocities of 10,000 km s@xmath9 , it would take over 28 yr for this material to reach the edge of the dust - free cavity .
given that the nir excess is present @xmath31 yr following explosion , and that the sn ejecta could not reach the dust in such a short time , collisional heating of dust by sn ejecta can not explain the observed nir excess .
the emission features that arise in sne iin , as in , originate from the interaction of the sn ejecta with a dense csm .
shocked gas , located between the forward shock which is plowing into the csm and a reverse shock of the sn ejecta , can cool , and if the density is large enough , it may form new dust grains .
the cooling time for such dust is short ( @xmath310100 s ) , but a persistent heat source can generate an extended period of nir emission . the physical conditions for post - shock dust formation are not necessarily easy to produce : strongly interacting sne are often x - ray sources , which could prevent the formation of new dust grains in the post - shock gas . in a few cases ,
however , evidence for the formation of dust in the post - shock region has been found for sne 1998s @xcite , 2006jc @xcite , and 2005ip @xcite , and possibly also sn 2006tf @xcite .
this newly formed dust is then heated by the energy from the shock as it continues to propagate into the csm , which gives rise to the prolonged nir excess .
this scenario is difficult to reconcile with the case of sn 2006gy , however .
spectra taken around day 200 show a decline in the luminosity of the h@xmath10 emission line , which indicates a reduction in the csm interaction at this time @xcite . furthermore ,
as detailed by @xcite , the lack of x - ray , h@xmath10 , and radio emission a little more than a year after the sn explosion implies that the observed nir luminosity can not be explained by shock - heated dust .
in fact , the radio nondetection of continues through day 648 @xcite .
@xcite do claim the detection of very weak h@xmath10 emission from on day 394 , but the inferred luminosity of @xmath310@xmath58 erg s@xmath9 is nearly three orders of magnitude lower than the observed nir luminosity at this time .
@xcite also claim an @xmath38-band detection of 19.4 mag on day 394 ( though we note that @xcite report an upper limit of @xmath59 20 mag , and @xcite give an upper limit of @xmath59 20.3 mag , at a similar epoch ) , which would constitute an optical luminosity similar to that observed in the nir .
the source of this optical luminosity could potentially be the heat source for newly formed dust at day @xmath3400 , though it can not explain the similar nir luminosity seen on day 736 . while was detected in the optical at @xmath0 800 day to have a luminosity similar to that seen in the nir , the very blue nature of optical data suggest that it is a scattered light echo ( see below ) of uv / optical emission from the sn peak .
the light echo must originate _ outside _ the dust - forming region and therefore can not heat the dust . while we can not strongly rule out dust formation in the post - shock region of it is implausible to explain the late - time nir luminosity via new dust , because there are no signs of a heating source that lasts for @xmath0 1 yr .
perhaps the most natural explanation for the late - time nir excess is a dust - heated echo , as first discussed by @xcite .
the qualitative behavior of ir echoes predicted by @xcite provides an excellent match to the observed behavior of .
the ir emission comes from dust near the explosion site which is heated by the early - time uv / optical emission from the sn .
the model predicts a fast rise in the ir , followed by an extended plateau ( see below ) before the ir luminosity follows a rapid decline
. exhibits a very slow decline in the nir , only a factor of @xmath32 in flux over the course of @xmath31 yr , consistent with being on the extended plateau phase as predicted by the model .
the plateau occurs because the hottest dust , which exists right at the edge of the dust - free cavity described above , dominates the emission .
according to the view of a distant observer , the emitting volume is a series of paraboloid light fronts that expand throughout the dust - free cavity ( see figure 1 of @xcite or figure 4 of @xcite ) .
as a paraboloid expands it continually heats dust at @xmath48 , which is what gives rise to the plateau in the light curve . after a time @xmath32@xmath48/@xmath60
, the vertex of a given paraboloid will reach the back edge of the cavity and will no longer be heating dust at @xmath61 .
emission from the dust will be dominated by the uv / optical radiation produced by the sn at peak , and once this radiation sweeps past the back edge of the cavity the ir luminosity begins to rapidly decline .
therefore , if we know the duration of the plateau phase we can determine the size of the dust - free cavity .
constraints on the cavity size can be determined by assuming that the echo starts at our first @xmath27 observation on day 405 and ends on our last observation on day 736 . in this case
we find that @xmath62 cm , however , it would be contrived if our observations were to perfectly bookend the plateau phase of the echo : the echo almost certainly started prior to day 405 and continued beyond day 736 . if we instead assume that the slight nir excess seen at day @xmath3100 ( see figure [ early - lc ] ) is the rise of an ir echo , then we find that @xmath63 cm , which shows reasonable agreement with the value we calculated for the dust - free cavity in @xmath11 3.2.2 , @xmath55 cm , considering the uncertainties in the dust temperature and the likelihood that the plateau extends beyond day 736 .
our nir observations are only sensitive to the warmest dust , but if we assume that the nir luminosity peaks in the @xmath27 band , then a lower limit to the total energy emitted in the ir is @xmath64 @xmath65 d , or @xmath5 4@xmath66 erg .
this is comparable to the canonical optical output of a normal sn ii .
typically , this would pose a significant problem for the ir echo hypothesis as it would require order unity of the original optical radiation from the sn to be absorbed and re - radiated in the nir .
indeed , this is precisely the argument used by @xcite to argue against the light - echo hypothesis for sn 2005ip . in the case of this is not a problem , however , because the nir emission falls well within the available energy budget to heat the dust . after applying a bolometric correction , @xcite find that @xmath67 erg were emitted by during the first @xmath3220 day of the light - curve evolution , which means that less than 2% of this energy has been absorbed and re - radiated by the dust .
this value is comparable to those found by @xcite for sne 1979c and 1980k .
if we assume that the dust radiates as a blackbody , we can estimate the optical depth of the dust shell as @xmath68 .
we note that like @xmath69 above , this value constitutes a lower limit for @xmath70 .
even if the nir luminosity is a factor of 10 larger than our lower limits , this would imply @xmath71 , which corresponds to an optical extinction of @xmath72 .
this is far less than the total observed extinction from the host , @xmath73 @xcite , suggesting that the dusty shell responsible for the ir echo can not produce all of the non - galactic reddening observed toward . that there is a considerable amount of dust along the line of sight , which is not in the csm of ,
should not come as a surprise given that there is a prominent dust lane in ngc 1260 ( see figure [ false - color ] ) . with this information in hand , it is now possible to estimate the total dust mass of the shell that creates the nir echo ; according to @xcite , it is simply related to the dust properties , the optical depth of the dust , and the inner and outer radii of the shell . above we estimate the inner radius of the dust shell to be @xmath38@xmath74 cm .
the value of the outer radius can be deciphered from the decay of the nir light curve following the plateau .
as we have not yet observed the nir decay , we adopt an outer radius of 2@xmath48 , which _ minimizes _ the dust mass in the shell . finally , we follow the same assumptions as @xcite about the dust properties : a grain density @xmath75 = 3 g @xmath76 , grain radius @xmath77 m , and mean absorption efficiency @xmath49 = 1 . with these asumptions
we find a total dust mass of @xmath78 . assuming a typical value for the dust - to - gas ratio , 1:100
, the total mass of the circumstellar shell is @xmath310 .
this value shows good agreement with the initial estimate of 510 from @xcite , which was based on only observations taken between 400 and 465 day after explosion .
we note that @xmath310 constitutes a lower limit to the total mass of the shell , because the actual value of the duration of the plateau and @xmath69 could potentially be much larger than the values we adopted above . at a distance of @xmath310@xmath6 cm from the sn explosion site ,
it is not immediately evident that the dust giving rise to the nir echo is part of the progenitor s csm or the local interstellar medium ( ism ) .
@xcite argue that the large dust mass needed at @xmath310@xmath6 cm could be explained if the progenitor passed through a phase of eruptive mass loss @xmath310001500 yr prior to explosion , in an event similar to observed outbursts from @xmath79 car .
pre - existing dust in the walls of a giant hii region , where a massive star like sn2006gy s progenitor might have lived , can not account for the ir echo because at typical distances of @xmath310 pc or more from the sn , the dust would be far too cold to reproduce the nir excess .
@xcite predicts that any sn with an ir echo due to pre - existing csm dust should also show a faint scattered light echo in the optical as well .
he shows that this dust can lead to optical emission characterized by a @xmath80 scattering law , where @xmath10 is some value between 1 and 2 .
more than a year after explosion had faded rapidly in the optical @xcite .
therefore , we obtained _ hst _ images to search for a possible late - time scattered light echo .
our optical detection on day 825 shows a significant change in the optical decline rate of . between day 200400 the sn faded by @xmath5 3 mag in the optical @xcite , whereas the decline between day 400825 was only @xmath31 mag .
in addition to this reduced rate of decline , the sn underwent significant color evolution . assuming that @xmath32 and @xmath34 approximate the @xmath81 and @xmath82 bands , respectively
, we find that @xmath83 mag on day 825 .
by contrast , with a 12000 k color temperature at peak @xcite , the observe color of would have been @xmath84 mag .
thus , the spectrum on day 825 is _ bluer _ than the spectrum at peak .
indeed , the @xmath30.4 mag color change shows excellent agreement with a @xmath85 scattering law .
a @xmath57 scattering law , on the other hand , would predict a @xmath86 mag change to the @xmath87 color , or roughly @xmath88 mag during the late - time epochs .
this behavior is similar to what one would expect from a scattered light echo : as in the case of an ir echo , when successive paraboloids sweep out to progressively larger radii , dust in the circumstellar ( or in some cases the interstellar ) environment can scatter that light toward the observer , creating a plateau in the optical light curve of the sn . at the same time , scattering preferentially selects shorter wavelengths , which results in a spectrum that is bluer than that of the sn at peak .
this is the precise behavior observed for the type ia sne 1991 t @xcite , 1998bu @xcite , and 2006x @xcite which both showed significant departures from the expected decline rate of type a sne . spectra taken roughly two years after maximum , for sne 1991 t and 1998bu , and @xmath310 months after maximum for sn 2006x , show emission features similar to those seen in the spectra near peak superposed on top of a blue continuum for both these sne .
light echoes have also been observed around a number of core - collapse sne , e.g. sn 2003gd @xcite and sn 1993j @xcite , though we note that late - time spectra of the light echoes do not exist in these cases . in figure [
late - sed ] we show the late - time sed of , including our nir detections on day 469 and 736 and the optical detections from day 825 .
the solid blue line represents a single - component blackbody at @xmath89 k , roughly the peak temperature of @xcite , normalized to our @xmath34 detection .
the sed clearly shows that the late - time emission is significantly bluer than a blackbody spectrum .
we note that the early - time spectra of show considerable line blanketing blueward of @xmath34000 , such that the emission decreases blueward of our @xmath31 detection .
the nir emission rapidly rises toward the ir , suggesting that there is considerable ir emission to which our nir measurements are not sensitive .
figure [ reflect - spec ] illustrates the observed _ hst _ photometry along with the spectrum of obtained near peak @xcite .
note that no correction for reddening has been made .
the photometry shows excellent agreement with a @xmath80 scattering law , where @xmath90 .
the lack of an actual reflection spectrum precludes a more detailed model of the scattering dust at this time .
given that the other typical scenarios for a large late - time luminosity from a sn , @xmath4co decay and extended csm interaction , have been ruled out , we conclude that the late - time _ hst _
detection is a scattered light echo .
the location of the scattering dust is of interest as it can help determine properties of the environment local to . above
we show that the csm dust responsible for the nir echo can not account for all the reddening observed toward , suggesting that the sn light passes through external , non - csm dusty regions in the host galaxy prior to reaching us . therefore , the scattered light echo could originate in a number of different locations . with only a single epoch of the unresolved scattered light echo
, a precise determination of the location of the scattering dust can not be obtained .
nevertheless , we can gain some insight into the dust location based on the total optical and nir emission .
for the case where the nir - echo dust also gives rise to the scattered light echo , and assuming the sn can be modeled as a single short pulse of uv / optical radiation , the ratio of the scattered light luminosity ( @xmath91 ) to the ir luminosity at late - times can be shown to be @xcite : @xmath92^{1/2}}-\frac{(1-g)^2+gct / r_2}{[(1-g)^2 + 2gct / r_2]^{1/2}}\right\ } \nonumber,\end{aligned}\ ] ] where @xmath93 is the dust albedo , @xmath60 is the speed of light , @xmath94 is the time since the sn explosion , @xmath95 is outer radius of the dust shell , and @xmath96 is a measure of the degree of forward scattering .
the case @xmath96 = 0 corresponds to isotropic scattering and @xmath96 = 1 is completely forward scattering .
note that in equation [ eqn2 ] we have assumed that @xmath97 , where @xmath48 is the inner radius of the dust shell , and that the albedo is independent of frequency .
empirical estimates and numerical calculations @xcite show @xmath98 0.6 . for the dust albedo
we adopt @xmath99 0.6 @xcite .
as mentioned above , we can not constrain @xmath95 with our current observations , but we do know that @xmath95 must be greater than @xmath48 , and the value of equation [ eqn2 ] does not change when @xmath100 . for reasonable limits to the outer radius , @xmath101 , we find that @xmath1020.60 . direct integration of the _ hst _ optical flux yields an optical luminosity of @xmath3([email protected])@xmath103 .
this value is somewhat uncertain because our data do not extend into the ultraviolet ; however , optical spectra taken near peak show that line blanketing severely reduces the flux blueward of @xmath34000 @xcite , so this uncertainty should not have a significant effect on the total scattered light luminosity .
if we assume that the nir flux continues to decline at the same rate as that observed between day 405 and 736 , then we find a lower limit to the ir luminosity ( again because our nir observations only probe emission from the hottest dust ) of @xmath104 on day 825 .
therefore , on day 825 , @xmath105 , consistent with the predicted values above in the case where the scattering dust is the same dust responsible for the ir echo .
the alternative possibility is that the scattering dust is not located in the csm , instead existing at some other location in ngc 1260 along the line of sight .
@xcite inferred that this was the case with sne 1991 t and 1998bu , and like those two sne we know that has a significant amount of dust located along the line of sight .
as shown by @xcite , if we approximate the sn light curve as a short pulse with duration @xmath106 , then the scattered echo luminosity is @xmath107 where @xmath108 can be obtained via direct integration of the observed light curve , @xmath109 , and @xmath110 is the fraction of light scattered to the observer . under the assumption that light is being scattered by dust in a sheet of thickness @xmath111 , where @xmath112 is the distance between the dust and the sn , then @xcite @xmath113\right\}^{3/2 } } , \label{eqn4}\ ] ] where @xmath70 is the optical depth of the dust . using the relation @xmath114 @xcite , and the fact that @xmath115 outside the dusty csm is @xmath31.6 mag , we find @xmath116 .
from @xcite we know that @xmath117 erg .
therefore , using @xmath118 erg s@xmath9 , we find ( from equation [ eqn4 ] ) that dust located at @xmath119 pc can explain the observed scattered light luminosity .
our observations appear to be consistent with one of two scenarios : either ( i ) the scattered light echo is due to csm dust , which has also been heated and is radiating in the ir , or ( ii ) a dusty region @xmath320 pc from the sn is responsible for the late - time optical luminosity .
this latter scenario seems very plausible if the progenitor of was a very massive star , @xmath5 100 @xcite .
if this were the case we might expect that the progenitor resides in the center of a giant hii region with multiple dense , dusty clouds nearby .
this scenario might be very similar to the carina nebula surrounding @xmath79 car , which has multiple dusty molecular clouds only 1020 pc from @xmath79 car ( e.g. @xcite ) . perhaps given a similar progenitor birth environment , evidence for patchy dust at the tens of pc scale around grb environments also appears to be secure ( e.g. @xcite ) .
if the progenitor of were surrounded by such a nebula then our assumption of a single , thin scattering surface would no longer be valid , and the scattering dust could potentially be located at a number of different locations . with only a single epoch of observations , we can not distinguish between the two possible dust locations .
one additional epoch of optical imaging should prove sufficient to determine which scenario is correct .
as long as @xmath120 , then @xmath110 is roughly constant , whereas the models of @xcite show a continuous decline in the scattered light echoes when @xmath96 = 0.6 and the scattering dust is in the csm .
a relatively flat optical light curve would put the dust at @xmath320 pc , whereas significant decline from the observed day 825 flux would suggest csm dust . finally , we note that the prospects for fully resolving the light echo are dim : at a distance of 73 mpc it would take several decades before the light echo could be resolved with high spatial resolution images ( i.e. , comparable to _ hst _ in the optical ) .
we have presented new observations of , including early - time nir data from pairitel , as well as optical and nir detections more than two years after the explosion .
the early nir data show good agreement with a single - component , cooling blackbody between day 54 and 133 .
there is tentative evidence for a slight nir excess toward the end of this period .
the late - time nir and optical detections show a rate of decline which is slower than that of @xmath4co , indicating that the luminosity is not generated via heating by radioactive decay .
this slow rate of decline is not expected in the pair - instability sn models .
rather , the nir observations can best be explained with an ir echo , and the optical detections are most compatible with a scattered light echo . both ir and scattered light echoes should be present if there is a massive dust shell external to the sn @xcite . at the time of discovery
was the most luminous sn known @xcite , which when combined with its proximity to the host - galaxy nucleus ( @xmath3300 pc ) , suggested that perhaps there was something unique about star formation and sne in that region . since the discovery of several other vlsne have been found , including sne 2005ap @xcite , 2006tf @xcite , 2008am @xcite , and 2008es @xcite . in each of these cases
the sn has been localized well outside the central kpc of the host galaxy , with the exception of sn 2008es , for which a definitive host has not yet been identified @xcite .
this suggests that the location of was not in fact special .
current and future surveys will dramatically increase the rate of discovery of vlsne , which should , in turn , allow for a more systematic study of their host environments ( for a discussion of the rates of vlsne see @xcite ) .
given our interpretation for the late - time optical and nir emission arising from a dust shell at @xmath38@xmath74 cm from the sn , what might we expect from future observations of ? the ir - echo models of @xcite show that after a time @xmath121 , the ir luminosity falls off the plateau and exhibits significant decline . based on our predicted value for the size of the dust - free cavity ( see @xmath113.2.2 )
, we would expect that the plateau phase of the ir echo has ended , and that the ir flux is now in steady decline .
future observations of this decline will place limits on the outer extent of the dust shell , which will in turn provide better limits on the total mass of the dust shell .
our ability to predict the behavior of the scattered light echo is limited by the current degeneracy concerning the physical location of the scattering dust .
if the same dust is responsible for both the ir and scattered light echoes , then we would expect that the optical emission is now fading , in a manner similar to what ought to be seen in the nir .
our observations are also consistent with dust located at @xmath320 pc from the sn , in which case we might expect the scattered light echo to continue at roughly a constant flux for several more years . continued observations with _ hst
_ would allow us to distinguish between these two situations
. the shell - shock model of @xcite may also explain the peak luminosity of several other vlsne : 2005ap , 2006tf , 2008es @xcite .
this model requires large csm densities at a distance of a few @xmath122 cm , which can be accomplished if the sn progenitor undergoes significant ( @xmath30.11 yr@xmath9 ) mass loss during the few decades prior to explosion @xcite .
the most luminous lbvs have observed mass shells with @xmath510 ( @xcite and references therein ) , indicative of giant eruptions @xcite but not lbvs in their more typical state with less violent wind variability ( see @xcite for a general reference on lbv winds ) . the possible connection between large csm densities and vlsne may suggest that their progenitors are lbvs @xcite . for the case of , @xcite argue that the dusty shell at @xmath310@xmath6 cm could exist if the progenitor underwent an lbv - like eruptive mass - loss phase @xmath31200 yr prior to the sn .
ir echoes for other vlsne have not been reported to date , but we strongly encourage a search for them . if a substantial fraction of them exhibit late - time characteristics similar to , this may be suggestive of a common timescale , @xmath31000 years and @xmath310 years , for extreme mass loss in the progenitors of vlsne . the combination of a vlsne and ir echo would point to multiple phases of eruptive mass loss , which is reminiscent of lbv behavior .
the instability driving these eruptions still remains unclear ( one possibility is the pulsation pair - instability described by @xcite to explain ) , and we encourage future observational and theoretical work to better characterize these systems .
we thank schuyler van dyk for his assistance in the preparation of the hst snapshot proposal .
would like to thank andy becker , maryam modjaz , and dovi poznanski for useful discussions that helped improve this paper .
we thank cullen blake , dan starr , and emilo falco for their assitance in the operations of pairitel .
this publication makes use of data products from the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by the national aeronautics and space administration and the national science foundation .
is supported by the nsf graduate research fellowship program .
j.s.b.s group is partially supported by nasa/_swift _ grant # nng05gf55 g and a hellman faculty award .
a.v.f.s group is supported by national science foundation ( nsf ) grant ast0607485 and the tabasgo foundation .
j. x. p. is partially supported by nasa/_swift _
grant nnx07ae94 g , and an nsf career grant ( ast0548180).the peters automated infrared imaging telescope is operated by the smithsonian astrophysical observatory ( sao ) and was made possible by a grant from the harvard university milton fund , the camera loan from the university of virginia , and the continued support of the sao and uc berkeley .
the pairitel project is partially supported by nasa/_swift _ guest investigator grant # nng06gh50 g .
some of the data presented herein were obtained at the w. m. keck observatory , which is operated as a scientific partnership among the california institute of technology , the university of california , and nasa ; it was made possible by the generous financial support of the w. m. keck foundation .
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 . ,
a. m. , fruchter , a. s. , hook , r. n. & hack , w. 2002 , in the 2002 hst calibration workshop : hubble after the installation of the acs and the nicmos cooling system , ed .
s. arribas , a. koekemoer , & b. whitmore ( baltimore , md : space telescope science institute ) , 337 | supernova ( sn ) 2006gy was a hydrogen - rich core - collapse sn that remains one of the most luminous optical supernovae ever observed .
the total energy budget ( @xmath0 2@xmath110@xmath2 erg radiated in the optical alone ) poses many challenges for standard sn theory .
we present new ground - based near - infrared ( nir ) observations of , as well as a single epoch of _
hubble space telescope ( hst ) _ imaging obtained more than two years after the explosion .
our nir data taken around peak optical emission show an evolution that is largely consistent with a cooling blackbody , with tentative evidence for a growing nir excess starting at day @xmath3100 .
our late - time keck adaptive optics ( ao ) nir image , taken on day 736 , shows little change from previous nir observations taken around day 400 .
furthermore , the optical _ hst _ observations show a reduced decline rate after day 400 , and the sn is bluer on day 825 than it was at peak .
this late - time decline is inconsistent with @xmath4co decay , and thus is problematic for the various pair - instability sn models used to explain the nature of .
the slow decline of the nir emission can be explained with a light echo , and we confirm that the late - time nir excess is the result of a massive ( @xmath510 ) dusty shell heated by the sn peak luminosity .
the late - time optical observations require the existence of a scattered light echo , which may be generated by the same dust that contributes to the nir echo .
both the nir and optical echoes originate in the proximity of the progenitor , @xmath310@xmath6 cm for the nir echo and @xmath720 pc for the optical echo , which provides further evidence that the progenitor of was a very massive star . |
let @xmath0 denote the fibonacci numbers ( where we have shifted the usual initial condition ) ; i.e. @xmath1 .
+ consider the formal power series @xmath2 let @xmath3 be the coefficient of @xmath4 in @xmath5 .
it is clear that @xmath6 , where @xmath7 is the number of partitions of @xmath8 into an even number of distinct fibonacci numbers , and @xmath9 is the number of partitions of @xmath8 into an odd number of distinct fibonacci numbers . in [ 2 ] , n. robbins proved that @xmath10 . in [ 1 ]
, f. ardila gave a simpler proof for robbins result by giving a recursion on @xmath3 .
in this paper , we consider _ quasifibonacci numbers _ , which serve as generalization of fibonacci numbers
. given a positive integer @xmath11 .
a sequence @xmath12 of positive integers is called _ quasifibonacci sequence of level @xmath13 _ if * @xmath14 for all @xmath15 , and * @xmath16 for all @xmath17 .
we also say that @xmath18 are quasifibonacci numbers .
in particular , ( shifted ) fibonacci numbers and lucas numbers are quasifibonacci . as we will see in this paper , robbins result can be generalized to _
quasifibonacci numbers_. more precisely , we shall prove the following theorem : let @xmath19 be a quasifibonacci sequence of even level .
consider the formal power series @xmath20 then @xmath10 similarly , we have @xmath6 where @xmath7 ( resp .
@xmath9 ) is the number of partitions of @xmath8 into an even ( resp .
odd ) number of distinct quasifibonacci numbers .
+ in this paper , we study the structure of the set of such partitions .
in fact , a digraph and poset structure on such sets will be constructed in section 3 . in section 4
, we will unveil intrinsic symmetry and recursive relations between these posets . finally , as an application
, we shall prove theorem 1.1 in section 5 .
the following notations will be used throughout the paper . * @xmath21 * given a quasifibonacci sequence @xmath19 , define @xmath22 @xmath23 represents the set of partitions of @xmath24 into distinct quasifibonacci numbers @xmath25 . * for @xmath26 ,
define @xmath27 we also say that @xmath28 is the _ representation _ of @xmath29 . * for @xmath30 , define @xmath31 * for any @xmath32 , define the _ length _
@xmath33 to be the largest @xmath34 such that @xmath35 .
( abusing the notation , we also identify @xmath28 with @xmath36 , which is called the _
reduced representation_. ) the next lemma gives some important arithmetic properties of the @xmath25 s which will be used frequently throughout the paper .
let @xmath19 be a quasifibonacci sequence
. then 1 .
@xmath37 for any @xmath15 2 . for @xmath38 and any @xmath39 , we have @xmath40 3 . for @xmath41 and any @xmath42 , we have @xmath43 .
@xmath44 5 .
@xmath45 6 .
@xmath46 1 . by the definition of quasifibonacci numbers ,
it is clear that @xmath47 when @xmath48 .
the case for @xmath49 follows immediately from induction and the following inequality : @xmath50 2 .
it is clear that @xmath51 .
suppose @xmath52 . by the previous lemma , @xmath53 , a contradiction .
therefore , @xmath54 .
this is straightforward .
4 . write @xmath55 ( @xmath56 )
. then @xmath57 5 .
it is obvious that @xmath58
. + by ( 4 ) , @xmath59 6 . by ( 4 ) , we have @xmath60
for each @xmath61 , we construct a digraph @xmath62 in the following way : 1 .
set @xmath63 .
in particular , set @xmath64 if @xmath65 .
2 . for @xmath66 , @xmath67 ,
let @xmath68 if there exists @xmath15 such that * @xmath69 , @xmath70 * @xmath71 , @xmath72 * @xmath73 for all @xmath74 + ( here @xmath75 represents the directed edge @xmath76 . ) the digraph structure induces a natural partial order on @xmath23 as follows : + _ for @xmath77 , set @xmath78 if there exists a path in @xmath79 from @xmath28 to @xmath80 .
_ in other words , @xmath28 covers @xmath80 if and only if @xmath81 .
this makes @xmath23 into a poset @xmath82 .
the following examples show the corresponding hasse diagrams for @xmath83 and @xmath84 , where @xmath85 and @xmath86 denote the fibonacci numbers and lucas numbers , respectively .
first 24 hasse diagrams for @xmath83 : + first 24 hasse diagrams for @xmath84 : + for @xmath38 , define @xmath87 by lemma 2.1(2 ) , we have @xmath88 .
+ furthermore , let @xmath89 , @xmath90 denote the subposet of @xmath82 restricted on the vertex set @xmath91 and @xmath92 , respectively .
( abusing the notation , @xmath89 , @xmath90 also denote the corresponding subdigraphs of @xmath79 . )
the figure below shows that @xmath93 can be decomposed into @xmath94 and @xmath95 .
+ now we study the structure of the posets @xmath82 in detail .
suppose @xmath96 .
then there is a unique maximal element @xmath97 in @xmath82 .
more precisely , @xmath97 is the only element in @xmath23 which does not contain @xmath13 consecutive 1 s .
+ furthermore , if @xmath98 , then @xmath99 .
suppose @xmath42 is a maximal element in @xmath82 .
it is clear that @xmath28 does not contain @xmath13 consecutive 1 s .
( otherwise , assume that @xmath100 and @xmath101 .
set @xmath102 where @xmath103 , @xmath104 and @xmath105 for @xmath106 . then @xmath107 but @xmath108 . )
+ now we show the uniqueness .
suppose both @xmath28 and @xmath109 are maximal elements in @xmath82 .
let @xmath110 be the largest index such that @xmath111 .
without loss of generality , assume that @xmath112 , @xmath113 and @xmath114 for @xmath115 .
since @xmath28 is maximal , it does not contain @xmath13 consecutive 1 s .
it follows that @xmath116 which is the largest possible value of length @xmath117 with no @xmath13 consecutive 1 s .
+ however , lemma 2.1(4 ) gives @xmath118 a contradiction .
hence the maximal element is unique .
denote it by @xmath97 . +
now suppose @xmath38 . by lemma 2.1(2 ) , @xmath119 . as proved above , any element of length @xmath117 with no @xmath13 consecutive 1 s is smaller than @xmath25 .
so we must have @xmath99 .
suppose @xmath120 and @xmath96
. then @xmath121 .
assume that @xmath122 then we must have @xmath123 .
thus@xmath124 a contradiction .
similarly , we have the following dual result : suppose @xmath96 .
then there is a unique minimal element @xmath125 in @xmath82 .
more precisely , @xmath125 is the only element in @xmath23 which does not contain @xmath13 consecutive 0 s in reduced representation .
however , we do nt have @xmath126 in general .
actually , we will show that @xmath82 is a modular lattice in section 5 .
if we viewing @xmath127 as a sequence , then there exists local symmetry relations between the posets .
for instance , in example 3.1 , the posets are central symmetric from @xmath128 to @xmath129 , and from @xmath129 to @xmath130 . in general
, similar symmetry appears for all quasifibonacci sequences . + in order to describe this special symmetry relation , we recall the definition of _ dual posets_. two posets @xmath131 , @xmath132 are _ dual posets _ to each other if there exists an order - reversing bijection @xmath133 whose inverse is also order - reversing ; that is @xmath134 for @xmath98 , let @xmath135 . then @xmath82 is dual to @xmath136 .
define @xmath137 by setting @xmath138 .
( note that @xmath139 is not necessarily nonzero ) + it is easy to check that @xmath81 if and only if @xmath140 .
hence @xmath82 is dual to @xmath136 via @xmath141 .
being more careful , we can derive similar symmetry on @xmath89 and @xmath90 . for @xmath142 ,
let @xmath143 .
then @xmath89 is dual to @xmath144 , and @xmath90 is dual to @xmath145 .
it suffices to show that @xmath146 and @xmath147 .
+ let @xmath148 , ( @xmath149 ) . since @xmath150 , @xmath151 .
thus @xmath152 , @xmath153 .
hence @xmath154 .
+ on the other hand , for any @xmath155 , we must have @xmath156 .
so @xmath157 .
+ therefore , @xmath146 and @xmath158 , as desired .
other than symmetry , there are intrinsic recursive relations in the poset sequence @xmath159 . in order to describe the recursion clearly
, we introduce the following notations .
suppose @xmath66 , @xmath160 satisfies @xmath161 whenever @xmath162 .
define @xmath163 similarly , suppose @xmath66 , @xmath164 satisfies @xmath165 whenever @xmath166 .
define @xmath167 let @xmath131 be a finite subset of @xmath168 .
the poset structure on @xmath131 is defined as usual .
let @xmath169 .
suppose that for any @xmath170 , we have @xmath171 whenever @xmath162 .
define @xmath172 regarded as a poset , with the natural partial order .
the following figure gives an example for @xmath173 and @xmath174 : + we can also define the addition between the posets : let @xmath131 , @xmath132 be disjoint finite subsets of @xmath168 with natural digraph and poset structure .
let @xmath175 , @xmath176 be subposets of @xmath131 , @xmath132 , respectively .
suppose there is a bijection @xmath177 such that @xmath28 covers @xmath178 for all @xmath179 .
+ define @xmath180 regarded as a digraph with @xmath181 @xmath182 we also treat @xmath183 as a poset if the partial order determined by the directed edges in @xmath184 is exactly the natural one . in particular , @xmath185 .
+ from now on , let @xmath186 denote the only element in @xmath168 with the @xmath110-th entry being 1 and the others being 0 ; i.e. , @xmath187 .
+ let @xmath188 denote the only element @xmath189 in @xmath168 with @xmath190 and @xmath191 ; i.e. , @xmath192 .
now we describe the recursion explicitly .
we show that each @xmath82 with @xmath98 can be expressed in terms of @xmath193 . if @xmath194 , then @xmath82 is isomorphic to @xmath195 .
more precisely , we have @xmath196 , @xmath197 and @xmath198 . if @xmath199 , then @xmath200 @xmath201 and @xmath202 . if @xmath203 , then @xmath204 @xmath201 and @xmath205 .
the only remaining case is @xmath206 . by proposition 4.1
, @xmath82 is dual to @xmath136 where @xmath143 .
+ by lemma 2.1(1 ) and lemma 2.1(6 ) , we have @xmath207 hence @xmath136 can be determined by the propositions above .
moreover , by proposition 4.2 @xmath89 and @xmath90 are dual to @xmath144 and @xmath145 , respectively .
so they can also be determined by recursions .
assume @xmath96 .
( otherwise the proposition is trivially true . )
let @xmath97 be the maximal element in @xmath82
. then @xmath99 ( proposition 3.1 ) .
so @xmath208 , which implies @xmath209 .
note that @xmath210 by corollary 3.2 , we have @xmath211 .
+ for any @xmath39 , @xmath212 hence @xmath43 for all @xmath39 . in particular , it is valid to do the subtraction @xmath213
. + therefore , the map @xmath214 gives an isomorphism from @xmath82 to @xmath195 and @xmath215 .
assume @xmath96 .
in this case , @xmath216 .
+ applying a similar argument , we obtain @xmath217 . furthermore , by proposition 3.1 , @xmath218 . + define @xmath219 then @xmath220 gives an isomorphism from @xmath89 to @xmath195 and @xmath201 .
+ on the other hand , for any @xmath221 , we claim that @xmath222 . + assume the contrary . then @xmath223 a contradiction .
+ hence , the subtraction @xmath224 is valid .
+ note that @xmath225 .
thus @xmath226 for all @xmath227 .
so the map @xmath228 gives an isomorphism from @xmath90 to @xmath229 and @xmath230 .
+ therefore , @xmath231 via the natural map @xmath232 : @xmath233 .
the proof is almost the same .
assume @xmath96 .
+ in this case , we have @xmath234 and @xmath235 .
+ for @xmath227 , we still have @xmath236 .
so the map @xmath237 gives an isomorphism from @xmath90 to @xmath195 .
this completes the proof .
@xmath82 , @xmath89 , @xmath90 are modular lattices .
we prove the statement by induction on @xmath24 . + * base case : * when @xmath238 , @xmath82 is either @xmath239 or a single - element set .
the statement is trivially true . +
* inductive step : * consider @xmath38 ( @xmath240 ) .
it is clear that the dual poset of a modular lattice is also a modular lattice .
so , by symmetry , it suffices to consider @xmath241 . 1 .
if @xmath242 , then @xmath82 , @xmath89 , @xmath90 are isomorphic to @xmath195 , @xmath243 and @xmath229 , respectively .
so they are modular lattices by the induction hypothesis .
if @xmath244 , then @xmath89 , @xmath90 are isomorphic to @xmath195 and @xmath229 , respectively .
so they are lattices by induction . to show that @xmath82 is a lattice , we need to show that for any @xmath245 , @xmath246 and @xmath247 exist . + indeed , if @xmath248 or @xmath249 , then @xmath246 and @xmath250 exist by the induction hypothesis . + if @xmath251 and @xmath252 , it is easy to check that @xmath246 is simply @xmath253 in @xmath195 . similarly , @xmath247 is simply @xmath254 in @xmath229 .
hence , @xmath82 is a lattice .
+ to show modularity , we need to show that @xmath255 both cover @xmath247 if and only if @xmath256,@xmath257 are both covered by @xmath246 . + indeed , if @xmath258 and @xmath259 , the statement follows by the induction hypothesis . + now suppose @xmath260 and @xmath261
. without loss of generality , assume that @xmath246 covers both @xmath256 , @xmath257 .
+ obviously , @xmath262 .
so @xmath263 and @xmath264 covers @xmath256 .
+ therefore , @xmath265 is covered by both @xmath256 and @xmath257 , as desired .
the case @xmath266 is similar to case(2 ) . in this section
we will prove theorem 1.1 . as mentioned in the introduction section , @xmath3 is the difference of the number of partitions into an even number of @xmath25 s and the number of partitions into an odd number of @xmath25 s .
+ to distinguish these two kinds of partitions , we define the _ sign function _ @xmath267 by setting @xmath268 if @xmath28 contains an even number of 1 s and @xmath269 otherwise . in general , for any finite subset @xmath131 of @xmath168 with natural partial ordering , define @xmath270 it is clear that @xmath271 .
we also define @xmath272 and @xmath273 .
then , obviously , @xmath274
. the following lemma will be useful in the proof below .
* let @xmath275 .
then @xmath276 * let @xmath277 and @xmath131 a finite subset of @xmath168 .
then @xmath278 * let @xmath131 , @xmath132 be finite subsets of @xmath168
. then @xmath279 .
straightforward . to visualize the relation of odd and even partitions , we color the digraph with two colors . in the corresponding hasse diagram , a vertex @xmath280
is colored blue if @xmath268 and colored red if @xmath269 .
the figure below shows the first 24 colored hasse diagrams for @xmath83 : + suppose @xmath28 covers @xmath80 in @xmath82 . by the definition ,
@xmath80 has @xmath281 more 1 s than @xmath28 .
hence adjacent vertices in the corresponding digraph have different colors if @xmath13 is even .
now we prove theorem 1.1 by proving the following stronger result : for any @xmath282 , @xmath283 , @xmath284 , @xmath285 .
we perform induction on @xmath24 . + * base case : * when @xmath238 , @xmath82 is either @xmath239 or a single - element set .
the statement is true . +
* inductive step : * consider @xmath38 @xmath286 . 1 .
if @xmath242 , then @xmath287 , @xmath288 , @xmath196 . hence we have @xmath289 similarly , we have @xmath290 and @xmath291 .
2 . if @xmath199 , then @xmath201 and @xmath292 and @xmath293 .
hence we have @xmath294 @xmath295 and @xmath296 3 . if @xmath297 , then @xmath201 , @xmath205 and @xmath298 .
hence we have @xmath299 @xmath300 and @xmath301 4 . if @xmath302 , then @xmath89 , @xmath90 , @xmath82 are dual to @xmath144 , @xmath145 and @xmath136 , respectively , where @xmath303 .
hence we have @xmath304 @xmath305 and @xmath306 this completes the proof .
to write down the recursion explicitly , we have @xmath307 @xmath308 where @xmath309 .
_ comment : _ if we take @xmath310 , then the recursion above is precisely the one in [ 1 ] . in this case
, @xmath13 is an odd number .
hence for each pair of adjacent vertices @xmath311 in @xmath79 , the parity of the number of 1 s in @xmath28 and @xmath80 must be the same .
in other words , every adjacent pair of vertices have same color .
since @xmath82 are lattices , @xmath79 are connected graphs . therefore all vertices in @xmath82 have same color .
it is easily seen that @xmath312 is not bounded in this case .
instead we have the following estimate .
let @xmath313 be an integer .
for any @xmath314 , @xmath315 .
* base case * when @xmath238 , @xmath82 is either @xmath239 or a single - element set .
so @xmath316 .
the statement is true . + * inductive step * consider @xmath317 @xmath318 . with a similar argument
, we can derive the following recursion : @xmath319 where @xmath320 .
+ note that @xmath321 @xmath322 . hence , in any of the six cases , @xmath323 .
this completes the proof . _
comment : _ this upper bound is the best possible because there exists @xmath317 satisfying @xmath324 for all @xmath325
. however , it is possible to improve the result by spliting the intervals into more pieces and refining the estimate .
the author would like to thank professor richard stanley for valuable suggestions on this paper . | in this paper , we study partitions of positive integers into distinct quasifibonacci numbers .
a digraph and poset structure is constructed on the set of such partitions .
furthermore , we discuss the symmetric and recursive relations between these posets .
finally , we prove a strong generalization of robbins result on the coefficients of a quasifibonacci power series . |
more than 54% of all wolf - rayet ( wr ) stars are in binary systems that consist of a wr and an ob - type star @xcite .
the collision between two stellar winds in these systems produces characteristic signatures in different wavelength bands that can include non - thermal radio emission @xcite , copious and variable x - rays @xcite , and dust emission in the infrared ( ir ) @xcite .
subject of this _ letter _ is the wc9d star wr65 .
@xcite reported the presence of warm dust in this object .
they discussed that in early - type wc stars an increase in the wind density provided by e.g. shock compression in colliding winds in binary systems may be sufficient to result in grain formation .
@xcite found no need to invoke binarity to explain the presence of dust in late wc stars .
he fitted the observed ir spectrum of assuming a single star with stellar mass - loss in the form of dust with @xmath0 and surrounded by a dust shell with an inner radius of 520@xmath1 .
@xcite detected radio emission from , but could not constrain its spectral index . in a previous paper @xcite we demonstrated that single wc type stars are x - ray quiet and proposed that all x - ray active wc stars must be in binary systems .
x - ray emission from was detected and tentatively explained as a result of wind - wind collision . in this paper
we report on the x - ray light curve of and changes in its x - ray spectra that unambiguously confirm its status as a colliding - wind binary ( cwb ) .
.coordinates of from different sources [ cols="^,^,^,^,^",options="header " , ]
the coordinates of from different sources are compiled in table[tab : coor ] .
the coordinates from simbad and the usno - b1.0 catalog agree well with the source location in the hri - i image . except for
there are no other known objects at this position .
wr65 is a probable member of the cluster pismis20 , which contains one more wr star , wr67 , and has a distance of @xmath2pc @xcite .
m ) . is the bright object at the bottom , close to the detector edge .
the coordinates are galactic . ]
we fit the optical spectrum of and its available photometry using the potsdam wolf - rayet ( powr ) stellar atmosphere models @xcite .
the deduced stellar parameters are listed in table[tab : par ] .
a wind - blown bubble with a radius of @xmath3 is known around @xcite .
@xcite discovered a hi shell with a radius of 22pc surrounding and wr67 .
the infrared _ spitzer _ irac image of at @xmath4 m is shown in fig.[fig : irac ] . despite its unfavorable location close to the edge of the detector ,
a rather compact nebula around is visible .
the radius of the nebula is @xmath5 or 1.4pc .
given the strong mass - loss rate of and its strong uv field , it is plausible that the nebula is physically associated with this star .
cccc @xmath6 & @xmath7 & @xmath8 [ kms@xmath9 & @xmath10^{\rm b}$ ] + 2.4 & 5.8 ... 6.4 & @xmath11 & @xmath12 + + + in the x - ray sky , is located in a very interesting neighbourhood ( see fig[fig : acis ] ) . the star is @xmath13 away from the supernova remnant ( snr ) g320.4 - 1.2 and @xmath14 away from the x - ray bright `` cir pulsar '' and its associated pulsar wind nebula ( pwn ) .
the snr and the cir pulsar are at the distance @xmath15kpc @xcite .
this region of the sky is often observed , and serendipitous x - ray observations of are in archives . the most homogeneous data set found in archives is from _ chandra _ acis - i observations conducted from year 2000 to 2005 . in 2005
was also observed on one occasion by _
hrc - i and twice by _ xmm - newton _ , but was not detected during _ xmm - newton _ observations affected by high radiative background .
the local x - ray background is high because is projected on the peripheral region of the pwn mentioned above .
therefore we paid special attention to the background subtraction.conveniently , the x - ray emission from the pwn is well studied @xcite .
@xcite find that it is constant over at least twelve years .
we defined background regions as annuli around a stellar point source .
figure[fig : lc ] shows the background - subtracted x - ray light - curve of that displays a surprisingly high level of variability .
such strong x - ray variability in massive stars is observed only in binary systems @xcite .
the x - ray light curve gives strong evidence that the star is a colliding wind binary ( cwb ) . unfortunately
, the light - curve is too sparse to search for the period . given the time interval between the two recorded maxima , one may conclude that the period is not longer than about five years .
five of the observations of yielded enough counts for a crude x - ray spectral analysis .
the background - subtracted spectra obtained at different epochs are displayed in fig.[fig : time ] .
the spectral energy distribution is harder than for single wr stars @xcite .
the unresolved emission features coincide with the location of lines of fe , ca , ar , s , si , and mg ( cf .
figs.[fig : time ] , [ fig:2sp ] ) in collisionally ionized x - ray spectra .
interestingly , the iron lines at @xmath16kev ( 1.8 ) are weaker at some epochs compared to others ( e.g. they seem to disappear completely on 2005 february 07 ) . if real , this may reflect the change of the temperature in the hot plasma . the x - ray spectrum of in high states displays strong emission features in the 2 - 4 range ( see top panel in fig.[fig : time ] )
. weak lines of caxix(@xmath17 ) , caxx(@xmath18 ) , and arxviii(@xmath19 ) are sometimes observed in the x - ray spectra of cwbs , e.g. the wc8 binary @xmath20vel @xcite , the wn6 binary wr25 @xcite , and the lbv binary @xmath21car @xcite .
it seems plausible to assume that the unresolved ca and ar lines contribute to the emission `` bump '' at 2 - 4 in the spectrum .
however , the strength of this emission complex in is much higher than observed in any other cwb
. we rule out its origin from the background pwn emission , because there are no emission features at 2 - 4 in the spectra extracted from an annulus region around .
in colliding wind binaries consisting of an ob star and a wc star , the latter has a significantly denser , c , o - enriched wind that is basically opaque for x - rays @xcite .
when in the course of the orbital motion the wind collision region ( wcr ) is at superior conjunction relative to the wc star ( with respect to the observer ) , the x - rays generated in the wcr suffer strong absorption .
despite many still unclear details , this general picture is observationally confirmed by the x - ray spectral analysis of the wc+o type binaries wr140 and @xmath20vel . in these systems
the strongest photo - attenuation is seen in low x - ray states , while the lowest photo - attenuation is found when the x - ray flux is high @xcite . to investigate how x - ray
spectral variations are coupled with the variation of the absorption column , we fit a model to the observed spectra and check how the model parameters vary between the different epochs . following a method used in @xcite to successfully analize x - ray spectra of wr140
, we tried to fit the observed spectra alternatively with non - equilibrium ionization , pshock , and power - law models using the xspec software @xcite .
however , no meaningful constraints on model parameters could be obtained because of the poor quality of the spectra .
@xcite found that only models with non - equilibrium ionization can reproduce the _ detailed _ x - ray spectra of cwbs .
however , he noticed that these spectra can be crudely described by thermal plasma emission with some `` average '' temperature .
finally , we select a two - temperature thermal plasma model ( * ? ? ?
* apec ) where each component is allowed to suffer different degrees of attenuation the `` 2t2@xmath22 '' model .
this model provides a first insight into the properties of the system by constraining the model parameters .
we refer to the plasma with lower temperature @xmath23 as the `` soft '' model component , and to the plasma with higher temperature @xmath24 as the `` hard '' model component . as demonstrated in figs.[fig : fit],[fig : fit1 ] the 2t2@xmath22-model sufficiently well reproduces the shape of the spectra observed at different epochs .
we assume solar composition . selecting wc - wind abundances results in scaling down of the parameter that describes normalization of the emission measure ( em ) , but the effect of abundances on _ spectral shape _
can not be discriminated given the poor quality of available data .
the spectrum of with highest signal - to - noise and its best - fit model are shown in fig.[fig : fit ] .
the temperatures inferred from spectral fitting are @xmath25kev and @xmath26kev .
such temperatures can be expected from the collision of metal enriched winds in wide binaries .
the temperature is highest at the point of head - on collision , and is decreasing along the shock cone @xcite . the interstellar absorption column density towards the cir pulsar , @xmath27@xmath28 , is well constrained and is consistent with the color excess inferred from our fitting of the spectral energy distribution .
the neutral hydrogen column densities are @xmath29@xmath28 for the soft component and @xmath30@xmath28 for the hard component .
the column density to the soft component , @xmath31 , exceeds the interstellar column by a factor of more than two , but it is lower than the column density to the hard component , @xmath32 .
this indicates that the hotter plasma is more deeply embedded . from the spectrum
obtained at 2005/10/18 the model flux is @xmath33ergs@xmath34@xmath28 , which corresponds to @xmath35ergs@xmath34 , or in terms of stellar bolometric luminosity to @xmath36 .
the flux of the soft spectral component , @xmath37ergs@xmath34@xmath28 , corresponds to @xmath38ergs@xmath34 which is similar to the x - ray luminosity of a single o dwarf @xcite .
spectra observed at other epochs were also fitted with the 2t2@xmath39-model . for illustration ,
the spectrum obtained in the low state and its corresponding best - fit model are shown in fig.[fig : fit1 ] .
figure[fig : nh ] shows the model parameters depending on the time of observation . within the error bars , the temperatures @xmath23 and @xmath24 , and the absorption column to the source for the soft component , @xmath31 are similar at all observed epochs , and therefore are not plotted in fig.[fig : nh ] .
in contrast , the column density to the hard component , @xmath32 , changes at different epochs ( panel a of fig.[fig : nh ] ) .
variations in the em of the hard and soft components are shown in panels b and c of fig.[fig : nh ] .
the em of the soft component changes strongly , reflecting the x - ray light - curve variations , while the changes in the em of the hard component are less pronounced . .
_ panel a : _ variations of absorption column density of the hard component obtained from spectral fits .
_ panel b : _ normalization factor that characterizes the emission measure of the hard component .
_ panel c : _ normalization factor of the soft component . ]
the main findings from the analysis of x - ray spectra obtained at different epochs are : ( i ) the spectra display emission features at the wavelengths of the lines of strongly ionized fe , ca , ar , s , si , and mg ; ( ii ) the star shows strong spectral x - ray variability ; ( iii ) the spectra can be fitted with a two - temperature model , consisting of a soft component with @xmath40kev and a hard component with @xmath41kev ; ( iv ) the absorption for the hard component is strong and variable , while the absorption for the soft component is less severe and nearly constant ; ( v ) the changes in the em are stronger for the soft component . can these observations be explained in the framework of a colliding wind model ?
@xcite compare the 2.0 - 10.0kev light curves of wr140 and @xmath21 car ( both are systems with high eccentricity and high inclination ) and note that they are very similar .
@xcite explain the light - curve ( 2.010.0kev ) of @xmath21car using a hydrodynamic simulation of colliding winds .
they show that the data are broadly consistent with the x - ray emission being moderately low throughout most of the orbit .
however , a sharp peak occurs near periastron , when a greater fraction of the stellar winds collides , and the higher pre - shock densities and the lower pre - shock velocities result in stronger radiative cooling .
the eclipse occurs immediately after periastron passage when the companion of @xmath21car swings behind the lbv star with its dense wind .
it is hard to see how a similar model can be used to explain the x - ray variability of wr65 .
strong variations in the absorbing column indicate that the orbit of has high inclination too . considering the light - curve ( fig.[fig : lc ] ) we can rule out that has a period of a few days : the two data points obtained in april 2003 indicate constant count rates over at least a week .
moreover , two data points in the light - curve separated by two months ( 12/2004 and 02/2005 ) are similar , while there is an increase of emission by nearly 70% in the next two months . therefore , has most likely an orbital period of a few years . in this case
the long duration of the minimum is inconsistent with its interpretation as attenuation by the wind of the wc star immediately after periastron passage . quite different from the short - duration rise of column density observed in wr140 and @xmath21 car ,
the column density in is persistently very high , @xmath42 , sharply decreasing only at the maximum x - ray light .
the temporary variations of column density in are similar to those measured in the wc8+o binary @xmath20vel @xcite .
its hard x - ray emission increases by a factor of @xmath43 when column density decreases by factor of @xmath44few during the passage of weak - wind inner side of the shock cone across the line of sight @xcite . in the soft x - ray emission
is present at all observed epochs .
its variation can be explained by changing emission measure , while the absorption is nearly constant and comparably low . a soft spectral component in the low state
is also observed in @xmath20vel , v444cyg and @xmath21car .
the origin of soft x - rays in these stars is attributed either to individual stellar winds or to a circumstellar nebula .
accordingly , in these objects the absorbing column to the soft component is similar to the interstellar absorbing column .
in contrast , we find for that @xmath45 , and the em of the soft component changes with the x - ray light - curve .
thus it is unlikely that the intrinsic wind emission of a companion is solely responsible for the soft x - rays from wr65 .
wr65 is a persistent dust maker @xcite .
@xcite notice that binary stars that are persistent dust makers have nearly circular orbits .
we can assume that the same holds for .
there is a growing number of wc9d binary stars that constitute the `` pinwheel '' class of objects @xcite . in these binary systems
the dust is formed in the wind collision region and is distributed along a wide ( compared to the orbital separation ) archimedean spiral .
the peak of dust production occurs at some distance downstream from the colliding wind stagnation point .
so far there is no direct evidence that is a pinwheel .
however , the topology and structure of circumstellar matter in a pinwheel could help to explain x - ray spectral variability in .
@xcite presented a model for the absorption of x - rays that includes a treatment of dust grains .
their work confirmed that large - grain absorption for soft x - rays ( @xmath46kev ) is reduced due to the self - blanketing of grains .
interestingly , from the point of view of radiative transfer this is identical to the reduction of opacity for the x - rays in clumped stellar winds @xcite .
winds of wc stars are extremely rich in c and o and could have enhanced abundance of mg and ne .
c and o are dominant absorbers for the softer x - rays ( @xmath47kev ) , while mg and ne strongly contribute to the absorption of harder x - rays ( @xmath48kev ) . in the dusty shell around a wc9d star ,
c would be depleted from the gaseous phase and form grains , reducing the opacity for the softer x - rays .
this may help to explain the large differences between the column densities that we determine from soft and hard x - rays .
recently , @xcite presented 3-d hydrodynamical simulations of colliding winds . scaling the results of their simulation to the parameters of ,
soft x - ray emission could originate from a large volume extending above the orbital plane .
this emission would show little variation with orbital phase and suffer comparably mild absorption in the outer regions of the wc star wind , similarly to what is observed in wr65 . on the other hand ,
a high postshock temperature of @xmath49kev is predicted for a confined region near the point where the winds collide head - on .
densities up to @xmath50g@xmath51 can be expected in the dusty spiral arms . assuming the wind opacity for hard x - rays as @xmath52@xmath53g@xmath34 @xcite
, a spiral arm width of @xmath54 cm would explain the observed high absorption .
this width is consistent with the observed width of the dusty spiral arms in wr104 @xcite .
the lowest absorbing column can be expected at the phase when the line of sight crosses only the outer low density part of the spiral arms .
it is possible that the observations of at high state have caught this orbital phase . to conclude
, our first brief study of the variable x - ray emission from indicates that it is a colliding wind system , where the dust and complex geometry of its colliding wind region are pivotal in explaining its x - ray properties . at present , is the only known wc9d star where x - ray emission is detected .
future observations are needed to constrain the orbital parameters of and allow for detailed modeling .
the authors are grateful to p. m. williams for initiating this study and to n. s. schulz for useful discussions on the acis - i spectra .
the authors thank the referee ( a.f.j .
moffat ) for detailed and helpful comments . based on data : obsid 754 ,
6117 , 6116 , 5535 , 5534 , 5515 , 4384 , 3834 , 3833 .
this research has made use of software provided by the chandra x - ray center and data obtained through the nasa / ipac infrared science archive and high energy astrophysics science archive research center .
the simbad database , operated at cds , strasbourg , france , and nasa s astrophysics data system were extensively used during this work .
corcoran m. f. , pittard j. m. , stevens i. r. , henley d. b. , pollock a. m. t. , 2005 , in : x - ray and radio connections , eds .
sjouwerman and k.k dyer . published electronically by nrao , http://www.aoc.nrao.edu/events/xraydio | we report the discovery of variability in the x - ray emission from the wolf - rayet type star wr65 . using archival data spanning over 5yr
we detect changes of the x - ray flux by a factor of 3 accompanied by changes in the x - ray spectra .
we believe that this x - ray emission originates from wind - wind collision in a massive binary system .
the observed changes can be explained by the variations in the emission measure of the hot plasma , and by the different absorption column along the binary orbit .
the x - ray spectra of display prominent emission features at wavelengths corresponding to the lines of strongly ionized fe , ca , ar , s , si , and mg .
wr65 is a carbon rich wc9d star that is a persistent dust maker .
this is the first investigation of any x - ray spectrum for a star of this spectral type .
there are indications that the dust and the complex geometry of the colliding wind region are pivotal in explaining the x - ray properties of .
[ firstpage ] stars : early - type stars : wolf - rayet stars : individual : wr65 x - ray : stars |
it is well known that metallic nanoparticles can sustain localized surface plasmon ( lsp ) oscillations , whose resonance frequencies in the quasi - static limit depend solely on the geometry of the nanoparticle , the permittivity of the metal and the surrounding permittivity .
the dependency of the lsp resonance ( lspr ) on the surrounding medium makes metallic particles extremely good sensors , progressing towards the detection of single molecules @xcite .
however , the weak effect of retardation on the lsp resonance in nanosized metal particles leaves only one parameter to truly engineer : the geometry . by modifying the structure of the metal nanoparticle to have a dielectric core with a metal shell
, an increased tunability is achieved due to the plasmon hybridization of the inner and outer surfaces of the metal @xcite .
especially the spherical core - shell structure has received a considerable amount of attention in recent years @xcite due to its excellent and tunable sensing properties , which show great promise in biological studies such as cancer therapy @xcite .
the plasmon hybridization allows one to position the lsp resonance of the nanoshell as desired by simply varying the core size @xmath0 and/or outer radius @xmath1 appropriately @xcite .
the hybridization of the inner and outer surface plasmons increases when the metal shell becomes thinner @xcite , which gives rise to significantly altered lsp resonances compared to usual homogeneous metal nanoparticles .
studies of the hybridization between two spherical @xcite or cylindrical @xcite metal nanoparticles in few - nm proximity reveal that effects of nonlocal response increase with increasing hybridization .
furthermore , nanosized metal particles @xcite and metal films @xcite are also strongly affected by nonlocal effects .
the core - shell particle thus calls for a nonlocal description , since it features an ultra - thin metallic shell with resulting strong plasmon hybridization .
the use of arrays of nanotubes with high aspect ratio for biosensing @xcite and hydrogen sensing @xcite has yielded impressive results , yet only few theoretical studies have been performed on the nanotube @xcite .
_ investigate the plasmonic modes and dispersion relations of the nanotube @xcite , while zhu _ et al .
_ perform calculations using the discrete dipole approximation to discuss the changes of the resonance wavelength of the nanotube due to variations of the aspect ratio @xcite .
thus , to our knowledge no systematic study has yet been performed that addresses which parameters determine the lspr refractive - index sensitivity of a nanotube - based sensor . in this paper , we fill this gap with a systematic study of the sensing and scattering properties of a single infinitely long cylindrical core - shell nanowire ( see inset of fig .
[ fig : fig1 ] ) , which is a good description of dilute arrays of non - interacting nanotubes with high aspect ratio . on the basis of this study
, we propose how to optimize a nanotube - based sensor to achieve the utmost sensitivity for the refractive - index sensing of both gases and liquids .
the outline of this paper is as follows . in sec .
[ sec : theory ] we discuss the physical principles of local and nonlocal response , and introduce the sensitivity and figure - of - merit ( fom ) as quantitative measures of the performance of a lspr - based sensor . section [ sec : results ] is dedicated to the study of a nanotube with a silica core and gold shell .
we determine the dependency of the sensitivity and fom on the shape and size of the nanotube , using both local and nonlocal theory to model the response of the gold shell .
our conclusions and outlook on nanotube - based sensors is given in sec .
[ sec : con ] , and details on the analytical calculations in the appendix .
the ability of lspr - based sensors to detect changes in the refractive index of their surrounding medium is usually quantified by the sensitivity and fom @xcite .
the sensitivity @xmath2 is determined as the shift in wavelength of the considered lsp resonance in the extinction spectrum of the sensor , when varying the background refractive index @xmath3 , while the fom is given as @xmath4 where @xmath5 is the resonance linewidth , calculated as the fwhm of the considered lsp resonance in the extinction spectrum .
thus , to determine the performance of the nanotube as a lspr sensor , we must calculate its extinction cross section , as this quantifies the extinction spectrum and therefore allows us to determine the sensitivity and fom .
predictions for the extinction cross section depend on how the optical response of electrons in the metal is modeled . the common approach to describe the response of metals is by making the local approximation which assumes that the response field at a certain position is proportional to the driving field at that position , with the proportionality function being a position- and frequency - dependent dielectric function .
this approach has the rather unphysical consequence that all surface charges reside on an infinitely thin layer on the boundaries of the metal , thereby neglecting the actual extent ( or wave nature ) of the electrons .
while the local approximation is justified as long as the metal boundaries are far apart such that the interaction between electrons due to their extent can be neglected ( i.e. large metallic structures ) , it can not be safely assumed for nanosized metal particles where the wavelength of the electron becomes comparable in size to the metal particle . describing the metal using the semiclassical hydrodynamic drude model @xcite , we relax the local approximation by allowing the existence of local inhomogeneity in the density of the electron gas , which gives rise to pressure waves .
the electron - gas pressure waves provide a means to transport energy in the metal in addition to the electromagnetic waves , which gives rise to nonlocal response : the response of the metal at a certain spatial point can depend on the driving field at other nearby points ( on the length scale of the fermi wavelength ) in the metal . in the appendix , we provide an analytical expression for the extinction cross section in the cases of both nonlocal and local response , for a normally incident tm - polarized wave , see the inset of fig .
[ fig : fig1 ] .
we have checked the analytical expression with our numerical implementation of the hydrodynamic drude model @xcite , which showed perfect agreement ( not shown in this paper ) .
we consider the specific core - shell structure , where the core is silica ( @xmath6 ) with dielectric constant @xmath7 and the shell is gold ( au ) modeled with the data by raki _
_ @xcite . to clearly show the difference in extinction cross section in local and nonlocal response , we start by examining the case where interband effects in au are neglected .
figure [ fig : fig1 ] depicts the extinction cross section for a @xmath8 silica - au cylinder in vacuum comparing the local and nonlocal model .
the local approximation shows three distinct peaks , two at low frequencies ( dipole and quadrupole peaks ) and one at a high frequency ( near 7ev ) .
these are due to the interaction between the localized plasmons at the inner and outer surface of the nanoshell , or equivalently , the interaction between a cavity mode and a cylinder mode @xcite .
the nonlocal description allows the same classification of peaks as the local approximation @xcite , although the high - frequency peak is blueshifted compared to the local model . since sensing depends on peak shifts , it is important to take possible nonlocal blueshifts into account
however , the low - frequency resonances show no noticeable blueshift , because the strength of the nonlocal blueshift does not only increase with decreasing thickness of the metal layer @xcite but it also depends on the frequency , with a decreasing blueshift for lower frequencies .
thus we find that there is an intricate interplay between plasmon hybridization and nonlocal response : since a thinner metal shell produces stronger plasmon hybridization , the dipole and quadrupole peaks are pushed to such low frequencies that the nonlocal blueshift effect due to nanosized metallic features is counteracted by the low frequency of the resonances .
the panel on the right of fig .
[ fig : fig1 ] shows the nonlocal normalized intensity distribution in the metal at the dipole and quadrupole resonance frequencies , illustrating the expected dipole and quadrupole nature of the resonances . above the plasma energy @xmath9
we see the characteristic additional resonances in the nonlocal model due to the excitation of longitudinal modes , as previously reported for different metal nanoparticles @xcite .
the difference between the red and green curves in fig .
[ fig : fig1 ] shows the importance of taking into account interband transitions in the response of the metal shell .
the implications on the dipole and quadrupole resonances are that they are redshifted and damped due to interband transitions , with greatest impact on the quadrupole peak . in the remaining part of this paper
, we will always use measured values for the dielectric function @xcite , i.e. we take interband transitions into account .
we will concentrate on the dipole resonance , since this peak is the strongest , is close to visible and infrared frequencies and can be affected by the shape and size of the cylinder and the background permittivity .
there are two geometrical properties that can be modified in the nanotube structure : the first is the shape defined by the @xmath10 ratio and the second is the overall size , that is , varying the outer radius @xmath1 but keeping @xmath10 constant . in fig .
[ fig : fig2 ] we show the effect of shape variations of the nanotube on its sensing abilities , which is quantified through the change in the dipole resonance wavelength when the background refractive index is increased .
we see that regardless of the shape , the dependency is always approximately linear . however , as shown in fig .
[ fig : fig2](i ) there is no significant dependency on the background refractive index for low @xmath10 ratios , indicating the lack of ability to sense . only when the shell becomes thin ( @xmath11 ) does the resonance wavelength shift with the refractive index .
the thinner the shell , the greater is the average slope of the curves .
relaxing the nonlocal description to a local one does not change this trend , because the dipole resonances occur at too low energies for the nonlocal blueshift to kick in .
furthermore , the resonance wavelength shifts to higher wavelengths when the shell becomes thinner , because the coupling between the cavity and cylinder modes increases .
thus , even though fig .
[ fig : fig2](iv ) represents a nanotube with a @xmath12 thin metal shell , where nonlocal blueshifts are expected to be very prominent , the local approximation predicts sensitivities that are almost identical to the nonlocal description .
so , as in fig .
[ fig : fig1 ] , here in fig .
[ fig : fig2 ] we see that for ultra - thin nanotubes the usual observation of larger nonlocal blueshifts for smaller structures does not occur .
the nonlocal blueshift cancels out with the decrease of the resonance energy due to increased hybridization .
.sensitivity and figure - of - merit calculations eq .
( [ eq : fom ] ) in the nonlocal description at the refractive index of air @xmath13 ( for gas sensing ) and water @xmath14 ( for liquid sensing ) for the four different shapes of fig .
[ fig : fig2 ] . [ cols="^,^,^,^,^ " , ] [ tab : tab1 ] for a more quantitative description of the sensitivity of the nanotube , we present sensitivity and fom calculations of the nanotube structures shown in fig . [
fig : fig2 ] at the refractive index of air and water in table [ tab : tab1 ] . as in fig .
[ fig : fig2 ] , it is again clear from table [ tab : tab1 ] that increased sensitivity can be achieved for thinner metal shells . comparing
the sensitivity of the nanotube with other lspr sensors based on different nanoparticle geometries @xcite , where the sensitivity is in the range @xmath15 nm per refractive index unit ( riu ) , shows that the nanotube is comparable in sensitivity for ratios @xmath16 , while it is superior for very high @xmath10 ratios .
comparison of the fom with other nanoparticle lsp sensors also shows equally good performance by the nanotube , although the fom is mainly dependent on the properties of au and not easily improved by changing the geometry @xcite .
the sensitivity values in table [ tab : tab1 ] also reveal that the nanotube has a high sensitivity at both the refractive index of air and water , which shows the versatility of a nanotube - based sensor and its applicability as both a gas and liquid sensor .
besides shape variations , we also varied the size @xmath1 of the nanotube , while keeping @xmath10 constant .
figure [ fig : fig3 ] depicts the dipole resonance wavelength as a function of the background refractive index for three different sizes with @xmath17 .
the sensing ability of the nanotube is not as dependent on size as it is on shape , which can be seen by the three almost parallel lines in fig .
[ fig : fig3 ] . even though the sensitivity does not change much with increasing size , there is still an optimum size which occurs at @xmath18 and @xmath19 for liquid and gas sensing , respectively , see the inset of fig .
[ fig : fig3 ] .
the fact that it is neither the smallest nor the biggest nanotube size that gives the highest sensitivity can be explained by a trade - off between the total structure size and the shell thickness .
if the size of the structure is too small , then we have a weak lsp excitation and thereby poor sensing ability , but if the structure size is too big ( with the shape kept constant ) the absolute shell thickness increases , which also decreases the sensing ability , as we saw in fig .
[ fig : fig2 ] .
therefore , for a larger @xmath10 value the optimum size will also be larger . in fig .
[ fig : fig3 ] we also show the calculations using the local approximation .
as seen , effects are surprisingly well accounted for even with a local description , despite the fact that we actually consider very thin metallic shells , for instance a 3 nm shell in fig .
[ fig : fig3](iii ) , with concomitant strong plasmon hybridization .
the strong hybridization in ultra - thin metal shells shifts the dipole resonance to very low energies , where the nonlocal blueshift is weak .
the sensitivity and consequently the fom are therefore weakly influenced by nonlocal response . although it is hardly visible in fig .
[ fig : fig3 ] , the local resonances do in fact occur at slightly longer wavelengths than in the nonlocal description , revealing a small nonlocal blueshift .
we have examined the infinite single dielectric - metal nanotube structure as an approximation for a dilute array of nanotubes with high aspect ratio .
we calculate the extinction properties of a silica - gold nanotube analytically for both local and nonlocal response by extending the mie theory for nanowires to nanotube geometries .
our investigation reveals that in contrast to the spherical nanoshell @xcite , the sensing ability of the nanotube is highly dependent on the shape of the structure , where few - nm thin shells produce extreme sensitivities .
the sensitivity is shown to be less dependent on the overall structure size .
the sensitivity at the refractive index of air and water of ultra - thin nanotubes are superior to other nanoparticle geometries , making nanotubes very promising for both gas and liquid sensing .
our results also show unexpectedly that nonlocal response has negligible influence on the extinction and sensing properties of the nanotube , even though the metal shell is ultra - thin ( a few nm ) , because the hybridization in the nanotube is so strong that the dipole resonance is pushed to very low energies .
the strength of the nonlocal blueshift is an interplay between the metal thickness and the resonance energy , where a thinner shell produces a stronger blueshift while a lower energy produces a weaker blueshift .
this interplay is surprisingly well - balanced in the nanotube structure , because a thinner shell gives rise to lower resonance energies . with the high sensitivity and good fom of the nanotube geometry
, we propose a sensor based on ultra - thin nanotubes .
the robustness of the sensitivity of the nanotube to size variations provides desirable advantages , since fluctuations in size due to imperfect fabrication will have less impact . in the special case of gas sensing , the sensitivity may be further improved by a factor of two by designing the nanotube to have a hollow core . with a hollow core , the inner surface of the metal shell is also exposed to the surrounding medium , which significantly improves the sensitivity .
however , mechanical stability is sacrificed with a hollow core if for instance the nanotubes are to stand vertically on a substrate .
the nonlocal optical properties of the nanotube are determined by solving maxwell s wave equation coupled to the hydrodynamic equation for the current @xcite .
we solve the coupled set of equations by extending the mie theory for wires of ref .
@xcite to core - shell structures . by expanding the electromagnetic fields in the dielectric core , metal shell and surrounding medium in cylindrical bessel functions
, we can most easily take into account maxwell s boundary conditions along with the additional boundary condition of a vanishing normal component of the current in the nonlocal case @xcite .
although quantum tunneling is not taken into account with this treatment , we do not expect any such effects to be important in this structure @xcite . to determine the extinction property of the infinite cylindrical nanotube we calculate the extinction cross section @xcite @xmath20 where @xmath21 is the background wave vector , @xmath22 is the background permittivity and @xmath23 is a cylindrical bessel - function expansion coefficient for the scattered electromagnetic field .
we consider a normally incident electric - field polarization perpendicular to the cylinder axis ( tm ) , as sketched in the inset of fig .
[ fig : fig1 ] .
the nonlocal - response scattering coefficient is calculated analytically as @xmath24 - \sqrt{\epsilon } j_n^\prime(k_0 r_2 ) \left[j_n p_n - h_n q_n\right ] } { \sqrt{\epsilon_\text{b } } h_n(k_0 r_2 ) \left[c_n+
j_n^\prime p_n - h_n^\prime q_n\right ] - \sqrt{\epsilon } h_n^\prime(k_0 r_2 ) \left[j_n p_n - h_n q_n\right]}. \label{eq : annl}\ ] ] here , @xmath25 and @xmath26 are the bessel and hankel functions of the first kind , @xmath27 and @xmath28)$ ] is the drude local - response function that includes interband effects through @xmath29 .
the argument of the bessel and hankel functions are @xmath30 unless written explicitly otherwise .
the coefficients @xmath31 , @xmath32 and @xmath33 are given by @xmath34 , \label{eq : pn } \\
q_n & = q_n \alpha_n + j_n(k_\text{c}r_1 ) \left[j_n(k_\text{t}r_1 ) \delta_n + j_n \tau_n \right ] , \label{eq : qn } \\ c_n & = \frac{i n}{k_0 r_2 } \left [ h_n(k_\text{l}r_2 ) c_n - j_n(k_\text{l}r_2 ) d_n \right ] , \label{eq : cn}\end{aligned}\ ] ] where @xmath35 and @xmath36 is the dielectric constant of the core
. furthermore , @xmath37 and @xmath38 with @xmath39 being the fermi velocity of the metal shell .
the coefficients @xmath40 , @xmath41 , @xmath42 , @xmath43 and @xmath44 of eqs . (
[ eq : pn]-[eq : qn ] ) are given as @xmath45 , \label{eq : alphan } \\
\delta_n & = -\frac{k_\text{l}n^2 \sqrt{\epsilon_\text{c } } \epsilon_\text{other } ( \epsilon-\epsilon_\text{other})}{k_\text{t}k_0 ^ 2 r_1 ^ 2 } \left [ j_n^\prime(k_\text{l}r_2 ) h_n(k_\text{l}r_1)- h_n^\prime(k_\text{l}r_2 ) j_n(k_\text{l }
r_1 ) \right ] , \label{eq : deltan } \\
\tau_n & = -\frac{k_\text{l}n^2 \sqrt{\epsilon_\text{c } } \epsilon_\text{other } ( \epsilon-\epsilon_\text{other})}{k_\text{t}k_0 ^ 2 r_1 r_2 } \left [ h_n^\prime(k_\text{l}r_1 ) j_n(k_\text{l}r_1)- j_n^\prime(k_\text{l}r_1 ) h_n(k_\text{l } r_1 ) \right ] , \label{eq : taun}\end{aligned}\ ] ] while the coefficients @xmath46 and @xmath47 of eq .
( [ eq : cn ] ) are given as @xmath48 + j_n^\prime(k_\text{l}r_1 ) g_n \left [ j_n p_n - h_n q_n \right ] , \label{eq : cn } \\
d_n & = f_n \left [ h_n^\prime(k_\text{l}r_2 ) \eta_n + h_n(k_\text{l } r_1 ) \kappa_n \right ] + h_n^\prime(k_\text{l}r_1 ) g_n \left [ j_n p_n - h_n q_n \right ] , \label{eq : dn}\end{aligned}\ ] ] where @xmath49 , \label{eq : etan } \\ \kappa_n & = \frac{n^2 ( \epsilon-\epsilon_\text{other})}{k_\text{t } r_2 r_1 } \left [ j_n(k_\text{t}r_1 ) h_n - h_n(k_\text{t}r_1 ) j_n \right ] .
\label{eq : kappan}\end{aligned}\ ] ] the local - response result can be retrieved in the limit of a vanishing fermi velocity for which @xmath50 , @xmath51 and @xmath52 . | we study the refractive - index sensing properties of plasmonic nanotubes with a dielectric core and ultra - thin metal shell .
the few - nm thin metal shell is described by both the usual drude model and the nonlocal hydrodynamic model to investigate the effects of nonlocality .
we derive an analytical expression for the extinction cross section and show how sensing of the refractive index of the surrounding medium and the figure - of - merit are affected by the shape and size of the nanotubes .
comparison with other localized surface plasmon resonance sensors reveals that the nanotube exhibits superior sensitivity and comparable figure - of - merit .
example.eps gsave newpath 20 20 moveto 20 220 lineto 220 220 lineto 220 20 lineto closepath 2 setlinewidth gsave .4 setgray fill grestore stroke grestore |
the construction of an internally consistent and falsifiable theory of quantum gravity is one of the major challenges of modern theoretical physics . as the perturbative quantisation of the einstein - hilbert action yields a non - renormalisable theory @xcite several alternative approaches
have been proposed : a change in the degrees of freedom and of the microscopic action , as well as a different approach to quantisation , or assumptions about a discrete nature of spacetime offer possible routes to a predictive theory of quantum gravity .
however , the possibility remains that the apparent nonrenormalisability is not a failure of einstein gravity but rather of the simple perturbative quantisation scheme .
this is the underlying viewpoint of , e.g. , lattice simulations of the gravitational path integral @xcite . within analytical continuum approaches to a non - perturbative quantisation of gravity , weinberg s asymptotic - safety scenario @xcite represents a possible way for a predictive theory of quantum gravity :
weinberg argued that , if the non - perturbative renormalisation group trajectory of a quantum field theory approaches a non - gauian fixed point ( ngfp ) in the ultraviolet ( uv ) , the uv limit can safely be taken .
if furthermore the ngfp has a finite number of relevant directions , the theory has predictive power .
if moreover an rg trajectory emanating from the ngfp can be connected continuously with a regime in the infrared , where both the cosmological constant and newton s constant are positive and small , the asymptotic - safety scenario is consistent with our universe as we observe it now . a suitable tool to study this scenario is the functional renormalisation group ( rg ) .
formulated in terms of the wetterich equation @xcite , the functional rg describes the change of the effective average action @xmath0 as a function of a momentum scale @xmath1 : @xmath2^{-1}(\partial_tr_k)\ } , \quad \partial_t = k\frac{d}{dk}. \label{eq : wetterich}\ ] ] here , @xmath3 denotes the second functional derivative of this effective action with respect to the fields , which is the full inverse propagator at the scale @xmath1 .
the quantity @xmath4 is an infrared regulator , which suppresses contributions of modes with momenta @xmath5 to the supertrace .
projecting the equation onto a given operator on the left- and right - hand side yields the non - perturbative @xmath6 function of the coupling associated with this operator .
the functional rg for gravity has been pioneered by reuter @xcite .
it proceeds in many ways similar to the corresponding quantisation of yang - mills theories , see , e.g. , @xcite .
whereas various systematic and consistent nonperturbative approximation schemes , such as derivative or vertex expansions have been devised and successfully applied in a large variety of cases ranging from critical phenomena to strong - coupling problems in gauge theories , the choice of truncations of the effective action in gravity has partly been guided by the available computational tools , most notably the heat - kernel expansion .
early calculations have focussed on the einstein - hilbert truncation where the existence of a ngfp has been observed for the first time @xcite .
the fixed point has remained remarkably stable upon the inclusion of higher orders of the curvature @xcite , even up to @xmath7 @xcite .
recently , new techniques have allowed to go beyond the class of @xmath8 truncations and include , e.g. , the operator @xmath9 involving the weyl tensor among the gravitational interactions @xcite .
not only the existence of the non - gauian fixed point has been confirmed in a variety of studies , but also the critical exponents classifying the relevant directions at the fixed point show a satisfactory convergence upon increasing the truncation .
all these investigations have accumulated a substantial body of evidence for asymptotic safety of quantum einstein gravity ( qeg ) ( for reviews see @xcite ) .
this stimulated , of course , also the study of phenomenological implications of the existence of a fixed point ( see e.g. @xcite for references ) . as is known from certain gauges in yang - mills theories , e.g. , the landau gauge or the coulomb gauge
, ghosts can play an important if not dominant role in the strong - coupling regime of a gauge theory @xcite .
this is because the strong - coupling regime can be entropically dominated by field configurations near the gribov horizon which induce an enhancement of the ghost propagator .
as the gribov ambiguity is also present in standard gauges in general relativity @xcite , similar mechanisms may become relevant in qeg .
however , investigations of the non - gauian fixed point in qeg have neglected running couplings in the ghost sector so far , mainly for technical reasons . only a classical ghost term has been considered . in this work ,
we take a first step in this direction and examine the rg flow in the ghost sector . more specifically , we truncate the theory space of qeg down to the following action : @xmath10 where @xmath11f_{\nu}[\bar{g},h]\label{eq : ggf},\\ \gamma_{\mathrm{gh}}&= & - \sqrt{2}\,z_{\mathrm{c}}(k)\int d^d x \ , \sqrt{\bar{g}}\ , \bar{c}_{\mu } \mathcal{m}^{\mu}{}_{\nu}c^{\nu}\label{eq : ggh},\\ \gamma_{\mathrm{r\,gh}}&= & \bar\zeta ( k ) \int d^d x \sqrt{\gamma}\ , \bar{c}^{\mu}r c_{\mu } , \label{eq : grgh}\end{aligned}\ ] ] and @xmath12 .
the einstein - hilbert action @xmath13 contains the dimensionful newton constant @xmath14 , the running graviton wave function renormalisation @xmath15 , and the running cosmological constant @xmath16 . for the necessary gauge fixing of metric fluctuations ,
we apply the background - field method @xcite , where the full metric @xmath17 is decomposed into the background metric @xmath18 with its compatible covariant derivative @xmath19 .
@xmath20 denotes the fluctuations around this background which do not have to obey any constraints concerning their amplitude .
this is a crucial difference to a perturbative treatment on a fixed background .
the curvature scalar constructed from the full metric is denoted by @xmath21 , the one pertaining to the background metric by @xmath22 .
the gauge - fixing action @xmath23 contains the gauge - fixing condition @xmath24 $ ] , which reads for the background - covariant generalisation of the harmonic gauge : @xmath25= \sqrt{2 } \bar{\kappa } \left(\bar{d}^{\nu}h_{\mu \nu}-\frac{1+\rho}{d}\bar{d}_{\mu}h^{\nu}{}_{\nu } \right).\ ] ] in eq .
, we ignore any independent running of @xmath26const . the appropriate ghost term @xmath27 contains the faddeev - popov operator @xmath28 and @xmath29 denotes a wave function renormalisation for the ghosts , which we will later set to @xmath30 in our calculations ; this amounts to a classical treatment of the ghost `` kinetic '' term . the @xmath31 term in eq .
contains the running curvature - ghost coupling @xmath32 which is of central interest in this work . in any dimension
, it corresponds to a marginal coupling in a perturbative power - counting classification , as the ghosts carry canonical dimension @xmath33 .
since such a classification does no longer hold at a non - gauian fixed point , it is a crucial question whether this curvature - ghost coupling becomes relevant or irrelevant because of the interactions . in the remainder of this work ,
we drop the argument @xmath1 of the couplings which are implicitly understood as running couplings .
our truncation defines a hypersurface in theory space .
the solution of the flow equation provides us with an rg trajectory in this hypersurface , once an initial condition is specified . for unspecified initial conditions ,
the flow equation defines a vector field on this hypersurface . in order to arrive at an explicit representation of this vector field in terms of the flow of the couplings , we need to project the right - hand side of the flow equation onto the operators of our truncation . for this
, we perform the computation on a maximally symmetric background metric of a @xmath34 dimensional sphere of radius @xmath35 in euclidean space where @xmath36 whereas this leads to an enormous simplification of the calculations , this background does not allow to disentangle the flow of @xmath37 from that of , e.g. , an operator of the form @xmath38 . moreover ,
as we set @xmath39=@xmath40 in the flow equation , an operator of the form @xmath41 can not be disentangled from @xmath37 .
our calculation shares these ambiguities with most of the other works on the asymptotic - safety scenario for qeg , as the derivation of the flow equation with two distinct metrics and more complex backgrounds is highly non - trivial .
a simplifying but sufficiently unique choice for the background ghost fields is given by covariantly constant ghosts , @xmath42 which allows to disentangle the flow of @xmath43 from that of the ghost wave function renormalisation @xmath44 .
past works in qeg evaluated the trace on the right - hand side of the flow equation by invoking a propertime representation and using heat - kernel techniques . for our purposes , we explicitly invert @xmath45 on a @xmath34 sphere to get the graviton propagator in terms of a basis of hyperspherical harmonics . the trace operation over the eigenvalues can then be evaluated explicitly .
this technique has the advantage that it allows for a straightforward inclusion of external ghost fields , and hence the flow of the coupling parameter of the curvature - ghost term @xmath43 is accessible with our method . for this
, we use a decomposition of @xmath46 into an inverse propagator matrix contribution @xmath47+r_k$ ] containing the regulator but no external ghost fields and a fluctuation matrix contribution @xmath48 containing external ghost fields .
the components of @xmath48 are either linear or bilinear in the ghost fields . with this decomposition , the right - hand side of the flow equation can be expanded as follows : @xmath49^{-1}(\partial_t r_k)\}\label{eq : flowexp}\\ & = & \frac{1}{2 } { \rm str}\tilde{\partial}_t \ln ( \mathcal{p}_k+\mathcal{f } ) \nonumber\\ & = & \frac{1}{2 } { \rm str } \tilde{\partial}_t\ln \mathcal{p}_k + \frac{1}{2}\sum_{n=1}^{\infty}\frac{(-1)^{n-1}}{n } { \rm str } \tilde{\partial}_t(\mathcal{p}_k^{-1}\mathcal{f})^n,\nonumber\end{aligned}\ ] ] where the derivative @xmath50 in the second line by defintion acts only on the @xmath1 dependence of the regulator , @xmath51 . on the background of @xmath34 spheres ,
the left - hand side of the flow equation for the scalar curvature - ghost coupling after setting @xmath52 takes the form @xmath53 as @xmath48 contains terms linear and bilinear in the ghosts , the flow of @xmath43 is fully determined by the @xmath54 and @xmath55 terms of the expansion of the flow equation eq . .
we evaluate the fluctuation matrix entries in the landau - dewitt gauge @xmath56 , as this is a fixed point of the rg flow @xcite , and we set the second gauge parameter @xmath57 for our evaluation of the einstein - hilbert sector ; see app.[eh ] .
then we apply the york decomposition @xcite to the graviton @xmath58 to identify its trace part @xmath59 , its transverse traceless part @xmath60 , and the traceless longitudinal vector and scalar parts @xmath61 and @xmath62 : @xmath63 in the landau - dewitt gauge , the contribution of all graviton modes except for the transverse traceless tensor mode is zero , as can be seen by considering the generic form of the flow equation : the full propagator ( including a regulator ) of vector , scalar and conformal mode is @xmath64 in the background version of the harmonic gauge eq . . thereby
, each term in the @xmath65 expansion containing the vector , scalar , or conformal mode will always be @xmath64 , too .
hence , we arrive at the five diagrams displayed in fig .
[ diagrams ] , in which only the transverse traceless graviton mode propagates : .
four diagrams on the bottom : two - vertex contribution to the wave function renormalisation of the ghost and the curvature - ghost coupling @xmath66 .
spiral lines denote gravitons , dashed lines denote ghosts .
the regulator insertion in the internal propagators is implicitly understood.,title="fig : " ] .
four diagrams on the bottom : two - vertex contribution to the wave function renormalisation of the ghost and the curvature - ghost coupling @xmath66 .
spiral lines denote gravitons , dashed lines denote ghosts .
the regulator insertion in the internal propagators is implicitly understood.,title="fig : " ] the two - vertex diagram appears in four different versions , as two different antighost - ghost - graviton vertices exist in our truncation , namely one @xmath67 and one @xmath68 .
accordingly , there is a diagram with two @xmath43 vertices , one with two @xmath44 vertices and two mixed ones .
now we can make use of the fact that the equation of motion for the transverse traceless part of @xmath69 vanishes identically on @xmath34 spheres for vanishing ghost fields , @xmath70 hence , the antighost - ghost - graviton vertex which is @xmath71 does not receive a contribution from the transverse traceless mode , when evaluated on the euclidean desitter background .. hence , a different choice of background would require the evaluation of some of the ghost self - energy diagrams .
however , this does not imply that the final result for the @xmath6 function is background dependent , as the fluctuation matrix entering the tadpole diagram would also change on a different background .
a maximally symmetric background simply corresponds to a very efficient organisation of the flow equation also in terms of diagrams . ] thereby , three of the four sunset diagrams are zero .
finally , the diagram with two @xmath68 vertices vanishes for covariantly constant ghost background fields , as the ghost kinetic term can be brought into the following symbolic form by partial integration @xmath72 upon deriving the antighost - ghost - graviton vertex connected to the antighost from this equation , the result is always proportional to @xmath73 , and hence the diagram vanishes on the covariantly constant ghost background , cf .
this simply reflects the fact that this diagram contributes solely to the running of the ghost wave function renormalisation .
to summarise , only the tadpole diagram contributes to the running of @xmath43 , corresponding to the @xmath54 term in the expansion of the flow equation eq . .
to evaluate the tadpole diagram , the necessary fluctuation matrix entry projected onto a @xmath34 sphere is @xmath74 next , we need the explicit form of the transverse traceless graviton propagator . the second variation of the einstein - hilbert action including the gauge fixing with respect to the metric takes the form @xcite @xmath75 in the landau - dewitt gauge , we need only the inverse propagator for the transverse traceless tensor on the @xmath34 sphere : @xmath76
now we can invert the two - point function to arrive at the full regularised @xmath1-dependent propagator , @xmath77 where symmetrisation of @xmath78 and @xmath79 on the left - hand side is understood implicitly . at this stage , we take advantage of the existence of a basis for traceless tensor functions on the @xmath34 sphere : the tensor hyperspherical harmonics @xmath80 .
they fulfil a completeness and an orthogonality relation , and are eigenfunctions of the covariant laplacean @xmath81 , @xmath82 for the tensor mode there is a @xmath83-fold degeneracy of the hyperspherical harmonics for fixed @xmath84 but different @xmath85 @xcite where @xmath86 the eigenvalues of the laplacean are given by @xmath87 as the hyperspherical harmonics form a basis for functions on the @xmath34 sphere , we can expand the green s function as follows : @xmath88 with expansion coefficients @xmath89 .
we insert our expression eq . into eq . , and use the eigenvalue equation eq . .
as the regulator is some function of @xmath90 , it turns into the same function of @xmath91 in the hyperspherical - harmonics basis . applying the completeness relation allows to rewrite the right - hand side of the definition of the green s function eq . .
by a comparison of coefficients with respect to the hyperspherical - harmonics basis , we obtain @xmath92 for @xmath93 . here
, we have assumed that the argument of the regulator @xmath94 is a function of the laplacean , @xmath95 , such that @xmath96 from the expression eq .
, it is obvious that the cosmological constant is similar to a wrong - sign mass term for the graviton modes.the @xmath54 term in the expansion of the flow eq .
finally reads @xmath97 where the trace implies an integration with measure @xmath98 . invoking the orthogonality relation eq . and
evaluating the @xmath99 derivative , we end up with the following expression : @xmath100 we parameterise the regulator function for the transverse traceless mode by @xmath101 where the shape function @xmath102 specifies the details of the wilsonian momentum - shell integration .
different choices correspond to different rg schemes . as we need to expand in the curvature radius @xmath35 in order to project the flow onto the truncation , an analytic shape function is required .
here , we work with an exponential shape function @xmath103 as an example .
moreover , the regulator can be adjusted to the flow of the spectrum of the full propagator @xcite , by choosing @xmath104 evaluated on the background field . for the graviton ,
this yields @xmath105 where the prime denotes the derivative of @xmath102 with respect to @xmath106 .
in addition , we have introduced the graviton anomalous dimension @xmath107 ( our choice of a spectrally adjusted regulator corresponds to type iii in @xcite . ) for the trace in eq .
, i.e. , the sum over @xmath84 , we invoke the euler - maclaurin formula that transforms the sum over @xmath84 into an integral .
we expand the result in powers of the @xmath34-sphere curvature @xmath35 amd project the result onto the power of @xmath108 in order to perform a comparison of coefficients with respect to the scalar curvature - ghost term , cf .
. incidentally , all non - integral terms in the euler - maclaurin formula do not contribute to the scalar curvature - ghost coupling .
our method also applies straightforwardly to the einstein - hilbert sector , where our results confirm the asymptotic - safety scenario of qeg obtained in other gauges and with other regulators , see app .
as we use an unprecedented combination of the landau - dewitt gauge @xmath56 together with a spectrally adjusted regulator , our results in this sector represent an independent confirmation of asymptotic safety .
for the discussion of the fixed - point structure of qeg , we introduce dimensionless renormalised couplings @xmath109 , @xmath110 , and @xmath111 which are related to the bare quantities by @xmath112 the running of the wave function renormalisations of graviton and ghost , @xmath113 and @xmath44 , respectively , are governed by the corresponding anomalous dimensions , @xmath114 ( see eq . ) , and @xmath115 . from this point on ,
we confine ourselves to @xmath116 , even though the calculations in the remainder are straightforwardly generalisable to @xmath117 , see also @xcite @xcite . with these prerequisites , we can now state our result for the graviton - tadpole induced flow of the coupling @xmath111 : @xmath118 this flow equation has a gauian fixed point @xmath119 .
let us investigate the properties of this fixed point in the ghost sector for the case that the remaining system is at the non - gauian fixed - point of the einstein - hilbert sector , @xmath120 , @xmath121 .
for this , we evaluate eq . at the ngfp and
obtain @xmath122 in the present truncation involving a classical ghost kinetic term , we have @xmath123 . together with the fact that @xmath124 in the physical domain , this implies that the sign of @xmath125 decides about the sign of the linearised flow of @xmath111 near its gauian fixed point .
indeed , this function is negative for all @xmath110 , see fig .
[ fig : f ] .
is negative for all values of @xmath110 and asymptotically tends to zero for @xmath126 , whereas it is unbounded for positive @xmath110 . ] the negative sign signals that @xmath111 is asymptotically free . inserting the fixed - point values from the einstein - hilbert sector in the landau - dewitt gauge @xmath56 with a spectrally adjusted cutoff ( see app .
[ eh ] ) , we get @xmath127 .
as long as the ghost anomalous dimension @xmath128 remains sufficiently small , our conclusion persists also for a larger truncation with a running ghost kinetic term .
our result implies that ( i ) the non - gauian fixed point in the graviton sector is not influenced by the scalar curvature - ghost coupling , as the latter is zero at the fixed point , and ( ii ) this curvature - ghost coupling is rg relevant for the flow towards the ir .
the latter property relates the initial value of this coupling to a physical parameter that has to be fixed by an rg condition ( i.e. , by an experiment ) .
this does not necessarily imply that the scalar curvature - ghost coupling gives rise to an _ independent _ physical parameter .
since the background - field effective action has to satisfy ( regulator - modified ) slavnov - taylor and background - field identities @xcite , this operator may be related to other purely gravitonic operators .
an answer to this question requires to resolve the difference between background - metric and fluctuation - metric dependencies which is beyond the scope of this work .
note that our arguments straightforwardly generalise to the coupling of any operator of the form @xmath129 , where @xmath130 is a scalar operator of some fields , e.g. , matter fields .
the interaction part of the corresponding flow will always have a contribution @xmath131 which supports an anti - screening flow .
of course , other contributions such as other interaction terms , the anomalous dimensions of the matter fields , and dimensional rescaling terms can eventually win out over the gravitational contributions .
we have contributed another building block to the asymptotic - safety scenario for quantum einstein gravity by computing the rg flow of a scalar curvature - ghost coupling @xmath132 with the aid of the functional renormalisation group . in our present truncation involving an einstein - hilbert sector ,
a classical ghost kinetic term , and the curvature - ghost interaction , the coupling @xmath111 is found to be asymptotically free and rg relevant .
therefore , it belongs to the conjectured finite set of physical parameters which have to be fixed for an otherwise fully predictive theory of quantum gravity .
as this curvature - ghost coupling is marginal in a perturbative power - counting scheme , the relevance of this coupling at the non - gauian fixed point of the einstein - hilbert system is another example of the tendency of the fixed - point theory to increase the critical exponents of operators towards the rg relevant regime . from a technical viewpoint
, we have shown that a direct integration of gravity fluctuations in the functional rg equation is possible without relying on heat - kernel traces and propertime representations .
as little is known about heat - kernel expansions with respect to ghost operators , a running ghost sector has not been included in an asymptotic - safety study of gravity up to now .
however , ghost operators or a more general gauge - fixing sector may carry important pieces of information about the flow of a theory in certain gauges , as it is the case , e.g. , in landau - gauge yang - mills theory in the strong - coupling domain .
we believe that a more detailed investigation of the ghost sector is important for a better understanding of the non - gauian fixed - point regime of quantum einstein gravity .
for instance , an evaluation of the ghost wave function renormalisation is still an open question , but at the same time of primary importance as it will both feed back into the einstein - hilbert sector as well as take influence on the curvature - ghost coupling studied here .
the authors would like to thank d.f .
litim , j.m .
pawlowski , and f. saueressig for helpful discussions .
this work was supported by the dfg under contract no .
gi 328/5 - 1 ( heisenberg program ) , for 723 , and gk 1523 .
in this appendix , we summarise the results obtained from applying our technique to the flow of the einstein - hilbert sector . as the combination of the landau - dewitt gauge @xmath56 together with the spectrally adjusted regulator has not been used before , our results represent an independent verification of the fixed - point scenario . where @xmath134 here , @xmath135 is the set of all modes of the york decomposition of the graviton as well as the ghost .
we have also introduced a normalising factor @xmath136 which is chosen such that numerical factors in front of the eigenvalue of @xmath90 in the inverse propagator cancel in @xmath106 .
this guarantees that all lanplacean momentum modes are regularised by the same effective ir cutoff scale . in order to regularise the conformal instability
, @xmath137 may also acquire a negative sign . inserting the propagators and using the euler - maclaurin formula for evaluating the spectral traces , we project onto the two running couplings in the einstein - hilbert sector .
e.g. , in four dimensions , @xmath138 is accompanied by a factor of the curvature radius squared , whereas the combination @xmath139 comes with a factor of @xmath140 .
this allows for a straightforward projection and yields the following @xmath6 functions of the dimensionless couplings : note that branch cuts in the logarithm and the polylogarithm contribute imaginary parts which cancel for @xmath142 .
thereby , the flow equations are completely real for all values of @xmath110 . at the fixed point , @xmath143 and @xmath144 , implying ( for @xmath142 ) @xmath145 the universal value @xmath147 is very close to the numerical values from other gauges and regulators @xcite .
the eigenvalues of the stability matrix correspond to the critical exponents at the fixed point ( apart from a minus sign ) . in the present truncation
, they appear as a complex conjugate pair , @xmath148 inducing a spiral shape of the flow lines in the vicinity of the ngfp , see fig .
[ flowspiral ] .
g. t hooft and m. j. g. veltman , annales poincare phys .
a * 20 * , 69 ( 1974 ) . m. h. goroff and a. sagnotti , phys
. lett .
b * 160 * , 81 ( 1985 ) .
a. e. m. van de ven , nucl .
b * 378 * , 309 ( 1992 ) .
s. m. christensen and m. j. duff , nucl .
b * 170 * , 480 ( 1980 ) .
j. ambjorn , j. jurkiewicz and r. loll , phys .
lett . * 95 * , 171301 ( 2005 ) [ arxiv : hep - th/0505113 ] . j. ambjorn , j. jurkiewicz and r. loll , phys .
b * 607 * , 205 ( 2005 ) [ arxiv : hep - th/0411152 ] .
j. ambjorn , j. jurkiewicz and r. loll , phys .
lett . * 93 * , 131301 ( 2004 ) [ arxiv : hep - th/0404156 ] . s. weinberg , _ in * hawking , s.w . , israel , w. : general relativity * , 790 - 831 _ s. weinberg , arxiv : hep - th/9702027 .
s. weinberg , arxiv:0903.0568 [ hep - th ] . c. wetterich , phys .
b * 301 * , 90 ( 1993 ) . c. wetterich , nucl .
b * 352 * , 529 ( 1991 ) .
m. reuter , phys .
d * 57 * , 971 ( 1998 ) [ arxiv : hep - th/9605030 ] .
m. reuter and c. wetterich , nucl .
b * 417 * , 181 ( 1994 ) .
m. reuter and c. wetterich , arxiv : hep - th/9411227 .
m. reuter and c. wetterich , phys .
d * 56 * , 7893 ( 1997 ) [ arxiv : hep - th/9708051 ] .
j. m. pawlowski , annals phys . * 322 * , 2831 ( 2007 ) [ arxiv : hep - th/0512261 ] .
h. gies , arxiv : hep - ph/0611146 . w. souma , prog .
phys . * 102 * , 181 ( 1999 ) [ arxiv : hep - th/9907027 ] .
o. lauscher and m. reuter , phys .
d * 65 * , 025013 ( 2002 ) [ arxiv : hep - th/0108040 ] .
m. reuter and f. saueressig , phys .
d * 65 * , 065016 ( 2002 ) [ arxiv : hep - th/0110054 ] .
d. f. litim , phys .
lett . * 92 * , 201301 ( 2004 ) [ arxiv : hep - th/0312114 ] .
l. n. granda and s. d. odintsov , grav .
cosmol .
* 4 * , 85 ( 1998 ) [ arxiv : gr - qc/9801026 ] . o. lauscher and m. reuter , class .
* 19 * , 483 ( 2002 ) [ arxiv : hep - th/0110021 ] .
o. lauscher and m. reuter , phys .
d * 66 * , 025026 ( 2002 ) [ arxiv : hep - th/0205062 ] .
p. f. machado and f. saueressig , phys .
d * 77 * , 124045 ( 2008 ) [ arxiv:0712.0445 [ hep - th ] ] .
a. codello , r. percacci and c. rahmede , annals phys .
* 324 * , 414 ( 2009 ) [ arxiv:0805.2909 [ hep - th ] ] .
p. forgacs and m. niedermaier , arxiv : hep - th/0207028 .
m. niedermaier , jhep * 0212 * , 066 ( 2002 ) [ arxiv : hep - th/0207143 ] .
m. niedermaier , nucl .
b * 673 * , 131 ( 2003 ) [ arxiv : hep - th/0304117 ] .
a. codello , r. percacci and c. rahmede , int .
j. mod .
a * 23 * , 143 ( 2008 ) [ arxiv:0705.1769 [ hep - th ] ] .
d. benedetti , p. f. machado and f. saueressig , arxiv:0901.2984 [ hep - th ] .
m. niedermaier and m. reuter , living rev .
* 9 * , 5 ( 2006 ) .
d. f. litim , arxiv:0810.3675 [ hep - th ] .
r. percacci , arxiv:0709.3851 [ hep - th ] .
e. manrique and m. reuter , arxiv:0905.4220 [ hep - th ] .
t. kugo and i. ojima , prog .
* 66 * ( 1979 ) 1 .
v. n. gribov , nucl .
b * 139 * , 1 ( 1978 ) .
d. zwanziger , phys .
d * 69 * ( 2004 ) 016002 .
r. alkofer and l. von smekal , phys .
* 353 * , 281 ( 2001 ) ; [ arxiv : hep - ph/0007355 ] . c. s. fischer , j. phys .
g * 32 * , r253 ( 2006 ) [ arxiv : hep - ph/0605173 ] .
a. k. das and m. kaku , nuovo cim . * 50b * , 303 ( 1979 ) . l. f. abbott , nucl . phys .
b * 185 * , 189 ( 1981 ) .
f. freire , d. f. litim and j. m. pawlowski , phys .
b * 495 * , 256 ( 2000 ) [ arxiv : hep - th/0009110 ] .
u. ellwanger , m. hirsch and a. weber , z. phys .
c * 69 * , 687 ( 1996 ) .
d. f. litim and j. m. pawlowski , phys .
b * 435 * , 181 ( 1998 ) .
j. w. york , j. math .
* 14 * , 456 ( 1973 ) . m. a. rubin and c. r. ordonez , j. math . phys . *
26 * , 65 ( 1985 ) .
m. a. rubin and c. r. ordonez , h. gies , phys .
d * 66 * , 025006 ( 2002 ) [ arxiv : hep - th/0202207 ] .
p. fischer and d. f. litim , aip conf .
* 861 * , 336 ( 2006 ) [ arxiv : hep - th/0606135 ] .
p. fischer and d. f. litim , phys .
b * 638 * , 497 ( 2006 ) [ arxiv : hep - th/0602203 ] . | we consider the asymptotic - safety scenario for quantum gravity which constructs a non - perturbatively renormalisable quantum gravity theory with the help of the functional renormalisation group .
we verify the existence of a non - gauian fixed point and include a running curvature - ghost coupling as a first step towards the flow of the ghost sector of the theory .
we find that the scalar curvature - ghost coupling is asymptotically free and rg relevant in the ultraviolet .
most importantly , the property of asymptotic safety discovered so far within the einstein - hilbert truncation and beyond remains stable under the inclusion of the ghost flow . |
gas and dust mixtures are ubiquitously present in nature , so a correct numerical prescription of its evolution turns out to be of the uttermost importance . in many astrophysical applications
, dust can be described as a set of particles immersed in a fluid phase ( gas ) .
mathematically , such a system can be described using the @xcite notation , by the following set of equations @xmath0
@xmath1 @xmath2 @xmath3 @xmath4 where @xmath5 and @xmath6 are the dust particles number density and mass respectively , @xmath7 is the gas density , @xmath8 and @xmath9 are the dust and gas velocities , @xmath10 is the gas thermal energy , @xmath11 is the drag coefficient for a single particle , @xmath12 represents the gas pressure , and @xmath13 stands for any external forces , like gravity or radiation pressure . note than in equation [ eu2 ] , the external force per unit volume @xmath14 is required for the gas .
@xmath15 is the lagrangian derivative , and its specific form will be discussed in section 2 .
the effects of forces related to the intrinsic volume of the dust particles have been ignored , since in normal astrophysical applications they become negligible . in the present work ,
we will concentrate on the study of drag forces .
the form of the drag force of gas on a single dust grain may vary considerably as a function of the grain and gas properties @xcite .
if the mean free path of the gas molecules is bigger than the dust particle radius s ( assuming spherical grains ) , the expression of the drag coefficient on a single dust grain becomes @xmath16 where @xmath17 is the velocity of the gas molecules due to thermal motion , @xmath18 the gas temperature , @xmath19 is the mean molecular weight and @xmath20 is the atomic mass of hydrogen .
if on the contrary , the mean free path of the gas molecules is smaller than the dust particle radius , the expression of the drag force on a single dust particle becomes @xmath21 where the dimensionless coefficient @xmath22 will be given by @xcite @xmath23 @xmath24 @xmath25 where @xmath26 is the reynolds number and @xmath27 is the molecular viscosity of the gas . under certain circumstances ( typically for small dust grain sizes ) , the acceleration experienced by the dust can become very large , leading to very short stopping times .
the occurrence of such short stopping times may become , under certain circumstances , a very severe problem in the numerical simulation of dust and gas mixtures .
in protoplanetary disks , for example , the typical range of body sizes spreads from micron - sized dust grains , up to kilometre - sized planetesimals .
consequently , the ranges of dust - gas coupling intensities and stopping times will be large , leading to a large range of dynamical time scales . the first attempt to study gas and dust mixtures in the framework of the sph method
was developed by @xcite , and was subsequently improved by @xcite by the inclusion of an implicit time - integration scheme .
the main problem with the method was its incapacity to guarantee a convergent solution under certain circumstances .
@xcite proposed a variation of @xcite method . despite being capable of providing stable and convergent solutions ,
their method still suffers three main difficulties , intrinsic to any two fluid approach : ( i ) an inclination to produce artificial dust clumps whenever the dust is concentrated below the gas resolution , due to the pressureless nature of the dust component , ( ii ) the necessity of a very high spatial resolution , in the high drag regime in order to avoid overdissipation , and ( iii ) the necessity of a very high number of iterations in the implicit time integration scheme , or a very high number of time - steps in the explicit scheme , for the high drag regime .
more recently , a new one fluid approach has been proposed by the same authors @xcite . in this new approach ,
both fluids are evolved as a single fluid by using the barycentric velocity as a common reference frame . through this approach ,
most of the aforementioned problems are avoided .
however , in its present state , the one fluid method struggles with the low drag regime when dust and gas are not well described as a mixture and the velocity field should be multi - valued @xcite , whereas a two - fluid method handles this situation with ease . in this paper , a new two - fluid sph method will be investigated in order to solve the third of the aforementioned problems .
a simple semi - analytical model is proposed , in order to approximate the time evolution of the dust component , and thus avoid the need for a numerical integration of its time evolution .
special attention will also be paid to the impact of overdissipation in the method . in particular
, it will be shown that the method is much better at resolving dusty shocks in the limit of short stopping times than other explicit or implicit two - fluid sph methods .
whenever possible , an estimation of the resolution requirements of the method will be provided .
this paper is organized as follows . in section 2 , the possibility of imposing an analytical decay model as an approximate solution for the small dusty grains evolution will be discussed . in section 3 ,
a series of numerical tests will be presented in order to compare the accuracy of the present method with more traditional approaches . finally , in section 4
, we will draw our conclusions . +
as mentioned in the introduction , the objective of the present work is to avoid the need for a full numerical integration of the velocity evolution of small dust grains , whenever the stopping time becomes prohibitively short . in order to do so
, one could try to estimate the total change in velocity of a dust particle , after having interacted through drag with the gas , for a certain time @xmath28 . as seen in the introduction , if we concentrate exclusively on the drag interaction , the equations of motion for the time - evolution of an arbitrary pair of dust and gas fluid elements ( represented in a two - fluid sph method by a pair of particles located at positions @xmath29 and @xmath30 ) are @xmath31 @xmath32 @xmath33 where @xmath34 is the volume density of the dust component , @xmath35 and we consider the epstein regime where we have defined @xmath36 .
in the present work , the adopted evolutionary equations for the dust and gas components are @xmath37 @xmath38 @xmath39 where @xmath40 equations [ vd_sol ] , [ vg_sol ] and [ ug_sol ] will constitute an approximate solution for the equations of motion , as long as dust and gas densities can be considered as approximately constant along the integration time - step @xmath28 , since @xmath41 @xmath42 @xmath43 the main attractive of equations [ vd_sol ] , [ vg_sol ] and [ ug_sol ] is that they can be used to approximately describe both strong and weak drag regimes . if @xmath44 , equations [ vd_sol ] to [ ug_sol ] become @xmath45 @xmath46 @xmath47 whereas if @xmath48 , equations [ vd_sol ] , [ vg_sol ] and [ ug_sol ] simply read @xmath49 @xmath50 @xmath51 which is the expected solution for the equations of motion of a strongly coupled dust and gas mixture .
another attractive feature of equations [ vd_sol ] to [ ug_sol ] is that they naturally incorporate , due to its fully lagrangian nature , perfect advection into the numerical scheme .
if one calculates the time evolution of the relative velocity between dust and gas , in the dust frame , one gets @xmath52 so , as long as the velocity evolution for each phase is calculated by using the local acceleration on each frame , the extra term related to the differential velocity of the frames , will be naturally included into the scheme .
this property , although not completely intuitive , can clearly be seen if one considers the case of ballistic particles and a gas that do not interact at all ( something that sph can treat very easily ) .
the lagrangian equations that describe the evolution of such a system , which are the equations that would be solved by an sph implementation , are @xmath53 @xmath54 if one now calculates the time variation of the relative velocity between the phases as in equation [ lagdg ] , we obtain @xmath55 where the second term on the right hand side just reflects that we have had to choose between the dust and the gas when defining our lagrangian derivative .
it is not an extra term that needs to be implemented .
note that in some recent sph one - fluid prescriptions , the extra advection terms do need to be explicitly calculated ( e.g. equation 14 of @xcite ) .
this characteristic should be clearly considered as an advantage of our method .
the key to our two fluid method for modelling a dusty gas is that we now operator split the differential equations that describe the evolution of gas and dust , so that we solve everything except the drag term using standard explicit integration methods , and subsequently modify the resulting velocities by applying the drag term separately .
for example , to include gas pressure and drag forces between the dust and the gas , we first use standard explicit sph integration to apply @xmath56 @xmath57 @xmath58 and then , we apply equations [ vd_sol ] to [ ug_sol ] to the obtained intermediate velocities and thermal energy @xmath59 , @xmath60 and @xmath61 @xmath62 @xmath63 @xmath64 + in order to apply equations [ vd_sol2 ] , [ vg_sol2 ] , and [ ug_sol2 ] in the sph two fluid approach , the gas and dust elements are discretized into a set of mass elements , often called particles .
any continuous quantity will be thus reconstructed by means of an interpolation method @xmath65 @xmath66 where @xmath67 is the mass of each sph particle , @xmath68 is the smoothing length of each sph particle , and @xmath69 is the interpolating function , called the kernel ( see for example * ? ? ?
* ) . in general , in the two fluid scheme
, the value of the gas velocity at a dust location ( and vice versa ) will be unknown , so in equations [ vd_sol2 ] , [ vg_sol2 ] , and [ ug_sol2 ] the use of equations [ interp1 ] and [ interp2 ] will be necessary .
in particular , using the @xmath70 index to refer to dust particles , @xmath71 to gas particles , and @xmath72 to the neighbours of opposite type , we can evaluate the difference between the dust and gas velocities as @xmath73 @xmath74 where @xmath75 , @xmath76 , @xmath77 , and @xmath78 . by using sph interpolation , equations [ vd_sol2 ] , [ vg_sol2 ] and [ ug_sol2 ]
can be discretized @xmath79 @xmath80 @xmath81 where @xmath82 and @xmath83 are normalisation factors ( see below ) . physically
, sph particles should be understood as finite mass elements of each one of the components .
in particular , sph dust particles should be interpreted as homogeneous ensembles of dust particles of radius @xmath84 , intrinsic mass @xmath6 , and number density @xmath5 . therefore , for each sph dust particle one can assign a volume density @xmath85 which will represent the total dust mass contained within the volume of the sph dust particle ( determined by its kernel support radius ) .
smoothing lengths and volume densities for both components can be calculated by a standard iterative sph manner , solving @xmath86 through a newton - raphson method @xcite , where @xmath87 for the standard cubic spline kernel , and the dust and gas densities are given by @xmath88 @xmath89 this procedure is equivalent to solving the continuity equations [ cont1 ] and [ cont2 ] .
sph particle masses will be assigned by dividing the total mass of each component present in the simulation , by the number of particles of the component . in order to calculate the dust - to - gas ratio at a given dust particle location
, we estimate the gas and dust mass fraction contained within the interpolation sphere of the sph dust particle .
that is , we take @xmath90 this prescription is chosen due to its greater stability , in comparison with the simpler dust and gas densities quotient .
we have found that , whenever discontinuities are present in the computational domain ( for example in the shock tube test ) , the fluctuations in the dust density can lead to high stopping time fluctuations if the ratio @xmath91 is used directly in equation [ ts ] . if equation [ dust - to - gas ] is used , because the mass of the sph dust particle is constant , the fluctuations are avoided .
furthermore , this approach allows us to calculate dust evolution even with a very low number of sph dust particles , since it does not rely on the validity of the fluid approximation for the dust component .
also , and in order to minimize fluctuations if a low number of neighbours is present , a normalization factor @xmath82 has also been included in the sph dust summation @xcite , equal to @xmath92 due to the symmetric structure of equations [ vd_sph1 ] and [ vg_sph1 ] linear momentum is preserved during the interaction , and a projection of the relative velocity along the line of sight of the particlesis introduced in order to guarantee angular momentum conservation @xcite . a normalization factor @xmath27 , equal to the number of the spatial dimensions of the system , is necessary to guarantee the equivalence of the projection method with equations [ vd_sol ] , [ vg_sol ] and [ ug_sol ] up to a second order approximation ( see @xcite for an excellent discussion ) .
for the same reason , energy can also be shown to be conserved .
the kinetic energy of the mixture , at @xmath93 , will be expressible as @xmath94 so the change in kinetic energy will be @xmath95 then , by assuming that @xmath96 one finds , after introducing the sph summations @xmath97 the method has been tested with two different integrators , a second order runge - kutta fehlberg @xcite , and a second order predictor - corrector @xcite .
the obtained results with the two integrators have been equivalent in all cases , except in the sound wave test ( section 3.2 ) where the runge - kutta scheme leads to a poorer energy and momentum conservation .
please , see appendices a and b for a detailed explanation of both integration methods . to investigate the stability of the numerical scheme , equations [ vd_sol ] and [ vg_sol ]
may be written in the following form @xmath98 @xmath99 where @xmath100 as can be seen , equations [ vd_stab ] and [ vg_stab ] can be interpreted as a forward euler method , where velocity is evolved with respect to @xmath101 instead of time . following @xcite ,
a von newmann analysis can be done .
if the dust and gas components are perturbed with a monochromatic plane wave @xmath102 [ vd_vn ] @xmath103 equations [ vd_stab ] and [ vg_stab ] may be written as the following linear system @xmath104 the corresponding two eigenvalues of the system are @xmath105 and the system will remain numerically stable ( @xmath106 ) whenever @xmath107 which will always occur , given the definition of @xmath101 , except in the limit @xmath108 . in this case ,
equation [ cour1 ] will act as a courant - like condition . in order to keep stability
, it will be enough to decrease the integration @xmath101-step by a factor of 2 , and evolve the system of equations [ vd_stab ] and [ vg_stab ] in two steps .
because of the existing linear relation between the velocity and @xmath101 , the accuracy of the solution will not be affected by the number of steps performed , like in an ordinary explicit integration scheme .
the present method possesses two very different regimes , depending on the ratio between the gas integration time - scale @xmath28 , and the dust stopping time @xmath109 .
if @xmath110 ( when an explicit integration could be used ) @xmath111 and equations [ vd_sph1 ] , [ vg_sph1 ] and [ ug_sph1 ] thus become @xmath112 @xmath113 @xmath114 in this limit , the stability condition [ cour1 ] becomes @xmath115 coinciding with the courant condition of an explicit integration as shown by @xcite .
therefore , equations [ vd_sph3 ] to [ ug_sph3 ] will be equivalent to an explicit sph two fluid method @xcite as long as very sharp density gradients are absent from the gas component . to quantify the errors produced by this approximation
, the behaviour of the algorithm in the presence of strong density gradients ( shocks ) will be tested in section 3.3 .
it is also interesting to check whether the method can reproduce the properties of the dust and gas mixture in the so - called _ terminal velocity approximation _
( see for example @xcite and references therein ) .
when dust and gas are strongly coupled , the dust reaches a constant relative velocity with respect to the gas , which is small but still finite .
such a relative velocity is proportional to the pressure gradient and the stopping time @xmath109 .
one can see this by using @xmath121w(r_{\rm kj } ) \end{aligned}\ ] ] where we have introduced @xmath122 to simplify the notation .
now , in order to reach the terminal velocity , pressure gradient and drag forces must balance each other , leading to @xmath123w(r_{\rm kj})\\ & = \sum_{\rm k}^{dust}\frac{m_{\rm k}}{\rho_{\rm k}}\left[-\frac{\nabla p_{\rm j}}{\rho_{\rm j}}+\frac{1}{t_{\rm s}}(\textbf{v}_{\rm kj}\cdot\hat{\textbf{r}}_{\rm kj})\hat{\textbf{r}}_{\rm kj } \right]w(r_{\rm kj})\end{aligned}\ ] ] which is simply the sph equivalent of @xmath124 the procedure followed in section 2.1 can be extended to the non - linear drag regimes as long as an approximate analytic solution can be found for the time evolution of the dust grains .
for example , in a full non - linear regime ( equations [ kst ] and [ cd3 ] ) , a procedure analogous to the one in section 2.1 can be followed . in such a regime ,
the equations of motion of the dust and gas components can always be expressed as @xmath125 @xmath126 where @xmath127 . in this case , the chosen equations for a pair of arbitrary dust and gas fluid elements located at points @xmath29 and @xmath30 are @xmath128 @xmath129 @xmath130 where @xmath131 by using sph discretization , equations [ vd_sol3 ] , [ vg_sol3 ] and [ ug_sol3 ] become @xmath132 @xmath133 @xmath134 where in this case , an additional sph summation is necessary to calculate @xmath135 , since it depends on the relative velocity of the components at the dust particle location .
to perform most of the numerical tests , the dragging algorithm was implemented in a purpose - built sph code .
the code included self - consistent @xmath136 and @xmath137 calculation , grad - h terms @xcite , and riemann solver - like artificial viscosity with thermal conductivity whenever neeeded @xcite . to perform the sedov test ,
the dragging algorithm was implemented into a well tested three - dimensional sph code . for the sake of conciseness
, the exact details of the sph code will not be presented here , but the interested reader is referred to @xcite . the dustybox test @xcite was performed in order to prove the capacity of the method to reproduce the expected asymptotic behaviour of the drag force .
a set of @xmath138 dust and @xmath138 gas particles with homogeneous densities @xmath7 and @xmath85 are placed in a periodic box with an initial velocity @xmath139 and @xmath140 . in order to construct the initial model , particles are evenly distributed along a cubic lattice with @xmath141 .
the dust lattice is shifted , with respect to the gas , by half of the gas particles separation in each direction .
the mass of each sph particle is equal to @xmath142 where @xmath143 is the computational domain volume , and @xmath144 the number of particles in each phase .
an isothermal equation of state is adopted ( @xmath145 ) , and in this case no artificial viscosity is used .
the physical units of the problem are chosen such that @xmath146 g @xmath147 , @xmath148 g @xmath147 , @xmath149 cm s@xmath150 .
these are the appropriate conditions for a dust particle at the mid - plane of a protoplanetary disk at 1 au from the central star ( see for example @xcite ) .
the computational domain comprises a total volume of 1 cubic au , and the total mass of gas inside the domain is @xmath151 g. the integration time - step @xmath28 is calculated by finding the minimum value , for all gas particles , of @xmath152 and @xmath153 where @xmath137 is the sph particle smoothing length and @xmath154 is the gas particle acceleration .
since the pressure gradient is zero , the exact solution for equation ( 4 ) is easy to find in this case @xmath155 allowing a direct comparison of the results obtained . in fig .
[ fig : test1a ] the time evolution of the velocity of a single sph dust particle is presented for several different values of the dust grain size @xmath84 .
left figure corresponds to a case with @xmath156 while right figure corresponds to a case with @xmath157 . as can be seen , irrespectively of the dust grain size , the correct terminal velocity between gas and dust components is reached in all cases .
whenever the gas integration time step becomes smaller than the dust stopping time , the algorithm is capable of following the velocity decay of the dust component towards its limiting velocity .
if the dust stopping time becomes much smaller than the gas integration time step , the algorithm simply tries to put both components on their barycentric velocity , right from the start .
if under any circumstance , resolving the velocity decay becomes essential , one can always artificially decrease the gas integration time - step by reducing the gas courant time condition ( equation [ courant0a ] ) by an arbitrary factor . in fig .
[ fig : test1b ] time evolution of the dust component velocity in the three - dimensional @xmath157 , @xmath158 mm case is shown , for different values of the gas integration time step . as can be seen ,
since the stopping time is much shorter than the gas integration time - step ( @xmath159 yrs ) , an artificially reduced gas integration time - step is needed to start resolving the dust component velocity decay . in order to better appreciate the precision of the adopted approximation , in fig .
[ fig : test1c ] , the relative errors in the @xmath157 test with @xmath158 m ( left plot ) and @xmath158 mm ( right plot ) are presented , for several different kernels .
we test the standard @xmath160 cubic spline kernel @xcite , the @xmath161 quintic spline kernel , and the double hump cubic kernel @xcite .
as can be seen , the correct terminal velocity is obtained , irrespectively of the used kernel , with very high precision ( the relative error between the numerical and analytical results is @xmath162 ) .
the greatest departures from the analytical solution are obtained during the velocity decay phase . in this phase , only the double hump kernel keeps errors under acceptable limits .
a similar result was also found by @xcite in their study .
one can also see from the right plot of fig .
[ fig : test1c ] , that the normalization condition ( equation [ norm ] ) , helps to reduce the errors further .
if the double hump kernel is used in conjunction with the normalization condition , the maximum relative error is @xmath163 in all the tested cases .
this is a very important results , since it shows that the trajectories of dust particles of arbitrary size can be accurately calculated , in a protoplanetary - like environment , without the need for excessive resolution .
the second test performed was the study of the propagation of a sound wave in a dust - gas mixture in a constant drag regime , also known as the dustywave test @xcite .
this can be done by setting the drag coefficients on a single grain , in the epstein regime , to be equal to @xmath164 as a consequence , the equations of motion for the dust and gas components become @xmath165 @xmath166 where we have introduced the drag coefficient per unit volume @xmath167 . this is a particularly interesting problem , because as in the previous case , it possesses an analytic solution @xcite . to set up the test , an ensemble of dust and gas particles with homogeneous densities @xmath85 and @xmath7 are evenly distributed over a periodic one - dimensional domain @xmath168
particle masses are assigned in the same way as in the previous section .
no artificial viscosity is used in this case in order to avoid introducing non - physical energy dissipation in the test .
the integration time - step @xmath28 is again calculated by finding the minimum value given by equations [ courant0a ] and [ courant0b ] .
an isothermal equation of state @xmath169 with @xmath170 is used in this case . in order to create the waves ,
a sinusoidal perturbation is introduced for each particle , both in position and velocity @xmath171 @xmath172 where @xmath173 is the original position of each particle , and @xmath174 , so that the velocity perturbation of the wave is @xmath175 .
the spatial perturbation @xmath176 will be different for every resolution , and is selected in each case so that the density perturbation of the wave is always @xmath177 . after introducing the perturbation , the propagation of the resulting sound wave within the domain is followed .
as previously mentioned , we used the predictor - corrector integrator for this test , since it gives better long - term energy and momentum conservation . + in fig .
[ fig : test2a ] , four different snapshots of the time evolution of the sound wave velocity in a @xmath178 , @xmath179 case with @xmath180 ( @xmath181 ) ( left panel ) , and @xmath182 ( @xmath183 ) ( right panel ) , are presented . in fig .
[ fig : test2b ] , four different snapshots of the time evolution of the sound wave velocity in a @xmath184 , @xmath185 case with @xmath186 ( @xmath187 ) ( left panel ) , and @xmath188 ( @xmath183 ) ( right panel ) , are presented . as can be seen ,
a good agreement with the analytical solutions has been obtained in both cases . in order to quantify the deviation from the analytical solution
several error norms are calculated ( see figure captions ) @xmath189 @xmath190^{1/2},\ ] ] @xmath191 where @xmath192 is the maximum value of the exact solution in the plotted region , @xmath193 is the analytical solution for the @xmath70-th point of the plot , and @xmath144 is the number of plotted points ( http://users.monash.edu.au/dprice/@xmath194splash/userguide/ ) .
as previously mentioned , one of the most important characteristics of dust and gas mixtures is that the local sound speed modification as a function of the dust / gas fraction ( equation [ cs ] ) . since the analytical solutions seen in fig .
[ fig : test2a ] and [ fig : test2b ] take into account such a modification , the test confirms the capacity of the algorithm to reproduce this feature of dust / gas mixtures .
the results of figs .
[ fig : test2a ] and [ fig : test2b ] also confirm that , whenever the amount of momentum transferred between the phases is small compared with the total momentum of the gas , an arbitrarily low number of dust particles can be used .
both in the lower drag case of fig .
[ fig : test2a ] , and in fig .
[ fig : test2b ] only 8 dust particles per wavelength are necessary to obtain reasonable results .
unfortunately , as can be seen in fig .
[ fig : test2c ] , a certain excess of energy dissipation by drag in the high drag regime becomes unavoidable .
this is not a new phenomena and was already found by @xcite in their simulations .
the sph two fluid scheme needs a minimum resolution ( @xmath195 ) in order to correctly resolve the small position and velocity differences between the dust and gas phases , otherwise overdissipation becomes unavoidable .
however , because our method treats only the gas as a fluid , and not the dust , this resolution criterion must only be satisfied by the gas component , not the dust . in fig .
[ fig : test2c ] , the effect of particle resolution is investigated in the high drag regime for two different dust - to - gas ratios . from top to bottom a total of 256 , 128 , 64 and 32 gas particles have been used . in the left panel of fig .
[ fig : test2c ] a @xmath156 case with @xmath183 is presented . in this case ,
equal numbers of gas and dust particles have been used . in the right panel of fig .
[ fig : test2c ] a @xmath157 case with @xmath183 is presented . in this case
only 8 dust particles have been used . as can be seen , and in complete agreement with the minimum resolution condition ,
only the ones with a minimum number of @xmath196 gas and dust particles is capable of matching the expected solution in the @xmath156 case .
however , in the @xmath157 case , overdissipation effects become much less dramatic , even with a very low gas and dust particle resolution .
this is important , since most astrophysical applications have low dust - to - gas ratios .
it is also important to note that the present method is less dissipative than the previous ones , because we need to perform many fewer integration time - steps , in order to evolve the simulation to a given time . in the present method ,
the interpolation error is only committed once per gas integration time step , in contrast with explicit or implicit methods where the error can be committed hundreds or thousands of times per gas integration time step .
in fact , in earlier versions of the present method an iterative procedure was tried in order to achieve a higher precision in the final relative velocities between the components , but instead it resulted in a degree of overdissipation comparable to the one using a standard integration method .
the next two tests are the shock tube test , and the sedov blast test @xcite .
they were both performed in order to test the behaviour of the scheme , in the presence of strong density and pressure gradients . since equation ( 16 )
will only be valid as long as no big changes in the density or the pressure gradient occur during the integration time - step , these experiments are critical to prove the usefulness of the method . in these experiments ,
thermal energy plays an essential role in the evolution of the system , so this time an adiabatic equation of state with @xmath197 and @xmath198 is used .
also , to correctly model the shocks , @xcite artificial viscosity is used in both cases with coefficients @xmath199 and @xmath200 for the thermal conduction parameter .
the signal velocities are @xmath201 and latexmath:[$\textbf{v}_{\rm sig , u}= time - step @xmath28 is calculated in both cases by finding the minimum value for all gas particles between @xmath203 and @xmath204 + where @xmath154 is the sph gas particle acceleration .
note that if additional forces affecting both phases ( like radiation pressure for example ) were introduced in the simulation , condition [ dt_a ] should also be taken into account for the dust particles . in these tests
, the more restrictive conditions will occur at the shock front . to set up the shock tube test , an ensemble of particles with @xmath205 , @xmath206 , @xmath207 , @xmath208 ,
are evenly distributed in a one - dimensional bounded domain @xmath168 . to model the density jump ,
a different number of particles is used at each side of the discontinuity . in particular , since it is a one - dimensional case @xmath209 where @xmath144 is the total number of particles .
particle masses are calculated as in the previous sections .
[ fig : test3 ] , presents the results for two weakly dragged cases in a constant drag regime .
the left panel represents a case with @xmath210 , while the right panel represents non linear drag regime ( equations [ kst ] and [ cd3 ] ) with @xmath211 . despite not having an analytical solution for the transient phase , in both figures
, the long - term analytical solution of the problem has been added ( dotted line ) as a guideline . in both cases , the obtained solution compares very favourably with the results previously obtained by @xcite through the use of explicit / implicit methods . in the left panel of fig .
[ fig : test3b ] , a strongly dragged case with @xmath212 is presented for two different resolutions in the non - linear regime . in this case , the analytical solution is known ( solid line ) , and as can be seen , it is well matched by the numerical results if enough resolution is used . in the right panel of fig . [
fig : test3b ] the same case is presented for a @xmath157 case with 569 gas and 50 dust particles .
as can be seen , despite the reduced dust resolution , the correct result is obtained and there is no evidence of overdissipation . .
such a calculation would be impossible with an explicit time step.,width=302 ] it is also very interesting to compare the results obtained using the semi - implicit method , with those obtained with an explicit integration scheme , for a very high drag regime with @xmath213 . as can be seen in fig .
[ fig : test3c ] , whereas excess dissipation in the explicit calculation gives an incorrect solution , our semi - implicit method avoids the problem .
as previously mentioned , the source of overdissipation in the semi - implicit method comes from the incapacity of the algorithm to estimate the local barycentric velocity , due to the lack of resolution .
however , and in contrast with an ordinary explicit method , if the semi - implicit method is used , the error in the barycentric velocity estimation is only committed once per gas integration time - step . on the contrary , if an explicit method is used , due to the courant condition of the drag interaction , the error in the estimation of the drag acceleration is committed a lot more times per gas integration time - step , leading to a very poor result .
this effect will also occur if a conventional implicit integration scheme is used ( e.g. * ? ? ? * ) . finally , in the uppermost panels of fig .
[ fig : test4a ] the result of a sedov blast test with @xmath184 , @xmath179 , and @xmath214 ( @xmath215 ) is presented . in this case
, the gas integration time - step is set by the courant time condition at the shock front . in the test
, a total of @xmath216 particles are evenly distributed in a three - dimensional cartesian grid with @xmath217 .
particle masses are calculated as in the previous sections .
the dust grid is displaced with respect to the gas one by half of the gas particle separation in each direction . to model the explosion a total thermal energy of @xmath218 code units
is distributed over the particles inside a certain radius ( @xmath219 ) . as a comparison ,
the sedov blast test performed with an ordinary explicit integration scheme , is also presented in the lower panels of fig .
[ fig : test4a ] . as can be seen , there are no significant differences .
.computational time increase factors as a function of the drag strength in the sedov test .
the computational time of each simulation with the explicit method is divided by the computational time of the semi - implicit method . in the semi - implicit method , since the time - step of the simulation is exclusively determined by the gas courant time condition , and the dust - to - gas ratio is small , no noticeable extra computational effort is needed if the drag strength is increased . [ cols="^,^",options="header " , ] again , in this case , no evidence of the resolution limitation has been found , due to the lower density ratio between the gas and dust components . since the typical gas - to - dust ratios in the interstellar medium are very similar to the ones used in the sedov test , we expect the method to be useful in realistic astrophysical simulations . additionally , we have used this test to compare the computational time of the method , with that of a traditional explicit integration . in table 1 , a comparison of the computational time for several different drag strengths , is presented for both cases . in each case , the computational time spent by each simulation is divided by the computational time of the semi - implicit method in the @xmath215 case . as can be seen , as the drag strength is increased , the explicit integrator computational time is increased , by several orders of magnitude , with respect to the computational time spent by semi - implicit method for @xmath215 . on the contrary ,
the computational time of the semi - implicit method remains stable , since the integration time - step is exclusively determined by the gas courant condition , and is independent of the drag strength . as can be seen in fig .
[ fig : test4b ] , arbitrarily large values for the drag coefficient can be used .
this value would be completely prohibitive in any two - fluid explicit integration method .
the final test performed was the mid - plane settling of dust particles in a one - dimensional vertical section of an isothermal disk with @xmath169 and @xmath170 . to set up the test , 100 gas and 100 dust particles , with @xmath184 and @xmath220 ,
are evenly distributed over a one - dimensional domain ( @xmath221 ) .
particle masses are assigned following the same procedure as in the previous sections .
an external acceleration @xmath222 is used to simulate the vertical component of the gravitational field from the star at the centre of the disk , where @xmath223 is the angular frequency .
( see appendices a and b for a detailed explanation about how to implement external forces in the integration scheme ) .
no boundaries have been used , and since one does not expect shocks to be important , the use of artificial viscosity is avoided .
the evolution equations of the system are given by @xmath224 @xmath225 in order for the system to relax , the gas particles are evolved under gravitational and pressure forces , until the hydrostatic equilibrium condition is attained . whenever hydrostatic equilibrium is reached , equations [ eu_disk1 ] and [ eu_disk2 ] can be solved @xmath226 @xmath227 giving the gas hydrostatic density profile : @xmath228 which is valid as long as @xmath229 . in fig .
[ fig : test5a ] , the initial gas density profile of the isothermal disk is presented . as can be seen
, the gas perfectly reproduces a gaussian density profile with @xmath230 , @xmath231 and @xmath232 . after gas relaxation , drag forces
are switched on , and evolution is started again . if the drag coefficient @xmath233 is high enough , dust particles reach a limiting velocity , given by the solution of equations [ eq1 ] and [ eq2 ] .
@xmath234 in fig .
[ fig : test5b ] , the dust component velocity as a function of @xmath235 is presented for two cases ( @xmath236 and @xmath237 ) . as can be seen ,
the correct limiting velocity of the dust component is reached in both cases . because @xmath238 , the momentum transferred between the dust and gas phases is rather small , and the gas component remains very close to the hydrostatic equilibrium . as can be seen in the @xmath237 case ( right plot )
, dust particles almost instantaneously reach its limiting velocity . on the contrary , if @xmath236 ( left plot ) , particles need more time to reach the limiting velocity and the transitory state can be seen for @xmath239 . in order to check whether the algorithm is capable of correctly reproducing such a transitory regime , the velocity as a function of @xmath235 for a single sph dust particle can be compared with the numerical solution of equations [ eu_disk1 ] and [ eu_disk2 ] . in fig .
[ fig : test5c ] the evolution of a single sph dust particle is plotted for three different @xmath240 values .
circles represent the velocity of the particle , for different time steps , as it falls down towards the disk mid - plane .
dashed lines represent the numerical solution of equations [ eu_disk1 ] and [ eu_disk2 ] for each case , while solid lines represent the limiting velocity for each case as given by equation [ v_lim ] .
if @xmath241 , the particle does not have time to reach the limiting velocity , and simply suffers velocity damping while it oscillates around the disk mid - plane .
as can be seen , a perfect agreement is achieved with the theoretical behaviour . if @xmath236 , the dust particle reaches the limiting velocity at @xmath242 , in perfect agreement with the numerical solution of equations [ eu_disk1 ] and [ eu_disk2 ] , and explaining the global velocity profile of the dust component ( as seen in fig .
[ fig : test5b ] ) . for @xmath237 , although the theoretical solution is approximately obtained , some oscillations of the particle velocity can be observed .
such oscillations occur due to the low number of gas particles present in the outermost parts of the disk . if a higher resolution simulation is performed ( 1000 gas particles ) , the oscillations disappear , and the velocity of the dust particle closely matches the analytical solution .
.,width=302 ] +
a new method has been proposed to avoid explicit integration of the time evolution equations of small dust grains in the two fluid sph approach . through the use of semi - analytic solutions for the decay of the gas and dust relative velocity ,
the present method has been able to reproduce all the features of the previous two fluid sph approach of @xcite , with the advantage of a considerable gain in computational time in strong drag regimes . due to its strictly dissipative nature
, the velocity changes induced by the drag force can be estimated without the need for explicit acceleration recalculations or iterative procedures , even when the stopping time becomes much shorter than the gas evolutionary time - scale .
the method is numerically stable , and always provides convergence towards the analytical solutions as the resolution is increased .
the method has also been capable of reproducing the correct behaviour of the drag force for all regimes . in the weak drag regime ,
the method is theoretically equivalent to a standard explicit integration , both in accuracy and computational efficiency , as long as strong gradients are not present in the immediate neighbourhood of dust particles . in the high drag regime ,
the method is capable of reproducing all the expected features of dust / gas mixtures .
the results obtained in the test cases are completely analogous to those found by @xcite through the use of standard explicit and implicit methods . in agreement with previous studies @xcite ,
a resolution limit has been found for the method in the dustywave experiment . for high drag regimes with dust - to - gas ratios of order unity , the resolution should exceed @xmath195 in order to avoid overdissipation .
however , in the shock tube experiment , our method avoids the effects of overdissipation , which until now has been considered to be one of the main limitations of the two - fluid sph approach .
furthermore , it has also been demonstrated that for low drag regimes , and even for high drag regimes with low dust - to - gas ratios , the number of dust particles present in the simulation becomes irrelevant , and the accuracy of the solution is only dependent on having sufficient gas resolution . since in the vast majority of astrophysical applications
the dust - to - gas ratio is expected to be rather low , only a good gas resolution will be necessary to avoid overdissipation .
however , special attention must be payed to this limitation , since it will be very difficult to completely avoid overdissipation in complex global simulations , especially if one expects abrupt changes in the dust - to - gas ratios .
we thank the anonymous referee , whose thoughtful and thorough report not only resulted in substantial improvements to the paper , but as a by - product also meant that we improved the numerical algorithm itself .
we also thank joe monaghan and daniel price for very useful discussions .
figures 2 to 9 have been created using splash @xcite , a sph visualization tool publicly available at http://users.monash.edu.au/@xmath194dprice/splash .
the calculations for this paper were performed on the dirac complexity machine , jointly funded by stfc and the large facilities capital fund of bis , and the university of exeter supercomputer , a dirac facility jointly funded by stfc , the large facilities capital fund of bis and the university of exeter .
this work was also supported by the stfc consolidated grant st / j001627/1 .
23 armitage p. j.,2010 , cambridge university press ayliffe b. a. , laibe g. , price d. j. , bate m. r , 2011 , mnras , 423 , 1450 bate m. , 1995 , phd thesis , univ .
cambridge fehlberg e. , low - order classical runge - kutta formulas with step size control and their application to some heat transfer problems , nasa technical report 315 fulk d.a . ,
quinn d.w . , 1996 , j. comput .
phys . , 17 ,
19 laibe g. , price d. j. , 2011 , mnras , 418 , 1491 laibe g. , price d. j. , 2012a , mnras , 420 , 2345 laibe g. , price d. j. , 2012b , mnras , 420 , 2365 laibe g. , price d. j. , 2014 , mnras , 440 , 2136 laibe g. , price d. j. , 2014 , mnras , 440 , 2147 marble , f. , 1970 , annual review of fluid mechanics , 2 , 397 monaghan j. j. , 1992 , annual review of astronomy and astrophysics , 30 , 543 monaghan j. j. , 1997 , journal of computational physics , 136 , 298 monaghan j. j. , 2002 , mnras , 335 , 843 monaghan j. j. , 1997 , journal of computational physics , 138 , 801 monaghan j. j. , kocharyan a. , 1995 , computer physics communications , 87 , 225 price d. j. , 2007 , publications of the astronomical society of australia , 24 , 159 price d. j. , 2008 , journal of computational physics , 227 , 10040 price d.j . , monaghan j. j. , 2004 , mnras , 348 , 139 randles p.w . , and libersky l.d . , 1996 , smoothed particle hydrodynamics some recent improvements and applications , computer methods in applied mechanics and engineering , 138 , 375 saffman p.g.,on the stability of laminar flow of a dusty gas , j. fluid mechanics , 13 , 120 serna a. , alimi j .- m . , chieze j .-
, 1995 , astrophysical journal , 461 , 884 sedov l. , 1959 , similarity and dimensional methods in mechanics , new york : academic springel v. , hernquist l. , 2002 , mnras , 333 , 649 weidenschilling , s.j . , 1977 , mnras , 180 , 57 whipple , f.l .
, 1972 , from plasma to planet , ed a. elvius , wiley , london , p.211 wtezstein m. , nelson a. f. , naab t. , burkert a. , 2009 , apjs , 184 , 298
in order to capture both the gas and dust evolution described by equations [ eu1 ] , [ eu2 ] , [ eu3 ] , [ cont1 ] , and [ cont2 ] , the dust - gas drag equations must be coupled with an explicit hydrodynamical integrator . combination with an integrator incorporating the gas pressure gradients is needed .
one chosen integrator is a second order runge - kutta - fehlberg scheme and the combined scheme can be summarized as follows @xmath243 for the first half of the time - step , and @xmath244 for the full time - step .
we have introduced the dust and gas components external accelerations @xmath245 and @xmath246 , in order to account for forces like gravity , physical viscosity or radiation pressure .
this method relies on the possibility of considering pressure and drag forces as separable interactions .
as the performed tests have shown , it seems to be a good assumption . +
another particularly useful property of the present method is its capacity to predict the correct modified sound speed of the dust / gas mixture , as a function of the dust / gas ratio . by substituting the pre - dragged quantities
@xmath247 and @xmath248 and the expression for the @xmath101 parameter into the @xmath249 and @xmath250 equations , one can convert the two - step method into an equivalent one - step method given by the set of equations for the first half of the time - step @xmath251 where @xmath252 , @xmath253 and @xmath254 for the full time - step , where we have defined @xmath255 as we can see , in this set of equations , dust can be no longer considered pressureless .
it suffers an acceleration due to pressure gradient , and possesses an effective density @xmath256 .
this result can be understood if one realizes that a purely dissipative force does not always lead to a velocity decrease .
because drag is a purely dissipative force , it will always lead to a decrease in the relative velocity between dust and gas components .
but sometimes , the only way to decrease such a relative velocity is to accelerate the dust component .
also , the effective densities @xmath256 and @xmath257 can be understood as the effective inertial response of the dust and gas components to the effective pressure terms .
the weaker the drag force is , the higher the pressure gradient must be to accelerate the dust component . if @xmath258 , @xmath259 , @xmath260 , and the equations for the change in velocity of the dust and gas components become @xmath261 for the first half time - step and @xmath262 for the full time - step .
the effective dust density term @xmath256 has become infinitely big , so the dust does not respond at all to the pressure gradient terms .
that is , gas and dust decouple , and gas evolves as a single component fluid with sound speed @xmath263 .
if , on the contrary , @xmath264 , @xmath265 , @xmath266 , and the equations for the change in velocity of the dust and gas components become this time @xmath267 for the first half time - step and @xmath268 for the full time - step . both effective density terms @xmath256 and @xmath257 become equal , so both dust and gas components evolve as a single component fluid , with the total mass of the mixture being advected
however , and since only the gas component can produce real pressure , they travel with a modified sound speed @xmath269 , exactly as predicted by theory ( see for example @xcite ) . as can be seen in equations a9 , both phases adopt in this regime the barycentric velocity in just one time - step , as it corresponds to a case where @xmath270 .
despite being particularly useful to visualize the behaviour of dust and gas mixtures , and to show that the effective sound speed of the mixture is the expected one in the strong drag regime , we still recommend using the two - step method given by equations a1 and a2 .
it is clearly technically easier to implement into a pre - existing sph code .
the second chosen integrator was a modification of the predictor - corrector scheme of @xcite and it can be summarized as follows @xmath271 for the predictor phase and @xmath272\delta t/2,\\ \tilde{\textbf{v}}^{t+1}_{\rm g } & = \textbf{v}^{t+1/2}_{\rm g } - \left[\left(\frac{\mathbf{\nabla } p}{\rho_{\rm g}}\right)_{t+1/2 } - \left(\frac{\mathbf{\nabla } p}{\rho_{\rm g}}\right)_{t}\right]\delta t/2 + \left[\textbf{a}_{\rm ext , g}^{t+1/2 } - \textbf{a}_{\rm ext , g}^{t}\right]\delta t/2 , \\
\textbf{v}^{t+1}_{\rm d } & = \tilde{\textbf{v}}^{t+1}_{\rm d } -\xi\left[(\tilde{\textbf{v}}^{t+1}_{\rm d}-\tilde{\textbf{v}}^{t+1}_{\rm g})-(\textbf{v}^{t+1/2}_{\rm d}-\textbf{v}^{t+1/2}_{\rm g})\right],\\ \textbf{v}^{t+1}_{\rm g } & = \tilde{\textbf{v}}^{t+1}_{\rm g } + \frac{\rho_{\rm d}}{\rho_{\rm g}}\xi\left[(\tilde{\textbf{v}}^{t+1}_{\rm d}-\tilde{\textbf{v}}^{t+1}_{\rm g})-(\textbf{v}^{t+1/2}_{\rm d}-\textbf{v}^{t+1/2}_{\rm g})\right],\\ \textbf{r}^{t+1}_{\rm d } & = \textbf{r}^{t+1/2}_{\rm d } + ( \textbf{v}^{t+1}_{\rm d}-\textbf{v}^{t+1/2}_{\rm d})\delta t/3,\\ \textbf{r}^{t+1}_{\rm g } & = \textbf{r}^{t+1/2}_{\rm g } + ( \textbf{v}^{t+1}_{\rm g}-\textbf{v}^{t+1/2}_{\rm g})\delta t/3 , \end{aligned}\ ] ] for the corrector phase . | a method to avoid the explicit time integration of small dust grains in the two fluid gas /dust smoothed particle hydrodynamics ( sph ) approach is proposed . by assuming a very simple exponential decay model for the relative velocity between the gas and dust components , all the effective characteristics of the drag force can be reproduced .
a series of tests has been performed to compare the accuracy of the method with analytical and explicit integration results .
we find that the method performs well on a wide range of tests , and can provide large speed ups over explicit integration when the dust stopping time is small .
we have also found that the method is much less dissipative than conventional explicit or implicit two - fluid sph approaches when modelling dusty shocks .
[ firstpage ] hydrodynamics - methods : numerical - planets and satellites : formation - protoplanetary discs - dust , extinction . |
since the recent rediscovery of low - density parity - check ( ldpc ) codes , a great deal of effort has been devoted to constructing ldpc codes that can work well with the belief - propagation iterative decoder .
the studies of long block - length ldpc codes are very much established .
the recent works of @xcite , @xcite have shown that , for long block - lengths , the best performing ldpc codes are irregular codes and these codes can outperform turbo codes of the same block - length and code - rate .
these long ldpc codes have degree distributions which are derived from differential evolution @xcite or gaussian approximation @xcite .
it can be shown that , using the concentration theorem @xcite , the performance of infinitely long ldpc codes of a given degree distribution can be characterised by the average performance of the ensemble based on cycle - free assumption .
this assumption , however , does not work for short and moderate block - length ldpc codes due to the inevitable existence of cycles in the underlying tanner graphs .
consequently , for a given degree distribution , the performance of short block - length ldpc codes varies considerably from the ensemble performance .
various methods exist for the construction of finite block - length irregular codes @xcite,@xcite,@xcite .
in addition to irregular ldpc codes , algebraic constructions exist and the resulting codes are regular and usually cyclic in nature .
some examples of algebraic ldpc codes are the euclidean and projective geometry codes @xcite .
it has been noticed by the authors that , in general , there is a performance association between the code minimum distance ( @xmath2 ) and decoding convergence .
the irregular ldpc codes converge very well with iterative decoding , but their @xmath2 are reasonably low . on the other hand , the algebraically constructed ldpc codes , which have high @xmath2 , tend not to converge well with the iterative decoder .
it is not surprising that algebraically constructed codes may outperform the irregular codes .
the latter have error - floor which is caused by the @xmath2 error - events . on the encoding side ,
the existence of algebraic structure in the codes is of benefit . rather than depending on the parity - check or generator matrices for encoding , as in the case of irregular codes
, a low - complexity encoder can be built for the algebraic ldpc codes .
one such example is the linear shift - register encoder for cyclic ldpc codes . assuming that @xmath3 and @xmath4 denote the codeword and information length respectively , algebraic codes that are cyclic offer another decoding advantage .
the iterative decoder has @xmath3 parity - check equations to iterate with instead of @xmath5 equations , as in the case of non - cyclic ldpc codes , and this leads to improved performance .
it has been shown that the performance of ldpc codes can be improved by going beyond the binary field @xcite , @xcite .
_ showed that , under iterative decoding , the non binary ldpc codes have better convergence properties than the binary codes @xcite
. they also demonstrated that a coding gain of @xmath6db is achieved by moving from @xmath7 to @xmath8 .
non binary ldpc codes in which each symbol takes values from gf(@xmath0 ) offer an attractive scheme for higher - order modulation .
the complexity of the symbol - based iterative decoder can be simplified as the extrinsic information from the component codes can be evaluated using the frequency domain dual codes decoder based on the fast - walsh - hadamard transform . based on the pioneering works of macwilliams @xcite,@xcite on the idempotents and the mattson - solomon polynomials , we present a generalised construction method for algebraic gf(@xmath0 ) codes that are applicable as ldpc codes .
the construction for binary codes using idempotents has been investigated by shibuya and sakaniwa @xcite , however , their investigation was mainly focused on half - rate codes . in this paper , we construct some higher code - rate non binary ldpc codes with good convergence properties .
we focus on the design of short block - length ldpc codes in view of the benefits for thin data - storage , wireless , command / control data reporting and watermarking applications .
one of the desirable features in any code construction technique is an effective method of determining the @xmath2 and this feature is not present in irregular code construction methods . with our idempotent - based method , the @xmath2 of a constructed code can be easily lower - bounded using the well - known bch bound .
the rest of the paper is organised as follows . in section [ sec : cyclotomic_coset ] , we briefly review the theory of the cyclotomic cosets , idempotents and mattson - solomon polynomials .
based on the theory , we devise a generalised construction algorithm and present an example in section [ sec : construction ] .
we also outline an efficient and systematic algorithm to search for algebraic ldpc codes in section [ sec : construction ] . in section [
sec : performance ] , we demonstrate the performance of the constructed codes by means of simulation and section [ sec : conclusion ] concludes this paper .
we briefly review the theory of cyclotomic cosets , idempotents and mattson - solomon polynomials to make this paper relatively self - contained .
let us first introduce some notations that will be used throughout this paper .
let @xmath9 and @xmath10 be positive integers with @xmath11 , so that @xmath12 is a subfield of @xmath13 .
let @xmath3 be a positive odd integer and @xmath13 be the splitting field for @xmath14 over @xmath12 , so that @xmath15 .
let @xmath16 , @xmath17 , @xmath18 be a generator for @xmath13 and @xmath19 be a generator for @xmath12 , where @xmath20 .
let @xmath21 be the set of polynomials of degree at most @xmath22 with coefficients in @xmath23 .
[ def : ms ] if @xmath24 , then the finite - field transform of @xmath25 is : @xmath26 where @xmath27 .
this transform is widely known as the mattson - solomon polynomial .
the inverse transform is : @xmath28 [ def : idempotent ] consider @xmath29 , @xmath30 is an idempotent if the property of @xmath31 is satisfied . in the case of @xmath32 ,
the property of @xmath33 is also satisfied .
an @xmath34 cyclic code @xmath35 can be described by the generator polynomial @xmath36 of degree @xmath5 and the parity - check polynomials @xmath37 of degree @xmath4 such that @xmath38 .
it is widely known that idempotents can be used to generate @xmath35 .
any @xmath12 cyclic code can also be described by a unique idempotent @xmath39 which consists of a sum of primitive idempotents .
this unique idempotent is known as the generating idempotent and , as the name implies , @xmath40 is a divisor of this idempotent , i.e. @xmath41 , where @xmath42 contains the repeated factors or non - factors of @xmath14 .
[ lemma : ms_idempotent ] if @xmath29 is an idempotent , @xmath43 .
* ch 8) ) since @xmath31 , from equation [ eqn : mattson_solomon ] , it follows that @xmath44 , @xmath45 for some integers @xmath46 and @xmath47 . clearly , @xmath48 implying that @xmath49 is a binary polynomial .
[ def : cyclotomic_cosets ] if @xmath50 is a positive integer , the binary cyclotomic coset of @xmath51 is : @xmath52 where we shall always assume that the subscript , @xmath50 , is the smallest element in the set @xmath53 , and @xmath54 is the smallest positive integer with the property that @xmath55
. if @xmath56 is the set consisting of the smallest elements of all possible cyclotomic cosets then @xmath57 [ lemma : nonbinary_idempotent ] let @xmath58 and let @xmath59 represents the @xmath60th element of @xmath53 .
let the polynomial @xmath61 be given by @xmath62 where @xmath63 is the number of elements in @xmath53 and @xmath64 is defined below . )
if @xmath32 , @xmath65 , if @xmath66 , @xmath64 is defined recursively as follows : @xmath67 the polynomial so defined , @xmath68 , is an idempotent . we term @xmath68 a _ cyclotomic _ idempotent .
[ def : parity_check_idempotent ] let @xmath69 and let @xmath70 be @xmath71 then ( refer to lemma [ lemma : nonbinary_idempotent ] ) @xmath72 is an idempotent and we call @xmath72 a _ parity - check _ idempotent . the parity - check idempotent @xmath72 can be used to describe the code @xmath35 , the parity check matrix being made up of the @xmath3 cyclic shifts of the polynomial @xmath73 .
if @xmath74 denotes the greatest common divisor of @xmath75 and @xmath76 then , in general , @xmath77 is much lower than @xmath78 denotes the weight of polynomial @xmath79 . ] .
based on this observation and the fact that @xmath72 contains all the roots of @xmath80 , we can construct cyclic codes that have a low - density parity - check matrix .
[ def : difference_enumerator ] let the polynomial @xmath81 .
the difference enumerator of @xmath79 , denoted as @xmath82 , is defined as follows : @xmath83 where we assume that @xmath82 is a modulo @xmath84 polynomial with real coefficients .
[ lemma : orthogonal ] let @xmath32 and let @xmath85 for @xmath86 denote the coefficients of @xmath87 . if @xmath88 , @xmath89 , the parity - check polynomial derived from @xmath72 is orthogonal on each position in the @xmath3-tuple .
consequently ( i ) the @xmath2 of the resulting @xmath35 is @xmath90 and ( ii ) the underlying tanner graph has girth of at least @xmath91 .
\(i ) ( cf .
* theorem 10.1 ) ) let a codeword @xmath92 and @xmath93 .
for each non zero bit position @xmath94 of @xmath95 where @xmath96 , there are @xmath77 parity - check equations orthogonal to position @xmath94 .
each of the parity - check equation must check another non zero bit @xmath97 @xmath98 so that the equation is satisfied .
clearly , @xmath99 must equal to @xmath90 and this is the minimum weight of all codewords.(ii ) the direct consequence of having orthogonal parity - check equation is the absence of cycles of length @xmath100 in the tanner graphs . it can be shown that there exists three integers @xmath75 , @xmath101 and @xmath102 , such that @xmath103 for @xmath104 .
if these three integers are associated to the variable nodes in the tanner graphs , a cycle of length @xmath91 can be formed between these variable nodes and some check nodes .
from lemma [ lemma : orthogonal ] we can deduce that @xmath72 is the parity - check polynomial for one - step majority - logic decodable codes if @xmath88 , @xmath105 or the parity - check polynomial for difference - set cyclic codes if @xmath106 , @xmath105 .
[ lemma : nonbinary_dmin_bound ] for the non binary gf(@xmath0 ) cyclic codes , the @xmath2 is bounded by : @xmath107 where @xmath108 denotes the maximum run of consecutive ones in @xmath109 taken cyclically modulo @xmath3 .
the lower - bound of the @xmath2 of a cyclic code , bch bound is determined from the number of consecutive roots of @xmath110 and from lemma [ lemma : ms_idempotent ] , it is equivalent to the run of consecutive ones in @xmath109 .
based on the mathematical theories outlined above , we devise an algorithm to construct gf(@xmath0 ) @xmath35 which are applicable for iterative decoding .
the construction algorithm can be described in the following procedures : 1 .
given the integers @xmath9 and @xmath3 , find the splitting field ( @xmath13 ) of @xmath14 over @xmath12 .
we can only construct @xmath12 cyclic codes of length @xmath3 if and only if the condition of @xmath11 is satisfied .
2 . generate the cyclotomic cosets modulo @xmath111 and denote it @xmath112 .
3 . derive a polynomial @xmath113 from @xmath112 .
let @xmath114 be the smallest positive integer such that @xmath115 .
the polynomial @xmath113 is the minimal polynomial of @xmath116 : @xmath117 construct all elements of gf(@xmath0 ) using @xmath113 as the primitive polynomial .
4 . let @xmath118 be the cyclotomic cosets modulo @xmath3 and @xmath56 be a set containing the smallest number in each coset of @xmath118 .
assume that there exists a non empty set @xmath119 and following definition [ def : parity_check_idempotent ] , construct the parity - check idempotent @xmath72 .
the coefficients of @xmath72 can be assigned following lemma [ lemma : nonbinary_idempotent ] .
generate the parity - check matrix of @xmath35 using the @xmath3 cyclic shifts of @xmath73 .
compute @xmath46 and @xmath47 , then take the mattson - solomon polynomial of @xmath72 to produce @xmath109 . obtain the code dimension and the lower - bound of the @xmath2 from @xmath109 . note that care should be taken to ensure that there is no common factor between @xmath3 and all of the exponents of @xmath72 , apart from unity , in order to avoid a degenerate code . [ cols="^,<,^,^,^,^",options="header " , ] ^@xmath120^the code minimum distance in binary level .
+ ^@xmath121^distance to the sphere - packing - bound constrained for binary transmission .
[ ex : code - construction ] let us assume that we want to construct a @xmath122 @xmath123 cyclic idempotent code .
the splitting field for @xmath124 over @xmath122 is @xmath122 and this implies that @xmath125 , @xmath126 and @xmath127 .
let @xmath118 and @xmath112 denote the cyclotomic cosets modulo @xmath3 and @xmath111 respectively .
@xmath128 and therefore the primitive polynomial @xmath113 has roots of @xmath129 , @xmath130 , i.e. @xmath131 . by letting @xmath132 , all of the elements of @xmath122 can be defined . if we let @xmath72 be the parity - check idempotent generated by the sum of the cyclotomic idempotents defined by @xmath53 where @xmath133 and @xmath134 , @xmath135 be @xmath136 , @xmath137 and @xmath137 respectively , @xmath138 and its mattson - solomon polynomial @xmath109 tells us that it is @xmath139 cyclic code with @xmath140 .
a systematic algorithm has been developed to sum up all combinations of the cyclotomic idempotents to search for all possible @xmath12 cyclic codes of a given length .
the search algorithm is targeted on the following key parameters : 1 .
sparseness of the resulting parity - check matrix .
since the parity - check matrix of @xmath35 is directly derived from @xmath72 which consists of the sum of the cyclotomic idempotents , we are only interested in low - weight cyclotomic idempotents .
let us define @xmath141 as the maximum @xmath77 then the search algorithm will only choose the cyclotomic idempotents whose sum has total weight less than or equal to @xmath141 .
high code - rate .
the number of roots of @xmath72 which are also roots of unity define the dimension of @xmath35 and let us define @xmath142 as the minimum information length of @xmath35 .
we are only interested in the sum of the cyclotomic idempotents whose mattson - solomon polynomial has at least @xmath142 zeros .
high @xmath2 .
let us define @xmath143 as the minimum value of the @xmath2 of @xmath35 .
the sum of the cyclotomic idempotents should have at least @xmath144 consecutive powers of @xmath19 which are roots of unity but not roots of @xmath72 .
the search algorithm can be relaxed to allow the existence of cycles of length 4 in the resulting parity - check matrix of @xmath35 .
the condition of cycles - of - length-4 is not crucial as we will show later that there are codes that have good convergence properties when decoded using iterative decoder .
clearly , by eliminating the cycles - of - length-4 constraint , we can construct more codes .
following definitions [ def : ms ] and [ def : parity_check_idempotent ] : @xmath145 and hence it is possible to maximise the run of the consecutive ones in @xmath109 if the coefficients of @xmath68 are aligned appropriately .
it is therefore important that all possible non zero values of @xmath134 , @xmath146 are included in the search in order to guarantee that we can obtain codes with the highest possible @xmath2 or at least to obtain a better estimate of the @xmath2 .
as an example of the performance attainable from an iterative decoder , computer simulations have been carried out for several @xmath12 cyclic ldpc codes .
we assume bpsk signalling and the iterative decoder used is the modified belief - propagation decoder which approximates the performance of a maximum - likelihood decoder @xcite,@xcite .
the frame - error - rate ( fer ) performance of the @xmath147 cyclic ldpc code is shown in fig .
[ fig : fer - gf64 - 21 - 15 ] and is compared with the sphere - packing - bound @xcite,@xcite for binary codes of length @xmath148 bits offset by the binary transmission loss .
we can see that the performance of the code is within @xmath149db away from this bound at @xmath150 fer .
the binary level minimum - distance of this @xmath139 cyclic ldpc code is @xmath151 .
frame error performance of the @xmath147 cyclic ldpc code , width=216 ] frame error performance of the @xmath152 cyclic ldpc code , width=216 ] frame error performance of the @xmath153 cyclic ldpc code , width=216 ] fig .
[ fig : fer - gf4 - 255 - 175 ] shows the fer curve of the @xmath152 cyclic ldpc code which is equivalent to @xmath154 binary code . at @xmath150 fer ,
the performance of this code is approximately @xmath155db away from the sphere - packing - bound of length @xmath156 bits .
while both of the codes mentioned above are free from cycles of length 4 , good convergence codes exist even if they have cycles of length 4 in the underlying tanner graph .
one such example is the @xmath153 cyclic code whose fer performance is shown in fig .
[ fig : fer - gf8 - 91 - 63 ] . at @xmath150 fer
, the code performs around @xmath157db away from the sphere - packing - bound of length @xmath158 bits .
the parameters of the codes in fig .
[ fig : fer - gf4 - 255 - 175 ] and [ fig : fer - gf8 - 91 - 63 ] are available in table [ tbl : code - parameter ] .
some other examples of the non binary @xmath12 cyclic ldpc codes with their parameters and distance from the sphere - packing - bound are also shown in table [ tbl : code - parameter ] .
an algebraic construction technique for gf(@xmath0 ) ( @xmath1 ) ldpc codes based on summing the cyclotomic idempotents to define the parity - check polynomial is able to produce a large number of cyclic codes .
the fact that we consider step - by - step summation of the cyclotomic idempotents , we are able to control the sparseness of the resulting parity - check matrix . the lower - bound of the @xmath2 and the dimension of the codes can be easily determined from the mattson - solomon polynomial of the resulting idempotent . for gf(@xmath159 ) case where the parity - check polynomials are orthogonal on each bit position , we can even determine the true @xmath2 of the codes regardless of the code length .
in fact , this special class of binary cyclic codes are the difference - set cyclic and the one - step majority - logic decodable codes which can be easily constructed using our method . for non - binary cases , if the constructed code has low @xmath2 , we can concatenate this code with an inner binary code to trade improvement in @xmath2 with loss in code - rate .
simulation results have shown that these codes can converge well under iterative decoding and their performance is very close to the sphere - packing - bound of binary codes for the same code length and rate . the excellent performance of these codes coupled with their low - complexity encoder offers an attractive coding scheme for applications that required short block - lengths such as thin data - storage , wireless , command / control data reporting and watermarking .
this research is partially funded by the uk overseas research students award scheme .
s. y. chung , g. d. forney , jr .
, t. j. richardson , and r. l. urbanke , `` on the design of low - density parity check codes within 0.0045 db of the shannon limit , '' _ ieee comm .
letters _ , vol . 3 , pp . 5860 , feb .
s. y. chung , t. j. richardson , and r. l. urbanke , `` analysis of sum - product decoding of low - density parity - check codes using a gaussian approximation , '' _ ieee trans .
inform . theory _
47 , pp . 657670 , feb .
2001 .
c. j. tjhai , e. papagiannis , m. tomlinson , m. a. ambroze , and m. z. ahmed , `` improved iterative decoder for ldpc codes with performance approximating to a maximum likelihood decoder . ''
uk patent application 0409306.8 , apr .
e. papagiannis , m. ambroze , and m. tomlinson , `` improved decoding of low - density parity - check codes with low , linearly increased added complexity . ''
submitted to 4^th^ iasted international conference on communication systems and networks , 2005 . | based on the ideas of cyclotomic cosets , idempotents and mattson - solomon polynomials , we present a new method to construct gf(@xmath0 ) , where @xmath1 cyclic low - density parity - check codes .
the construction method produces the dual code idempotent which is used to define the parity - check matrix of the low - density parity - check code . an interesting feature of this construction method is the ability to increment the code dimension by adding more idempotents and so steadily decrease the sparseness of the parity - check matrix .
we show that the constructed codes can achieve performance very close to the sphere - packing - bound constrained for binary transmission .
coding , idempotent , non binary ldpc , mattson - solomon polynomial |
in this paper , we address a number of outstanding classical questions on the geometry of embedded surfaces of constant mean curvature and more generally , laminations and foliations of @xmath3 and other three - manifolds , where the leaves of these laminations are surfaces with constant mean curvature ( possibly varying from leaf to leaf ) . in the foliation case , we call every such foliation a _ cmc foliation . _
the first of these classical problems is to classify codimension one cmc foliations of @xmath3 or @xmath4 ( with their standard metrics ) in the complement of a closed countable set @xmath6 .
the simplest such examples in @xmath3 are families of parallel planes or concentric spheres around a given point .
a slightly more complicated example appears when considering a family of pairwise disjoint planes and spheres as in figure [ figspheres ] , where the set @xmath6 consists of two points . by spheres and planes with two singularities .
, width=453 ] we solve this classification problem in complete generality ( see theorem [ thmspheres ] for a solution of an even more general problem , where the leaves of the `` foliation '' are allowed to intersect in a controlled manner . ] ) : [ thmspheresintrod ] suppose @xmath7 is a cmc foliation of @xmath8 with a closed countable set @xmath9 of singularities is a foliation of @xmath10 , and it does not extend to a cmc foliation of @xmath11 for any proper closed subset @xmath12 .
] then , all leaves of @xmath7 are contained in planes and round spheres .
furthermore if @xmath13 is empty , then @xmath14 is a foliation by planes . in the case of the unit three - sphere @xmath15 with its standard metric of constant sectional curvature
, we obtain a similar result : we remark that the special case of theorem [ thmspheresintrod ] where the singular set @xmath6 of @xmath7 is empty is a classical result of meeks @xcite .
also , we note that in the statement of the above theorem , we made no assumption on the regularity of the foliation @xmath7 . however , the proofs in this paper require that @xmath7 has bounded second fundamental form on compact sets of @xmath16 or @xmath4 minus the singular set @xmath6 .
this bounded curvature assumption always holds for a topological cmc foliation for this concept .
] by recent work of meeks and tinaglia @xcite on curvature estimates for embedded , non - zero constant mean curvature disks and a related 1-sided curvature estimate for embedded surfaces of any constant mean curvature ; in the case that all of the leaves of the lamination of a three - manifold are minimal , this 1-sided curvature estimate was given earlier by colding and minicozzi @xcite .
consider a foliation @xmath7 of a riemannian three - manifold @xmath2 with leaves having constant absolute mean curvature .
after possibly passing to a four - sheeted cover , we can assume @xmath2 is oriented and that all leaves of @xmath7 are oriented consistently , in the sense that there exists a continuous , nowhere zero vector field in @xmath2 which is transversal to the leaves of @xmath7 .
in this situation , the mean curvature function of the leaves of @xmath7 is well - defined and so @xmath17 is a cmc foliation .
therefore , when analyzing the structure of such a cmc foliation @xmath7 , it is natural to consider for each @xmath1 , the subset @xmath18 of @xmath7 of those leaves that have mean curvature @xmath0 .
such a subset @xmath18 is closed since the mean curvature function is continuous on @xmath7 ; @xmath18 is an example of an @xmath0-lamination .
a cornerstone in proving the above classification results is to analyze the structure of an @xmath0-lamination @xmath19 ( or more generally , a weak @xmath0-lamination , see definition [ definition ] ) of a punctured ball in a riemannian three - manifold , in a small neighborhood of the puncture .
this local problem can be viewed as a desingularization problem . in our previous paper
@xcite , we characterized the removability of an isolated singularity of a minimal lamination @xmath19 of a punctured ball in terms of the growth of the norm of the second fundamental form of the leaves of @xmath19 when extrinsically approaching the puncture .
we will extend this local removable singularity theorem to the case of a weak @xmath0-lamination , see theorem [ tt2 ] below .
next we set some specific notation to be used throughout the paper , which is necessary to state the next local removable singularity theorem . given a riemannian three - manifold @xmath2 and a point @xmath20 , we denote by @xmath21 the distance function in @xmath2 and by @xmath22 , @xmath23 the open metric ball of center @xmath24 and radius @xmath25 , its closure and boundary sphere , respectively .
in the case @xmath16 , we use the notation @xmath26 , @xmath27 and @xmath28 , @xmath29 , where @xmath30 .
furthermore , @xmath31 stands for the distance function to the origin @xmath32 . for a codimension - one lamination @xmath33 of @xmath2 and a leaf @xmath34 of @xmath19 ,
we denote by @xmath35 the norm of the second fundamental form of @xmath34 .
since leaves of @xmath19 do not intersect , it makes sense to consider the norm of the second fundamental form as a function defined on the union of the leaves of @xmath19 , which we denote by @xmath36 . in the case of a weak @xmath0-lamination @xmath37 of @xmath2 , given @xmath38 there
exist at most two leaves of @xmath19 passing through @xmath24 ( by the maximum principle for constant mean curvature surfaces ) , and thus , we define @xmath39 to be the function on @xmath19 that assigns to each @xmath38 the maximum of the norms of the second fundamental forms of leaves @xmath34 of @xmath19 such that @xmath40 .
observe that @xmath39 is not necessarily continuous .
[ local removable singularity theorem ] [ tt2 ] a weak @xmath0-lamination @xmath19 of a punctured ball @xmath41 of a riemannian three - manifold @xmath2 extends to a weak @xmath0-lamination of @xmath42 if and only if there exists a positive constant @xmath43 such that @xmath44 in some subball . in particular under this hypothesis , 1 .
the second fundamental form of @xmath19 is bounded in a neighborhood of @xmath24 .
2 . if @xmath19 consists of a single leaf @xmath45 which is a properly immersed weak @xmath0-surface is allowed to intersect itself only in the way that the leaves of a weak @xmath0-lamination might intersect , see definition [ definition ] . ] with @xmath46 , then @xmath47 extends smoothly across @xmath24 .
we remark that in the case @xmath48 , a weak @xmath0-lamination is just a minimal lamination , see the first paragraph just after definition [ definition ] . in this way
, theorem [ tt2 ] generalizes the minimal case of the local removable singularity theorem proven in @xcite . besides the above local removable singularity theorem , a second key ingredient is needed in the proof of theorem [ thmspheresintrod ] : a universal scale - invariant curvature estimate valid for any cmc foliation of a compact riemannian three - manifold with boundary , solely in terms of an upper bound for its sectional curvature .
the next result is inspired by previous curvature estimates by schoen @xcite and ros @xcite for compact stable minimal surfaces with boundary , and by rosenberg , souam and toubiana @xcite for stable constant mean curvature surfaces .
[ thm5.7 ] there exists a positive constant @xmath49 such that the following statement holds .
given @xmath50 , a compact riemannian three - manifold @xmath2 with boundary whose absolute sectional curvature is at most @xmath51 , a weak @xmath52 foliation @xmath7 of @xmath2 and a point @xmath53 , we have @xmath54 where @xmath55 is the function that assigns to each @xmath20 the supremum of the norms of the second fundamental forms of leaves of @xmath7 passing through @xmath24 .
the above curvature estimate is an essential tool for analyzing the structure of a weak cmc foliation of a small geodesic riemannian three - ball punctured at its center . among other things
we prove that if the mean curvatures of the leaves of such a weak cmc foliation are bounded in a neighborhood of the puncture , then the weak cmc foliation extends across the puncture to a weak cmc foliation of the ball .
a global application of the above theorem is that a compact , orientable riemannian three - manifold not diffeomorphic to the three - sphere @xmath4 or to the real projective three - space @xmath56 does not admit any weak cmc foliation with a non - empty countable closed set of singularities ; see @xcite for this and other related results . in a different direction , it is natural to ask whether every closed , orientable three - manifold admits a riemannian metric together with a smooth cmc foliation .
the existence of such foliations follows from the next main theorem in @xcite and the facts that every closed three - manifold has a differentiable structure and the euler characteristic of any closed manifold of odd dimension is always zero ; the proof of this theorem relies heavily on the seminal works of thurston @xcite on the existence of smooth codimension - one foliations of smooth closed @xmath57-manifolds and of sullivan @xcite , which explains when such foliations are minimal with respect to some riemannian metric .
[ main ] a smooth closed orientable @xmath57-manifold @xmath58 admits a smooth cmc foliation @xmath17 for some riemannian metric if and only if its euler characteristic is zero .
furthermore , @xmath17 can be chosen so that the mean curvature function of its leaves changes sign .
the paper is organized as follows . in section [ sec3 ]
we extend to @xmath0-surfaces the stability lemma proved in @xcite for the minimal case . in section [ sec10 ] we discuss some regularity aspects of @xmath0 and cmc laminations , and define weak @xmath0 and cmc laminations .
section [ sec4 ] contains the proof of theorem [ thm5.7 ] .
section [ secprooftt2 ] is devoted to prove the local removable singularity theorem [ tt2 ] based on the previously proven minimal case ( theorem 1.1 in @xcite ) . in section [ sectspheres ]
we demonstrate theorem [ thmspheresintrod ]
. finally , in section [ sec : localstructure ] we apply a rescaling argument and theorem [ thmspheresintrod ] to obtain theorem [ corol6.6 ] which describes the structure of any singular cmc foliation of a riemannian three - manifold in a small neighborhood of any of its isolated singular points .
in @xcite we extended the classical characterization of planes as the only orientable , complete , stable minimal surfaces in @xmath3 ( do carmo and peng @xcite , fischer - colbrie and schoen @xcite , pogorelov @xcite ) to the case that completeness is only required outside a point , and called this result the _ stability lemma_. this result was a crucial tool in the proof of the minimal case of the local removable singularity theorem ( theorem 1.1 in @xcite ) .
next we extend the stability lemma to the case of constant mean curvature surfaces , although this extension will not be directly used in the proof of theorem [ tt2 ] : rather than this , the proof of theorem [ tt2 ] will be based on the validity of the minimal case of the local removable singularity theorem proven in @xcite , which in turn only uses the minimal case of the stability lemma .
we remark that the notion of stability used in this paper is that the first eigenvalue of the stability operator for compactly supported variations is non - negative , in contrast with the usual stability notion related to the isoperimetric problem , where only compactly supported variations that preserve infinitesimally the volume are considered .
[ defcompl ] an immersed surface @xmath59 is _ complete outside the origin , _ if every divergent path in @xmath47 of finite length has as limit point the origin .
let @xmath60 be an isometric immersion of an orientable surface @xmath47 .
the mean curvature function @xmath0 of @xmath61 is constant with value @xmath62 ( in short , @xmath63 is an immersed @xmath64-surface ) if and only if for all smooth compact subdomains of @xmath47 , @xmath61 restricted to the subdomain is a critical point of the functional @xmath65 . given an @xmath0-surface @xmath63 and a function @xmath66 , the second variation of @xmath67 for any compactly supported variation of @xmath61 whose normal component of the variational field is @xmath68 , is well - known to be @xmath69 where @xmath70 is the _ jacobi operator _ on @xmath47 (
here @xmath71 is the norm of the second fundamental form and @xmath72 is the gaussian curvature ) .
the immersion is said to be _ stable _ if @xmath73 is a non - negative operator on @xmath47 , i.e. , @xmath74 for every @xmath66 .
more generally , a schrdinger operator @xmath75 with @xmath76 is called _ non - negative _ if @xmath77 [ stability lemma for @xmath0-surfaces ] [ lema1 ] let @xmath59 be a stable , immersed constant mean curvature ( orientable if minimal ) surface , which is complete outside the origin .
then , the closure @xmath78 of @xmath47 is a plane . if @xmath79 , then @xmath47 is complete and so , it is a plane @xcite .
assume now that @xmath80 .
consider the metric @xmath81 on @xmath47 , where @xmath82 is the metric induced by the usual inner product @xmath83 of @xmath3 and @xmath84 is the distance to the origin in @xmath3 .
note that if @xmath47 were a plane through @xmath32 , then @xmath85 would be the metric on @xmath86 of an infinite cylinder of radius 1 with ends at @xmath32 and at infinity .
since @xmath87 with @xmath88 , is isometric to @xmath89 , then @xmath90 is complete .
the laplacians and gauss curvatures of @xmath91 are related by the equations @xmath92 and @xmath93 .
thus , the stability of @xmath94 implies that the following operator is non - negative on @xmath47 : @xmath95 where @xmath96 . since @xmath97 where @xmath98 is the unitary normal vector field to @xmath47 ( with respect to @xmath82 ) for which @xmath0 is the mean curvature , then @xmath99 viewing the right - hand - side of ( [ eq : parabolah ] ) as a quadratic polynomial in the variable @xmath0 , its discriminant is @xmath100 .
since the coefficient of @xmath101 on the right - hand - side of ( [ eq : parabolah ] ) is @xmath102 , we deduce that @xmath103 on @xmath47 . applying theorem 2.9 in @xcite to the operator @xmath104 on @xmath105
, we deduce that @xmath105 has at most quadratic area growth .
this last property implies that that every bounded solution of the equation @xmath106 has constant sign on @xmath47 ( see theorem 1 in manzano , prez and rodrguez @xcite ) .
arguing by contradiction , suppose that @xmath94 is not flat .
then , there exists a bounded jacobi function @xmath107 on @xmath94 which changes sign ( simply take a point @xmath108 and choose @xmath107 as @xmath109 where @xmath110 is a non - zero tangent vector to @xmath47 at @xmath24 ) .
then clearly @xmath107 satisfies @xmath106 on @xmath47 .
this contradiction proves the lemma .
in this section we start by recalling the classical notion of lamination and discuss some previous results on regularity of these objects when the leaves have constant mean curvature
. then we will enlarge the class to admit _ weak _ laminations by allowing certain tangential intersections between the leaves .
these weak laminations and foliations will be studied in subsequent sections .
[ deflamination ] a _ codimension - one lamination _ of a riemannian three - manifold @xmath2 is the union of a collection of pairwise disjoint , connected , injectively immersed surfaces , with a certain local product structure .
more precisely , it is a pair @xmath111 satisfying : 1 . @xmath112 is a closed subset of @xmath2 ; 2 .
@xmath113 is an atlas of coordinate charts of @xmath2 ( here @xmath114 is the open unit disk in @xmath115 , @xmath116 is the open unit interval in @xmath117 and @xmath118 is an open subset of @xmath2 ) ; note that although @xmath2 is assumed to be smooth , we only require that the regularity of the atlas ( i.e. , that of its change of coordinates ) is of class @xmath119 ; in other words , @xmath120 is an atlas with respect to the topological structure of @xmath2 .
3 . for each @xmath121
, there exists a closed subset @xmath122 of @xmath116 such that @xmath123
. we will simply denote laminations by @xmath112 , omitting the charts @xmath124 in @xmath125 unless explicitly necessary .
a lamination @xmath112 is said to be a _ foliation of @xmath2 _ if @xmath126 .
every lamination @xmath112 decomposes into a collection of disjoint , connected topological surfaces ( locally given by @xmath127 , @xmath128 , with the notation above ) , called the _ leaves _ of @xmath112 .
note that if @xmath129 is any collection of leaves of @xmath19 , then the closure of the union of these leaves has the structure of a lamination within @xmath19 , which we will call a _ sublamination .
_ a codimension - one lamination @xmath19 of @xmath2 is said to be a _
cmc lamination _ if each of its leaves is smooth and has constant mean curvature ( possibly varying from leaf to leaf ) .
given @xmath1 , an _
@xmath0-lamination _ of @xmath2 is a cmc lamination all whose leaves have the same mean curvature @xmath0 . if @xmath48 , the @xmath0-lamination is called a
_ minimal lamination .
_ since the leaves of a lamination @xmath19 are disjoint , it makes sense to consider the second fundamental form @xmath130 as being defined on the union of the leaves .
a natural question to ask is whether or not the norm @xmath39 of the second fundamental form of a ( minimal , @xmath0- or cmc ) lamination @xmath19 in a riemannian three - manifold is locally bounded . concerning this question ,
we make the following observations . o.1 .
: : if @xmath19 is a minimal lamination , then the 1-sided curvature estimates for minimal disks by colding and minicozzi ( * ? ?
* ; * ? ? ?
* ) imply that @xmath39 is locally bounded ( to prove this , one only needs to deal with limit leaves , where the 1-sided curvature estimates apply ) . o.2 .
: : as a consequence of recent work of meeks and tinaglia @xcite on curvature estimates for embedded disks of positive constant mean curvature and a related 1-sided curvature estimate , a cmc lamination @xmath19 of a riemannian three - manifold @xmath2 has @xmath39 locally bounded .
given a sequence of cmc laminations @xmath131 of a riemannian three - manifold @xmath2 with uniformly bounded second fundamental form on compact subsets of @xmath2 , a simple application of the uniform graph lemma for surfaces with constant mean curvature ( see colding and minicozzi @xcite or prez and ros @xcite from where this well - known result can be deduced ) and of the arzel - ascoli theorem , gives that there exists a limit object of ( a subsequence of ) the @xmath131 , which in general fails to be a cmc lamination since two `` leaves '' of this limit object could intersect tangentially with mean curvature vectors pointing in opposite directions ; nevertheless , if @xmath131 is a _ minimal _
lamination for every @xmath57 , then the maximum principle ensures that the limit object is indeed a minimal lamination , see proposition b1 in @xcite for a proof .
still , in the general case of cmc laminations , such a limit object always satisfies the conditions in the next definition .
[ definition ] a ( codimension - one ) _ weak cmc lamination _
@xmath19 of a riemannian three - manifold @xmath2 is a collection @xmath132 of ( not necessarily injectively ) immersed constant mean curvature surfaces called the _ leaves _ of @xmath19 , satisfying the following three properties . 1 .
@xmath133 is a closed subset of @xmath2 .
if @xmath20 is a point where either two leaves of @xmath19 intersect or a leaf of @xmath19 intersects itself , then each of these local surfaces at @xmath24 lies at one side of the other ( this can not happen if both of the intersecting leaves have the same signed mean curvature as graphs over their common tangent space at @xmath24 , by the maximum principle ) .
the function @xmath134 given by latexmath:[\[\label{eq : sigma } is uniformly bounded on compact sets of @xmath2 . furthermore : * if @xmath136 , then we call @xmath137 a _ weak cmc foliation _ of @xmath2 . *
if the leaves of @xmath19 have the same constant mean curvature @xmath0 , then we call @xmath19 a
_ weak @xmath0-lamination _ of @xmath2 ( or @xmath0-foliation , if additionally @xmath136 ) . _
_ : : 1 . : : the function @xmath39 defined in ( [ eq : sigma ] ) for a cmc lamination is not necessarily continuous , as demonstrated by the following example : consider the union in @xmath3 of @xmath138 and the sphere @xmath139 where @xmath140 .
also note that this example can be modified to create a weak cmc foliation @xmath7 of @xmath3 minus a point with non - continuous related function @xmath141 : add to the previous example @xmath19 all planes @xmath142 , and foliate the open set @xmath143 by the spheres @xmath144 with @xmath145 where @xmath146 , together with the spheres @xmath147 , @xmath148 . in either case , @xmath39 or @xmath141 is not continuous at the origin .
: : a weak @xmath0-lamination for @xmath48 is a minimal lamination in the sense of definition [ deflamination ] .
3 . : : as a consequence of observation o.2 above , every cmc lamination ( resp .
cmc foliation ) of @xmath2 is a weak cmc lamination ( resp .
weak cmc foliation ) .
the following proposition follows immediately from the definition of a weak @xmath0-lamination and the maximum principle for @xmath0-surfaces .
[ prop10.2 ] any weak @xmath0-lamination @xmath19 of a three - manifold @xmath2 has a local @xmath0-lamination structure on the mean convex side of each leaf .
more precisely , given a leaf @xmath149 of @xmath19 and given a small disk @xmath150 , there exists an @xmath151 such that if @xmath152 denotes the normal coordinates for @xmath153 ( here @xmath154 is the exponential map of @xmath2 and @xmath98 is the unit normal vector field to @xmath149 pointing to the mean convex side of @xmath149 ) , then the exponential map @xmath154 is an injective submersion in @xmath155 , and the inverse image @xmath156 is an @xmath0-lamination of @xmath157 ) in the pulled back metric , see figure [ fighlamin ] .
-lamination with @xmath158 can intersect each other or themselves , but only tangentially with opposite mean curvature vectors .
nevertheless , on the mean convex side of these locally intersecting leaves , there is a lamination structure .
, width=192,height=151 ] [ def - limset ] let @xmath47 be a complete , embedded surface in a riemannian three - manifold @xmath2 .
a point @xmath20 is a _
limit point _ of @xmath47 if there exists a sequence @xmath159 which diverges to infinity in @xmath47 with respect to the intrinsic riemannian topology on @xmath47 but converges in @xmath2 to @xmath24 as @xmath160 .
let lim@xmath161 denote the set of all limit points of @xmath47 in @xmath2 ; we call this set the _ limit set of @xmath47_. in particular , lim@xmath161 is a closed subset of @xmath2 and @xmath162 , where @xmath78 denotes the closure of @xmath47 . the above notion of limit point
can be extended to the case of a lamination @xmath19 of @xmath2 as follows : a point @xmath163 is a _
limit point _
if there exists a coordinate chart @xmath164 as in definition [ deflamination ] such that @xmath165 and @xmath166 with @xmath167 belonging to the accumulation set of @xmath122 . the notion of limit point can be also extended to the case of a weak @xmath0-lamination of @xmath2 , by using that such an weak @xmath0-lamination has a local lamination structure at the mean convex side of any of its points , given by proposition [ prop10.2 ] .
it is easy to show that if @xmath24 is a limit point of a lamination @xmath19 ( resp . of a weak @xmath0-lamination ) , then the leaf @xmath34 of @xmath19 passing through @xmath24 consists entirely of limit points of @xmath19 ; in this case , @xmath34 is called a _
limit leaf _ of @xmath19 .
we now prove the universal curvature estimates for weak cmc foliations stated in theorem [ thm5.7 ] .
let @xmath2 be a compact riemannian three - manifold possibly with boundary , whose absolute sectional curvature is at most @xmath50 .
let @xmath7 be a weak cmc foliation of @xmath2 and @xmath53 . recall that @xmath141 is the function defined in ( [ eq : sigma ] ) .
note that @xmath168 is invariant under rescaling of the ambient metric .
this invariance implies that we can fix @xmath169 and prove that @xmath168 is bounded independently of the compact riemannian three - manifold @xmath2 with boundary whose absolute sectional curvature is at most @xmath51 and independently of the weak cmc foliation @xmath7 of @xmath2 .
we fix @xmath170 .
arguing by contradiction , assume that there exists a sequence of weak cmc foliations @xmath171 of compact three - manifolds @xmath172 with boundary , such that the absolute sectional curvature of @xmath172 is at most @xmath51 , and there exists a sequence of points @xmath173 in leaves @xmath174 of @xmath171 with @xmath175 for all @xmath57 , where @xmath176 .
after replacing @xmath177 by @xmath178 and @xmath172 by the closed metric ball @xmath179 , we can assume @xmath180 .
since the sectional curvature of @xmath172 is at most @xmath51 , a standard comparison argument for zeros of jacobi fields on geodesics gives that the exponential map @xmath181 is a local diffeomorphism from the ball of radius @xmath177 centered at the origin in the tangent space to @xmath172 at @xmath173 ( endowed with the ambient metric at @xmath173 ) onto the metric ball @xmath182 .
after lifting the ambient metric of @xmath172 to @xmath183 , we can consider the above map to be a local isometry ( note that the metric on @xmath183 depends on @xmath57 ) .
we can also consider the weak cmc foliation @xmath171 to be a weak cmc foliation of @xmath183 with the pulled back metric .
next consider the homothetic expansion of the metric in @xmath183 centered at the origin with ratio @xmath184 .
after this new normalization , we have the following properties : ( p1 ) : : @xmath183 becomes a metric ball of radius @xmath185 , which we will denote by @xmath186 , since its riemannian metric still depends on @xmath57 .
( p2 ) : : the absolute sectional curvature of @xmath186 is less than or equal to @xmath51 .
( p3 ) : : we have a related cmc foliation @xmath187 on @xmath186 so that the second fundamental form of some leaf @xmath188 of @xmath187 passing through @xmath32 satisfies @xmath189 for all @xmath190 . by lemma 2.2 in @xcite , the injectivity radius function @xmath191 of @xmath186 satisfies @xmath192 for all points @xmath193 . by theorem 2.1 in @xcite , given @xmath194 there exists @xmath195 ( only depending on @xmath51 but not on @xmath57 ) so that we can pick harmonic coordinates in the metric ball @xmath196 centered at any point @xmath197 of radius @xmath198 . for the notion of harmonic coordinates , see @xcite ; the only property we will use here about these harmonic coordinates is that the metric tensor on @xmath196 ( which depends on @xmath57 ) is @xmath199-controlled .
fix @xmath190 and let @xmath200 be a supremum of the function @xmath201 , @xmath202 and @xmath203 denotes extrinsic distance in @xmath204 .
we observe that the following properties hold : * the supremum in ( [ eq : fn ] ) exists since the second fundamental form of all leaves of @xmath187 is uniformly bounded in @xmath204 by definition of weak cmc foliation .
* @xmath205 may not be continuous ( as @xmath206 might fail to be continuous ) but still it is bounded , and it vanishes at @xmath207 . a similar argument as in the observation
above shows that the supremum of @xmath205 is attained at an interior point of @xmath204 . *
the value of @xmath205 at @xmath32 tends to @xmath208 as @xmath209 .
let @xmath210 and let @xmath211 .
after rescaling the metric of the ball @xmath212 of radius @xmath213 centered at @xmath214 by the factor @xmath215 , we have associated weak cmc foliations @xmath216 of @xmath217 such that the norm of the second fundamental form of @xmath216 is at most @xmath218 everywhere and is equal to 1 at the center @xmath214 of this ball . using the techniques described in @xcite
, it follows that a subsequence of the @xmath216 converges to a weak cmc foliation @xmath219 of @xmath8 ; next we sketch an explanation of these techniques : first , one uses the above harmonic coordinates to show that the coordinatized riemannian manifolds @xmath220 converge uniformly on compact subsets of @xmath3 in the @xmath221-euclidean topology to @xmath3 endowed with its euclidean metric .
then one shows that the leaves in @xmath216 can be locally written as graphs of functions defined over euclidean disks of uniformly controlled size ( this property follows from the uniform graph lemma for surfaces with constant mean curvature , note that for this we need an uniform bound on the second fundamental form of the leaves of @xmath216 , which is obtained by the blow - up process ) .
another consequence of the uniform graph lemma is that one obtains uniform local @xmath222-bounds for the graphing functions .
the next step consists of using that the graphing functions satisfy the corresponding mean curvature equation , which is an elliptic pde of second order whose coefficients have uniform @xmath223-estimates ( this follows from the @xmath222-bounds for the graphing functions and the @xmath199-control of the ambient metric on @xmath220 , see lemma 2.4 in @xcite ) , together with schauder estimates to conclude local uniform @xmath224-bounds for the graphing functions of the leaves of @xmath216 .
finally , these local uniform @xmath224-bounds for the graphing functions allow us to use the arzel - ascoli theorem to obtain convergence ( after extracting a subsequence ) in the @xmath222-topology to limit graphing functions of class @xmath222 . using that the convergence is @xmath222 one can pass to the limit the mean curvature equations satisfied by the graphing functions in the sequence , thereby producing a ( local ) limit weak cmc foliation of an open set of @xmath3 .
finally , a diagonal argument produces a global limit weak cmc foliation @xmath219 of @xmath8 of a subsequence of the @xmath216 . for details , see @xcite .
the above process insures that the limit weak cmc foliation @xmath219 of @xmath8 satisfies the following properties : 1 .
the second fundamental form of the leaves of @xmath219 is bounded in absolute value by @xmath225 ( in particular , there is a bound on the mean curvature of every leaf of @xmath219 ) , and there is a leaf @xmath226 of @xmath219 passing through the origin which is not flat .
@xmath219 is not a minimal foliation ( otherwise @xmath219 would consist entirely of planes , contradicting the existence of @xmath226 ) .
since the leaves of @xmath219 have uniformly bounded second fundamental forms , after a sequence of translations of @xmath219 in @xmath3 , we obtain another limit weak cmc foliation @xmath227 of @xmath8 with a leaf @xmath228 passing through the origin which has non - zero _ maximal _ mean curvature among all leaves of @xmath227 .
but the two - sided surface @xmath228 is then stable by proposition 5.4 in @xcite and since it is also complete , then @xmath228 must be flat ( see for instance lemma [ lema1 ] for a proof of this well - known result ) .
this contradiction finishes the proof of the theorem .
10000 in section [ sectspheres ] we will use the next corollary , which follows immediately from theorem [ thm5.7 ] after scaling the estimate for balls of radius 1 . [
cor : scale ] there exists an @xmath49 such that if @xmath7 is a weak @xmath52 foliation of a ball @xmath229 , then @xmath230 , where @xmath141 is given by ( [ eq : sigma ] ) .
let @xmath19 be a weak @xmath0-lamination of a punctured ball @xmath42 in a riemannian three - manifold @xmath2 , such that @xmath231 for some @xmath232 .
to prove theorem [ tt2 ] , it suffices to check that @xmath19 extends to a weak @xmath0-lamination of @xmath42 for a smaller @xmath25 . throughout this section
, we will assume without loss of generality that @xmath233 is sufficiently small so that the exponential map @xmath234 restricted to @xmath235 induces @xmath8-coordinates on @xmath42 .
[ ass : flat ] let @xmath19 be a weak @xmath0-lamination of a punctured ball @xmath42 in a riemannian three - manifold @xmath2 , such that @xmath231 for some @xmath232 . then for every sequence of positive numbers @xmath236 , there exists another sequence @xmath237 such that @xmath238 arguing by contradiction , suppose that the lemma fails .
it follows that there exists an @xmath239 and a sequence of points @xmath240 such that @xmath241 and @xmath242 for each @xmath57 .
let @xmath243 and consider the sequence of rescaled weak @xmath244-laminations @xmath245 ; here by @xmath246 we mean @xmath42 endowed with the riemannian metric @xmath247 , where @xmath248 is the metric on @xmath2 .
as @xmath0 is fixed , @xmath249 , @xmath231 and @xmath250 is invariant under homothetic rescalings of the metric around @xmath24 , then there exists a subsequence of the weak @xmath251-laminations @xmath131 that converges to a minimal lamination @xmath252 of @xmath253 , which furthermore satisfies @xmath254 in @xmath255 ( recall that @xmath84 stands for radial distance in @xmath3 to the origin ) . as @xmath256 for all @xmath57 , then @xmath257 for some @xmath258 ; thus the lamination @xmath252 is not flat . in this
setting , corollary 6.3 in @xcite gives that the closure @xmath259 of @xmath252 in @xmath3 consists of a single leaf @xmath260 , which is a non - flat minimal surface with finite total curvature ( clearly @xmath261 where @xmath262 ) .
* claim a : * _ @xmath260 contains the origin @xmath32 . _ to see this , it suffices to show that the distance sphere @xmath263 intersects @xmath19 for every @xmath264 sufficiently small .
otherwise , there exists @xmath265 such that the following properties hold : 1 .
2 . for all @xmath267 $ ]
, the mean curvature function of @xmath268 is strictly greater than @xmath0 .
3 . the family @xmath269\ }
$ ] foliates the punctured closed ball @xmath270 .
since @xmath271 intersects @xmath19 and @xmath19 is a closed subset of @xmath270 , then there exists a largest @xmath272 such that @xmath273 .
this contradicts the mean curvature comparison principle , which proves claim a. * claim b : * _ there exists @xmath274 such that the following properties hold : _
( b1 ) : : the intersection of @xmath19 with @xmath275 consists of a single leaf @xmath34 of the induced lamination .
furthermore , @xmath276 is a simple closed curve along which @xmath34 and @xmath277 intersect almost orthogonally .
( b2 ) : : @xmath278 is properly embedded in @xmath279 , with @xmath24 in its closure . to see that claim b holds , first note that by claim a , the intersection of @xmath280 with any closed ball @xmath281 centered at @xmath32 of sufficiently small radius @xmath282 is a punctured disk which is almost orthogonal to @xmath283 .
since @xmath280 is not flat , then the convergence of the laminations @xmath284 to @xmath280 has multiplicity one ( by lemma 4.2 in @xcite ) .
thus , we deduce that for @xmath57 large and @xmath285 , there exists a unique leaf @xmath286 of @xmath19 that intersects @xmath287 , and this intersection is a simple closed curve along which @xmath286 and @xmath287 intersect almost orthogonally . if @xmath286 were not the unique leaf of @xmath19 that intersects @xmath288 , then @xmath289 would contain a non - empty sublamination which does not intersect @xmath287 . a similar comparison argument for the mean curvature as in the proof of claim a shows that this is impossible for @xmath57 sufficiently large .
hence item * ( b1 ) * above holds by taking @xmath290 and @xmath291 for @xmath57 large .
suppose that @xmath34 is a limit leaf of @xmath19 .
in this case , item * ( b1 ) * implies that every point in @xmath276 is the limit in @xmath2 of a sequence of points of @xmath34 itself .
this is impossible , since @xmath34 and @xmath277 intersect almost orthogonally in a jordan curve .
thus , @xmath34 is not a limit leaf of @xmath19 .
we next prove item * ( b2 ) * : if @xmath34 were not proper in @xmath292 , then @xmath293 would contain a limit leaf , which therefore would not be @xmath34 ; this contradicts * ( b1)*. finally , if @xmath24 is not in the closure of @xmath34 , then @xmath24 is at positive distance from @xmath34 .
this contradicts * ( b1 ) * together with @xmath294 as @xmath173 converges to @xmath24 .
now * ( b2 ) * is proved , as well as claim b. we next finish the proof of the lemma . since @xmath34 is properly embedded in @xmath292 , then @xmath34 is a locally rectifiable current in @xmath295 .
as @xmath34 has bounded mean curvature ( actually constant ) , then theorem 3.1 in harvey and lawson @xcite implies that @xmath34 has finite area .
since @xmath34 has bounded mean curvature and finite area , the monotonicity formula in corollary 5.3 of allard @xcite implies that @xmath34 has a well - defined finite density at @xmath24 . in this setting
, we can apply theorem 6.5 in @xcite to deduce that under any sequence of homothetic expansions @xmath296 of @xmath34 , the surfaces @xmath297 converge ( up to a subsequence ) to a cone @xmath298 ( depending on the sequence ) , which is the cone over a stationary , integral one - dimensional varifold @xmath299 in the unit two - sphere of @xmath3 , and @xmath300 is flat at its smooth points .
but the blow - up limit @xmath280 is smooth and not flat , which is a contradiction .
this contradiction proves the lemma . by lemma [
ass : flat ] , it follows that the following property holds : * ( p ) * _ under rescaling by every sequence @xmath301 with @xmath302 , a subsequence of the weak @xmath251-laminations @xmath303 converges in @xmath255 to a lamination @xmath252 of @xmath8 by parallel planes .
( note that @xmath252 might depend on @xmath304 ) .
_ [ propos5.2 ] theorem [ tt2 ] holds in the particular case @xmath305 and @xmath306 .
_ as property * ( p ) * holds , it follows that for @xmath307 sufficiently small , in the annular domain @xmath308 , the normal vectors to the leaves of @xmath309 are almost parallel , and after a rotation ( which might depend on @xmath310 ) , we will assume that the unit normal vector to the leaves of @xmath309 lies in a small neighborhood of @xmath311 .
hence , for such a sufficiently small @xmath312 , each component @xmath43 of @xmath309 that intersects @xmath313 is of one of the following four types , see figure [ fig4new ] : .,width=529,height=253 ] type 1 .
: : @xmath43 is a compact disk with boundary @xmath314 in @xmath315 .
: : @xmath43 is a compact annulus with one boundary curve @xmath314 in @xmath316 and the other boundary curve in @xmath317 .
: : @xmath43 is a compact planar domain whose boundary consists of a single closed curve @xmath314 in @xmath316 together with at least two closed curves in @xmath317 , and where @xmath314 bounds a compact disk in @xmath318 ; type 4 . : : @xmath43 is an infinite multigraph whose limit set consists of two compact components of @xmath309 of type 2 ( any such spiraling component does not intersect the intersection of @xmath319 with an open slab of small width around height @xmath320 ) .
we also define @xmath321 for each @xmath322 ( so @xmath323 ) .
note that @xmath324 . given a component @xmath43 of @xmath309 ,
let @xmath325 be the leaf of @xmath326 $ ] that contains @xmath43 .
given @xmath327 fixed , the above division of components @xmath43 of @xmath328 can be directly adapted to components @xmath329 of @xmath330 .
we make the following elementary observations : ( o1 ) : : if for some @xmath322 , @xmath331 is of type 4 , then @xmath332 is of type 4 for every @xmath333 , and @xmath325 has @xmath32 in its closure .
( o2 ) : : if for some @xmath322 , @xmath331 is either empty , of type 1 or of type 3 , then @xmath325 is a disk which is at positive distance from @xmath32 .
( o3 ) : : if neither * ( o1 ) * nor * ( o2 ) * occur , then @xmath331 is of type 2 for every @xmath57 and thus , @xmath325 is a proper annulus limiting to @xmath32 .
[ ass5.3 ] there exists a leaf of @xmath334 having @xmath32 in its closure . if @xmath309 contains a component of type 4 , then
* ( o1 ) * implies that our claim holds .
if @xmath309 contains a component @xmath43 of type 2 such that @xmath331 is of type 2 for every @xmath57 , then * ( o3 ) * insures that @xmath325 contains @xmath32 in its closure and we are also done .
we will prove that the remaining case is impossible and this will finish the proof of this assertion .
the remaining case is that for every component @xmath43 of @xmath309 , there exists @xmath327 such that @xmath331 is empty , of type 1 or of type 3 ; in the case that @xmath331 is of type 3 , then @xmath332 is empty for every @xmath335 . by * ( o2 ) * , we have that @xmath336 $ ] consists of an ( infinite ) collection of pairwise disjoint compact disks .
since @xmath32 is in the closure of @xmath19 , there exists a sequence of points @xmath337 in compact disk leaves @xmath338 of @xmath336 $ ] , such that the @xmath339 converge to @xmath32 as @xmath340 .
we define for every @xmath341 , @xmath342 we claim that there exists @xmath341 such that @xmath343 is not closed in @xmath344 .
otherwise , @xmath345 is a collection of closed subsets of the compact space @xmath319 , which clearly satisfies the finite intersection property ; therefore , there exists a point @xmath346\cap a$ ] .
in particular , @xmath347 for some @xmath348 .
but as @xmath349 is disjoint from @xmath350 , we arrive to a contradiction .
this contradiction proves that there exists @xmath341 such that @xmath343 is not closed in @xmath344 .
since @xmath343 is not closed in @xmath344 , then there exists a point @xmath351 which is in the closure of @xmath343 but not in @xmath343 . as @xmath352 is closed in @xmath344 and @xmath353 , then @xmath354 .
thus , there exists a leaf component @xmath355 of @xmath356 passing through @xmath61 , which is disjoint from @xmath343 as @xmath357 .
furthermore , by our previous arguments , @xmath355 is topologically a closed disk which is at positive distance from @xmath32 .
choose a compact neighborhood @xmath358 of @xmath355 in @xmath359 which does not contain the origin . since @xmath356 is
a lamination and @xmath355 is a closed disk leaf , @xmath358 can be chosen so that every leaf of @xmath356 which intersects @xmath358 is entirely contained in @xmath358 .
if follows that there exists a subsequence of the disks @xmath360 which is contained in @xmath358 .
this is clearly a contradiction , as the @xmath338 contain points that converge to @xmath32 .
this proves assertion [ ass5.3 ] .
[ ass5.4 ] there exists a leaf @xmath34 of @xmath334 which is a proper annulus having @xmath32 in its closure .
moreover , every such @xmath34 extends smoothly across @xmath32 . by assertion [ ass5.3 ] ,
there exists a component @xmath43 of @xmath361 such that @xmath325 has @xmath32 in its closure . by observation *
( o2 ) * above , @xmath331 is of type 2 or 4 for all @xmath190 . if @xmath331 is of type 2 for some @xmath57 ( hence for all @xmath57 ) , then @xmath325 is a proper annulus limiting to @xmath32 .
otherwise , @xmath331 is of type 4 for some @xmath57 ( hence for all @xmath57 ) , and thus the limit set of @xmath43 in @xmath319 produces two compact components @xmath362 of @xmath361 each of which is of type 2 and such that @xmath363 are proper annuli limiting to @xmath32 .
this proves the first sentence of the assertion .
we next choose a proper annular leaf @xmath34 of @xmath334 with @xmath364 and check that @xmath34 extends smoothly across @xmath32 . since every blow - up limit of @xmath34 is a lamination of @xmath8 by parallel planes ( property * ( p ) * ) , then @xmath34 intersects small spheres @xmath365 of radius @xmath366 almost orthogonally in a curve of length no greater than @xmath367 , and thus , @xmath34 has finite area .
the conformal structure of @xmath34 must be the one of a punctured disk , as follows from the fact that under the conformal change of metric @xmath368 , @xmath369 has linear area growth ( here , @xmath370 and @xmath83 is the inner product in @xmath3 , recall that our present goal is to prove proposition [ propos5.2 ] , which is the @xmath3-case of theorem [ tt2 ] ) .
since @xmath34 has finite area and is conformally a punctured disk , then @xmath34 can be conformally parameterized by a mapping from a punctured disk into @xmath3 with finite energy . in this
setting , the main theorem in @xcite ( which holds true even if we exchange our current ambient manifold @xmath3 by any riemannian three - manifold whose sectional curvature is bounded from above and whose injectivity radius is bounded away from zero , conditions which will be satisfied in the general setting for @xmath2 that will be dealt with in proposition [ propos5.7 ] below , since we work in an arbitrarily small ball @xmath371 , @xmath274 ) , implies that @xmath34 extends @xmath372 through @xmath24 and so , standard elliptic theory gives that @xmath34 extends smoothly across @xmath32 as a mapping .
since @xmath34 is embedded around @xmath32 , then the extended image surface is also smooth .
this completes the proof of assertion [ ass5.4 ] .
[ ass5.5 ] there are no type 4 components of @xmath373 .
arguing by contradiction , if there exists a type 4 component @xmath43 of @xmath373 , then the limit set of @xmath43 in @xmath319 produces two compact components @xmath362 of @xmath361 each of which is of type 2 and such that @xmath363 are proper annuli limiting to @xmath32 . by assertion [ ass5.4 ] ,
both @xmath363 extend smoothly through @xmath32 by the previous paragraph .
this contradicts the usual maximum principle for @xmath0-surfaces , as both @xmath374 , @xmath375 have the same orientation at @xmath32 since the orientation of the multigraph component @xmath43 of type 4 induces the orientation of both @xmath363 . by observation * ( o2 ) * and assertions [ ass5.4 ] and [ ass5.5 ] , every leaf of @xmath19 which limits to @xmath32 is a proper annulus which extends smoothly across @xmath32 ( hence by the maximum principle there are at most two of them , with common tangent plane @xmath376 at @xmath32 and oppositely pointing mean curvature vectors ) and there exists at least one such proper annulus .
therefore , property * ( p ) * can now be improved to the following property : * ( p) * _ under rescaling by every sequence @xmath301 with @xmath302 , a subsequence of the weak @xmath251-laminations @xmath377 converges in @xmath255 to a lamination @xmath252 of @xmath8 by planes parallel to @xmath376 . _ let @xmath378 be one of the at most two proper annular leaves in @xmath19 limiting to @xmath32 .
let @xmath379 be the extended @xmath0-disk obtained after attaching the origin to @xmath378 .
consider an intrinsic geodesic disk @xmath380 in @xmath379 centered at @xmath32 with radius @xmath381 , and let @xmath98 be the unit normal vector field to @xmath382 .
pick coordinates @xmath383 in @xmath380 and let @xmath384\mapsto \g _ q(t)=q+t\eta ( q)$ ] be the straight line in @xmath3 passing through @xmath385 with velocity vector @xmath386 ( here @xmath387 is small and independent of @xmath388 so that the straight lines @xmath389 do not intersect each other ) .
then for some @xmath387 small , @xmath390 produces `` cylindrical '' normal coordinates in a neighborhood @xmath391 of @xmath32 in @xmath3 , and we can consider the natural projection @xmath392 since by * ( p) * every blow - up limit of @xmath19 from @xmath32 is a lamination of @xmath3 by planes parallel to @xmath376 , we conclude that for @xmath381 and @xmath393 sufficiently small , the angle of the intersection of any leaf component @xmath394 of @xmath395 with any straight line @xmath389 as above can be made arbitrarily close to @xmath396 .
taking @xmath381 much smaller than @xmath393 , a monodromy argument implies that any leaf component @xmath394 of @xmath395 which contains a point at distance at most @xmath397 from @xmath32 is a graph over @xmath382 ; in other words , @xmath398 restricts to @xmath394 as a diffeomorphism onto @xmath382 .
[ ass5.6 ] there exists a uniform bound around @xmath32 for the function @xmath141 defined in ( [ eq : sigma ] ) ( note that this property will complete the proof of proposition [ propos5.2 ] ) .
reasoning by contradiction , assume that there exists a sequence of points @xmath173 in leaves @xmath174 of @xmath19 converging to @xmath32 , such that @xmath399 diverges . without loss of generality , we can assume that @xmath400 and @xmath173 is a point where the following function attains its maximum : @xmath401 where @xmath402 denotes the intrinsic distance in @xmath379 to the boundary @xmath403 .
now expand the above coordinates @xmath404 centered at @xmath32 with ratio @xmath405 . under this expansion
, @xmath391 converges to @xmath3 with its usual flat metric and the straight lines @xmath389 converge to parallel lines .
the graphical property that @xmath398 restricts to any leaf component @xmath394 of @xmath395 @xmath397-close to @xmath32 as a diffeomorphism onto @xmath382 gives that after passing to a subsequence , the @xmath0-graphs @xmath406 converge after expansion of coordinates to a minimal surface in @xmath3 which is an entire graph . by the bernstein theorem ,
such a limit surface is a flat plane .
this contradicts that the ratio of the homothetic expansion coincides with the norm of the second fundamental form of @xmath406 at @xmath173 for all @xmath57 .
this contradiction finishes the proof of assertion [ ass5.6 ] , and completes the proof of proposition [ propos5.2 ] .
[ propos5.7 ] theorem [ tt2 ] holds in the general case for the ambient manifold @xmath2 . in the manifold setting for @xmath2 , under rescaled exponential coordinates from @xmath24
we have the same description as in the proof of proposition [ propos5.2 ] , and the arguments in that proof adapt with straightforward modifications ; also see cases iv and v of the proof of theorem 1.1 in @xcite .
this completes the proof of theorem [ tt2 ] .
we next extend corollary 7.1 in @xcite to the case of a weak @xmath0-lamination in a riemannian three - manifold .
we remark that the statements in items 4 , 5 of the corollary below do not have corresponding statements in corollary 7.1 in @xcite .
regarding item 5 of corollary [ corrs ] and using its notation , we make the following definition .
[ def5.8 ] the absolute mean curvature function of a weak cmc foliation @xmath7 of @xmath407 is the function @xmath408 defined by @xmath409 note that as in the case of @xmath141 given by ( [ eq : sigma ] ) , the function @xmath410 is not necessarily continuous .
the hypothesis of boundedness of @xmath410 in item 5 of corollary [ corrs ] is essential : take @xmath16 , @xmath411 and @xmath7 the foliation of @xmath255 by concentric spheres .
[ corrs ] let @xmath1 .
suppose that @xmath2 is a riemannian three - manifold , not necessarily complete .
if @xmath412 is a closed countable subset and @xmath19 is a weak @xmath0-lamination of @xmath407 such that for every @xmath413 there exists positive constants @xmath414 ( possibly depending on @xmath24 ) satisfying the following curvature estimate : @xmath415 then @xmath19 extends across @xmath416 to a weak @xmath0-lamination of @xmath2 . in particular : 1 .
the closure of any collection of the stable leaves of a weak @xmath0-lamination of @xmath407 extends across @xmath416 to a weak @xmath0-lamination of @xmath2 consisting of stable @xmath0-surfaces .
the closure in @xmath2 of any collection of limit leaves of a weak @xmath0-lamination @xmath19 of @xmath407 is a weak @xmath0-lamination of @xmath2 , all whose leaves are stable @xmath0-surfaces .
if @xmath7 is a weak @xmath0-foliation of @xmath407 , then @xmath7 extends across @xmath416 to a weak @xmath0-foliation of @xmath2 .
if @xmath7 is a weak cmc foliation of @xmath407 and @xmath1 , then the closure in @xmath407 of any collection @xmath18 of leaves of @xmath7 with constant curvature @xmath0 extends across @xmath416 to a weak @xmath0-lamination of @xmath2 .
if @xmath7 is a weak cmc foliation of @xmath407 with bounded absolute mean curvature function , then @xmath7 extends across @xmath416 to a weak cmc foliation of @xmath2 .
_ let @xmath19 be a weak @xmath0-lamination of @xmath407 satisfying the curvature estimate ( [ eq : ce ] ) , where @xmath416 is closed and countable .
since the extension of @xmath19 across @xmath416 is a local question , it suffices to extend @xmath417 in small , open extrinsic balls in @xmath2 . since @xmath416 is countable , we can take these balls so that each of their boundaries are disjoint from @xmath416 , and their closures in @xmath2 are compact .
it follows that for every such ball @xmath418 , the set @xmath419 is a complete countable metric space . by baire s theorem ,
the set @xmath420 of isolated points of the locally compact metric space @xmath419 is dense in @xmath419 .
[ ass5.9new ] in the above situation , @xmath421 extends across @xmath420 to a weak @xmath0-lamination of @xmath422 .
consider an isolated point @xmath423 . by hypothesis
, there exist @xmath424 such that the inequality ( [ eq : ce ] ) holds . taking @xmath151 smaller if necessary
, we can assume that the closed ball @xmath425 is compact and contained in @xmath418 , its boundary @xmath426 is disjoint from @xmath416 and that @xmath427 . by theorem [ tt2 ] ,
the induced local weak @xmath0-lamination @xmath428 $ ] extends across @xmath24 to a weak @xmath0-lamination of @xmath429 .
this proves the assertion .
consider the collection @xmath430 of open subsets @xmath358 of @xmath418 such that @xmath431 and there exists a weak @xmath0-lamination @xmath432 of @xmath358 whose restriction to @xmath433 coincides with @xmath434 . by assertion [ ass5.9new ] , @xmath435 .
we claim that @xmath436 .
note that if @xmath437 , then the related weak @xmath0-lamination @xmath432 is unique ( since leaves of @xmath432 are analytic surfaces that coincide with the leaves of @xmath434 ) . given @xmath438 and given a point @xmath439 , the related weak @xmath0-laminations @xmath440 that extend @xmath434 to @xmath441 satisfy @xmath442 by the above uniqueness property .
therefore , @xmath443 and our claim is proved .
we want to prove that if @xmath444 , then @xmath445 , which will finish the proof of the first statement of the corollary .
arguing by contradiction , suppose @xmath446 . since @xmath447 is a non - empty closed subset of @xmath419
, then @xmath448 is a complete countable metric space and so , baire s theorem again insures that the set @xmath449 of its isolated points is dense in @xmath448 . by assertion [ ass5.9new ] , the @xmath0-lamination @xmath450 obtained by extension of @xmath19 to @xmath391 extends through every isolated point of @xmath448 ; hence @xmath451 . by definition of @xmath391
, this implies that @xmath452 , hence @xmath453 . as @xmath454 , then @xmath455 which contradicts that @xmath449 is dense in @xmath448 .
now the proof of the first statement of corollary [ corrs ] is complete .
item 1 of the corollary follows from the already proven first statement and from curvature estimates for stable @xmath0-surfaces ( schoen @xcite , ros @xcite , see also theorem 2.15 in @xcite ) . by theorem 4.3 of @xcite ( see also theorem 1 in @xcite ) , limit leaves of a weak @xmath0-lamination are stable ( if @xmath158 they are two - sided ; in the minimal case , the two - sided cover of every limit leaf is stable ) . as the collection of limit leaves of a weak @xmath0-lamination
is closed , then item 2 of the corollary follows from item 1 .
item 3 is a direct consequence of item 2 , as every leaf of a weak @xmath0-foliation is a limit leaf . to prove item 4 ,
let @xmath7 be a weak cmc foliation of @xmath407 , where @xmath416 is closed and countable , and let @xmath1 .
reasoning as in the case of a weak @xmath0-lamination , we can reduce the proof of the extendability of any collection @xmath18 of leaves of @xmath7 with constant mean curvature @xmath0 to the case in which @xmath7 is a weak cmc foliation of a small open extrinsic ball @xmath418 with compact closure in @xmath2 , such that @xmath456 .
also the above argument based on baire s theorem allows one to reduce the proof to the case that @xmath457 where @xmath413 is an isolated point of @xmath416 and @xmath458 . to prove that @xmath18 extends across @xmath24 , it suffices to show that for some small @xmath239 , the induced local weak @xmath0-lamination @xmath459 $ ] extends across @xmath24 to a weak @xmath0-lamination of @xmath429 .
consider the weak cmc foliation @xmath460 $ ] . by the universal curvature estimate in theorem [ thm5.7 ] applied to each of the compact three - manifolds with boundary @xmath461 , @xmath341
, there exists a constant @xmath49 independent of @xmath462 such that for each @xmath462 , we have @xmath463,\ ] ] where @xmath50 is an upper bound of the sectional curvature of @xmath2 in @xmath464 and @xmath465 is defined in ( [ eq : sigma ] ) .
taking @xmath310 smaller if necessary ( this does not change the constant @xmath51 ) , we can assume that @xmath466 for all @xmath341 .
thus , given @xmath467 and @xmath468 , we have @xmath469 hence , the weak @xmath0-lamination @xmath470 satisfies the curvature estimate in the hypothesis of theorem [ tt2 ] , and thus , @xmath470 extends across @xmath24 as desired .
this proves item 4 of the corollary .
finally we prove item 5 .
let @xmath7 be a weak cmc foliation of @xmath407 with bounded absolute mean curvature function , where @xmath416 is closed and countable .
similar arguments as in the previous cases show that we can reduce the proof of item 5 to the proof of the extendability of a weak cmc foliation @xmath7 of a small open extrinsic ball @xmath471 with compact closure in @xmath2 , where @xmath413 is isolated in @xmath416 and @xmath472 . to prove that @xmath7 extends across @xmath24
, it suffices to prove that for some smaller @xmath239 , the induced local weak cmc foliation @xmath473 $ ] extends across @xmath24 to a weak cmc foliation of @xmath429 .
since by hypothesis the absolute mean curvature function of the leaves of @xmath474 is bounded , we can choose @xmath151 sufficiently small so that for all @xmath475 $ ] , the absolute mean curvature function of the ( smooth ) distance sphere @xmath268 is strictly greater than the maximum value of the absolute mean curvature of the leaves of @xmath474 .
as an application of the mean curvature comparison principle , we conclude that the closure of every leaf of @xmath474 intersects @xmath476 ( see claim a in the proof of lemma [ ass : flat ] for a similar argument ) .
take a sequence @xmath477 converging to @xmath24 as @xmath209 . as @xmath474 is a weak cmc foliation of @xmath478 , for each @xmath190 there exists at least one leaf @xmath174 of @xmath474 with @xmath479 .
let @xmath480 be the ( constant ) mean curvature of @xmath174 and let @xmath481 .
after replacing by a subsequence , we may assume that @xmath482 .
let @xmath470 be the weak @xmath0-lamination of @xmath478 consisting of all leaves of @xmath474 whose mean curvature is @xmath0 .
we claim that @xmath24 lies in the closure of @xmath470 in @xmath471 . to see this it suffices to show that given @xmath483 , some leaf of @xmath470 intersects @xmath484 .
fix @xmath341 . as @xmath485 , then for @xmath57 sufficiently large @xmath486 . as the closure of @xmath174 intersects @xmath426 and @xmath174
is connected , then @xmath174 also intersects @xmath487 . for each @xmath190 large , pick a point @xmath488 . since @xmath487 is compact , after extracting a subsequence , the @xmath489 converge as @xmath160 to a point @xmath490 .
as the mean curvatures of the @xmath174 converge to @xmath0 , then there passes a leaf @xmath228 of @xmath470 through @xmath61 , and our claim is proved .
[ ass5.11 ] the weak cmc foliation @xmath474 extends across @xmath24 to a weak cmc foliation of @xmath471 ( and thus , the proof of item 5 of corollary [ corrs ] is complete ) . by the last claim and the already proven item 4 of this corollary
, @xmath470 extends across @xmath24 to a weak @xmath0-lamination of @xmath491 .
let @xmath492 be the leaf of the extended weak @xmath0-lamination @xmath493 passing through @xmath24 ( thus , @xmath494 is a leaf of @xmath474 ) .
after possibly choosing a smaller @xmath312 , we may assume that @xmath492 is a smooth embedded disk in @xmath495 with compact boundary in @xmath426 .
using again the curvature estimates ( [ eq : handbook ] ) we get that for any sequence of positive numbers @xmath496 , a subsequence of the punctured balls @xmath497 $ ] converges as @xmath209 to @xmath498 with its usual metric , and the weak cmc foliations @xmath499 converge to a limit weak cmc foliation @xmath500 of @xmath255 , which is in fact a minimal foliation since the mean curvature of the leaves of @xmath474 is bounded .
note that one of the leaves of @xmath500 is the punctured plane @xmath501 passing through @xmath32 , corresponding to the blow - up of the tangent plane to the disk @xmath492 at @xmath24 . by item 3 of this corollary
, @xmath500 extends across the origin to a minimal foliation of @xmath3 ; since every leaf of this extended minimal foliation is a complete stable minimal surface in @xmath3 , then every such leaf is a plane , which must be parallel to @xmath501 . in particular
, the limit foliation @xmath500 is independent of the sequence @xmath496 . in this situation
, it follows that for @xmath312 sufficiently small , the leaves in @xmath474 can be uniformly locally expressed as non - negative or non - positive normal graphs with bounded gradient over their projections to @xmath492 .
in particular , there is a weak cmc foliation structure on @xmath502 , and assertion [ ass5.11 ] is proved .
we start by proving the @xmath8-version of theorem [ thmspheresintrod ] in the more general setting of weak cmc foliations .
[ thmspheres ] suppose that @xmath503 is a weak cmc foliation of @xmath8 with a closed countable set @xmath9 of singularities ( these are the points where the weak cmc structure of @xmath7 can not be extended ) .
then , each leaf of @xmath7 is contained in either a plane or a round sphere , and @xmath504 .
furthermore if @xmath13 is empty , then @xmath14 is a foliation by planes . _ proof .
_ note that if all leaves of @xmath7 are minimal , then @xmath7 is a minimal foliation of @xmath505 , hence by item 3 of corollary [ corrs ] , @xmath7 extends to a minimal foliation of @xmath3 , which must then consist entirely of parallel planes and the theorem holds in this case .
therefore , in the sequel we may assume that @xmath7 contains a leaf which is not minimal .
[ ass6.2 ] every non - minimal leaf of @xmath7 is proper in @xmath506 .
furthermore , if @xmath6 is bounded , then every non - minimal leaf of @xmath7 is contained in a ball .
consider a leaf @xmath34 of @xmath7 , with mean curvature @xmath0 . by item 4 of corollary [ corrs ]
, the collection @xmath18 of @xmath0-leaves in @xmath7 extends across @xmath6 to weak @xmath0-lamination of @xmath3 and so , the closure of @xmath34 in @xmath3 is a weak @xmath0-lamination of @xmath3 . if @xmath158 and @xmath34 is not proper in @xmath506 , then @xmath492 contains a limit leaf @xmath507 , which is complete since @xmath492 is a weak @xmath0-lamination of @xmath3 . by theorem 4.3 of @xcite
( see also theorem 1 in @xcite ) applied to the weak @xmath0-lamination @xmath492 , @xmath507 is stable ( @xmath507 is two - sided since its mean curvature is non - zero ) .
this contradicts that there are no stable complete @xmath0-surfaces in @xmath3 for any @xmath158 .
this proves the first sentence in the assertion .
next suppose @xmath6 is bounded and take a non - minimal leaf @xmath508 .
if there exists an extrinsically divergent sequence of points @xmath509 , then the extrinsic distance @xmath203 from @xmath173 to @xmath6 tends to infinity as @xmath209 and hence , corollary [ cor : scale ] applied to the weak cmc foliation @xmath510 of @xmath511 implies that the second fundamental form of @xmath34 at @xmath173 decays to zero in norm , as @xmath512 . in particular , the trace of the second fundamental form of @xmath34 must be zero since @xmath34 has constant mean curvature , which gives a contradiction .
therefore , every non - minimal leaf @xmath34 of @xmath7 lies in some ball of @xmath3 .
[ ass6.3 ] if a leaf @xmath34 of @xmath7 is contained in a ball of @xmath3 , then its closure @xmath492 is a round sphere . as in the proof of the previous assertion ,
the closure of @xmath34 in @xmath3 has the structure of a weak @xmath0-lamination of @xmath3 by item 4 of corollary [ corrs ] .
since there are no bounded minimal laminations in @xmath3 by the maximum principle , then @xmath158 . by assertion [ ass6.2 ] , @xmath34 is proper in @xmath506 , and thus , @xmath492 consists of a single compact immersed surface which does not intersect itself transversely , and whenever @xmath492 intersects itself , it locally consists of two disks with opposite mean curvature vectors .
hence , @xmath492 is alexandrov - embedded .
in this situation , alexandrov @xcite proved that @xmath492 is a round sphere .
[ ass6.4 ] if @xmath6 is bounded , then theorem [ thmspheres ] hold .
as @xmath6 is bounded , then assertions [ ass6.2 ] and [ ass6.3 ] give that the closure of every non - minimal leaf of @xmath7 is a round sphere .
consider the collection @xmath120 of all spherical leaves of @xmath14 union with @xmath6 .
then , the restriction of @xmath7 to the complement of the closure of @xmath120 is a minimal foliation @xmath474 of the open set @xmath513 . applying item 3 of corollary [ corrs ] to @xmath474 , @xmath514 and @xmath515 ,
we conclude that @xmath474 extends across @xmath516 to a minimal foliation of @xmath517 . by item 4 of corollary [ corrs ] , @xmath474
extend across @xmath6 to a minimal lamination of @xmath3 , and thus the extended leaves of @xmath474 are complete . as @xmath474 consists of stable leaves by theorem 1 in @xcite ,
then item 1 of corollary [ corrs ] insures that the extended leaves of @xmath7 across @xmath6 are complete stable minimal surfaces in @xmath3 , hence planes .
as the weak cmc foliation @xmath7 is now entirely formed by punctured spheres and planes , then it is clear that @xmath6 consists of one of two points .
this completes the proof of assertion [ ass6.4 ] . to prove theorem [ thmspheres ] in the general case of a closed countable set @xmath518
, we next analyze the structure of @xmath7 in a neighborhood of an isolated point @xmath519 ( recall that the set of isolated points in @xmath6 is dense in @xmath6 by baire s theorem ) . since @xmath24 is isolated in @xmath6 , we can choose a sphere @xmath520 such that @xmath521 . as @xmath6 is closed , then @xmath520 is at positive distance from @xmath6 . since @xmath141 is locally bounded in @xmath522 ( by definition of weak cmc lamination )
, an elementary compactness argument shows that there is a uniform upper bound for the restriction to @xmath523 of the norms of the second fundamental forms of all leaves in @xmath7 which intersect @xmath524 ; in particular the absolute mean curvature of every such leaf satisfies @xmath525 for some @xmath526 . by item 5 of corollary [ corrs ] ,
the mean curvature of the leaves of @xmath7 is unbounded in every neighborhood of @xmath24 , since @xmath519 .
therefore , there exist leaves of @xmath7 which intersect @xmath527 and whose mean curvatures satisfy @xmath528 .
every such leaf @xmath34 is entirely contained in @xmath529 and thus , assertion [ ass6.3 ] implies that the closure @xmath492 of @xmath34 in @xmath3 is a round sphere .
note that either @xmath530 or @xmath24 lies in the open ball @xmath531 enclosed by @xmath492 ( otherwise a monodromy argument shows that @xmath532 is a `` product '' foliation by spheres , which produces a singularity @xmath533 ; this contradicts that @xmath534 ) .
the above arguments show that for every isolated point @xmath24 of @xmath6 , one of the two following possibilities holds : ( a ) : : there exists an open neighborhood @xmath535 of @xmath24 in @xmath3 such that @xmath7 restricts to @xmath536 as a weak cmc foliation by round spheres and @xmath537 , see figure [ figure5new ] left .
( b ) : : there exists an open ball @xmath538 such that @xmath539 and the weak cmc foliation @xmath7 restricts to @xmath540 as a union of round spheres punctured at @xmath24 , all tangent at @xmath24 . in this case
, we call @xmath541 , see figure [ figure5new ] right . .
right : case * ( b)*. in both cases , the dot represents an isolated singular point @xmath519.,height=188 ] if possibility * ( a ) * occurs for an isolated point @xmath24 of @xmath6 , we define @xmath542 to be the maximal such open set @xmath535 , with the ordering given by the inclusion .
note that in this case , we have two mutually exclusive possibilities : ( a1 ) : : the boundary @xmath543 of @xmath542 is empty ; in this case , @xmath544 and theorem [ thmspheres ] is proved with @xmath545 .
( a2 ) : : the boundary @xmath543 is non - empty ; in this case @xmath543 is either a round sphere ( and @xmath542 is an open ball of @xmath3 containing @xmath24 ) , or @xmath543 is a plane ( and @xmath542 is an open half - space containing @xmath24 ) . in both of these cases
, @xmath542 only intersects @xmath6 at the point @xmath24 .
if possibility * ( b ) * holds for an isolated point @xmath24 of @xmath6 , we define @xmath542 to be the union of the maximal open 1-parameter family of spheres in @xmath7 , possibly punctured at @xmath24 , that contains the open ball @xmath535 described in possibility * ( b ) * , together with the point @xmath24 if this union contains a spherical leaf of @xmath7 that does not pass through @xmath24 . as in case * ( a ) * , we have two mutually exclusive possibilities : ( b1 ) : : the boundary @xmath543 of @xmath542 is empty ; in this case , @xmath544 and theorem [ thmspheres ] is again proved with @xmath545 .
therefore , in the sequel we will assume that for each isolated point @xmath519 , we have @xmath546 .
( b2 ) : : the boundary @xmath543 is non - empty ; in this case @xmath543 is either a round sphere ( and in @xmath542 is an open ball of @xmath3 with @xmath24 in its closure ) , or @xmath543 is a plane ( and @xmath542 is an open halfspace with @xmath24 in its closure ) . in both of these cases ,
@xmath542 only intersects @xmath6 in at most the point @xmath24 .
note that the case of two simultaneous such maximal open sets @xmath547 can occur in case * ( b ) * ; for instance when @xmath7 is the foliation of @xmath255 given by the @xmath548-plane together with all spheres passing through @xmath306 and tangent to the @xmath548-plane ( in this case @xmath549 and @xmath550 ) .
in the case that we have two possibilities for choosing @xmath542 , we will simply arbitrarily choose one such @xmath542 in our discussions below .
we next collect some elementary properties of these open sets @xmath542 , which easily follow from the fact that @xmath7 is a foliation outside @xmath6 and from the description in possibilities * ( a ) * , * ( b ) * above .
( p1 ) : : if @xmath551 are distinct isolated points of @xmath6 , then @xmath552 . ( p2 ) : : if @xmath553 is a converging sequence of distinct isolated points of @xmath6 , then for @xmath57 sufficiently large , @xmath554 is an open ball and the radii of the @xmath554 converge to zero .
note that by maximality and a standard monodromy argument , if @xmath543 is a sphere then @xmath555 .
[ ass10.5 ] given an isolated point @xmath519 , suppose that @xmath542 is an open ball .
then , @xmath556 contains at least one point which is not isolated in @xmath6 .
recall that @xmath555 .
arguing by contradiction , suppose that @xmath556 consists only of isolated points of @xmath6 .
in particular , @xmath557 is finite , say @xmath558 .
note that @xmath24 lies in @xmath557 if and only if possibility * ( b ) * above holds for @xmath24 .
then , the above arguments show that around every point @xmath559 , necessarily case * ( b ) * occurs ( exchanging @xmath24 by @xmath560 ) , and that the following additional property holds : ( p3 ) : : if @xmath561 , then the related maximal open set @xmath562 is disjoint from @xmath542 and every leaf in the restriction of @xmath7 to @xmath563 is a punctured sphere or punctured plane whose closure only intersects @xmath564 at @xmath560 .
analogously , if @xmath565 is an isolated point where we have two possibilities @xmath566 for choosing @xmath542 , then the same property * ( p3 ) * holds for @xmath567 and @xmath568 , see figure [ newfigure ] . .
on the right , possibility * ( b ) * holds for @xmath24 and there are two possible choices @xmath542,@xmath569 for @xmath542 . in both figures ,
we have represented one of the points in @xmath556 ( namely , @xmath570 ) so that the related set @xmath571 is a halfspace.,width=453 ] let @xmath572 be pairwise disjoint , small open balls centered at the points @xmath573 . as the @xmath574 are isolated in @xmath6 , we can assume that @xmath575 .
we denote by @xmath576 , which is a spherical disk .
next we will prove the following property .
( p4 ) : : given @xmath577 , if the radius of the ball @xmath578 is small enough , then the intersection of @xmath7 with the region @xmath579 $ ] consists of a collection of annuli , each of which can be expressed as a normal graph over its projection to @xmath580 . to see why * ( p4 )
* holds , we first prove that under blow - up from @xmath574 , the induced foliation @xmath581 converges smoothly to the punctured tangent plane to @xmath543 at @xmath574 .
note that the region @xmath582 converges after such a blow - up to the punctured tangent plane to @xmath543 at @xmath574 .
since we have a scale - invariant uniform bound on the second fundamental of @xmath583 ( given by theorem [ thm5.7 ] , see the proof of assertion [ ass6.2 ] for a similar argument ) , then the leaves of the induced foliation @xmath581 are locally graphical over small geodesic disks of @xmath584 contained in @xmath585 .
it remains to check that these local graphs , when glued together inside a leaf of @xmath581 , do not define multi - valued graphs over the punctured spherical disk @xmath586 .
arguing by contradiction , suppose that there exists a leaf @xmath587 of @xmath581 which can be expressed by a multi - valued graph ( not univalent ) over @xmath586 .
consider a small compact solid cylinder @xmath588 whose axis is the normal line to @xmath543 passing through @xmath574 , such that both @xmath543 and @xmath589 intersect @xmath588 in compact closed disks bounded by distinct parallel circles .
then , the intersection of @xmath587 with @xmath590 contains a spiraling curve @xmath299 ( with infinite length in both directions ) which is trapped between the circles @xmath591 , @xmath592 .
the curve @xmath299 limits to two disjoint closed curves @xmath593 , which are topologically parallel to @xmath591 , @xmath592 . let @xmath594 be the leaves of @xmath7 that contain @xmath595 , @xmath596 , respectively ( note that a priori , @xmath597 could coincide with @xmath598 ) .
as @xmath599 consist of limit points of @xmath299 , then @xmath597 , @xmath598 are limit leaves of the sublamination @xmath18 of @xmath7 consisting of the leaves of @xmath7 with the same mean curvature @xmath0 as @xmath587 . by item 4 of corollary [ corrs ] , both @xmath597 , @xmath598 extend smoothly across @xmath574 .
since the @xmath0-surfaces @xmath600 , @xmath601 have the same mean curvature vectors at @xmath574 ( their orientations are induced by the one of the multi - valued graph inside @xmath587 ) , we contradict the maximum principle for @xmath0-surfaces . now property * ( p4 ) * is proved . since * ( p4 )
* holds , it follows that @xmath602 , considered to be a leaf of the weak cmc foliation @xmath7 of @xmath506 , has trivial holonomy on the exterior side of @xmath543 .
this implies that the leaves of @xmath7 nearby @xmath602 and outside @xmath542 are topologically punctured spheres which are graphs over @xmath602 .
therefore , these graphs extend smoothly to embedded topological spheres with constant mean curvature ; hence the extended graphs are themselves round spheres .
this contradicts the maximality of @xmath542 .
this contradiction finishes the proof of assertion [ ass10.5 ] .
[ ass6.6 ] theorem [ thmspheres ] holds .
consider the set @xmath603 of those points of @xmath6 which are isolated ; recall that @xmath603 is an open dense subset of @xmath6 by baire s theorem .
if for every @xmath604 the related maximal open set @xmath542 given just before assertion [ ass10.5 ] is a halfspace , then clearly @xmath6 only consists of one or two singularities , and by assertion [ ass6.4 ] , the theorem holds in this case .
so it suffices to show that for every @xmath604 , @xmath542 can not be an open ball .
arguing by contradiction , suppose that there exists @xmath604 such that @xmath542 is an open ball .
since @xmath605 by assertion [ ass10.5 ] and @xmath606 is closed in @xmath6 , then the set @xmath607 of isolated points in @xmath606 is non - empty ( in fact , @xmath607 is dense in @xmath606 by baire s theorem applied to the complete metric space @xmath608 together with the collection of open dense subsets of @xmath606 given by @xmath609 , where @xmath610 is an enumeration of the countable set @xmath611 ) .
pick a point @xmath612 , which must be a limit of a sequence of points @xmath613 . since the @xmath173 converge to @xmath385 , property * ( p2 ) *
insures that for @xmath57 large , @xmath554 is an open ball and the radii of @xmath554 converge to zero as @xmath209 . since each @xmath614 contains at least one point in @xmath606 ( by assertion [ ass10.5 ] )
, then we contradict that @xmath385 is isolated in @xmath606 .
this contradiction proves that @xmath542 can not be an open ball , and therefore finishes the proof of assertion [ ass6.6 ] .
we next prove the @xmath4-version of theorem [ thmspheres ] .
[ thmspheres2 ] let @xmath503 be a weak cmc foliation of @xmath4 with a closed countable set @xmath9 of singularities ( as in theorem [ thmspheres ] , these singularities are the points where the weak cmc foliation structure of @xmath7 can not be extended ) .
then , each leaf of @xmath7 is contained in a round sphere in @xmath4 and the number of singularities is @xmath615 or @xmath218
. this theorem can be proven with minor modifications of the proof in the @xmath3 case .
one starts by proving that if @xmath7 is a weak cmc foliation of @xmath616 with @xmath6 closed , then @xmath6 can not be empty .
this holds since otherwise , as @xmath4 is compact and the absolute mean curvature function of @xmath7 is locally bounded , there exists a leaf in @xmath7 of maximal absolute mean curvature , which by proposition 5.4 in @xcite must be stable .
this contradicts the non - existence of complete , stable surfaces with constant mean curvature in @xmath4 . by item 4 of corollary [ corrs ] , for any @xmath617
, the weak @xmath0-lamination @xmath18 of @xmath7 consisting of all leaves of constant mean curvature @xmath0 extends smoothly across the singular set @xmath6 of @xmath7 to a collection @xmath618 of compact immersed @xmath0-surfaces .
the non - existence of limit leaves of @xmath619 implies that this collection of surfaces in finite .
next we show that @xmath492 is alexandrov - embedded for every leaf @xmath34 of @xmath18 .
observe that @xmath492 can not have transversal intersections , as @xmath7 is a weak cmc foliation .
if @xmath492 is not embedded , then @xmath158 and there exists @xmath530 such that locally around @xmath24 , @xmath492 consists of two disks that lie at one side of each other with opposite mean curvature vectors .
as @xmath492 is compact , there exists some small @xmath151 such that @xmath620 is an embedded @xmath310-neighborhood on the mean convex side of @xmath492 , where @xmath2 stands for the unit normal vector field to @xmath492 for which @xmath0 is the mean curvature of @xmath492 .
in particular , the parallel surface @xmath621 at distance @xmath622 from @xmath492 inside @xmath623 is embedded . as every compact embedded surface separates @xmath4
, then @xmath621 divides @xmath4 into two open domains , one of which , called @xmath624 , contains @xmath625 .
the union of the closure of @xmath624 in @xmath4 with the @xmath626-neighborhood @xmath627 of @xmath492 can be viewed as the image of a submersion of a 3-manifold with boundary into @xmath4 , with its boundary image being @xmath492 .
this proves that @xmath492 is alexandrov - embedded .
let @xmath519 be an isolated point in @xmath6 , which exists by baire s theorem . as @xmath24 is isolated in @xmath6 , we can choose a geodesic sphere @xmath628 in @xmath4 of small radius @xmath629 centered at @xmath24 such that the geodesic ball @xmath630 enclosed by @xmath628 satisfies @xmath631 . as @xmath6 is closed , then @xmath628 is at positive distance from @xmath6 .
since @xmath141 is locally bounded in @xmath632 , an elementary compactness argument shows that there is a uniform upper bound for the restriction to @xmath628 of the norms of the second fundamental forms of all leaves in @xmath7 which intersect @xmath628 ; in particular the absolute mean curvature of every such leaf satisfies @xmath633 for some @xmath232 . by item 5 of corollary [ corrs ] ,
the mean curvature of the leaves of @xmath7 is unbounded in every neighborhood of @xmath24 , since @xmath519 .
therefore , there exist leaves of @xmath7 which intersect @xmath630 and whose mean curvatures satisfy @xmath634 .
every such leaf @xmath34 is then entirely contained in @xmath630 . as @xmath635 ,
then @xmath492 is contained in a hemisphere , and by a standard application of the alexandrov moving plane technique , we conclude that @xmath492 is a sphere . as we did in the case of @xmath3
, one can consider the maximal collection @xmath542 of round spheres in @xmath7 around every isolated point of @xmath6 . in this
setting , assertion [ ass10.5 ] remains valid with only straightforward modifications in its proof , as well as the baire s theorem argument in the proof of assertion [ ass6.6 ] .
we leave the remaining details for the reader . _ recall that by theorem [ thmspheres ] , if a weak cmc foliation @xmath17 of @xmath8 has a countable set of singularities @xmath636 ( as a weak foliation ) , then @xmath636 consists of at most two points .
if @xmath637 , then @xmath17 is a foliations by parallel planes .
if @xmath638 , then up to a translation we can assume @xmath639 and we have two cases : _ 1 .
some leaf of @xmath17 has the origin in its closure . in this case , up to a rotation of @xmath8 fixing @xmath32 , there are exactly two examples @xmath640 , where @xmath641 is the set of horizontal planes @xmath642 together with the spheres @xmath144 with @xmath146 for all @xmath643 , and @xmath644 is the set of spheres tangent to the @xmath548-plane at @xmath32 together with the @xmath548-plane . in particular , in this case 1 the weak foliation @xmath17 of @xmath645 is actually a foliation of @xmath645 .
2 . no leaf of @xmath17 has the origin in its closure , as in the particular foliation @xmath646 of @xmath255 consisting of the set of spheres centered at the origin . in this case 2
, the weak foliation @xmath17 can be isotoped to @xmath646 fixing the origin along the isotopy , and the spherical leaves can intersect themselves as leaves of a weak foliation . in theorem [ corol6.6 ] below we will describe the structure of a singular weak cmc foliation @xmath17 of a three - manifold in a neighborhood of an isolated singularity @xmath24 as being modeled on a weak singular cmc foliation of @xmath645 with @xmath638 ;
in fact we will show that the weak foliation @xmath17 is closely approximated by exactly one of the examples in the two cases above .
theorem [ thmspheres ] implies that the two possibilities represented in figure [ figure5new ] give canonical models for every weak cmc foliation of @xmath255 as a 1-parameter collection of spheres and planes with one singularity occurring at the origin .
in fact , this description is a good model for the local structure of any weak cmc foliation in a riemannian three - manifold around an isolated singularity , as the following theorem demonstrates . 1 .
@xmath7 extends across @xmath24 to a weak cmc foliation of @xmath42 if and only if the absolute mean curvature function of @xmath7 ( see definition [ def5.8 ] ) is bounded .
2 . suppose that the absolute mean curvature function of @xmath7 is unbounded in @xmath647 .
then , there exists @xmath648 such that : + _ ( 2a ) _ : : for every sequence @xmath649 with @xmath249 , the blow - up weak cmc foliations @xmath650 $ ] of @xmath651 converge ( after extracting a subsequence ) to a non - flat weak cmc foliation of @xmath255 . _ ( 2b ) _ : : there exists @xmath652 such that the closure in @xmath653 of every leaf of @xmath654 with absolute mean curvature @xmath655 is an embedded @xmath0-sphere that bounds a subball of @xmath42 that contains @xmath24 in its closure . in the sequel
we will assume that the absolute mean curvature function of @xmath7 is unbounded in @xmath647 , and so , for every @xmath656 the absolute mean curvature function of @xmath654 is also unbounded .
choose @xmath656 sufficiently small so that @xmath198 is less than the injectivity radius function of @xmath2 at @xmath24 and so that the geodesic spheres in @xmath653 centered at @xmath24 have positive ( not necessarily constant ) mean curvature with respect to the inward pointing normal vector .
by theorem [ thm5.7 ] , @xmath654 satisfies the curvature estimate @xmath657 for some constant @xmath232 independent of @xmath658 , where @xmath141 is given by ( [ eq : sigma ] ) ( see the proof of item 4 of corollary [ corrs ] for a similar argument ) .
it follows from this curvature estimate that blow - up rescalings of @xmath654 of the form @xmath659\subset \lambda_n [ b_n(p , r_0)-\{p\}]$ ] for @xmath302 , have subsequences that converge to a weak cmc foliation @xmath500 of @xmath645 , which gives item ( 2a ) of the corollary , except for the condition that the limit cmc foliation @xmath660 is non - flat , whose proof we postpone for the moment .
fix @xmath195 small , satisfying the properties previous to the assertion . by the main theorem in @xcite ,
there exists a uniform bound for the second fundamental form of any complete @xmath0-surface in @xmath661 whose two - sided cover is stable , which is independent of the value of @xmath0 . in particular , @xmath662 is also bounded from above by a universal constant @xmath663 .
now choose @xmath664 small enough so that the distance spheres @xmath268 , @xmath665 $ ] , all have mean curvature strictly greater than @xmath666 , and suppose that @xmath667 contains a complete @xmath0-surface @xmath226 whose two - sided cover is stable .
then , there exists a point @xmath385 in the closure of @xmath226 which is at maximal distance from @xmath24 , and a sequence of small intrinsic open disks @xmath668 that converge to an open @xmath0-disk @xmath669 passing through @xmath385 .
as @xmath669 lies in the closure of @xmath226 , then @xmath669 lies at one side of the distance sphere @xmath670 at @xmath385 .
the mean comparison principle applied to @xmath669 and @xmath670 gives a contradiction as the mean curvature of @xmath670 is strictly less than @xmath0 .
this proves the assertion , after relabeling @xmath671 by @xmath198 .
[ ass6.8 ] take @xmath195 as in assertion [ ass6.8new ] .
let @xmath672 be the supremum of the mean curvatures of the leaves of @xmath7 that intersect @xmath673 .
let @xmath674 where @xmath675 is the absolute mean curvature of @xmath34 .
then , the closure @xmath492 in @xmath653 of any @xmath676 is a compact immersed surface .
take @xmath676 . as @xmath34 is a leaf of a weak cmc foliation ,
item 4 of corollary [ corrs ] implies that the weak @xmath677-lamination @xmath678 of @xmath679 extends across @xmath24 to a weak @xmath677-lamination of @xmath653 , all whose leaves are complete ( note that the extrinsic distance from the lamination @xmath492 to @xmath673 is positive ) .
if @xmath34 were not proper , then the lamination @xmath492 would contain a complete leaf whose two - sided cover is stable , which contradicts assertion [ ass6.8new ] .
hence , @xmath34 is proper and so , @xmath492 is a connected , compact immersed @xmath677-surface in @xmath653 . first observe that if a codimension - one cmc foliation @xmath680 of a riemannian manifold @xmath2 is transversely orientable of a manifold @xmath2 is called _ transversely orientable _ if @xmath2 admits a continuous , nowhere zero vector field whose integral curves are transverse to the leaves of @xmath680 . ] , then the leaves of @xmath7 are embedded ; this follows from the fact that locally around a point @xmath385 of self - intersection of a leaf @xmath34 of @xmath7 with itself , @xmath34 consists of two small embedded disks tangent at @xmath385 with non - zero opposite mean curvature vectors at @xmath385 .
this contradicts the fact that the mean curvature vector of @xmath34 equals @xmath681(up to sign ) , where @xmath682 is a continuous unitary vector field on @xmath2 orthogonal to the leaves of @xmath7 .
we now prove the assertion .
since @xmath683 is simply connected , then @xmath684 $ ] is transversely oriented .
thus , assertion [ ass6.8 ] and the observation in the last paragraph imply that if @xmath676 , then @xmath492 is a compact immersed surface and @xmath34 is embedded .
so it remains to show that for @xmath198 sufficiently small , every leaf @xmath34 in @xmath685 with @xmath530 satisfies that @xmath492 is embedded at @xmath24 .
arguing by contradiction , suppose that for each @xmath190 , there exists a leaf @xmath174 in @xmath686 such that the compact immersed surface @xmath687 is embedded outside @xmath24 and @xmath24 is a point of self - intersection of @xmath687 .
in particular , @xmath688 has non - zero mean curvature and the mean curvature vectors of @xmath688 point in opposite directions at the two points of the abstract surface @xmath688 occurring at the self - intersection point @xmath24 ; this means that we can consider two local intrinsic disks @xmath689 , that intersect only at @xmath24 , and which have opposite mean curvatures with respect to a fixed orientation of their common tangent planes at @xmath24 . in particular , for some @xmath690 large , the surface @xmath691 intersected with @xmath692 consists of two disks , each of whose boundary curves lies in @xmath693 ; let @xmath694 be one of these two disks .
choose @xmath695 such that the absolute mean curvature of @xmath696 is greater than the absolute mean curvature of @xmath694 .
then , @xmath697 are tangent to @xmath694 at @xmath24 and lie on the same side of @xmath694 , since @xmath698 is connected and lies in one of the two closed complements of @xmath694 in @xmath692 .
but then at the point @xmath24 , the disk @xmath694 lies on the mean convex side of one of the disks @xmath699 , which contradicts the mean curvature comparison principle since the absolute mean curvature of @xmath694 is less than that of @xmath698 .
this contradiction finishes the proof of the assertion .
arguing by contradiction , suppose that @xmath701 .
note that the mean convex side of the compact embedded surface @xmath492 is @xmath702 ( this follows from considering the innermost distance sphere @xmath703 such that @xmath704 and comparing the mean curvature vectors of @xmath492 and of @xmath268 ) . as @xmath7 does not have any singularities in @xmath705
, then there exists a leaf @xmath507 of @xmath7 which maximizes the mean curvature function of @xmath7 restricted to @xmath705 . if @xmath706 , then proposition 5.4 in @xcite implies that @xmath34 is stable , which contradicts assertion [ ass6.8new ] .
if @xmath707 , then we can apply the same proposition 5.4 in @xcite on the mean convex side of @xmath507 , which is strictly contained in @xmath705 to contradict assertion [ ass6.8new ] again .
this proves the assertion .
suppose first that we are in case ( a ) and pick @xmath676 .
the restriction @xmath711 is strictly positive as @xmath34 is compact and @xmath712 . for simplicity
, we will assume that @xmath68 is a morse function , although this is not strictly necessary ( also , one could perturb slightly @xmath68 to a morse function so that the argument that follows remains valid ) .
let @xmath713 be the restriction of @xmath714 to @xmath715 . if @xmath34 is not topologically a sphere , then @xmath715 is not a closed topological ball and thus , there exists a critical point @xmath385 of @xmath68 in @xmath34 such that @xmath716 $ ] is not homeomorphic to @xmath717 $ ] for all sufficiently small @xmath151 .
at such a critical point @xmath385 , @xmath34 is tangent to the distance sphere @xmath718 at @xmath385 and the mean curvature vector of @xmath718 points outward from the mean convex side of @xmath34 at @xmath385 ( as @xmath68 is assumed to be a morse function , then @xmath68 has index 1 at @xmath385 ) . by contradiction
, suppose that the assertion fails in this case ( a ) .
thus there exists a sequence @xmath237 , leaves @xmath719 , none of which is a sphere , and critical points @xmath720 of @xmath721 such that the mean curvature vector of @xmath174 at @xmath385 points outward @xmath722 at @xmath214 .
after rescaling by @xmath723 and extracting a subsequence , the weak cmc foliations @xmath724 converge as @xmath209 to a weak cmc foliation @xmath725 of @xmath255 , whose leaves have closures being spheres or planes by theorem [ thmspheresintrod ] . furthermore , @xmath725 contains a leaf @xmath726 that passes through the limit point @xmath727 as @xmath209 of the points @xmath728 .
in particular , the closure @xmath729 of @xmath726 is a plane or a sphere passing through @xmath730 .
as @xmath174 is tangent to @xmath731 at @xmath214 , then @xmath726 is tangent to @xmath732 at @xmath730 .
therefore , we have three possibilities for @xmath726 : 1 .
@xmath726 is a plane passing through @xmath730 , orthogonal to this position vector .
@xmath726 is a sphere passing through @xmath730 with mean curvature vector pointing in the direction of @xmath730 .
3 . @xmath729 is a sphere passing through @xmath730 with mean curvature vector pointing in the direction of @xmath733 . also , observe that as the mean curvature vectors of @xmath174 , @xmath734 at their common point @xmath214 point in opposite directions , then case 3 above can not occur .
cases 1 , 2 can not occur either , as then the distance function in @xmath3 to @xmath730 has a non - degenerate ( global ) minimum at @xmath730 , which contradicts that @xmath205 has index 1 at @xmath214 for all @xmath57 .
hence the assertion holds for @xmath198 sufficiently small in case ( a ) .
we next finish the proof of theorem [ corol6.6 ] .
assertions [ ass6.9bis ] and [ ass6.10 ] imply that item ( 2b ) of the corollary holds .
it only remains to show that given @xmath649 with @xmath249 , the weak cmc foliation @xmath500 of @xmath645 obtained as a limit of a subsequence of @xmath735 $ ] in the first paragraph of the proof of this corollary , is non - flat .
fix @xmath195 small enough so that assertion [ ass6.10 ] holds .
assume that @xmath57 is chosen sufficiently large so that @xmath738 and so that any leaf of @xmath685 that is contained in @xmath739 has mean curvature greater than @xmath740 ( recall that @xmath741 was defined in the statement of assertion [ ass6.8 ] ) ; this choice of @xmath57 is possible since every leaf in @xmath685 contained in @xmath742 is compact embedded sphere and for @xmath57 large , the mean curvature such a leaf is at least @xmath743 . consider the non - empty open set @xmath744 , \quad \mbox{where}\quad { \cal a}_-=\ { l\in { \cal a}(r_0)\ | \ \delta ( l)\subset \overline { b}_n(p,{\textstyle \frac{1}{\l _ n}})\ } .\ ] ] we claim that there exists @xmath745 such that @xmath746 and @xmath747 . to see this ,
take a point @xmath748 and a sequence of points @xmath749 converging to @xmath385 as @xmath750 .
then , @xmath751 lies in the interior of @xmath752 for some @xmath753 with @xmath754 . as the norm of the second fundamental form of leaves of @xmath7 is uniformly bounded locally around @xmath385 , then all @xmath755 can be locally expressed around @xmath385 as graphs of uniform size over their tangent planes at @xmath751 . by the arzel - ascoli theorem
, a subsequence of these graphs converges to a graph of constant mean curvature at least @xmath740 , and a monodromy argument shows that this limit graph is contained in a spherical leaf @xmath745 with @xmath756 , so our claim is proved .
we will show that @xmath507 intersects @xmath757 ( so the assertion will be proved by taking @xmath758 ) . arguing by contradiction ,
suppose @xmath759 .
now consider the family @xmath760 which we claim is non - empty .
the arguments that demonstrate that this set is non - empty are more or less the same as the ones given in the previous paragraph ; one shows that for some point @xmath761 there is a sequence of points @xmath762 converging to @xmath385 and contained in respective leaves @xmath755 with mean curvatures converging to the mean curvature of @xmath507 , and so the leaves @xmath755 lie in @xmath763 for @xmath462 large .
since the family of balls @xmath764 is totally ordered under inclusion , it follows that for every @xmath765 and @xmath766 , we have @xmath767\neq \mbox{\o } $ ] . since @xmath757 is compact ,
we conclude that there exists a point @xmath768 such that @xmath769.\ ] ] as @xmath7 is a weak cmc foliation , there exists a leaf @xmath770 passing through @xmath768 , although @xmath771 might not be unique .
let @xmath772 be the collection of leaves of @xmath7 passing through @xmath768 .
since @xmath773 $ ] and @xmath774 , then every such leaf @xmath775 lies in @xmath685 ( and thus , the closure of @xmath771 is a sphere ) .
also , every two leaves @xmath776 intersect tangentially , with one at one side of the other ; therefore either @xmath777 or vice versa . as there exists a uniform local bound around @xmath768 for the norms of the second fundamental forms of leaves in @xmath772
, we conclude that the union of the leaves in @xmath772 is a compact set of @xmath2 .
thus , we can choose @xmath778 with the following properties : as before , there exists @xmath782 such that @xmath783 .
note that @xmath784 ( otherwise @xmath785 , which contradicts ( p2 ) above ) .
therefore , ( p2 ) implies that @xmath786 . in turn , this implies by ( [ ass7.15a ] ) that @xmath787 . as @xmath782 ,
then by definition of @xmath763 we have @xmath788 and so , @xmath789 . by definition of @xmath790
, this means that @xmath791 , from where int@xmath792\subset a_-$ ] .
this is impossible , as @xmath793 which contradicts ( p1 ) .
this contradiction finishes the proof of the assertion .
consider a sequence @xmath649 with @xmath249 . as we have explained , there exists @xmath195 small such that after extracting a subsequence , the weak cmc foliations @xmath794 $ ] converge as @xmath209 to a weak cmc foliation @xmath725 of @xmath255 , and it only remains to show that @xmath725 is not flat .
consider the every @xmath190 the distance sphere @xmath757 . by assertion [ ass7.15 ] , for @xmath57 sufficiently large there exists @xmath795 such that @xmath796 and @xmath797 .
thus , the rescaled leaves @xmath798 of the weak cmc laminations @xmath735 $ ] stay inside the closed ball @xmath799 with a point in @xmath800 , and so , the limit foliation @xmath500 has a leaf contained in the closed unit ball of @xmath801 with some point in the unit sphere ; in particular , @xmath725 is non - flat .
w. h. meeks iii , j. prez , and a. ros .
stable constant mean curvature surfaces . in _ handbook of geometrical analysis _ , volume 1 , pages 301380 . international press , edited by lizhen ji ,
peter li , richard schoen and leon simon , isbn : 978 - 1 - 57146 - 130 - 8 , 2008 .
mr2483369 , zbl 1154.53009 .
j. prez and a. ros .
properly embedded minimal surfaces with finite total curvature . in _ the global theory of minimal surfaces in flat spaces _ ,
pages 1566 .
lecture notes in math 1775 , springer - verlag , 2002 .
g. p. pirola , editor .
mr1901613 , zbl 1028.53005 . | in this paper we generalize the local removable singularity theorem in @xcite for minimal laminations to the case of weak @xmath0-laminations ( with @xmath1 constant ) in a punctured ball of a riemannian three - manifold .
we also obtain a curvature estimate for any weak cmc foliation ( with possibly varying constant mean curvature from leaf to leaf ) of a compact riemannian three - manifold @xmath2 with boundary solely in terms of a bound of the absolute sectional curvature of @xmath2 and of the distance to the boundary of @xmath2 .
we then apply these results to classify weak cmc foliations of @xmath3 and @xmath4 with a closed countable set of singularities .
_ mathematics subject classification : _
primary 53a10 , secondary 49q05 , 53c42 _ key words and phrases : _ minimal surface , @xmath0-surface , stability , curvature estimates , finite total curvature , minimal lamination , weak @xmath0-lamination , weak cmc foliation , removable singularity .
the classification of cmc foliations of @xmath3 and @xmath5 with countably many singularities .2 in william h. meeks iii joaqun p ' erez antonio ros |
let @xmath0 be a non - empty _ lattice polygon _ ,
i.e. the convex hull of a finite number of lattice points in @xmath1 , and consider a _ lattice direction _ @xmath2 , i.e. a non - zero primitive vector . the _ lattice width of @xmath3 in the direction @xmath4 _ is @xmath5 the _ lattice width _ of @xmath3 is defined as @xmath6 . throughout this paper
we will assume that @xmath3 is two - dimensional , hence @xmath7 .
a lattice direction @xmath4 that satisfies @xmath8 is called a _ lattice width direction _ of @xmath3 .
two lattice polygons @xmath3 and @xmath9 are called _ ( unimodularly ) equivalent _ if and only if there exists a _ unimodular transformation _
@xmath10 , i.e. a map of the form @xmath11 such that @xmath12 .
equivalent lattice polygons have the same lattice width . the lattice width of a polygon can be seen as a specific instance of the more general notion of _ lattice size _ , which was introduced in @xcite .
let @xmath13 be a subset with positive jordan measure .
then the lattice size @xmath14 of a non - empty lattice polygon @xmath3 is the smallest @xmath15 for which there exists a unimodular transformation @xmath10 such that @xmath16 .
note that @xmath17 , where @xmath18 $ ] .
this paper is concerned with polygons @xmath3 that are _ minimal _ in the following sense : @xmath19 for each lattice polygon @xmath20 .
equivalently , a two - dimensional polygon @xmath3 is minimal if and only if for each vertex @xmath21 of @xmath3 , we have that @xmath22 , where @xmath23 . our main result is a complete classification of minimal polygons up to unimodular equivalence , see theorem [ thm_clas ] . as a corollary
, we provide a sharp upper bound on the number of lattice points of these minimal polygons .
first , we show in lemma [ lemma2 ] that each minimal polygon @xmath3 satisfies @xmath24 , where @xmath25 .
the latter can also be proven using results on lattice width directions of interior lattice polygons ( see ( * ? ? ?
* lemma 5.3 ) ) , but we choose to keep the paper self - contained and have provided a different proof .
moreover , we use the technical lemma [ newlemma ] in the proofs of both lemma [ lemma2 ] and theorem [ thm_clas ] . in the joint paper @xcite with castryck and demeyer
, we study the betti table of the toric surface @xmath26 for lattice polygons @xmath3 .
in particular , we present a lower bound for the length of the linear strand of this betti table in terms of @xmath27 , which we conjecture to be sharp . for showing this conjecture for polygons of a fixed lattice width
, it essentially suffices to prove it for the minimal polygons ( see ( * ? ? ?
* corollary 5.2 ) ) .
hence , theorem [ thm_clas ] allows us to check the conjecture using a computer algebra system .
of course , the question of classifying minimal polytopes can also be asked in higher dimensions .
for instance , it can be shown that each three - dimensional minimal polytope @xmath28 with @xmath29 is equivalent to a tetrahedron of the form @xmath30 with @xmath31 and @xmath32 .
these include the reeve tetrahedrons ( where @xmath33 ) . for comparison
, there is only one minimal polygon with lattice width one up to equivalence , namely the standard simplex @xmath34 . in all dimensions
@xmath35 , among the minimal polytopes we will find back the so - called _ empty lattice simplices _ @xmath36 , i.e. convex hulls of @xmath37 lattice points without interior lattice points . if @xmath38 , not all empty lattice simplices have lattice width @xmath39 .
for more information , see @xcite .
throughout this section , we will use the notations from section [ sec_1 ] .
the following result appears already in ( * ? ? ?
* remark following lemma 5.2 ) , but can be proven in a shorter way .
if @xmath4 and @xmath44 do not form a @xmath45-basis of @xmath1 , we take a primitive vector @xmath46 such that @xmath4 and @xmath47 form a @xmath45-basis .
let @xmath48 be lattice points of @xmath3 such that @xmath49 .
write @xmath50 with @xmath51 and @xmath52 .
now @xmath53 so @xmath54 . after applying a unimodular transformation
, we may assume that @xmath55 and @xmath56 , and that @xmath3 fits into @xmath57 , hence @xmath43 .
since @xmath61 , we have that either @xmath63 by replacing @xmath4 by @xmath64 , we may assume that we are in the first case .
moreover , we may choose @xmath4 such that the difference @xmath65 is minimal but greater than 1 , and such that @xmath60 .
we apply a unimodular transformation so that @xmath66 and @xmath67 .
let @xmath68 ( resp .
@xmath69 ) be the smallest ( resp .
greatest ) @xmath70-coordinate occurring in @xmath71 .
note that @xmath72 and @xmath73 , hence @xmath74 and @xmath75 .
define the cone @xmath76 . since @xmath77 and @xmath74 , the polygon @xmath3 is contained in a cone @xmath78 for some @xmath79 . using the unimodular transformation @xmath80
, we may assume that @xmath81 , i.e. @xmath82 .
in fact , we then have that @xmath83 if @xmath84 , we have @xmath85 .
the strict inequality @xmath86 is impossible as the horizontal width @xmath87 would be less than @xmath88 .
so we have that @xmath89 and @xmath90 .
since @xmath91 for @xmath92 , we must have @xmath93 .
this is equivalent to @xmath94 via @xmath95 . from now on , assume that @xmath96 . then @xmath97 which means that either @xmath98 we can reduce to the latter case using the transformation @xmath99 .
in fact , we can keep subdividing this cone until we find a cone @xmath100 containing @xmath3 that does not contain any lattice point with @xmath70-coordinate in @xmath101 .
let @xmath102 be such that @xmath100 passes in between @xmath103 and @xmath104 .
then @xmath105 if @xmath106 ( resp .
@xmath107 ) is the smallest ( resp .
greatest ) @xmath108-coordinate occurring in a lattice point of @xmath71 , then @xmath109 and @xmath110 , so @xmath111 .
but this means that @xmath112 and @xmath113 contradicting the minimality of @xmath4 . by lemma [ lemma1 ]
, we only have to show that there are two linearly independent lattice width directions .
suppose that @xmath4 is a lattice width direction and that @xmath115 such that @xmath116 .
now let @xmath21 be a vertex of @xmath3 different from @xmath48 .
by minimality of @xmath3 , we have that @xmath117 .
that means there exists a direction @xmath44 such that @xmath118 . because @xmath119 and @xmath120 are still in @xmath71 , @xmath44 can not be @xmath4 or @xmath64 , so @xmath44 must be linearly indepentent of @xmath4 .
if @xmath121 , we are done . if @xmath122 , then by lemma [ newlemma ] , @xmath3 is equivalent to @xmath123 . 1 .
@xmath124 where @xmath125 satisfy @xmath126 ; 2 .
@xmath127 where @xmath128 satisfy @xmath129 and @xmath130 ; 3 .
@xmath131 with @xmath132 , @xmath133 , @xmath134 ; 4 .
@xmath135 with @xmath132 , @xmath136 , @xmath137 ; 5 .
@xmath138 with @xmath132 , @xmath139 , @xmath140 ; [ remarkh ] see figure [ figure types ] for a picture of the five types .
the minimal polygons appearing in the types ( t3 ) , ( t4 ) and ( t5 ) are inscribed in the hexagon @xmath141 this is also the case for the triangles of type ( t1 ) with @xmath142 ( where we allow @xmath143 ) and for the quadrangles of type @xmath144 with @xmath145 . if @xmath146 , then @xmath3 consists of a single point and it is of shape @xmath147 .
so assume @xmath148 .
because of lemma [ lemma2 ] , we may assume that @xmath149\times [ 0,d]$ ] .
moreover , we may assume that @xmath150 since @xmath94 is of type @xmath147 .
let @xmath21 be any vertex of @xmath3 .
by lemma [ newlemma ] , if @xmath60 for some primitive vector @xmath59 , then @xmath151 and @xmath152 , hence @xmath4 is a lattice width direction . by minimality , we know that there always exists a lattice direction @xmath4 satisfying @xmath60 . we claim that we can always take @xmath153 . indeed
, suppose that @xmath154 satisfies @xmath155 after a unimodular transformation , we may assume that @xmath156 , hence @xmath157 .
using a similar trick as in lemma [ lemma1 ] , we get that @xmath158 , which proves the claim .
let @xmath159 be set consisting of vectors @xmath160 for which there exists a vertex @xmath21 of @xmath3 with @xmath60 .
if @xmath161 , then @xmath3 has @xmath162 different lattice width directions , namely @xmath163 and @xmath164 . by (
* lemma 5.2(v ) ) or @xcite , this means that @xmath165 for some even @xmath88 , hence it is of type @xmath144 .
if @xmath166 , we claim that @xmath3 is of type @xmath147 or @xmath144 . indeed , for every vertex @xmath21 of @xmath3 , we have that either @xmath167 or @xmath168 is smaller than @xmath88 . in particular
, this means that there has to be a side of @xmath57 with @xmath21 as its only point in @xmath3 .
one then easily checks the claim : if @xmath3 is a triangle , then it will be of type @xmath147 ; if it is a quadrangle , then it is of type @xmath144 . from now on , suppose that @xmath159 is not equal to @xmath169 or @xmath170 , hence @xmath171 or @xmath172 .
we can suppose that @xmath172 by using the transformation @xmath173 if necessary .
hence , for each vertex @xmath21 of @xmath3 , there is a vector @xmath174 with @xmath60 .
since @xmath175 , there exists an integer @xmath176 such that @xmath177 $ ] for all @xmath178 .
if @xmath143 , then @xmath3 is a triangle whose vertices are vertices of @xmath57 , so it is of the form @xmath147 .
now assume that @xmath179 , hence @xmath3 is contained in the hexagon @xmath180 from remark [ remarkh ] .
each side of @xmath180 contains at least one lattice point of @xmath3 , and if it contains more than one point , it is also an edge of @xmath3 .
otherwise , there would be a vertex @xmath21 lying on exactly one side of @xmath180 , while not being the only point of @xmath3 on that side of @xmath180 .
but then there is no @xmath181 with @xmath60 ( as every side of @xmath180 contains a point of @xmath71 ) , a contradiction .
denote by @xmath182 the set of sides that @xmath3 and @xmath180 have in common .
then @xmath182 can not contain two adjacent sides @xmath183 : otherwise for the vertex @xmath184 , each side of @xmath180 would have a non - empty intersection with @xmath71 , contradicting the fact that there is a @xmath181 with @xmath118 .
assume that @xmath185 and take @xmath186\in\mathcal{s}$ ] .
its adjacent sides of @xmath180 contain no points of @xmath3 except from @xmath187 and @xmath188 .
this implies that @xmath189 where @xmath183 are opposite edges of @xmath180 , and that @xmath3 is the convex hull of these two edges .
hence @xmath3 is equivalent to the quadrangle @xmath190 , which is of type @xmath144 .
if @xmath182 consists of a single side @xmath191 , we may assume that @xmath192 $ ] is the bottom edge of @xmath180 .
let @xmath193 ( resp .
@xmath194 ) be the vertex of @xmath3 on the upper left diagonal side ( resp .
the right vertical edge ) of @xmath180 .
if @xmath193 is also on the top edge of @xmath180 ( i.e. @xmath195 ) , then @xmath3 has only four vertices , namely @xmath196 .
applying the transformation @xmath197 , we end up with a quadrangle of type @xmath144 . by a similar reasoning ,
if @xmath194 is on the top edge of @xmath180 ( i.e. @xmath198 ) , we end up with type @xmath144 .
if neither @xmath193 nor @xmath194 are on the top edge of @xmath180 , then there is a fifth vertex @xmath199 on that top edge , and we are in case @xmath200 .
the only remaining case is when @xmath201 , hence each edge of @xmath180 contains only one point of @xmath3 .
if @xmath180 and @xmath3 have no common vertex , then @xmath3 is of type @xmath202 .
if they share one vertex , we can reduce to type @xmath203 using a transformation if necessary .
note that two common vertices of @xmath180 and @xmath3 can never be connected by an edge of @xmath180 as that edge would be in @xmath182 , so there are at most three common vertices .
if there are three shared vertices , then @xmath3 is a triangle of type @xmath147 , again using a transformation if necessary .
so assume @xmath180 and @xmath3 share two vertices .
together these two points occupy four edges of @xmath180 and each of the other two edges of @xmath180 ( call them @xmath204 and @xmath205 ) contains exactly one vertex of @xmath3 .
take two pairs of opposite sides of @xmath180 ( so four sides in total ) that together contain @xmath204 and @xmath205 , then they contain all vertices of @xmath3 : since any common vertex of @xmath180 and @xmath3 lies on two sides of @xmath180 , they can not lie both on the sides we did nt choose , as they are parallel .
we can find a unimodular transformation mapping these sides into the four sides of @xmath57 , hence @xmath3 is of type @xmath144 . from the classification in theorem [ thm_clas ]
, one can easily deduce the following result from @xcite : @xmath206 for each lattice polygon @xmath0 , and equality holds for minimal polygons of type @xmath147 with @xmath88 even and @xmath207 . for odd @xmath88
, this inequality can be sharpened to @xmath208 , and equality holds for minimal polygons of type @xmath147 with @xmath209 and @xmath210 .
note that there exist minimal polygons attaining the bound ( see figure [ upperbound ] ) : the simplex @xmath214 is of type @xmath147 and has @xmath215 lattice points , and the quadrangle @xmath216 is of type @xmath144 and has @xmath217 lattice points .
now let s show that we indeed have an upper bound . if @xmath3 is minimal of type @xmath144 , @xmath203 or @xmath202 , then @xmath218 , since there are at most @xmath162 lattice points of @xmath3 on the boundary of @xmath57 and all the others are in @xmath219 this also holds for triangles of type @xmath147 with @xmath108 and @xmath70 non - zero .
if @xmath3 is of type @xmath200 , we obtain the same upper bound @xmath217 after applying a unimodular transformation that maps the bottom edge of @xmath3 to the left upper diagonal edge of @xmath180 .
we are left with triangles of type @xmath147 with either @xmath108 or @xmath70 zero .
assume that @xmath220 ( the case @xmath221 is similar ) .
then @xmath3 has the edge @xmath222 $ ] in common with @xmath57 and its other vertex is @xmath223 . for each @xmath224 , the intersection of @xmath3 with the horizontal line on height
@xmath225 is a line segment of length @xmath226 , hence it contains at most @xmath227 lattice points .
so in total , @xmath3 has at most @xmath228 lattice points .
99 m. barile , d. bernardi , a. borisov and j .- m .
kantor , _ on empty lattice simplices in dimension 4 _ , proc .
139 ( 2011 ) , 42474253 .
w. castryck and f. cools , _ linear pencils encoded in the newton polygon _
, accepted for publication in int .
res . not .
w. castryck and f. cools , _ the lattice size of a lattice polygon _ ,
theory ser .
a , vol . 136
( 2015 ) , 64 - 95 .
w. castryck , f. cools , j. demeyer and a. lemmens , _ computing graded betti tables of toric surfaces _ , preprint 2016 ( arxiv:1606.08181 ) .
j. draisma , t. b. mcallister and b. nill , _ lattice - width directions and minkowski s @xmath229-theorem _ , siam j. discrete math . 26
( 2012 ) , 11041107 . c. haase and g.m .
ziegler , _ on the maximal width of empty lattice simplices _ , europ
. j. combinatorics 21 ( 2000 ) , 111 - 119 .
a. seb , _ an introduction to empty lattice - simplices _ , lecture notes in computer science 1610 , ( cornujols , burkard , woeginger eds . ) , springer , june 1999 .
l. fejes tth and e. makai jr .
, _ on the thinnest non - separable lattice of convex plates _ ,
studia scientiarum mathematicarum hungarica 9 ( 1974 ) , 191 - 193 . | we classify the unimodular equivalence classes of inclusion - minimal polygons with a certain fixed lattice width . as a corollary
, we find a sharp upper bound on the number of lattice points of these minimal polygons .
+ _ msc2010 : _ primary 52b20 , 52c05 , secondary 05e18 |
although robots are finding their place in our society as artificial pets , entertainers , and tools for therapists , current technologies have yet to reach the full emotional and social capabilities necessary for rich and robust interaction with human beings . to achieve this potential
, research must imbue robots with the emotional and social capabilities -both verbal and non - verbal- necessary for rich and robust interaction with human beings .
this article describes research in which robotic heads can model natural face - to - face communication with individuals .
human face - to - face communication is based on multiple communication channels including auditory and visual .
facial expressions and gaze direction are the most important visual channels in human face - to - face communication . these channels of communications are considered in our research . despite significant progress towards development of realistic robotic heads ,
a number of problems remain to be solved .
social robots such as paro @xcite have the robustness and cost effectiveness for large scale , but lack the sophistication for deep social interaction .
other robots such as kismet @xcite exhibit facial expressions and head and ear movement using mechanical components , however , once these mechanical platforms are built , they are fixed and can not be readily modified . on the other hand ,
more human - like robots such as simon @xcite possess profound capabilities for social interaction , but due to large number of actuators in their mechatronic faces , they are expensive and maintenance - intensive for large - scale trials .
another potential problem in the design of robotic heads is the `` uncanny valley '' effect @xcite , where the effect of aesthetic design of a robot may influence the user s experience , perception , and acceptance of the robot .
given the tremendous effort required to develop robot heads and the number of design choices that must be made , including aesthetic face design , mechanical design and construction , hardware implementation , etc , it is often difficult to redesign or rebuild the head based on user experiences and limitations discovered during research . an alternative approach that overcomes many of these problems is to use state - of - the - art character animation technologies to create 3d avatar models that produce natural speech and facial expressions , and project these models onto a robotic face that can move like a human head .
projecting an animated facial character on a 3d human - like head has several advantages over displaying it on a 2d flat screen .
the first advantage is the physical embodiment and movement of the 3d head that makes the robot be more believable and better perceived by the users @xcite .
the other advantage is perception of mutual eye gaze which plays a vital role in face - to - face communication .
it is known that , the perception of 3d objects that are displayed on 2d surfaces are influenced by the mona lisa effect @xcite . in other words ,
the orientation of the object in relation to the observer will be perceived as constant regardless of observer s position @xcite .
for instance , if the 2d projected face is gazing forward , mutual gaze is perceived with the animation , regardless of where the observer is standing / sitting in relation to the display .
this effect is significantly reduced by projecting the face on a 3d human - like head .
this article describes our current progress towards designing , manufacturing , and evaluating a lifelike robotic head , called expressionbot , with the capability of showing facial expressions , visual speech , and eye gaze .
our eventual goal is to provide the research community with a low - cost portable facial display hardware equipped with a software toolkit that can be used to conduct research leading to a new generation of robotic heads that model the dynamics of face - to - face communication with individuals in different social , learning , and therapeutic contexts .
the remainder of this paper is organized as follows .
section [ sec : relatedwork ] reviews the related work on the design of social robotic faces .
section [ sec : mechanism and design ] presents the design , development , and production of the proposed robotic face .
section [ sec : hri experiments and results ] presents the results of our pilot studies on user experiences with the robotic face , while section [ sec : conclusion ] presents our conclusions .
there are many designs for robotic faces ranging from 2d graphical avatars to mechanically controlled robotic faces .
these designs fit into four main categories : mechatronic faces , android faces , onscreen avatars , and light - projected ( retro - projected ) physical avatars .
these categories are discussed in the following .
mechatronic robotic faces are physically implemented robots that use mechatronic devices and electric actuators to control facial elements . kismet @xcite is one the first and famous expressive mechatronic robots with many features such as eye lids , eye brows , lips and even expressive ears .
another example is the philips icat @xcite which has a cat - like head and torso with mechanical lips , eye lids , and eye brows .
mechatronic robotic faces have the advantage of being 3d , but they are inflexible , unrealistic , and have limited ability to display facial expressions and speech .
these faces look very much like a stereotypical robot rather than a human face .
android faces are other physically implemented robots that are originated from animatronics .
they have a larger number of mechatronic actuators controlling a flexible elastic skin ; therefore they look more realistic and seem more like a human rather than a robot .
example of android faces are albert hubo @xcite , hrp-4c @xcite , and zeno @xcite .
it is an interesting research question as to whether android faces that closely model human looks and behaviors will enter the uncanny valley as their realism mimics humans . due to larger number of actuators and their interaction with skin ,
they look more expressive than mechatronic faces .
however , they are mechanically very complex , expensive to design , build and maintain .
the on screen avatar class , such as grace @xcite and second life @xcite are the simplest and earliest robotic faces .
animations for these models can be made by developing a model for each expression , morphing between them , and then rendering the result to a computer screen . despite their low cost and high flexibility , they naturally have several limitations due to using a flat display as an alternative to a three dimensional physical head .
for example , aside aesthetic unpleasantness , they suffer from lack of establishing mutual gaze ( due to mona lisa effect ) and physical embodiment that both play vital roles in natural and realistic face - to - face communication . the final category , and the focus of our research , is the light - projected physical avatars ; these consist of translucent 3d masks with 2d/3d avatars projected onto them .
since the robotic face is projected onto a mask , the robotic face can range from cartoon - like to photorealistic .
light - projected physical avatars are thus a highly flexible research tool , relative to mechatronic and android faces .
factors such as engagement , embodiment , believability , credibility and realism can be investigated based on the appearance and behaviors of the 3d animated models , and the robotic head and neck movement . moreover ,
such a system can avoid the mona lisa effect @xcite and hence users can correctly perceive the robot s eye gaze direction .
additional features of robotic avatars include relatively low development cost , low power consumption , potentially low weight and fast reaction .
one of the early examples of light - projected physical avatars is the dome robot @xcite where a cartoonish animated face is projected onto a dome - shaped mask .
the dome mask makes the image and display calibration process easy .
however , it lacks human face realism and the results appear cartoonish .
the lighthead robotic face @xcite is another example that also projects an animation onto a face - shaped translucent mask , resulting in a more realistic appearance .
it is capable of displaying a wide range of facial expressions and emotions . in
@xcite the authors presented a mask head robot , called mask - bot , that generates visual speech , head movements , facial expressions and eye movements .
then later in @xcite , mask - bot 2i was introduced with an automatic approach for projection calibration by using a series of gray - coded patterns in a calibration booth which supports interchangeable masks .
both mask - bot and mask - bot 2i use human talking head animation that is photo realistic , which is not as flexible as computer animation .
furhat robot @xcite is another human - like light - projected robotic head that utilizes computer animation to deliver facial movements . in @xcite , the designers of furhat had also a study on perception of animation s gaze on 3d projected against flat screens and demonstrated the limitations of flat screens in delivering accurate direction of gaze due to mona lisa effect , which limits having situated , multiparty interaction in onscreen avatars .
furhat uses a pan - tilt unit for the neck system which has only 2 dof pitch and yaw . also furhat uses a mirror instead of a fish eye lens which makes the robot to have a larger form factor .
socibot @xcite is a commercialized retro - projected social robotic head .
it has an advanced neck system with 3 dofs with force / torque sensory measurement that can capture and replicate the head movement trajectory .
it uses a universal mask to show different avatar characters . compared to our system ,
the quality of animation and facial expression generation are low .
the visual speech is also nave ( i.e. open and close mouth ) .
moreover , their system is equipped with a speech synthesizer but it is not able to align recorded voices with accurate visual speech .
the expressionbot consists of three main sections , the neck control system , the display system and the animation application .
the neck system controls the projector and mask position allowing it to be rotated by the application to track faces and head gestures .
the display system consists of a small projector with a fish eye lens that projects the animation on a human like ( head shaped ) face mask .
the animation application displays a face animation along with speech and emotion to be projected on to the mask .
the neck mechanism of our existing prototype has three degrees of freedom ( dof ) providing a total of 150@xmath0 of yaw ( x - y plane ) , 30@xmath0 of pitch ( x - z plane ) and 30@xmath0 of roll ( y - z plane ) .
solidworks cad tool was used to design and maximize the range of motion , resulting in a light , compact , and quiet mechanism .
these design constraints were achieved using a 6@xmath16 footprint , low friction plastic gears and brushless servomotors .
also , the small footprint allows the neck and projector to be easily shrouded by the mask , and allows the user to control the distance from the mask to the lens in order to project the clearest possible image .
our system uses a dell dlp m110 portable projector .
the projector is capable of up to 300 ansi lumens under normal indoor illumination conditions , can display a maximum resolution of 1200@xmath1800 , and has a 10000:1 contrast ratio .
attached to the projector is a nikon fisheye converter fc - e8 which provides a viewing angle of approximately 183 degrees .
this allows the projector to display to the whole mask from a relatively close distance . to create the mask we designed a mold using the 3d model of the neutral face in autodesk maya .
we 3d printed this mold and used it to vacuum form a 1/8 inch sheet of white translucent acrylic plastic .
then , we added a metal band from top of the mask to the projector , which allows us to mount a wig on the robot s head .
this makes the expressionbot more aesthetically pleasant and natural , and covers the lights coming out from the sides of the mask due to fish eye lens wide projection - angle ( see fig . [
fig : proposed robot system ] ) .
we developed a face animation in c # .net for accurate natural visual speech and show expression based on multi - target morphing method @xcite .
recorded utterances are processed by the bavieca speech recognizer @xcite , which receives the sequence of words and the speech waveform as input , and provides a time - aligned phonetic transcription of the spoken utterance .
the aligned phonemes are represented using the international phonetic alphabet ( ipa ) , a standard that is used to provide a unique symbolic notational for the realization of phonemes in all of the world s languages @xcite . as ipa
is intended as a standard for the phonemic and phonetic representation of all spoken languages , having ipa in our system will allow us to add other languages easily as long as the speech recognizer is trained for that language . for a given language ,
visually similar phonemes are grouped into units called visemes .
for example the consonants /b/ , /p/ and
/m/ in the words `` buy , '' `` pie , '' and `` my '' form a single viseme class .
we categorized english phonemes into 20 viseme classes .
these classes represent the articulation targets that lips and tongue move to during speech production . a graphic artist designed 3d models of these viseme classes in maya .
figure [ fig : somevisemesexpression ] , demonstrates some visemes used in our animation system .
finally , natural visual speech is obtained by blending the proper models corresponding to each part of speech with different weights . to achieve a smooth and realistic look
, we used a kernel smoothing technique . during speech production ,
the avatar system receives the time - aligned phonetic input from bavieca system , converts the phonetic symbols into the corresponding visemes , which specifies the movements of the mouth and tongue , synchronized with the recorded or synthesized speech .
the algorithm models coarticulation by smoothing across adjacent phonemes . using
the kernel technique resulted in smoother and more natural looking animations ; however , when utterances included the labial phonemes /b/ , /m/ , /p/ , which are accompanied by lip closure , the smoothing algorithm prevented the lips from closing when the duration of the labial phoneme is very short ( e.g. , 5 msec . ) and the adjacent phoneme targets caused the lips to be open ( e.g. , // as in `` mama '' ) . to force lip closure for the labials
, we extended the duration of labial visemes to include the closure interval ( the period of relative silence before the sound is released , thus increasing the chance that at least one frame consisting of just the labial viseme will appear .
we designed the models in three portions : eyes , face and hair .
this design allows them to be interchangeable and customizable , and gives us the ability to design any number of characters to easily change the robot s appearance .
the system has the ability to control eye gaze independently of the visual speech and facial expression animation , and thus enables functionality to control eye gaze ( e.g. , in concert with face tracking ) . in order to blend the expressions with the lip movement
, the animation uses the following formula to generate facial expressions based on the current viseme and emotion morph targets : @xmath2 where @xmath3 represents the current viseme , @xmath4 is the desired expression model at the maximum intensity , @xmath5 is the neutral model .
the parameter @xmath6 $ ] is the intensity of the @xmath7 expression model @xmath8 .
the graphic artist designed 3d models of six basic expressions ( i.e. , anger , disgust , fear , joy , sadness and surprise ) in maya based on facial action coding system ( facs ) @xcite .
for example joy involves cheek raiser ( au 6 ) and lip corner puller ( au 12 ) and sadness involves inner brow raiser ( au 1 ) , brow lowerer ( au 4 ) and lip corner depressor ( au 15 ) . in order to blend the expressions with the lip movement , adding weight to the emotion morph targets without regard to the movements of the face caused by speech production will result in unnatural looking facial expressions .
for example , combining the surprise expression which causes the mouth to be fully open , conflicts with the production of phonemes like /b/ , /f/ and
/v/ that are produced with the lips closed or nearly closed .
combining the joy emotion with puckered mouth visemes such as /o/ will also result in visual speech and expressions that are not natural and are perceived as abnormal or creepy . to overcome this problem , we designed a table that provides a viseme weight factor and a maximum emotion weight for every viseme emotion combination .
these values are adjusted empirically for each combination .
we separated the facial expression morph targets into upper and lower face morph targets ; the upper face includes everything from the tip of the nose upwards .
the lower face includes the region below the nose ; mainly the lips , lower cheeks and chin .
this partitioning of the face enables us to adjust the weight of just the lower face morph target weights so that the upper face remains consistent with the morph targets of desired expressions .
in addition , for labial and labiodental visemes ( those for the letters m , b , p , f and v ) that require the avatar s lips to be closed or nearly closed to look natural , we developed visemes pre - blended with the open mouthed emotions .
these are used to replace the viseme and lower face expression when they come up in combination .
due to the projector and fish eye lens distortions , the resulting direct projection of the animated face model on the mask appears distorted ( see fig . [
fig : resultcalibration ] ) .
hence , we need to rectify the projection so the image appears undistorted and the facial regions ( e.g. , eyes , mouth ) of the facial model are projected to the desired position on the mask . in order to achieve a smooth animation displaying at 30 fps , we decide to distort the original maya models rather than rectifying the projection at run time in each frame .
+ [ fig : calibrationprocess ] assuming n is the neutral model in model coordinates ( fig .
[ fig : calibrationprocess].a ) , @xmath9 is the displayed projection in the screen coordinates where @xmath10 matrix is the multiplication of the world , view and projection matrices , respectively ( fig .
[ fig : calibrationprocess].b ) . assuming @xmath11 is the projected model on the mask ( fig .
[ fig : calibrationprocess].c ) , we aim to find @xmath12 ( the distorted neutral model ) such that @xmath13 looks undistorted on the mask , where @xmath14 . in order to estimate the function @xmath15
, we create a checkerboard in the screen coordinates ( fig .
[ fig : calibrationprocess].b ) and projected on the mask ( fig .
[ fig : calibrationprocess].c ) .
then , we define a piecewise homography mapping between the corresponding rectangles of the mask and the triangles displayed on the screen . to find the undistorted neutral model on the mask , @xmath16
, we apply an affine transformation to place the mold model , used to create the mask in the vacuum machine , on an image of the mask ( fig . [
fig : calibrationprocess].d ) .
afterwards , we apply the piecewise homography on the mold model and replace the corresponding vertices of the neutral model in the screen coordinates , @xmath17 , with distorted mold model to estimate @xmath18 .
we finally use @xmath19 as the neutral model in our application .
figure [ fig : resultcalibration ] shows the results of our calibration .
in order to evaluate individuals experiences and impressions of the expressionbot , we designed and conducted three experiments .
participants were 23 typical adults , 9 female and 14 males , with age range 18 - 51 years ( mean= 27.26 , sd=7.79 ) and a variety of ethnicities ( 19 caucasian , 2 asian , 2 hispanic ) .
hereafter , we refer to the 3d computer character on the computer screen as the screen - based agent , and the projection of the 3d model onto the robotic head as the physical agent . we used a 23 lcd display to display the screen - based agent at the same size as the physical agent .
the objective of the first experiment was to assess how accurately subjects were able to interpret the projected facial expressions .
participants watched the robotic agent and the screen - based agent in two different sessions randomly ( i.e. some participants observed the physical agent first while the others watched the screen - based agent first ) .
a series of six basic emotions ( joy , sadness , surprise , disgust , fear and anger ) were displayed in random order .
each expression was displayed one time for about 5 seconds .
the subject was then asked to select one of the six categories .
they could also respond `` none , '' if they were unable to assign the facial expression to one of the six categories .
tables [ tbl : confusion_physical_agent ] and [ tbl : confusion_screen - based_agent ] present confusion matrices of the intended and classified expressions displayed on the physical agent and the screen - based agent , respectively . comparing the percentages reported in these tables shows that the surprise and sad emotions were recognized perfectly ( 100% recognition rate ) in both agents by the participants .
the joy emotion was recognized perfectly when displayed on the physical agent but was recognized 92% of the time on the screen - based agent .
interestingly , anger was recognized correctly 85% of the time for the physical agent , and only 38% of the time for the screen - based agent .
disgust was classified as anger more often than it was classified as disgust for both agents , and fear was recognized correctly over 50% of the time in both agents , and confused most often with sadness . in sum , the results showed high recognition rates for joy , sadness and surprise in both agents , lower and similar recognition rates for disgust and fear in the two agents , and superior performance for anger when displayed on the physical agent .
.confusion matrix of recognized expression on the physical agent [ cols="^,^,^,^,^,^,^,^ " , ] in @xcite , the mona lisa effect on an animation displayed on a 2d screen and its 3d projection on a head model was studied by shifting eye gaze every 2 degrees from left to right .
then the participants were asked to report their perception of the agents eye gaze direction .
their results showed a clear mona lisa effect in the 2d setting and many subjects perceived a mutual gaze with the screen - based agent s head at the same time for frontal and near frontal gaze angles .
for the physical agent , the mona lisa effect was completely eliminated .
inspired by the experimental setting in @xcite , we evaluated the perception of the eye gaze direction of the physical agent and screen - based agent . in this experiment ,
five subjects were simultaneously seated around an animated agent in two separate sessions ; one session to examine the screen - based agent and another session for the physical agent ( see fig .
[ fig : experimentalsetup ] ) .
the seats were positioned at -45@xmath0 , -25@xmath0 , 0@xmath0 , 25@xmath0 , 45@xmath0 , where 0@xmath0 is the seat in front of the agent .
the distance of the subjects to the agent was five feet .
in the first setting ( called eye gaze only ) , the agent s head looked straight and only the eye gaze was shifted towards each subject .
after each shift , all the subjects were asked to report the subject number that the agent was looking at .
we repeated each experiment three times with 10 random eye gazes each time . in this
setting , the subjects perceived the direction of the eye gaze 50% ( sd=24% ) of the times correctly for the screen - based agent and 88% ( sd=13% ) of the times correctly for the physical agent ( [email protected] ) . in the second setting ( called eye gaze plus arbitrary head movement )
, the agent rotated its head and shifted its eye gaze randomly at the same time , but the head was not necessarily towards the subject of interest .
then , after each head movement and gaze shift , all the subjects reported the subject number that the agent was looking at .
we repeated each experiment three times with 10 random eye gazes each time . in this
setting , subjects perceived the direction of the eye gaze 43% ( sd=18% ) of the time correctly for the screen - based agent and 77% ( sd=15% ) of the times correctly for the physical agent ( [email protected] ) .
these results show that compared with the screen - based agent , the subjects perceived the eye gaze direction produced by the robotic face more accurately in both the `` eye gaze only '' and `` eye gaze plus arbitrary head movement '' settings .
this article described the design and creation of a low - cost emotive robotic head , called expressionbot , for natural face - to - face communication ( the cost of the hardware system is about $ 1500 ) .
the results of our hri studies on a group of participants illustrated that the subjects perceived the facial expression anger with a much greater accuracy in the robotic face than the screen - based face and they also rated the generated visual speech smooth and realistic on both robotic and screen - based systems .
in addition , we studied the perception of eye gaze s direction in two experiments , one in which the head was frontal and only the eye gaze was shifted , and the other with the head rotated but not necessarily correlated with the eye gaze direction . in both experiments , our results showed that participants perceived the robotic face mutual gaze more accurately .
the developed robotic head represents a new level of integration of emotive capabilities that enables researchers to study socially emotive robots / agents that can generate spoken - language , show emotions , and communicate effectively with people in a natural way as humans do .
such systems can be applied in many domains including health - care , education , entertainment , and home - care .
it will also be an ideal platform for designing a new generation of more immersive and effective intelligent tutoring and therapy systems , and robot - assisted therapeutic treatments .
j. wainer , d. j. feil - seifer , d. a. shell , and m. j. mataric , `` the role of physical embodiment in human - robot interaction , '' in _ robot and human interactive communication , 2006 .
roman 2006 .
the 15th ieee international symposium on _
, ieee.1em plus 0.5em minus 0.4em , 2006 , pp .
117122 .
f. delaunay , j. de greeff , and t. belpaeme , `` a study of a retro - projected robotic face and its effectiveness for gaze reading by humans , '' in _ human - robot interaction ( hri ) , 2010 5th acm / ieee international conference on _ , march 2010 , pp .
3944 . a. van breemen , x. yan , and b. meerbeek , `` icat : an animated user - interface robot with personality , '' in _ proceedings of the fourth international joint conference on autonomous agents and multiagent systems_.1em plus 0.5em minus 0.4emacm , 2005 , pp .
143144 .
d. hanson , s. baurmann , t. riccio , r. margolin , t. dockins , m. tavares , and k. carpenter , `` zeno : a cognitive character , '' in _ ai magazine , and special proc . of aaai national conference , chicago _ , 2009 .
r. gockley , r. simmons , j. wang , d. busquets , c. disalvo , k. caffrey , s. rosenthal , j. mink , s. thomas , and w. adams , `` grace and george : social robots at aaai , '' in _ proceedings of aaai _ , vol . 4 , 2004 , pp .
1520 .
m. hashimoto and d. morooka , `` facial expression of a robot using a curved surface display , '' in _ intelligent robots and systems , 2005.(iros 2005 ) .
2005 ieee / rsj international conference on _ , 2005 , pp .
765770 .
f. delaunay , j. de greeff , and t. belpaeme , `` towards retro - projected robot faces : an alternative to mechatronic and android faces , '' in _ robot and human interactive communication , 2009 .
ro - man 2009 .
the 18th ieee international symposium on _ , sept 2009 , pp .
306311 .
t. kuratate , y. matsusaka , b. pierce , and g. cheng , `` mask - bot : a life - size robot head using talking head animation for human - robot communication , '' in _ humanoid robots ( humanoids ) , 2011 11th ieee - ras international conference on _ , 2011 , pp . 99104 .
b. pierce , t. kuratate , a. maejima , s. morishima , y. matsusaka , m. durkovic , k. diepold , and g. cheng , `` development of an integrated multi - modal communication robotic face , '' in _ advanced robotics and its social impacts ( arso ) , 2012 ieee workshop on _ , 2012 , pp .
101102 . s.
al moubayed , j. beskow , g. skantze , and b. granstrm , `` furhat : a back - projected human - like robot head for multiparty human - machine interaction , '' in _ cognitive behavioural systems_.1em plus 0.5em minus 0.4emspringer , 2012 , pp . 114130 . | this article proposes an emotive lifelike robotic face , called expressionbot , that is designed to support verbal and non - verbal communication between the robot and humans , with the goal of closely modeling the dynamics of natural face - to - face communication .
the proposed robotic head consists of two major components : 1 ) a hardware component that contains a small projector , a fish - eye lens , a custom - designed mask and a neck system with 3 degrees of freedom ; 2 ) a facial animation system , projected onto the robotic mask , that is capable of presenting facial expressions , realistic eye movement , and accurate visual speech .
we present three studies that compare human - robot interaction with human - computer interaction with a screen - based model of the avatar .
the studies indicate that the robotic face is well accepted by users , with some advantages in recognition of facial expression and mutual eye gaze contact . |
many - body simulation methods , such as diffusion monte carlo ( dmc ) @xcite and auxiliary field quantum monte carlo ( afqmc ) , @xcite are capable of yielding highly accurate results for electronic systems . in extended systems , however , because of the use of periodic boundary conditions ( pbc ) , finite - size ( fs ) errors arise in many - body ( mb ) calculations , which are often larger than the statistical and other systematic errors .
the fs error reflects the spurious interactions of the system , modeled by a finite - size simulation cell ( supercell ) , with its own periodic images .
the long - range nature of the coulomb interaction in electronic systems exacerbates the problem , and makes the fs error more pronounced . to reduce this error
, calculations have to be performed using larger and larger simulation cells , in order to extrapolate the results to the infinite limit of the supercell size .
the computational cost of mb calculations is typically high .
for example , qmc methods scale with system size as @xmath0-@xmath1 , with large prefactors .
other mb methods typically have significantly worse scaling .
thus convergence of mb calculations with supercell size is difficult to achieve , and accurate extrapolation is often too costly .
@xcite fs correction schemes that can accelerate this convergence are therefore highly desirable .
the behavior of the fs error in a mb calculation is different from that in a standard density - functional theory ( dft ) calculation @xcite ( or other independent - electron calculations ) , as we further discuss in sec .
[ sec : prel ] .
the mb fs error consists of a part that is essentially parallel with the corresponding fs error in a dft calculation of the same supercell .
this part , which will be termed the one - body ( 1b ) fs error , is easier to correct . the residual error , which will be termed the two - body ( 2b ) fs error , is due to the approximate treatment of electron interactions in the extended system by the electrons in the simulation cell plus their images .
the 1b and 2b errors are largely separable ( see sec . [
sec : prel ] ) .
the correction of the 2b errors is the focus of fs correction methods .
it has been shown @xcite that using a modified periodic coulomb potential instead of the standard ewald form of the electron - electron repulsion can significantly reduce fs errors in the calculated ground - state energy and accelerate convergence .
more recently , chiesa _ et al . _
@xcite introduced a 2b fs correction based on the random phase approximation analysis of the momentum distribution and 2b structure factor , which achieves similar results .
both of these approaches are internal correction methods .
the former requires modification of the form of the electron - electron coulomb interaction , and the latter requires calculation of the structure factor in the mb calculation .
the method of kwee , zhang , and krakauer ( kzk ) is an external fs correction approach .
@xcite the idea is to have independent - electron calculations which mimic the fs behavior of mb calculations .
kzk developed a fs exchange - correlation ( xc ) functional which describes the effect of electron - electron interaction in the spirit of a local - density approximation ( lda ) in dft and which takes into account the finite size of the simulation cell .
a functional was parametrized for spin - unpolarized systems .
the functional has been included in the software package quantum espresso @xcite and the method is applicable to any mb total energy results @xcite as a simple , post - processing correction .
the fs functional of kzk does not include spin - dependence , however . in this paper
we generalize the method to arbitrarily spin - polarized systems , and parametrize a fs xc functional based on the local spin - density approximation ( lsda ) .
the new functional includes kzk , but will allow fs corrections in systems with spin - polarization or magnetic order , for example in many transition - metal oxide solids . in such systems
the effect of electron correlation is generally stronger , and the need for mb calculations is greater .
the complexity of the systems also means our ability to systematically simulate larger and larger supercells will be more limited .
accurate and simple fs corrections will therefore be very valuable . in the applications we illustrate how the new
fs lsda functional can help accelerate convergence of the mb results in various systems .
the remainder of the paper is organized as follows . in sec .
[ sec : prel ] , we describe the overall formalism of our fs correction and define some notations . in sec .
[ sec : xc ] , we present the fs xc functionals , with the exchange and correlation parts each in a separate subsection , and a summary of the parameter values of the result of our parametrization in the last subsection . in sec . [
sec : apps ] , several applications are presented to illustrate the fs correction method .
we further discuss several aspects of the method in sec .
[ sec : discuss ] before summarizing in sec .
[ sec : summary ] .
we will consider zero - temperature total energy calculations in a cubic supercell of side length @xmath2 , with @xmath3 atoms in it .
twist boundary conditions @xcite will usually be applied , the many - body generalization of @xmath4-points in independent - electron calculations . the fs error for a given twist @xmath4
is defined as @xmath5 where each @xmath6 denotes _ per atom _ energy , and @xmath7 is the corresponding result at the infinite supercell limit .
the fs error can be separated into 1b and 2b parts , which are also referred to as independent - particle and coulomb fs errors .
@xcite the 1b fs error arises from incomplete @xmath4-point integration , i.e. , a discrete momentum space mesh : @xmath8 where in the first term on the right the sum is over a sufficiently large set of @xmath4-points ( with appropriate weight if necessary ) to reach convergence . below we will also use the notation @xmath9 to denote this term . in a standard dft calculation , because of bloch s theorem , the 1b fs error is the only form of fs error : @xmath10 _ i.e. _ , @xmath11 in dft . in eq .
( [ eq : dft_fserror ] ) , different choices of the simulation cell lead to different requirements on the set @xmath12 over which to sum , but for each choice of @xmath2 , the infinite limit can be achieved with enough @xmath4-points .
the usual approach is to simply integrate over a dense @xmath4-point grid using the primitive cell .
-points : @xmath13 ( blue circle ) , @xmath2 ( red triangle ) , and the baldereschi @xcite mean - value @xmath4-point ( green square ) .
the inset shows the calculated ground - state energy .
the dashed lines are dft - lda results , while the solid lines are corresponding afqmc results .
the supercell sizes are indicated on the horizontal axes.,scaledwidth=43.0% ] the 2b fs error is the residual beyond the 1b fs error : @xmath14 thus @xmath15 is independent of the @xmath4 twist , _
i.e. _ , it is the fs error aside from @xmath4-point sampling errors .
this is illustrated in fig .
[ fig : si - qmc - dft ] , which shows the total energy difference of bulk silicon between qmc and standard dft lda calculations .
three special @xmath4-points are used : @xmath13 , @xmath2 , and the baldereschi @xcite mean - value @xmath4-point . the 1b fs errors are seen to be rather parallel between the independent - electron and mb calculations .
the energy difference between qmc and dft results , the 2b fs error in qmc , is essentially independent of @xmath4-points , consistent with eq .
( [ eq : def_2b ] ) .
the 1b fs error in the mb calculation can thus be removed by applying a correction @xmath16 from standard dft .
this is frequently done in practice .
( note that sometimes the 1b fs error has a different size dependence compared with that of the total @xmath17 , as for the l - point in fig .
[ fig : si - qmc - dft ] . applying this 1b fs correction would make the fs error _
larger_. @xcite ) a residual 2b fs error remains , however , which decays slowly with supercell size . to consider the 2b fs error in the framework of dft , specifically lsda , we review how standard lsda implicitly removes the 2b fs effect , leaving only 1b fs error , which can be eliminated using sufficiently dense @xmath4-point grids .
the many - electron hamiltonian in a simulation cell is ( in rydberg atomic units ) : @xmath18 where @xmath19 labels an electron , @xmath20 is the electron position , the ionic potential on @xmath19 can be local or non - local , and the sum runs over all electrons .
we have suppressed the electron spin index , but spin is assumed to be present .
the coulomb interaction @xmath21 between electrons depends on the supercell size @xmath2 ( and its shape ) , due to modification by the pbc , for example , via the ewald method .
@xcite the corresponding lsda , as usually formulated , introduces a fictitious mean - field system of electrons with the same external potential and the same spin configuration , with the single - electron hamiltonian : @xcite @xmath22 where the xc potential depends on the electronic densities for the two spin species , and the hartree term depends on the total density .
the densities @xmath23 are determined self - consistently , with fermi statistics imposed on the occupied orbitals ( eigenfunctions of @xmath24 ) .
the lsda xc functional in eq .
( [ eq : ks ] ) has an @xmath25 in the superscript : @xmath26 is supposed to be the xc functional of the _ infinite _ system .
the key is that using @xmath26 removes any 2b fs effect .
typically @xmath26 is obtained from qmc results on the homogeneous electron gas ( heg ) , extrapolated to infinite size .
@xcite because lsda yields an independent - electron problem , in which the hamiltonian is only dependent on the electronic densities , bloch s theorem can be applied .
the only requirement for size convergence is the convergence of the densities , which are fully defined by the periodicity of the system once the primitive cell is specified .
the extrapolation of fs heg data to the infinite size was important in that it eliminated the size dependence in the xc functional , making the theory consistent at the thermodynamic limit . to have an lsda theory for the fs supercell
, we must recover the fs dependence of the interaction term , and make it parallel to the mb hamiltonian : @xmath27 where the fs @xmath28 represents the xc effects in the _ finite _ simulation cell in which the mb calculation is carried out .
within dft , the xc energy functional @xmath29 is constructed based on the calculations of the heg , and then applied to realistic system by the lsda approximation .
@xcite below , we examine the electron gas in finite cubic supercells ( of linear size @xmath2 ) to obtain an ansatz for the fs xc energy @xmath30 , from which we can derive the fs xc potential @xmath28 .
a functional will be parametrized as a function of @xmath2 .
with such a functional , lsda calculations can be carried out in the same supercell as the mb calculation .
the difference between the fs lsda and the infinite lsda results ( e.g. , from usual dense @xmath4-point calculations ) , @xmath31 , will provide an estimate of the mb fs error .
the fs corrections used in the remainder of the paper can be summarized as : @xmath32
in this section we discuss the fs xc functional in the heg . we will use @xmath33 to denote the total number of electrons in the supercell of volume @xmath34 , and @xmath35 and @xmath36 for spin-@xmath37 and spin-@xmath38 , respectively . for convenience
, we will refer to the majority spin as @xmath37-spin below , i.e. , assuming @xmath39 .
the density is as usual specified by @xmath40 , the radius of a sphere containing one electron on average , which is related to the charge density by @xmath41 .
the fractional spin polarization parameter @xmath42 is defined as the ratio of spin density and charge density : @xmath43 where @xmath44 ( @xmath45 ) is the electronic density of spin component up ( down ) .
the total charge density is @xmath46=@xmath44+@xmath45 .
the spin unpolarized state corresponds to @xmath47 , while the fully spin polarized state to @xmath48 .
the desired fs xc functional @xmath30 is thus equivalently written as @xmath49 in the heg .
the total xc energy is @xmath50 .
we discuss the exchange and correlation energies separately in sec.s [ subsec : exchange ] and [ subsec : correlation ] and summarize our parametrized functional with numerical parameter values in sec .
[ subsec : xc - parameters ] .
the exact total energy of the heg can be separated into the hf energy , which is the sum of kinetic energy and the exchange energy , and the remainder , which is termed correlation energy .
the kinetic energy is simply given by a sum of the energies of independent electrons filled from the lowest single - particle energy state up to the fermi surface .
the exchange energy from the filled fermi sphere can be calculated analytically , and is of the form @xmath51 .
the asymptotic expression of the correlation energy in the high density region was derived by gell - mann and brueckner , @xcite which is used in combination with dmc results at various densities @xcite to obtain a parametrized fit @xcite of the lsda xc energy
. as mentioned , all of these are at the infinite size limit . in finite - sized supercells , the hf exchange energy of the heg scales as @xmath52 .
the fs error in the exchange energy per electron , @xmath53 , scales @xcite as @xmath54 , with a coefficient @xmath55 which is only dependent on @xmath33 ( i.e. , the ratio of @xmath56 ) , not on @xmath40 .
the fs exchange energy approaches the infinite size limit from below .
@xmath55 decays smoothly like @xmath57 when averaged over many @xmath4-points .
@xcite thus the first order correction of exchange energy will be of order @xmath58 .
we will include a next - order term in the parametrization , @xmath59 , which follows the same scaling relation . in a system with two spin components , each component contributes independently to the total exchange energy .
there is no exchange interaction between opposite spins .
thus , for a system of polarization @xmath42 , the exchange energy is given by @xmath60 where @xmath61 is the exchange energy of the unpolarized system with total density given by @xmath40 , and the density for the spin up ( down ) component is given by @xmath62 ( @xmath63 ) .
with this relation , we can obtain the fs exchange energy for an arbitrary spin polarization .
however , the fs exchange energy for spin unpolarized system is discontinuous as given by kzk .
@xcite at low densities , the exchange energy is forced to go to zero rapidly , because in the _ finite _ supercell there will only be less than one electron in each spin component when @xmath40 is larger than some threshold @xmath64 . in this regime , the functional form from scaling , i.e. , @xmath65 , would diverge .
a different functional form is used instead , and the two are joined at @xmath66 by requiring the exchange potential to be continuous , which resulted in a discontinuity in the exchange energy . with this form of the exchange energy , the use of eq .
( [ eq : ex ] ) would lead to multiple discontinuities in a partially polarized system , and an exchange that tends to be too high in the intermediate densities which connect the two density regimes . instead of the formula in eq .
( [ eq : ex ] ) , we will make an interpolation between spin unpolarized and fully polarized systems .
the interpolation formula @xcite is @xmath67 , \label{eq : ex - interpolation}\ ] ] with the function @xmath68 defined as @xmath69 the formula in eq .
( [ eq : ex - interpolation ] ) gives the exchange energy for any polarization by interpolation bwteeen the exchange energies for unpolarized and fully polarized systems .
when @xmath2 is infinite , this formula reduces to the exact relation in eq .
( [ eq : ex ] ) .
the use of this formula also makes the treatment of the exchange consistent with that of the correlation ( see sec .
[ subsec : correlation ] ) , for which there is no analogy of eq .
( [ eq : ex ] ) .
we will parametrize the exchange energies for the two end point systems in the following form : @xmath70 here @xmath71 has two discreet values : @xmath72 for spin unpolarized while @xmath73 for fully polarized .
the formula is the same as that in ref . for the high density region , but we have introduced two additional terms in the low density part .
the values of @xmath74 are determined by the infinite size heg exchange energies .
the other coefficients are obtained from a fit and by continuity conditions at @xmath64 . as described in sec .
[ subsec : xc - parameters ] , we choose the same @xmath64 for @xmath72 and @xmath73 , hence for all polarizations @xmath42 . at @xmath66 , we require the exchange potential and its first derivative to be continuous for all polarizations . the conditions are straightforward to impose at @xmath72 and @xmath73 . for an arbitrary spin polarization , the spin interpolation function @xmath68 in eq .
( [ eq : ex - interpolation ] ) depends on the spin density .
the exchange potential is given by @xmath75 \nonumber \\
& + n\frac{\partial f(\zeta)}{\partial n_{\uparrow(\downarrow ) } } [ \varepsilon_{x}(r_s , l,1)-\varepsilon_{x}(r_s , l,0 ) ] .
\label{eq : vx}\end{aligned}\ ] ] we see that , if the energy difference between the polarized and unpolarized systems , @xmath76-@xmath77 , is made continuous at @xmath66 , the exchange potential will be continuous for any @xmath42 .
it is easy to show that continuity of the first derivative of the exchange potential is also satisfied with these conditions .
we calculate the exchange energies for different supercell sizes by averaging over @xmath4-points .
the quality of the fits is consistent with that in ref . .
numerical values of the parameters are given in sec .
[ subsec : xc - parameters ] . unlike the exchange energy ,
the correlation energy consists of both intra- and inter - spin contributions .
there is no exact mapping of the correlation energy at an arbitrary polarization from spin unpolarized results . within standard dft
, the correlation energy for partial spin polarization is obtained from an interpolation between spin unpolarized and fully spin polarized systems , as we have done with the exchange energy .
we will follow the same procedure .
below , we first construct the energy functions for the two end - point systems of @xmath72 and @xmath73 . in ref .
, the correlation energy of unpolarized system of the heg is expressed separately for high , intermediate , and low density regions . in the high density region
, the formula is parameterized by fitting to ceperley and alder s extrapolation fit @xcite _ c(r_s , l ) & = _ c^(r_s)-_x(r_s , l ) + & + [ b_1(r_s)-1]_k(r_s , l ) + b_2(r_s)/n , [ eq : extrapolation ] where the parameters @xmath78 and @xmath79 were written as low order polynomials in @xmath80 and @xmath81 , respectively , and fitted to tabulated data for unpolarized systems .
@xcite ( note that here @xmath82 and @xmath83 are given by @xmath84-@xmath85 and @xmath86-@xmath87 , respectively . ) to avoid the divergence at sufficiently large @xmath40 or small @xmath33 , the correlation energy is set to zero . in the intermediate density region ,
a polynomial function is used whose parameters were completely determined from continuity conditions . for fully polarized systems , the available finite - size data from qmc were not as detailed .
we instead modify the procedure above , by generating fs heg data in the intermediate density regime and fitting them to obtain a unified function extending to higher densities . ) , while the dashed lines are from kzk .
b ) the parametrized correlation energy for fully spin polarized heg systems is compared with afqmc data .
solid lines represent the fits while symbols are from afqmc at finite sizes .
statistical error bars in the afqmc results are smaller than the symbol size .
the infinite size lsda result is also shown for @xmath73 .
, scaledwidth=48.0% ] we parameterize the spin unpolarized and fully spin polarized correlation energies by a new unified function with only two density regions : @xmath88 the functional form is chosen to approach the correct asymptotic value at high density , where we have used the perdew - zunger ( pz ) parametrization @xcite for the function @xmath89 .
the leading fs term @xmath90 exactly cancels its counterpart in the exchange energy to ensure that the total xc energy scales as @xmath91 as expected . for the @xmath92 term , we choose the functional form g(r_s , l , p)&g_1(l , p)r_s ( r_s ) + g_2(l , p)r_s + g_3(p)r_s^1/2 + & + g_4(p)r_s^3/2 ( r_s ) + g_5(p)r_s^3/2 + g_6(p)r_s^2 .
[ eq : gr_ec ] the last four parameters are determined by the fits , while the other two are determined by continuity conditions .
we require the correlation energy and its first derivative to be continuous at @xmath93 .
our choice of @xmath93 and the final parameter values are given in sec .
[ subsec : xc - parameters ] .
we perform phaseless afqmc @xcite calculations of the correlation energies of the heg , for @xmath73 and @xmath72 , in finite supercells at typical densities . with each choice of @xmath33 and @xmath2
, we average the total energy over multiple @xmath4-points .
the results are shown in fig .
[ fig : fit ] .
it is seen that the new parametrized functions fit all the afqmc data well .
figure [ fig : fit]a also shows a comparison between the new functional and that of kzk for unpolarized systems .
the two fits are almost indistinguishable , except in the intermediate region , where the new function shows a slightly better fit to afqmc results , especially for small lattice sizes . as mentioned , in the approach of kzk , the functional form at high density is fully determined by existing dmc data .
afqmc data were only used to guide the choice of the boundaries which separated the high density regime from the intermediate and low density regimes .
in contrast , the current approach only had afqmc data at intermediate densities to determine the functional form for @xmath94 .
the good agreement between the two parametrizations across the density range is thus very reassuring .
contrary to the exchange energy , the correlation energy approaches the thermodynamic limit from above , as illustrated in fig . [
fig : fit]b .
we have now obtained the fs correlation energy formulas for spin unpolarized ( @xmath72 ) and fully spin polarized ( @xmath73 ) systems , with which we can interpolate to arbitrary spin polarization @xmath42 .
as mentioned , the interpolation of the correlation energy is not exact . in standard dft
, two forms have been widely adopted : @xcite that of pz as we have used for the exchange in eq .
( [ eq : ex - interpolation ] ) , and that of perdew and wang ( pw ) : @xmath95 \nonumber \\ + [ \varepsilon_{c}(r_s , l,1)-\varepsilon_{c}(r_s , l,0 ) ] f(\zeta ) \zeta^4 , \label{eq : pw - interpolation}\end{aligned}\ ] ] where @xmath96 is the spin stiffness function which includes additional fitting parameters .
the interpolations for several different supercell sizes are shown in fig .
[ fig : ec - inter ] . in the pw results ,
we have fixed the @xmath96 at the infinite - size values without further parametrization .
it is seen that the two schemes give similar results .
compared to the actual fs data from afqmc calculations in polarized systems , both schemes provide reasonable estimates .
discrepancies are visible , especially at smaller supercell sizes . in the applications below
, we have used the pz interpolation scheme for the correlation energy , due to its simplicity , _ i.e. _ , using the same interpolation as in the exchange energy in eq .
( [ eq : ex - interpolation ] ) .
the pz form is equivalent to the special choice in eq .
[ eq : pw - interpolation ] .
this is further discussed in sec .
[ sec : discuss ] .
[ 1]d .. #1 & & & & + 0 & -0.9163 & & -1.1545 & + 1 & -2.2037 & & -1.7491 & + 2 & 0.4710 & & 0.2967 & + 3 & 0.2339 & 0.2109 & 0.1812 & 0.7528 + 4 & -0.4880 & 8.4987 & -0.4515 & 3.3314 + 5 & 0.1847 & -13.6840 & 0.1786 & -5.1050 + 6 & & -4.6977 & & -2.3048 + the functional forms of our parametrized fs lsda xc functional are given in eqs .
( [ eq : exnew ] ) , ( [ eq : ec ] ) , and ( [ eq : gr_ec ] ) . by performing afqmc calculations and fitting to the results with the procedures described above
, we obtain the values of the coefficients , which are given in table [ tab : xc - table ] .
the only remaining parameters are the densities at which the different pieces in the functional are joined . for the exchange energy , we used @xmath97 for both spin unpolarized and fully polarized states . for the correlation energy
, we chose @xmath98 for the fully polarized system and @xmath99 for the unpolarized system .
these choices are guided by the basic idea that beyond @xmath100 , we have a supercell of the heg in which there is only a fraction of an electron .
the exchange ( correlation ) energy needs to vanish rapidly in this regime .
clearly the location of @xmath100 can not be uniquely defined .
the results from the xc functional should not be sensitive to the precise value of @xmath100 .
we have verified that this is indeed the case in typical supercell sizes . in principle , @xmath64 should depend on the spin polarization @xmath42 . for the fully polarized system , @xmath101 signals the transition point , while for the unpolarized system , @xmath102 . we have taken them to be the same value , which simplifies the interpolation and makes the functional easier to implement , because all values of @xmath42 share the same @xmath64 . the form of the correlation energy is continuous and it is easier to have different @xmath100 values for different @xmath71 . we have thus chosen @xmath103 for @xmath72 to be the same as in kzk for consistency . in the correlation energy , there are six parameters in the formula of eq .
( [ eq : gr_ec ] ) .
we only need to fit the last four , since the other two , @xmath104 and @xmath105 , are fully determined by the continuity conditions .
this is why they have a dependence on lattice size @xmath2 . in the applications below ,
we have used the pz interpolation scheme in the correlation energy , i.e. , using the same interpolation as in the exchange energy .
the @xmath106 term is then exactly canceled at any polarization , leaving only @xmath107 and higher order terms in the fs errors .
we consider four examples here to illustrate the application of the new fs lsda functional . the four problems are chosen to cover different aspects of the application .
all of them have spin polarizations or magnetic ground states which require the new spin - dependent fs functional .
calculations in single atoms and molecules provide systematic mb results for arbitrary supercell sizes which allow a detailed study of the fs convergence .
the ionization energy of mn is calculated to study the effect of the fs lsda in supercells when there is a physical charge imbalance . in the last example
, transition metal oxide systems mno and feo are considered to study the application of the fs correction in extended systems with magnetic order . in the first three examples
, we use the phaseless afqmc method @xcite in a plane - wave basis , with a norm - conserving pseudopotential to carry out the mb calculations , before applying the fs corrections .
more systematic afqmc calculations of atomic and molecular systems using plane - wave basis exist .
@xcite our goal here is simply to use it as an accurate mb approach for model extended systems to study fs corrections . in the last example , we use existing dmc results @xcite and use the new fs functional as a post - processing correction . molecular dissociation energy vs. supercell size . calculated afqmc total energies of the p atom and p@xmath108 molecule are shown in panels ( a ) and ( b ) , respectively .
the corresponding @xmath109 of p@xmath108 is shown in ( c ) .
the infinite - size limits of dft and afqmc are shown by the dashed and dot - dashed lines , respectively .
afqmc statistical error bars are shown but many are smaller than the symbol size . the orange arrow in ( c ) points to the experimental value .
, scaledwidth=48.0% ] our first application is to calculate the dissociation energy ( @xmath109 ) of the p@xmath108 molecule , using a plane - wave basis and pbc with single @xmath4-point sampling , . in both afqmc and dft calculations ,
a norm - conserving kleinman - bylander @xcite separable nonlocal lda pseudopotential @xcite is used .
the p atomic ground state is magnetic , with an electronic configuration , a
close - shell " system for each spin .
the p@xmath108 molecular ground state is non - magnetic , with 5 @xmath37-spin and 5 @xmath38-spin electrons .
the total energy calculations are performed in cubic supercells of size @xmath110bohr .
figure [ fig : p ] shows the mb results from afqmc calculations , as well as dft results from both the standard , infinite - size lsda xc functional ( denoted by dft@xmath111 ) and the new fs functional ( denoted by dft@xmath112 ) .
the conventional dft@xmath111 energies converge to the infinite limit very rapidly , as expected .
the dft@xmath112 calculations exhibit a significantly slower convergence , reflecting the 2b fs effect which it is designed to capture . for the p atom ,
the overall fs effect is not very large , but a persistent fs error can be seen in the raw afqmc results . up to @xmath113bohr ,
the energy is still about @xmath114ev away from the infinite - size value .
the 1b correction with standard dft [ @xmath115 ; see eq .
( [ deltadftfs ] ) ] is in the wrong direction and leads to a larger fs error and slower rate of convergence .
by contrast , the dft@xmath112 total energies , which include both 1b and 2b corrections [ eq . ( [ deltadftfs ] ) ] , show size dependence similar to the afqmc results .
after @xmath116 correction , the afqmc total energy converges to the asymptotic value rapidly , as seen from the essentially flat curve beyond @xmath117bohr .
a similar trend is seen in the p@xmath108 molecule as shown in fig .
[ fig : p]b . in this case
the 1b correction does reduce the fs error , but only up to intermediate size @xmath2 , beyond which the standard dft has reached convergence .
the mb results remain unchanged for large @xmath2 , where the 2b fs effect dominates .
the dft@xmath112 calculation again tracks the fs dependence of the mb afqmc , and the total energy of the latter converges very rapidly to the infinite size limit after applying the full fs correction .
the p@xmath108 system is unpolarized , so the kzk parametrized fs functional can also be applied .
compared to the results there , @xcite the new functional seems to slightly overestimate the fs error .
this small difference is from the choice of exchange energy discontinuity point @xmath64 , as mentioned in sec .
[ subsec : xc - parameters ] . with the fs corrections ,
the calculated @xmath109 shows good convergence by about @xmath118bohr , as seen in fig . [
fig : p]c . molecular dissociation energy vs. supercell size .
same conventions as in fig .
[ fig : p ] .
, scaledwidth=48.0% ] our second application is for the si atom and molecule . these are open - shell " systems , and both systems are spin polarized in the ground state ( @xmath119 and @xmath120 , respectively ) .
the calculations are done in a similar way to the first application , with planewave basis and a norm - conserving non - local pseudopotential in a periodic supercell .
the dependence of total energies and the dissociation energy with lattice size @xmath2 is presented in fig .
[ fig : si ] . similar to the p atom ,
the usual dft 1b correction for the si atom has an opposite sign from the total fs error . with dft@xmath112 corrections ,
the convergence of the total energy of the atom is greatly improved across the entire range of lattice size studied .
similarly , for the si@xmath108 molecule , the total energy becomes essentially flat when the full correction is included , leading to rapid convergence to the infinite - size limit .
although the atoms and molecules are relatively simple systems for mb calculations , they provide excellent tests for the fs correction method .
they are model systems of a molecular solid in which the supercell size can be arbitrarily and systematically varied . because of the nature of the interactions and the highly inhomogeneous density distributions in the supercell , lsda calculations are not particularly accurate , as can be seen by the final dft @xmath109 s ( panel ( c ) in figs .
[ fig : p ] and [ fig : si ] ) . yet
the fs correction method based on lsda works extremely well .
this underscores an important point , namely that for the fs correction method to be effective , the system does not _ necessarily _ have to be weakly or moderately correlated .
the fs correction method requires lsda to provide a good approximation in capturing the _ difference _ between the system with interaction @xmath21 and that with the full interaction @xmath121 ( no pbc image interactions ) .
@xcite this is not the same as requiring lsda to work well in either system . as the supercell size increases , the correction decreases and approaches zero .
one would expect the correction to be ineffective if the supercell is less than the size of the xc hole , which would cause a distortion that is not captured in any lsda approximation . ) .
panel ( a ) shows the calculated afqmc total energies for mn , mn@xmath122 , and mn@xmath123 ; the corresponding first ( ip ) and second ( iip ) ionization energies are shown in panels ( b ) and ( c ) .
curves labeled qmc are from raw afqmc results [ for the charged ions , the leading order @xmath124 correction is included ( see text ) ] .
the @xmath125 and @xmath126 corrections are as in figs .
[ fig : p ] and [ fig : si ] [ for the charged ions , these corrections also add the @xmath127 correction ( see text ) . ] statistical error bars are shown and are about the size of the symbols .
the heavy black dash lines indicate the infinite size limits of the calculated results in panel ( a ) and the experimental ionization energies in panels ( b ) and ( c ) .
, scaledwidth=48.0% ] the third application is to the mn transition metal atom and its first and second ionization potentials .
the @xmath4-point used here is ( 0.25,0.25,0.25 ) . for charged systems such as mn@xmath128 or mn@xmath123
, there is an additional complication in a supercell calculation , since charged systems are incompatible with pbc .
a uniform neutralizing background is introduced to maintain charge neutrality , @xcite but this results in a slow @xmath129 size convergence to leading order.@xcite in independent - electron calculations such as dft , makov and payne ( mp ) @xcite argued that the dependence has the form : @xmath130 where @xmath131 is the neutralizing charge , @xmath132 is the madelung constant , and @xmath133 is the quadrupole moment of the supercell .
thus , adding the counter - terms @xmath134mp@xmath135 and @xmath134mp@xmath136 to @xmath137 accelerates the size - convergence of dft calculations for charged systems .
the @xmath134mp@xmath138 correction applies equally well to a many - body calculation with pbc , since it depends only on the neutralizing charge , and all afqmc results reported below for charged systems include this important correction .
the @xmath134mp@xmath139 correction depends on @xmath133 , which will generally differ , in a many - body calculation , from the dft value . in this work
we have not calculated @xmath133 in afqmc , but have used the dft value .
figure [ fig : mn ] shows the size convergence of total energies of mn , mn@xmath122 , and mn@xmath123 and the corresponding first and second ionization energies . for mn and mn@xmath122 total energies
, @xmath126 removes nearly all of the fs error , while it somewhat overcorrects for mn@xmath123 .
the @xmath125 corrections are smaller in all cases .
for the ionization energies , both @xmath125 and @xmath126 are seen to overcorrect the fs error . for ip ,
the residual error after @xmath126 correction is slightly smaller than the raw afqmc result , while @xmath125 increases the fs error .
for iip , both fs corrections increase the fs error .
extrapolation of these results to the infinite size limit , leads to ip and iip values of 7.27(08 ) and 23.15(08 ) ev , respectively .
these are in reasonably good agreement with the experimental values of 7.43 ev and 23.07 ev , respectively .
@xcite because of strong fs error cancellations , the raw qmc ip and especially iip energies show little fs effect .
this magnifies the residual errors in the fs correction for charged systems .
the most likely source of the residual error would appear to be the discrepancy in applying the @xmath134mp@xmath139 directly from dft .
the mb and dft estimates of the quadrupole moment @xmath133 are likely to differ .
we will leave further investigation of this effect to a future study .
the total energies , after fs correction , show much smaller fs error , especially in the neutral system of mn . given the strong magnetic nature of this transition metal system
, this is an encouraging result for our fs correction scheme .
in the supercell ) for a ) mno in the b1 phase , and b ) feo in the ib8 phase at volumes of 21.7 @xmath140/mno and 20.4 @xmath140/feo , respectively .
results from the dmc calculations are indicated by green circles ; corrected values using the structure factor method ( taken from ref . )
are shown by the blue squares .
dmc results corrected by @xmath126 are indicated by red triangles .
, scaledwidth=47.0% ] the last application is for crystalline mno and feo .
these materials are challenging to simulate due to the complex interplay of strong correlation effects , p - d hybridization , and magnetism .
we apply the dft fs correction directly to previous results from dmc calculations . @xcite
although the pseudopotentials used in our modified dft calculations are different from those used in dmc , this has only a weak effect on the fs corrections . as seen in fig .
[ fig : mno - feo ] , the @xmath126 correction greatly reduces the dmc fs error . in both materials , it somewhat over - corrects and approaches the infinite limit from above .
the structure factor correction of chiesa @xcite , also shown in the figure , somewhat under - corrects .
for mno the residual error is somewhat larger than for @xmath126 , while it is smaller for feo .
a possible difficulty for @xmath126 is the irregular supercells used for these materials .
the @xmath126 correction method is based on cubic supercells , parametrized by a single length scale @xmath2 .
this is a good approximation for other high symmetry systems such as fcc supercells . @xcite
if the supercell shape is far from cubic and has strong anisotropy , however , the fs total energy will increase , resulting in a smaller fs correction compared to that of a cubic system .
this is discussed further in the next section .
these applications demonstrate that the new fs error correction scheme can consistently remove most of the fs error and accelerate the size convergence of mb results to infinite size limit .
in this section we discuss several aspects of the new fs correction method , the parametrized lsda functional , and possible further improvements .
in , the fs lsda xc functional was separated into exchange @xmath141 [ eqs .
( [ eq : ex - interpolation])-([eq : exnew ] ) ] and correlation @xmath142 [ eqs .
( [ eq : ec ] ) and ( [ eq : gr_ec ] ) ] contributions .
this facilitates the modification of existing dft computer codes to generate the fs corrections . while useful , this separation somewhat complicates the fs treatments , which are based on xc functionals derived from the heg .
for example , while the total fs heg xc energy scales as @xmath92 , @xmath141 scales as @xmath143 , since the hartree - fock spectrum is gapless in this case .
the separation thus induces a canceling @xmath143 term in @xmath142 .
partly for this reason , we adopted the pz spin interpolation formula given by eq .
( [ eq : ex - interpolation ] ) for @xmath141 , rather than the exact relation in eq .
( [ eq : ex ] ) , since it makes the treatment of exchange consistent with that of the correlation , for which there is no analogy of eq .
( [ eq : ex ] ) , and ensures cancellation of the @xmath143 dependence . [ as mentioned , in the limit @xmath144 , eq .
( [ eq : ex - interpolation ] ) reduces to the exact relation in eq .
( [ eq : ex ] ) . ]
since the fs exchange energy for spin unpolarized system is discontinuous , @xcite another reason for using eq .
( [ eq : ex - interpolation ] ) is that it avoids multiple discontinuities in a partially polarized system , as discussed in sec .
[ subsec : exchange ] .
the continuity density @xmath64 in eq .
( [ eq : exnew ] ) was chosen for simplicity .
it could be made polarization - dependent , but the expressions would be more complex and not as easy to implement into dft codes . as discussed in sec .
[ subsec : xc - parameters ] , fs corrections are not sensitive to the precise value of @xmath64 , especially for extended systems .
similar considerations pertain to the treatment of correlation . since @xmath142 is continuous , it was easy to use different values for @xmath145 and @xmath146 in eq .
( [ eq : ec ] ) , as described in sec .
[ subsec : xc - parameters ] , when the pz interpolation form is used . if the pw interpolation of eq .
( [ eq : pw - interpolation ] ) were to be used , cancellation of the @xmath143 term would require more complicated expressions .
interpolation formulas , such as that of pz , allow us to map spin unpolarized and fully - polarized results onto the correlation energy at an arbitrary polarization .
for the fully polarized heg , the available finite - size data from qmc were not as detailed as for the unpolarized case .
our fits of the fully - polarized @xmath147 correlation functional are based on afqmc calculations of finite systems , mostly for @xmath40 greater than about 1.5 - 2 . while our correlation energy fits are acceptable ( see figs . [
fig : fit ] and [ fig : ec - inter ] ) , more qmc calculations for small @xmath40 might lead to improvements .
extensions of our fs error correction scheme are possible , for example based on orbital - dependent density functionals , @xcite such as hybrid dft functional calculations .
hybrid functionals @xcite augment the dft exchange - correlation energy with an exact exchange term calculated from hf theory .
hybrid functionals have become one of the most promising approaches for overcoming some of the shortcomings of local or semilocal dft approximations .
@xcite some complications would need to be addressed .
popular hybrid functionals such as pbe0 @xcite use a fraction @xmath148% of exact exchange and dft exchange for the remainder . in applications to semiconducting and insulating materials , the hf exchange converges as @xmath92 with pbc , since the single - particle spectrum has a gap .
[ the leading behavior is @xmath92 after a @xmath52 madelung - like dependence is removed , @xcite as is done in qmc calculations .
@xcite ] since dft fs exchange scales as @xmath143 for the heg , the fraction of the @xmath143 term in the fs correlation functional would have to be adjusted in order to obtain a net @xmath92 behavior .
the resulting @xmath142 would lead to a non - continuous behavior with increasing @xmath40 , which might require further modifications of the fits .
finally , our fs xc functional is based on heg calculations using cubic supercells of edge @xmath2 . for non - cubic applications , @xmath2 is determined by the volume of the supercell . in the heg , spatially isotropic supercells are favored .
for example , a small cubic to tetragonal distortion will cause the total energy to become more positive , scaling as @xmath149 ) , where @xmath150 is a constant and @xmath151 describes the aspect ratio @xmath152 of the tetragonal supercell .
therefore our fs xc functional tends to provide an upper limit to the magnitude of the fs correction for non - cubic supercells .
this may have contributed to the over - correction seen in fig .
[ fig : mno - feo ] , where anisotropic mno and feo supercells were used .
many - body calculational methods such as quantum monte carlo can potentially provide accurate results for extended systems where electron interaction effects are important .
these calculations are computationally more costly than independent - electron calculations , and are always accompanied by more significant fs errors . in this work ,
we have provided a framework within dft to understand and estimate the fs errors in mb calculations of spin - polarized systems .
a fs lsda functional is parametrized which leads to a simple , post - processing correction method .
the method is designed to approximately include 2b fs corrections in fs dft calculations of systems with arbitrary spin polarization .
the corrections to total energy , dissociation energy , and ionization energy calculations of several typical systems show a significant improvement on the convergence .
the formalism of the method and the parametrization of the fs functional are presented in detail .
the strengths and weaknesses of the approach are discussed .
there are several directions in which the method can be further developed .
the method can be easily incorporated in any standard dft computer program for periodic systems , and can be used to correct any mb results .
it is expected that the method will find applications in the study of a variety of realistic systems with magnetic ordering .
this work is supported by doe ( grant no .
de - fg02 - 09er16046 ) , nsf ( grant no .
dmr-1006217 ) , and onr ( grant no .
n00014 - 08 - 1 - 1235 ) .
calculations were performed at the center for piezoelectric by design .
we thank s. sorella for helpful discussions , w. purwanto for help with computing issues , and j. koloren and l. mitas for sending us the computational details of refs . .
d. m. ceperley and b. j. alder , phys .
* 45 * , 566 ( 1980 ) .
p. j. reynolds , d. m. ceperley , b. j. alder , and w. a. lester , j. chem .
phys . * 77 * , 5593 ( 1982 ) .
w. m. foulkes , l. mitas , r. j. needs , and g. rajagopal , rev .
phys . * 73 * , 33 ( 2001 ) .
s. zhang and h. krakauer , phys .
lett . * 90 * , 136401 ( 2003 ) .
a. j. williamson , r. q. hood , and j. c. grossman , phys .
lett . * 87 * , 246406 ( 2001 ) .
p. hohenberg , and w. kohn , phys . rev .
* 136 * , b864 ( 1964 ) . w. kohn , and l. j. sham , phys . rev . *
140 * , a1133 ( 1965 ) .
p. r. c. kent , randolph q. hood , a. j. williamson , r. j. needs , w. m. c. foulkes , and g. rajagopal , phys .
b * 59 * , 1917 ( 1999 ) . l. m. fraser , w. m. c. foulkes , g. rajagopal , r. j. needs , s. d. kenny , and a. j. williamson , phys .
b * 53 * , 1814 ( 1996 ) .
a. j. williamson , g. rajagopal , r. j. needs , l. m. fraser , w. m. c. foulkes , y. wang and m .- y .
chou , phys .
b * 55 * , 4851(r ) ( 1997 ) . n. d. drummond , r. j. needs , a. sorouri , and w. m. c. foulkes , phys .
b * 78 * , 125106 ( 2008 ) . s. chiesa , d. m. ceperley , r. m. martin , and m. holzmann , phys .
lett . * 97 * , 076404 ( 2006 ) .
h. kwee , s. zhang , and h. krakauer , phys .
lett . , * 100 * , 126404 ( 2008 ) .
p. giannozzi et al . , j. phys.:condens .
matter * 21 * , 395502 ( 2009 ) ; url http://www.quantum-espresso.org " .
w. purwanto , h. krakauer , and s. zhang , phys .
b * 80 * , 214116 ( 2009 ) .
e. sola and d. alf , phys .
lett . * 103 * , 078501 ( 2009 ) .
r. maezono , n. d. drummond , a. ma , and r. j. needs , phys .
b * 82 * , 184108 ( 2010 )
. s. j. binnie , s. j. nolan , n. d. drummond , d. alf , n. l. allan , f. r. manby , and m. j. gillan , phys .
b * 82 * , 165431 ( 2010 ) .
r. g. hennig , a. wadehra , k. p. driver , w. d. parker , c. j. umrigar , and j. w. wilkins , phys .
b * 82 * , 014101 ( 2010 ) .
s. sorella , m. casula , l. spanu , and a. d. corso , phys .
b * 83 * , 075119 ( 2011 ) . c. lin , f. h. zong , and d. m. ceperley , phys . rev .
e * 64 * , 016702 ( 2001 ) .
a. baldereschi , phys .
b * 7 * , 5212 ( 1973 ) .
d. m. ceperley , phys .
b * 18 * , 3126 ( 1978 ) .
j. p. perdew and a. zunger , phys .
b , * 23 * , 5048 ( 1981 ) .
u. von barth and l. hedin , j. phys .
c * 5 * , 1629 ( 1972 ) .
j. p. perdew and y. wang , phys .
b , * 45 * , 13244 ( 1992 ) . m. gell - mann and k. a. brueckner , phys . rev . * 106 * , 304 ( 1957 ) .
d. m. ceperley and b. j. alder , phys .
b * 36 * , 2092 ( 1987 ) .
b. tanatar and d. m .ceperley , phys .
b * 39 * , 5005 ( 1989 ) .
y. kwon , d. m. ceperley and r. m. martin , phys .
b * 58 * , 6800 ( 1998 ) . h. kwee , ph.d .
thesis , college of william and mary , ( 2008 ) .
g. ortiz and p. ballone , phys .
b * 50 * , 1391 ( 1994 ) .
m. suewattana , w. purwanto , s. zhang , h. krakauer , and e. j. walter , phys .
b * 75 * , 245123 ( 2007 ) .
j. koloren , s. hu , and l. mitas , phys . rev .
b * 82 * , 115108 ( 2010 ) .
j. koloren , and l. mitas , phys .
* 101 * , 185502 ( 2008 ) .
l. kleinman and d. m. bylander , phys .
* 48 * , 1425 ( 1982 ) . generated with opium : http://opium.sourceforge.net .
g. makov and m. c. payne , phys .
b * 51 * , 4014 ( 1995 ) . p. p. ewald , ann .
( leipzig ) * 64 * , 253 ( 1921 ) .
j. ziman , _ principles of the theory of solids _ ( cambridge university press , cambridge , 1964 ) .
j. sugar and c. corliss , j. phys .
chem . ref .
data * 14 * , suppl . 2 ( 1985 ) .
becke , j. chem . phys . * 98 * , 1372 ( 1993 ) s. kmmel and l. kronik , rev .
phys , * 80 * , 3 ( 2008 ) .
j. p. perdew , m. ernzerhof , and k. burke , j. chem . phys . * 105 * , 9982 ( 1996 ) f. gygi and a. baldereschi , phys . rev .
b * 34 * , 4405 ( 1986 ) .
nguyen and s. de gironcoli , phys .
b * 79 * , 205114 ( 2009 ) . | we extend the post - processing finite - size ( fs ) correction method , developed by kwee , zhang , and krakauer [ _ phys .
rev .
lett . _ * 100 * , 126404 ( 2008 ) ] , to spin polarized systems .
the method estimates the fs effects in many - body electronic structure calculations of extended systems by a modified density functional theory ( dft ) calculation , without having to repeat expensive many - body simulations .
we construct a unified fs dft exchange - correlation functional for spin unpolarized and fully spin polarized systems , under the local density approximation .
the results are then interpolated to arbitrary spin polarizations .
generalization to other functional forms in dft are discussed .
the application of this fs correction method to several typical magnetic systems with varying supercell sizes demonstrates that it consistently removes most of the fs errors , leading to rapid convergence of the many - body results to the infinite size limit . |
from the time when the quarks were predicted @xcite , their nature has remained rather elusive .
the reason is that direct experimental studies of quarks are very limited .
quarks do not exist in vacuum as free particles . under normal conditions ,
they are always confined inside hadrons .
the underlying theory of strong interactions quantum chromodynamics ( qcd ) predicts that quarks should become deconfined at very high temperatures and/or very high densities @xcite .
unfortunately , it is very difficult to achieve sufficiently high densities and/or temperatures in laboratory .
very high temperatures existed in the early universe during the first few microseconds of its evolution @xcite . nowadays , somewhat similar conditions , although at considerably smaller scales and for much shorter periods of time , are repeatedly recreated in the so - called `` little bangs '' at the heavy ion colliders in cern and at brookhaven .
sufficiently high densities may exist in the present universe inside central regions of compact stars .
recently , this possibility attracted a lot of attention when it was suggested that various color superconducting phases with rather large values of gaps in their quasiparticle energy spectra could appear at densities that exist inside stars @xcite .
if this turns out to be true , this would be of prime importance .
the presence of large gaps in the energy spectra can possibly be inferred from a detailed analysis of the observational data .
this would provide a confirmation of the existence of new ( quark ) states of matter inside compact stars . in theoretical studies of dense quark
matter , it should be appreciated that matter in the bulk of stars is neutral and @xmath0-equilibrated . under such conditions ,
the chemical potentials of different quarks should satisfy nontrivial relations .
these , in turn , affect the pairing dynamics between quarks which is reflected in a specific choice of the ground state of matter .
for example , it was argued in ref .
@xcite that a mixture of the two - flavor color superconducting ( 2sc ) phase and normal strange quarks is less favorable than the color - flavor - locked ( cfl ) phase after the charge neutrality condition is enforced .
a similar conclusion was also reached in ref .
@xcite .
assuming that the constituent medium modified mass of the strange quark is large ( i.e. , larger than the corresponding strange quark chemical potential ) , it was shown recently that neutral two - flavor quark matter in @xmath0-equilibrium can have another rather unusual ground state called the gapless two - flavor color superconductor ( g2sc ) @xcite . while the symmetry in the g2sc ground state is the same as that in the conventional 2sc phase , the spectrum of the fermionic quasiparticles is different .
in particular , two out of four gaped quasiparticles of the conventional 2sc phase become gapless in the g2sc phase .
in addition , the number densities of the pairing quarks in the g2sc phase are not equal at zero temperature @xcite .
for example , the density of red ( green ) up quarks is different from the density of green ( red ) down quarks .
the existence of gapless color superconducting phases was confirmed in refs .
@xcite , and generalized to nonzero temperatures in refs .
in addition , it was also shown that a gapless cfl ( gcfl ) phase could appear in neutral strange quark matter @xcite .
at nonzero temperature , the gcfl phase and several other new phases ( e.g. , the so - called dsc and usc phases ) were studied in refs . @xcite . if the surface tension is sufficiently small , as suggested in ref .
@xcite , the mixed phase composed of the 2sc phase and the normal quark phase will be more favored @xcite .
it was also suggested that a non - relativistic analogue of gapless superconducting phases could appear in a trapped gas of cold fermionic atoms @xcite .
( note that an alternative ground state for the atomic system , similar to the quark mixed phases in refs .
@xcite , was proposed in ref .
@xcite . ) while the basic properties of gapless color superconducting phases have been established in refs .
@xcite , there is probably much more to be learned about these phases in the near future . in this paper , we give a detailed derivation of the gluon and photon screening masses in neutral dense quark matter ( the results were briefly presented in ref .
we consider the general case of two - flavor quark matter , covering both the gapped and the gapless 2sc phases .
the gluon screening properties in the case of the ideal 2sc phase ( i.e. , without a mismatch between the fermi momenta of different quarks ) were considered in detail in refs .
some general features of the gluon screening in the g2sc phase were conjectured in refs .
@xcite without performing the calculation . as we shall see below , the actual results for the debye screening masses are in general agreement with the conjecture in refs .
the meissner ( magnetic ) screening properties , however , are very different @xcite .
the calculations in this paper indicate a chromomagnetic instability in neutral dense quark matter for a range of parameters in the model . as we shall argue in sec
. [ discussion ] , this instability may lead to a gluon condensation in dense quark matter .
this paper is organized as follows .
the linear response theory is briefly reviewed in sec .
[ lrt ] . after that , in sec .
[ qp ] , we introduce the model and set up the main notation .
there , we also present the general expression for the quark propagator in the color superconducting ground state of neutral dense quark matter . in sec .
[ gpt ] , we briefly discuss the general expression for the polarization tensor in dense quark matter . in sec .
[ gpt11 ] , we study the polarization tensor @xmath1 for @xmath2 and derive the corresponding expressions for the debye and meissner screening masses .
we show that , in accordance with the symmetry breaking pattern , there is no meissner effect ( i.e. , no higgs mechanism ) in this sector of the gauge theory .
the polarization tensor @xmath1 for @xmath3 ( i.e. , the 8th gluon and the photon ) is discussed in sec .
[ gpt88 ] .
there we derive the debye and meissner screening masses and briefly discuss their properties .
also , the mixing between the 8th gluon and the photon is discussed . in sec .
[ gpt44 ] , we study the polarization tensor @xmath1 for @xmath4 . the results for the debye and meissner screening masses are presented . as we show ,
the debye screening mass is given by a rather simple expression that naturally interpolates between the limits of the normal phase and the ideal 2sc phase .
the meissner mass , on the other hand , has an unexpected property .
its value squared is negative in a range of parameters , indicating a chromomagnetic instability in dense quark matter .
the discussion of the main results is given in sec .
[ discussion ] .
our findings are summarized in sec .
[ conclusion ] .
several appendices at the end contain useful formulas and some details of the calculation .
in order to present a self - contained discussion of the screening properties of dense quark matter , we start this paper with a brief introduction into the linear response theory and a discussion of the physical meaning of the polarization tensor in a gauge theory . the advanced reader , therefore , may skip this section and go directly to sec . [ qp ] .
the response of matter to an external perturbation is the main source of knowledge about properties of matter .
the linear response theory is the simplest framework that is often used to calculate such a response . in application to quark matter , for example , one studies a response of the system to an external source @xmath5 .
the source is coupled to the quantum gauge field .
the corresponding interaction part of the action reads @xmath6 where @xmath7 is the classical field associated with the external source @xmath5 , and @xmath8 is the inverse free gluon propagator . because of the presence of the external source , the expectation value of the gauge field becomes nonzero . in the linear response theory ,
it is given by the kubo s formula @xcite , @xmath9 where @xmath10 is the retarded gluon propagator .
in momentum space , this relation takes the following form : @xmath11 where @xmath12 is the energy - momentum four - vector . by making use of this result , it is instructive to derive an expression for the induced current
. it is given by @xmath13 \langle a^{b,\nu } ( p ) \rangle \nonumber\\ & \equiv & \pi_{\mu\nu}^{ab}(p ) \langle a^{b,\nu } ( p ) \rangle , \label{j - ind}\end{aligned}\ ] ] where @xmath14 is the gluon self - energy ( or the gluon polarization tensor ) . by definition , this is the one - particle irreducible part of the gluon two - point function .
the structure of this function is constrained by the gauge symmetry . to see this we consider the slavnov - taylor identity for the full gluon propagator @xcite .
the explicit form of this identity depends on a specific gauge fixing . in the covariant gauge ,
for example , it reads @xmath15 where @xmath16 is the gauge fixing parameter . in vacuum , where lorentz symmetry is not broken
, this relation implies that the gluon self - energy is transverse , i.e. , @xmath17 . because of this constraint , the tensor structure of @xmath14 in vacuum is fixed unambiguously , @xmath18 where the metric tensor is defined as @xmath19 . at nonzero temperatures and/or densities ,
the lorentz symmetry is broken down to its subgroup of spatial rotations @xmath20 .
then , the tensor structure of the gluon self - energy can have a more general form , @xmath21 where @xmath22 is a time - like four - vector that specifies the rest frame of the quark system and @xmath23 is the momentum three - vector with the absolute value @xmath24 .
the component functions @xmath25 , @xmath26 , @xmath27 , and @xmath28 are functions of @xmath29 and @xmath30 .
now , the slavnov - taylor identity in the covariant gauge takes the form @xmath31 ( for simplicity , the superscripts `` ab '' were omitted here ) .
it is easy to check that this is less restrictive than the transversality condition , @xmath32 , required in vacuum .
indeed , the transversality is equivalent to the following set of two relations between the component functions : @xmath33 while these are sufficient conditions to fulfil the slavnov - taylor identity in eq .
( [ st ] ) , they are not the necessary conditions in a non - abelian gauge theory when the lorentz symmetry is broken .
( it should be emphasized , however , that these are the necessary conditions in abelian gauge theories at nonzero temperatures and/or densities @xcite . ) in this paper , we study the polarization tensor @xmath14 in the case of dense quark matter which permits color superconductivity .
in particular , we discuss how the structure of the polarization tensor is affected by the ( gapless ) color superconductivity .
it is usually said that superconductivity is a result of a gauge symmetry breaking .
this common misleading statement may suggest that the polarization tensor @xmath14 does not need to satisfy the slavnov - taylor identity ( [ st1 ] ) .
in fact , this is not the case because a local ( gauge ) symmetry can never be truly broken @xcite . in practice , when doing specific calculations in gauge theories , one always breaks the local symmetry by a gauge fixing . as a result , it is only a global symmetry that may remain unbroken after the gauge choice is made .
for example , in the covariant gauge which we discussed above , the global color symmetry of the qcd action is left unbroken .
then , in a color superconducting phase of quark matter , this global symmetry is broken .
in such a description , the goldstone theorem requires the appearance of an appropriate number of the nambu - goldstone bosons ( cf . , collective excitations in ref .
obviously , the appearance of these additional degrees of freedom is an artifact of the gauge fixing . nevertheless , their inclusion in the analysis is important in order to insure that the general requirements of the gauge symmetry , such as the slavnov - taylor identity ( [ st1 ] ) , are fulfilled @xcite .
having said this , one should appreciate that the nambu - goldstone bosons in question are not the physical degrees of freedom .
this conclusion is easy to reach by noticing that there exist a gauge , namely the so - called unitary gauge , in which these bosons can be completely eliminated . in a way ,
their role is similar to the role of the faddeev - popov ghosts @xcite .
while both types of fields are unphysical , they are necessary for a consistent description of a gauge theory .
now let us further discuss the physical meaning of the polarization tensor . from a relation similar to that in eq .
( [ j - ind ] ) , it is clear that , in an abelian theory such as qed , this tensor is directly related to an observable quantity , namely to the induced current .
the corresponding current is a gauge invariant quantity in an abelian theory .
in contrast , the induced current in a non - abelian theory is not a gauge invariant quantity .
then , the physical meaning of the polarization tensor is not so clear . in spite of this difficulty
, we shall use the same interpretation of the polarization tensor in quark matter as in an abelian theory . in this paper
, we study static large - distance , electric and magnetic , screening properties of quark matter .
these describe the response of the system to a static perturbation from color / electric charges and currents .
the static limit means that @xmath34 . in this case
, the only nontrivial components of the polarization tensor will be @xmath35 and @xmath36 which depend only on @xmath30 .
note that @xmath37 [ pi - k - h ] let us denote the values of the two nontrivial component functions in the limit @xmath38 ( large distances ) as follows : @xmath39 [ m - k - h ] the quantities @xmath40 and @xmath41 are the debye and meissner screening masses , respectively .
it can be shown , see for example ref .
@xcite , that the quantity @xmath40 determines the large - distance behavior of the screened potential created by a static color / electric charge , i.e. , @xmath42 . by making use of the analogy with solid state physics systems
, we say that a system is a metal when @xmath40 is nonzero , and it is an insulator when @xmath40 is zero .
the quantity @xmath41 , when nonzero , determines the large - distance fall - off of the ( chromo-)magnetic field inside a color superconductor , i.e. , @xmath43 .
obviously , nonzero @xmath41 is an indication of the meissner effect . in the normal phase , on the other hand ,
the value of @xmath41 is vanishing .
in this paper , we continue the study of dense two - flavor quark matter constrained by the conditions of the charge neutrality and the @xmath0-equilibrium @xcite .
the use of phenomenological nambu - jona - lasinio ( njl ) type models has proved to be very convenient in such studies .
the njl model can be thought of as a low energy theory of qcd in which ( massive ) gluons , as independent degrees of freedom , are integrated out .
the gluons could be reintroduced back by gauging the color symmetry in the njl model , providing a semirigorous framework for studying the effect of the cooper pairing on the physical properties of gluons . in order to study the gluon screening properties in dense quark matter ,
we need to know the quark coupling to the gauge fields .
this is determined by the quadratic part of the quark lagrangian density @xmath44 where @xmath45 and @xmath46 are the generators of @xmath47 and @xmath48 gauge groups .
the coupling constants of the strong interactions and of the electromagnetism are denoted by @xmath49 and @xmath50 , respectively .
the up and down quark masses are assumed to be the same ( @xmath51 ) .
the quark spinor field has the following color - flavor structure : @xmath52 where @xmath53 is the flavor index and @xmath54 is the color index . in @xmath0-equilibrium
, the matrix of chemical potentials in the color - flavor space @xmath55 is given in terms of the quark chemical potential @xmath56 ( note that @xmath57 is the baryon chemical potential ) , the chemical potential for the electrical charge @xmath58 and the color chemical potential @xmath59 , @xmath60 in qcd the color chemical potential @xmath59 comes as a result of a nonzero expectation value of the 8th gluon field induced by the cooper pairing @xcite .
its absolute value is small because it is suppressed parametrically by the quark chemical potential , @xmath61 .
the explicit expressions for nontrivial elements of matrix @xmath55 read @xmath62 to study color superconducting phases , it is convenient to introduce the following @xmath63-component nambu - gorkov spinors : @xmath64 where @xmath65 is the charge - conjugate spinor , and @xmath66 is the charge - conjugation matrix . in this basis , the quadratic part of the quark lagrangian density @xmath67 becomes @xmath68 where the explicit form of vertex @xmath69 is @xmath70 in momentum space , the inverse free quark propagator @xmath71 reads @xmath72^{-1 } = \left(\begin{array}{cc } \left[g_0^{+}(k)\right]^{-1 } & 0 \\ 0 & \left[g_0^{-}(k)\right]^{-1 } \end{array}\right).\ ] ] the explicit form for the diagonal elements is @xmath73^{-1 } & = & \gamma^0 \left[(k_0-e_{ur}^{\mp})\lambda^{+}_{k } + ( k_0+e_{ur}^{\pm})\lambda^{-}_{k } \right]{\cal p}^{(1 ) } \nonumber\\ & + & \gamma^0 \left[(k_0-e_{dg}^{\mp})\lambda^{+}_{k } + ( k_0+e_{dg}^{\pm})\lambda^{-}_{k}\right]{\cal p}^{(2 ) } \nonumber\\ & + & \gamma^0 \left[(k_0-e_{ub}^{\mp})\lambda^{+}_{k } + ( k_0+e_{ub}^{\pm})\lambda^{-}_{k}\right]{\cal p}^{(3 ) } \nonumber\\ & + & \gamma^0 \left[(k_0-e_{db}^{\mp})\lambda^{+}_{k } + ( k_0+e_{db}^{\pm})\lambda^{-}_{k}\right]{\cal p}^{(4)},\label{g0}\end{aligned}\ ] ] with the notation @xmath74 and @xmath75 .
the four projectors @xmath76 ( with @xmath77 ) in the 6-dimensional color - flavor space are defined as follows : @xmath78 it is not difficult to see that @xmath79 projects out the red up and the green up quarks , and @xmath80 projects out the red down and the green down quarks . the projectors @xmath81 and @xmath82 project out the blue up and the blue down quarks , respectively . in eq .
( [ g0 ] ) , we also introduced the energy projectors , @xmath83 these projectors satisfy the following relations @xcite : @xmath84 where @xmath85 is an alternative set of energy projectors . in the chiral limit ,
the two sets of projectors in eqs .
( [ lambda - k ] ) and ( [ t - lambda - k ] ) coincide .
the full quark propagator in a color superconducting phase takes the following form : @xmath86^{-1 } = \left(\begin{array}{cc } \left[g_0^{+}(k)\right]^{-1 } & \delta^- \\ \delta^+ & \left[g_0^{-}(k)\right]^{-1 } \end{array}\right ) , \label{spinverse}\ ] ] with @xmath87 where @xmath88 is the antisymmetric tensor in the color subspace spanned by the red and green colors , while @xmath89 is the antisymmetric tensor in the flavor space .
the value of the gap parameter @xmath90 is determined from an appropriate gap equation , while the values of the chemical potentials @xmath58 and @xmath59 are determined from charge neutrality conditions @xcite .
the explicit form of the gap equation and the neutrality conditions is not important for the purposes of this paper . from eq .
( [ spinverse ] ) , we obtain @xmath91 where @xmath92^{-1 } , \\
\xi^{\pm } & = & -g_0^{\mp } \delta^{\pm } g^{\pm } , \end{aligned}\ ] ] with the free quark propagators @xmath93{\cal p}^{(1 ) } + \gamma^0 \left[\frac{\tilde{\lambda}^{+}_{k}}{k_0+e_{dg}^{\pm } } + \frac{\tilde{\lambda}^{-}_{k}}{k_0-e_{dg}^{\mp}}\right]{\cal p}^{(2 ) } \nonumber\\ & + & \gamma^0 \left[\frac{\tilde{\lambda}^{+}_{k}}{k_0+e_{ub}^{\pm } } + \frac{\tilde{\lambda}^{-}_{k}}{k_0-e_{ub}^{\mp}}\right]{\cal p}^{(3 ) } + \gamma^0 \left[\frac{\tilde{\lambda}^{+}_{k}}{k_0+e_{db}^{\pm } } + \frac{\tilde{\lambda}^{-}_{k}}{k_0-e_{db}^{\mp}}\right]{\cal p}^{(4)},\end{aligned}\ ] ] obtained from eq .
( [ g0 ] ) . by making use of the following relations : @xmath94 one can derive an explicit form of the nambu - gorkov components of the full propagator , @xmath95 the explicit form of the functions @xmath96 and @xmath97 reads @xmath98 [ g_i ] and @xmath99 [ xi_i ] where the following notation was used : @xmath100 in the following sections , we use the full quark propagator in eq . ( [ quarkpropagator ] ) to construct the polarization tensor for the gauge fields .
in dense quark matter , screening effects play a very important role at length scales larger than the average distance between quarks . in the normal phase , for example , the main effects are the debye screening and the landau damping .
these are the properties that can be extracted from the behavior of the polarization tensor .
the polarization tensor in dense matter is given by the so - called hard dense loop ( hdl ) approximation @xcite .
this approximation results from taking into account only the dominant one - loop quark contribution in which the internal quark momenta are hard ( i.e. , typical momenta are of order @xmath56 ) .
the density of quark states with hard momenta is proportional to @xmath101 ( i.e. , the density of states at the fermi surface ) . because of this large density of states ,
the quark hdl contribution is large compared to the contributions from the gluon and the ghost loops .
therefore , the gluon and the ghost contributions are not included in the hdl approximation . in the 2sc / g2sc phase of dense quark matter
, the polarization tensor is given approximately by the following one - loop expression @xcite : @xmath102 , \ ] ] where the trace runs over the dirac , color , flavor , and nambu - gokov indices .
the gluon part ( @xmath103 ) of this polarization tensor reduces to the standard hdl result in the normal phase ( @xmath104 ) , @xmath105 \nonumber\\ & & + \frac{\mathbf{p}^{\mu}\mathbf{p}^{\nu}}{p^2 } \frac{p_0 ^ 2}{p^2}q\left(\frac{p_0}{p}\right ) + \frac{p_0}{p}\left(u^{\mu } \frac{\mathbf{p}^{\nu}}{p } + \frac{\mathbf{p}^{\mu}}{p } u^{\nu}\right)q\left(\frac{p_0}{p}\right ) \bigg\ } , \label{pi - hdl}\end{aligned}\ ] ] where @xmath106 , and @xmath107 in the static limit ( @xmath34 ) , we obtain @xmath108 , and the polarization tensor becomes @xmath109 this gluon polarization tensor describes the static screening of quark color charges at large distances in the normal phase of dense quark matter . by comparing with eqs .
( [ pi - k - h ] ) and ( [ m - k - h ] ) , we derive the corresponding expression for the debye screening mass , @xmath110 as it should be , the meissner screening mass is zero in the normal phase .
in this section , we start with the screening properties of the @xmath111 gluons . these are the gluons of the unbroken @xmath112 subgroup which couple only to the red and green quarks .
the corresponding expression for the polarization tensor is diagonal , @xmath113 . after performing the traces over the color , flavor and nambu - gorkov indices , we arrive at @xmath114 , \label{pi11}\end{aligned}\ ] ] where @xmath115 is the temperature and @xmath116 .
here we use the imaginary time formalism , and the energy integration is replaced by the sum over the fermionic matsubara frequencies @xmath117 .
the explicit expressions for the components @xmath118 and @xmath97 of the quark propagator are given in eqs .
( [ g_i ] ) and ( [ xi_i ] ) . after the summation over the matsubara frequencies ,
we obtain the result in the following form : @xmath119 .
\label{pi11-n}\end{aligned}\ ] ] in this expression , we introduced the following notation for the two types of dirac traces : @xmath120 , \\ { \cal u}_{e_1e_2}^{\mu\nu } & = & { \rm tr}_{\rm d}[\gamma^{\mu}\gamma^5{\tilde \lambda}_{k}^{e_1 } \gamma^{\nu}\gamma^5{\tilde \lambda}_{k'}^{e_2}],\end{aligned}\ ] ] [ d - traces ] with @xmath121 . to leading order in @xmath122 ,
the results for these traces are given in eqs .
( [ tr2++])([tr1 + - ] ) in appendix [ tu - matrix ] .
the expressions for the coefficient functions @xmath123 at zero and nonzero temperatures are given in appendix [ cof - tu11 ] .
we write the integral over the three - momentum in eq .
( [ pi11-n ] ) as @xmath124 where @xmath125 is the polar angle and @xmath126 is the cosine of the angle between the three - momenta @xmath127 and @xmath128 . after performing the integral over the polar angle @xmath125 ,
see eqs .
( [ phi - int1 ] ) and ( [ phi - int2 ] ) , the corresponding traces in the integrand can be replaced by the following angular averaged expressions : @xmath129 [ traces - xi ] now we would like to note that there are two different types of coefficient functions in eq .
( [ pi11-n ] ) .
the coefficients @xmath130 , @xmath131 , @xmath132 and @xmath133 originate from particle - hole and antiparticle - antiparticle loops . in the leading order approximation
, we drop the antiparticle - antiparticle contributions to the polarization tensor .
these are suppressed by an inverse power of the quark chemical potential .
in addition , in the static long wavelength limit ( @xmath34 and @xmath134 ) , the particle - hole contributions simplify considerably .
the approximate expressions read @xmath135 - \delta(-e_{\delta , k}^- + \delta\mu ) \frac { ( e_{\delta , k}^-)^2 + ( e_k^-)^2 - 2 e_{\delta , k}^- e_k^- } { 4 ( e_{\delta , k}^-)^2 } , \\ c_{\pm\pm}^{22 } & \simeq & -\frac{\delta^2}{4 ( e_{\delta , k}^{-})^3}[1-\theta(-e_{\delta , k}^- + \delta\mu ) ] - \delta(-e_{\delta , k}^- + \delta\mu ) \frac{(e_{\delta , k}^-)^2 + ( e_{k}^-)^2 + 2 e_{\delta , k}^- e_k^- } { 4 ( e_{\delta , k}^-)^2 } , \\
c_{\pm\mp}^{12 } & = & c_{\pm\mp}^{21 } \simeq -\frac{\delta^2}{4 ( e_{\delta , k}^{-})^3}[1-\theta(-e_{\delta , k}^- + \delta\mu ) ] + \delta(-e_{\delta , k}^- + \delta\mu ) \frac{\delta^2}{4 ( e_{\delta , k}^-)^2}.\end{aligned}\ ] ] [ p - h ] here we neglected the corrections due to nonzero quark masses .
this is justified if the shift of the quark fermi momenta due to such masses is small , i.e. , @xmath136 .
we also used the following relations valid when @xmath134 : @xmath137 as is easy to check , the terms with the @xmath138- and @xmath139-functions in eq .
( [ p - h ] ) contribute only in the g2sc phase when @xmath140 ( this also includes the normal phase as a limiting case with @xmath104 ) .
now , the coefficients @xmath141 , @xmath142 , @xmath143 and @xmath144 are made of particle - antiparticle loops only .
their contributions to the polarization tensor are not negligible although they are formally suppressed by an inverse power of the quark chemical potential .
in fact , the corresponding contributions are ultraviolet divergent .
the divergences are removed by subtractions of the corresponding vacuum terms .
one can also check that a nonzero gap parameter ( @xmath145 ) leads only to a small correction to these particle - antiparticle contributions .
thus , we approximate the corresponding coefficient functions as follows : @xmath146 [ p - a ] while the traces in eq .
( [ traces - xi ] ) depend on the angular coordinate @xmath126 , the approximate coefficient functions @xmath123 in eqs .
( [ p - h ] ) and ( [ p - a ] ) do not .
therefore , the corresponding angular integration in eq .
( [ pi11-n ] ) can be easily performed , @xmath147 [ tu - integral ] the results on the right hand side are independent of the momentum @xmath30 .
therefore , in order to derive the final expression for the polarization tensor , we need to calculate only the following momentum integrals : @xmath148 , \label{ph - t}\\ \int dk k^2 c_{\pm\mp}^{12,21 } & \simeq & - \frac{1}{2 } ~ { \bar \mu}^2 \left[1- \frac{\delta\mu\ , \theta ( \delta\mu-\delta)}{\sqrt{{\delta\mu}^2-\delta^2 } } \right ] \label{ph - u},\\ \int d k k^2 c_{\pm\mp}^{11,22 } & \simeq & \frac{1}{2}{\bar \mu}^2 , \label{panp - t } \\
\int d k k^2 c_{\pm\pm}^{12,21 } & \simeq & 0 .
\label{panp - u}\end{aligned}\ ] ] [ c - integral ] the details of the calculation are given in appendix [ int-11 ] .
note that , in the 2sc phase when @xmath149 , the contributions from the normal ( @xmath150 ) and abnormal ( @xmath151 ) quark loops are equal , see eqs .
( [ ph - t ] ) and ( [ ph - u ] ) .
the additional contributions in the g2sc phase have equal absolute values but differ in sign . as for the particle - antiparticle quark loop ,
only the normal ( @xmath152 ) contribution is nonvanishing . by substituting the results in eqs .
( [ tu - integral ] ) and ( [ c - integral ] ) into eq .
( [ pi11-n ] ) , we derive the following expression for the polarization tensor @xmath153 : @xmath154 \nonumber\\ & \simeq & -\frac{4\alpha_s \bar\mu^2}{\pi } \frac{\delta\mu } { \sqrt{{\delta\mu}^2-\delta^2 } } \theta(\delta\mu-\delta ) u^{\mu}u^{\nu}. \label{scr11}\end{aligned}\ ] ] only the @xmath155-component of this polarization tensor is nonzero .
this component determines the debye screening mass , @xmath156 in the 2sc phase ( @xmath157 ) , where there are no gapless excitations charged with respect to the unbroken @xmath112 subgroup , this screening mass is identically zero @xcite .
in contrast , it is nonzero in the g2sc phase ( @xmath140 ) .
in fact , its value is proportional to the density of the gapless quasiparticle states . as for the meissner screening mass ,
its value vanishes in the 2sc phase as well as in the g2sc phase , @xmath158
in the 2sc / g2sc phase , the diquark condensate breaks the gauge symmetry @xmath159 down to the @xmath160 subgroup .
the structure of the condensate in the ground is such that the 8th gluon and the photon mix .
the new medium fields are given by the following linear combinations of the vacuum fields @xcite : @xmath161 [ tilde - a - gamma ] where the mixing angle @xmath162 is determined from the structure of the condensate by using simple group - theoretical arguments , @xmath163 the generator of the medium ( low - energy ) electromagnetism @xmath164 is defined as follows : @xmath165 . the 8th gluon can probe the cooper - paired red and green quarks , as well as the unpaired blue quarks .
after the traces over the color , the flavor and the nambu - gorkov indices are performed , the polarization tensor for the 8th gluon can be expressed as @xmath166 , \label{pi88 } \\
\pi_{88,b}^{\mu\nu}(p ) & = & \frac{g^2t}{4}\sum_n\int \frac{d^3 { \mathbf k}}{(2\pi)^3 } \mbox{tr}_{\rm d } \left [ \gamma^{\mu } g_{3}^+(k ) \gamma^{\nu } g_{3}^+(k ' ) + \gamma^{\mu } g_{3}^-(k ) \gamma^{\nu}g_{3}^-(k ' ) \right.\nonumber\\ & + & \left .
\gamma^{\mu } g_{4}^+(k ) \gamma^{\nu}g_{4}^+(k ' ) + \gamma^{\mu } g_{4}^-(k ) \gamma^{\nu}g_{4}^-(k ' ) \right ] .
\label{pi88b}\end{aligned}\ ] ] note that the contribution @xmath167 of paired quarks , up to a sign in front of the contribution from the abnormal quark loops , is the same as the expression for @xmath153 in eq .
( [ pi11 ] ) .
therefore , by following the same steps as in the previous section , we easily derive the result @xmath168 \nonumber\\ & \simeq & -\frac{4\alpha_s \bar\mu^2}{\pi } u^{\mu}u^{\nu } -\frac{4\alpha_s \bar\mu^2}{3\pi } \left [ 1- \frac{\delta\mu\ , \theta(\delta\mu-\delta)}{\sqrt{{\delta\mu}^2-\delta^2 } } \right ] \left ( g^{\mu\nu}-u^{\mu}u^{\nu}\right ) , \label{scr88 - 1}\end{aligned}\ ] ] [ c.f .
( [ scr11 ] ) ] .
the unpaired blue quarks give the standard normal phase hdl contribution @xmath169 , see eq .
( [ pi - hdl ] ) . in the static ( @xmath34 ) long - wavelength ( @xmath134 ) limit , this leads to the following result : @xmath170 thus , the final result for the polarization tensor of the 8th gluon reads @xmath171 \left ( g^{\mu\nu}-u^{\mu}u^{\nu}\right ) .
\label{scr88}\end{aligned}\ ] ] because of the symmetry considerations , one should expect a nontrivial gluon - photon mixing in the ground state .
so , we proceed to the calculation of the photon polarization tensor .
the general expression for this tensor is @xmath172 \right.\nonumber\\ & + & \left .
\gamma^{\mu } g_{2}^+(k ) \gamma^{\nu}g_{2}^+(k ' ) + \gamma^{\mu } g_{2}^-(k ) \gamma^{\nu}g_{2}^-(k ' ) \right.\nonumber\\ & + & \left .
2\left[\gamma^{\mu}\xi_{12}^-(k ) \gamma^{\nu}\xi_{21}^+(k ' ) + \gamma^{\mu}\xi_{12}^+(k ) \gamma^{\nu}\xi_{21}^-(k')\right ] \right.\nonumber\\ & + & \left .
2\left[\gamma^{\mu}\xi_{21}^-(k ) \gamma^{\nu}\xi_{12}^+ ( k ' ) + \gamma^{\mu}\xi_{21}^+(k ) \gamma^{\nu}\xi_{12}^-(k')\right]\right\ } , \label{pi99 - 2 } \\
\pi_{99,b}^{\mu\nu}(p ) & = & \frac{e^2t}{9}\sum_n\int \frac{d^3 { \mathbf k}}{(2\pi)^3 } \mbox{tr}_{\rm d } \left\ { 4\left[\gamma^{\mu } g_{3}^+(k ) \gamma^{\nu } g_{3}^+(k ' ) + \gamma^{\mu } g_{3}^-(k ) \gamma^{\nu}g_{3}^-(k')\right ] \right.\nonumber\\ & + & \left . \gamma^{\mu } g_{4}^+(k ) \gamma^{\nu}g_{4}^+(k ' ) + \gamma^{\mu } g_{4}^-(k ) \gamma^{\nu}g_{4}^-(k ' ) \right\}. \label{pi99b}\end{aligned}\ ] ] let us first start with the contribution of the unpaired blue quarks @xmath173 .
this is proportional to the standard hdl result , @xmath174 where @xmath175 is the fine structure constant and @xmath176 is given in eq .
( [ pi - hdl ] ) . at @xmath34 and @xmath134 , we arrive at @xmath177 to calculate the contribution of the paired quarks @xmath178 , we use the same approach as in the previous section . after performing the matsubara frequency summation
, we obtain the following representation : @xmath179 .
\label{pi99-n}\end{aligned}\ ] ] by substituting the results in eqs .
( [ tu - integral ] ) and ( [ c - integral ] ) into eq .
( [ pi99-n ] ) , we derive the following expression for the polarization tensor @xmath180 in the static long - wavelength limit : @xmath181 \nonumber\\ & \simeq & -\frac{8\alpha \bar\mu^2}{9\pi } \left[1+\frac{9\delta\mu\ , \theta(\delta\mu-\delta ) } { \sqrt{{\delta\mu}^2-\delta^2 } } \right ] u^{\mu}u^{\nu } - \frac{8\alpha \bar\mu^2}{27\pi } \left[1-\frac{\delta\mu\ , \theta(\delta\mu-\delta ) } { \sqrt{{\delta\mu}^2-\delta^2 } } \right ] \left(g^{\mu\nu}-u^{\mu}u^{\nu}\right ) .
\label{scr99 - 2}\end{aligned}\ ] ] by combining the results in eqs .
( [ scr99 - 1 ] ) and ( [ scr99 - 2 ] ) , we obtain the expression for the photon polarization tensor @xmath182 u^{\mu}u^{\nu } - \frac{4\alpha \bar\mu^2}{27\pi } \left[1-\frac{\delta\mu\ , \theta(\delta\mu-\delta ) } { \sqrt{{\delta\mu}^2-\delta^2 } } \right ] \left(g^{\mu\nu}-u^{\mu}u^{\nu}\right ) .
\label{scr99}\end{aligned}\ ] ] now , we calculate the gluon - photon mixed components of the polarization tensor .
the corresponding expression is given by @xmath183\right.\nonumber\\ & - & \left .
\gamma^{\mu } g_{2}^+(k ) \gamma^{\nu}g_{2}^+(k ' ) - \gamma^{\mu } g_{2}^-(k ) \gamma^{\nu}g_{2}^-(k ' ) \right.\nonumber\\ & + & \left .
\gamma^{\mu}\xi_{12}^-(k ) \gamma^{\nu}\xi_{21}^+(k ' ) + \gamma^{\mu}\xi_{12}^+(k ) \gamma^{\nu}\xi_{21}^-(k')\right.\nonumber\\ & - & \left
. 2 \left[\gamma^{\mu}\xi_{21}^-(k ) \gamma^{\nu}\xi_{12}^+ ( k ' ) + \gamma^{\mu}\xi_{21}^+(k ) \gamma^{\nu}\xi_{12}^-(k')\right]\right\ } , \label{pi89 - 2 } \\
\pi_{89,b}^{\mu\nu}(p ) & = & \frac{egt}{3\sqrt{3}}\sum_n\int \frac{d^3 { \mathbf k}}{(2\pi)^3 } \mbox{tr}_{\rm d } \left\ { -2\left[\gamma^{\mu } g_{3}^+(k ) \gamma^{\nu } g_{3}^+(k ' ) + \gamma^{\mu } g_{3}^-(k ) \gamma^{\nu}g_{3}^-(k')\right ] \right.\nonumber\\ & + & \left . \gamma^{\mu } g_{4}^+(k ) \gamma^{\nu}g_{4}^+(k ' ) + \gamma^{\mu } g_{4}^-(k ) \gamma^{\nu}g_{4}^-(k ' ) \right\}. \label{pi89b}\end{aligned}\ ] ] the contribution of the unpaired blue quarks is proportional to the hdl expression , @xmath184 with @xmath176 defined in eq .
( [ pi - hdl ] ) . at @xmath34 and @xmath134
, we obtain @xmath185 to calculate the contribution of the paired quarks , @xmath186 , we first perform the matsubara frequency summation .
the result is @xmath187 . \label{pi89-n}\end{aligned}\ ] ] by substituting the results in eqs .
( [ tu - integral ] ) and ( [ c - integral ] ) into eq .
( [ pi89-n ] ) , we get the following expression for the polarization tensor @xmath188 in the static long - wavelength limit : @xmath189 \nonumber\\ & \simeq & -\frac{8\sqrt{\alpha\alpha_s } \bar\mu^2}{3\sqrt{3}\pi } u^{\mu}u^{\nu } -\frac{8\sqrt{\alpha\alpha_s } \bar\mu^2}{9\sqrt{3}\pi } \left [ 1- \frac{\delta\mu\ , \theta(\delta\mu-\delta)}{\sqrt{{\delta\mu}^2-\delta^2 } } \right ] \left ( g^{\mu\nu}-u^{\mu}u^{\nu}\right ) , \label{scr89 - 2}\end{aligned}\ ] ] [ c.f .
( [ scr88 - 1 ] ) ] . by substituting the results in eqs . ( [ scr89 - 1 ] ) and ( [ scr89 - 2 ] ) into eq . ( [ pi89 ] )
, we obtain the expression for the mixed components of the polarization tensor @xmath190 \left ( g^{\mu\nu}-u^{\mu}u^{\nu}\right ) .
\label{scr89}\end{aligned}\ ] ] it is important to emphasize that the explicit result for the electrical @xmath155-component of this mixing gluon - photon polarization tensor is vanishing at @xmath34 and @xmath134 .
this is similar to the ideal 2sc case considered in ref .
( because of this , one should be careful when interpreting the results for the debye screening masses in a different basis of gauge fields @xcite . )
the expressions for the debye screening masses read @xmath191 [ m_d_8-gamma ] in order to obtain the meissner screening masses , we first derive all components of the polarization tensor that span the space of the 8th gluon and the photon . at @xmath34 and @xmath134 ,
the corresponding nonzero components , denoted as @xmath192 , read @xmath193 and @xmath194 .
the mixing disappears in the basis of the fields in eq .
( [ tilde - a - gamma ] ) .
the meissner screening mass for the @xmath195th gluon field is @xmath196 and the meissner screening mass for the medium photon @xmath197 is vanishing , @xmath198 .
this is consistent with the absence of the meissner effect for the unbroken @xmath199 . as is easy to see from eq .
( [ m_m_8 ] ) , the medium modified @xmath195th gluon has a chromomagnetic instability in the gapless 2sc phase .
this is because the meissner screening mass squared is _ negative _ when @xmath200 .
after performing the traces over the color , the flavor and the nambu - gorkov indices , the diagonal components of the polarization tensor @xmath202 with @xmath203 have the form @xmath204.\end{aligned}\ ] ] [ note that @xmath205 . ] apart from the diagonal elements , there are also nonzero off - diagonal elements , @xmath206 with @xmath207 , \end{aligned}\ ] ] the off - diagonal components of the gluon self - energy in eq .
( [ odd - diag ] ) are nonzero in general . the physical gluon fields in the 2sc / g2sc phase are the following linear combinations : @xmath208 and @xmath209 @xcite . these new fields ,
@xmath210 and @xmath211 , describe two pairs of massive vector particles with well defined electomagnetic charges , @xmath212 .
the components of the polarization tensor in the new basis read @xmath213 , \end{aligned}\ ] ] and @xmath214.\end{aligned}\ ] ] after matsubara frequency summation , the polarization tensors can be written in the following form : @xmath215 , \label{pi44-n } \\ \tilde{\pi}_{55}^{\mu\nu}(p ) & = & \pi\alpha_s \int \frac{d^3 { \mathbf k}}{(2\pi)^3 } \left[c_{++}^{55 } { \cal t}^{\mu\nu}_{++ } + c_{--}^{55 } { \cal t}^{\mu\nu}_{-- } + c_{+-}^{55 } { \cal t}^{\mu\nu}_{+- } + c_{-+}^{55 } { \cal t}^{\mu\nu}_{-+}\right ] .
\label{pi55-n}\end{aligned}\ ] ] [ pi44 - 55 ] the explicit expressions for the coefficient functions @xmath216 and @xmath217 at zero and nonzero temperatures are given in appendix [ cof - tu44 ] .
here we quote only the approximate zero temperature results at @xmath34 and @xmath218 .
the particle - hole contributions are @xmath219 \frac { e_{b , k}^- + e_{k}^-}{(e_{\delta , k}^-)^2 - ( e_{b , k}^-)^2 } \nonumber \\ & + & \frac{\theta(-e_{\delta , k}^-+\delta\mu)}{e_{\delta , k}^- } \frac{(e_{\delta , k}^-)^2 + e_{k}^- e_{b , k}^- } { ( e_{\delta , k}^-)^2 - ( e_{b , k}^-)^2},\label{cccc}\end{aligned}\ ] ] and the particle - antiparticle contributions are @xmath220 where @xmath221 and @xmath222 .
the vacuum contribution was subtracted in eq .
( [ c_44,55 ] ) in order to remove ultraviolet divergences in the polarization tensor . to proceed , we need to calculate the following two types of momentum integrals ( for the details of the calculation see appendix [ int-44 ] ) : @xmath223 \nonumber\\ & \approx & -2\bar\mu^2\left [ \frac{\delta^2 + 2(\delta\mu)^2}{2\delta^2 } -\frac{\delta\mu\sqrt{\delta\mu^2-\delta^2}}{\delta^2 } \theta\left(\delta\mu-\delta\right ) \right],\quad\mbox{for}\quad \mu_8\to 0 , \label{c4pp - integral}\end{aligned}\ ] ] and @xmath224 [ c4-all ] by substituting the results in eqs .
( [ tu - integral ] ) and ( [ c4-all ] ) into eq .
( [ pi44 - 55 ] ) , we derive expressions for the polarization tensors @xmath225 in the static long - wavelength limit : @xmath226 \nonumber\\ & = & - \frac{4\alpha_s\bar\mu^2}{\pi } \left [ 1-\frac{\delta^2}{4\mu_8 ^ 2}\ln \frac{\left(\delta^2+\mu_8 ^ 2\right)^2 - 4\left(\mu_8\delta\mu\right)^2}{\delta^4 } -\frac{\delta^2}{4\mu_8 ^ 2}\ln \frac{\delta^4-\mu_8
^ 2\left(\delta\mu-\sqrt{\delta\mu^2-\delta^2}\right)^2 } { \delta^4-\mu_8 ^ 2\left(\delta\mu+\sqrt{\delta\mu^2-\delta^2}\right)^2 } \theta\left(\delta\mu-\delta\right ) \right ] u^{\mu}u^{\nu}\nonumber\\ & -&\frac{\alpha_s\bar\mu^2}{3\pi } \frac{\delta^2}{\mu_8 ^ 2 } \left[\ln \frac{\left(\delta^2+\mu_8 ^ 2\right)^2 - 4\left(\mu_8\delta\mu\right)^2}{\delta^4 } + \ln \frac{\delta^4-\mu_8 ^ 2\left(\delta\mu-\sqrt{\delta\mu^2-\delta^2}\right)^2 } { \delta^4-\mu_8 ^ 2\left(\delta\mu+\sqrt{\delta\mu^2-\delta^2}\right)^2 } \theta\left(\delta\mu-\delta\right ) \right ] \left(g^{\mu\nu}-u^{\mu}u^{\nu}\right).\end{aligned}\ ] ] in the neutral 2sc / g2sc phase of quark matter , the value of the color chemical potential @xmath59 is small @xcite .
therefore , a good approximation for the above polarization tensors is obtained by taking the limit @xmath227 , @xmath228u^{\mu}u^{\nu } \nonumber\\ & & -\frac{2\alpha_s\bar\mu^2}{3\pi}\left [ \frac{\delta^2 - 2(\delta\mu)^2}{\delta^2 } + 2\frac{\delta\mu\sqrt{\delta\mu^2-\delta^2}}{\delta^2 } \theta\left(\delta\mu-\delta\right ) \right]\left(g^{\mu\nu}-u^{\mu}u^{\nu}\right).\end{aligned}\ ] ] therefore , the corresponding debye and meissner screening masses are @xmath229 , \label{debye-44 } \\
m_{m,44}^2 & = & m_{m,55}^2 = \frac{2\alpha_s\bar\mu^2}{3\pi}\left [ \frac{\delta^2 - 2(\delta\mu)^2}{\delta^2 } + 2\frac{\delta\mu\sqrt{\delta\mu^2-\delta^2}}{\delta^2 } \theta\left(\delta\mu-\delta\right ) \right ] .
\label{meissner-44 } \end{aligned}\ ] ] it is clear from eq . ( [ meissner-44 ] ) that the corresponding four gluons have chromomagnetic ( plasma type ) instabilities in the gapless 2sc phase ( @xmath200 ) , as well as in the gapped 2sc phase when @xmath230 .
this is indicated by a _
negative _ value of the meissner screening mass squared .
the results of our calculation for the screening masses in neutral two - flavor quark matter are summarized in table [ dm - masses ] . there
we list the squared values of the debye and meissner screening masses in units of @xmath231 .
.[dm - masses ] the debye and meissner screening masses for gauge bosons in neutral two - flavor quark matter . by definition , @xmath232 . [
cols="<,^,^ " , ] let us first discuss the results for the debye screening masses .
the values of @xmath233 with @xmath234 as functions of the dimensionless ratio @xmath235 are shown graphically in fig .
[ fig - d - masses ] ( we do not show the debye screening mass for the photon whose value is suppressed by the fine structure constant ) .
the vanishing value of the ratio @xmath235 corresponds to the normal phase of quark matter . in this limit , as is easy to check , our results for the debye screening masses coincide with the known results in refs .
@xcite . also , in the other limiting case , @xmath236
, our results coincide with those in the ideal 2sc phase @xcite . as before , by the ideal 2sc phase of quark matter , we mean the 2sc phase without any mismatch between the fermi momenta of the up and down quarks , i.e. , @xmath237 .
we see that the value of the debye screening mass for the gluons of the unbroken @xmath112 gauge subgroup is vanishing in the 2sc phase ( @xmath238 ) . this result can be understood in the same way as in the ideal 2sc case @xcite .
it reflects the fact that there are no gapless quasiparticles charged with respect to the @xmath112 subgroup in the ground state . in the g2sc phase ,
on the other hand , there are gapless charged quasiparticles around the two `` effective '' fermi momenta @xmath239 @xcite .
the densities of states at the corresponding fermi surfaces are @xcite @xmath240 as one can check from table [ dm - masses ] , the squared value of the debye screening mass @xmath241 is proportional to the sum of these densities of the gapless states ( up to higher order corrections ) . turning these arguments around
, one could obtain the result for the debye screening mass without calculating the polarization tensor .
indeed , one can write @xcite @xmath242 and determine the overall coefficient @xmath243 by matching the expression on the right hand side with the known expression in the normal phase ( @xmath244 )
. it is interesting to consider @xmath241 in the limit when the gap @xmath90 approaches @xmath245 from the side of the gapless phase .
as one can see , the formal value of @xmath241 goes to infinity as @xmath246 when @xmath247 .
this is the consequence of having a quadratic dispersion relation for gapless quasiparticles @xmath248 when @xmath249 .
however , the infinitely large value of the debye mass has little physical meaning .
this is because the distance scales , at which the corresponding screening is set in , also become infinite in the limit @xmath249 . from table
[ dm - masses ] , we see that there is no mixing between the debye screening masses of the 8th gluon and of the vacuum photon .
it is known , however , that the physical modes of the corresponding gauge bosons in the 2sc / g2sc phase are the linear combinations given in eq .
( [ tilde - a - gamma ] ) which are different from the vacuum fields .
moreover , one of the linear combinations , see eq .
( [ tilde - gamma ] ) , describes the medium modified photon of the unbroken electromagnetism .
one might think that the absence of mixing between the electric screening masses is in conflict with the gauge invariance of the 2sc / g2sc ground state with respect to @xmath199 .
however , the two propagating modes of the low - energy photon @xmath197 of @xmath199 have transverse polarizations and , therefore , should come from the magnetic sector .
the third , electrical mode of @xmath197 is not massless .
this latter decouples from the low - energy theory and its presence is irrelevant for the gauge invariance with respect to @xmath199 .
squared values of the meissner screening masses as functions of the dimensionless ratio @xmath235 for the gluons with @xmath111 ( solid line ) , for the gluons with @xmath201 ( long - dashed line ) , and for the @xmath195th gluon ( short - dashed line ) . by definition , @xmath232.,width=302 ] squared values of the meissner screening masses as functions of the dimensionless ratio @xmath235 for the gluons with @xmath111 ( solid line ) , for the gluons with @xmath201 ( long - dashed line ) , and for the @xmath195th gluon ( short - dashed line ) . by definition , @xmath232.,width=302 ] now we discuss the results for the meissner screening masses .
the values of @xmath250 with @xmath234 as functions of the dimensionless ratio @xmath235 are shown graphically in fig .
[ fig - m - masses ] ( the screening mass for the photon , suppressed by the fine structure constant , is not shown ) .
as it should be , the results in the two limiting cases , i.e. , @xmath244 and @xmath236 , coincide with the results in the normal phase ( i.e. , no meissner effect ) and with the results in the ideal 2sc phase @xcite , respectively . in agreement with the group
theoretical arguments , the meissner masses of the three gluons of the unbroken @xmath112 subgroup are vanishing in both the gapless and the gapped 2sc phases , see the solid line in fig .
[ fig - m - masses ] .
the vanishing meissner masses come as a result of the exact cancellation between the diamagnetic particle - antiparticle and the paramagnetic particle - hole contributions .
the results in table [ dm - masses ] show that the meissner screening masses in the subspace of the 8th gluon and the vacuum photon have a nontrivial mixing . as we saw in sec .
[ gpt88 ] , the mixing disappears after switching to the description in terms of the physical modes , defined in eq .
( [ tilde - a - gamma ] ) .
moreover , the meissner screening mass of the medium photon is vanishing .
this is consistent with the absence of the meissner effect for the unbroken electromagnetism .
the most interesting results of our calculation are the expressions for the meissner screening masses of the five gluons that correspond to the broken generators of the @xmath251 group .
we find that the squared values of these masses are _ negative _
( i.e. , the masses themselves are imaginary ) in some regions of parameters .
in particular , we obtain @xmath252 for @xmath253 when @xmath254 , and @xmath255 when @xmath200 .
the standard interpretation of such a result is the existence of a chromomagnetic ( plasma type ) instability in the corresponding phases of matter @xcite .
while the instability connected with the @xmath195th gluon appears only in the g2sc phase ( @xmath200 ) , the instability connected with the other 4 gluons also develops in the _ gapped _
2sc phase when @xmath230 .
this suggests that the gapless superconductivity itself is not the reason for the instability .
it is interesting to note , however , that the critical value of the mismatch , above which the instability starts to develop , is given by @xmath256 .
this happens to coincide with the value of the mismatch at which the 2sc phase becomes metastable when the neutrality condition is not enforced in quark matter ( i.e. , when the value of @xmath245 is treated as a free parameter and the coulomb effects are ignored ) . here ,
on the other hand , we consider neutral quark matter in the same way as in refs .
@xcite .
it is natural to suggest that the instability , indicated by negative values of the meissner screening masses squared , may result in some type of a gluon condensation . indeed , in terms of the gluon effective potential , the appearance of a negative gluon mass squared
could be viewed as considering a false vacuum that corresponds to a local maximum of the potential at @xmath257 . in this case
, the true vacuum should be given by the global minimum of the gluon potential .
it would be natural then if the minimum corresponds to another ( nonzero ) expectation value of the gluon field , i.e. , @xmath258 for @xmath259 , or for @xmath260 .
note that we do not exclude the possibility that the new stable ground state has a condensate that breaks the rotational symmetry of the system .
in fact , this might be the most natural outcome of the gluon condensation if its mechanism is the same , e.g. , as in ref .
the possibility of breaking the rotational symmetry is also suggested by the fact the it is the magnetic components of gluons @xmath261 rather than the electric gluons @xmath262 that drive the instability .
the expectation that the rotational symmetry is broken in the ground state may also hint at the possibility of a state with deformed quark fermi surfaces which was proposed in ref .
@xcite .
it may happen that not only the rotational symmetry but also the translational symmetry is broken in the true ground state of neutral two - flavor quark matter .
this would be the case when the diquark condensate is inhomogeneous like , for example , in the crystalline phase @xcite , or like in the abrikosov s vortex lattice phase @xcite .
regarding the crystalline phase ( also known as the larkin - ovchinnikov - fulde - ferrell phase @xcite ) in two - flavor quark matter , it might be appropriate to mention that this phase is claimed to appear precisely when @xmath263 .
it is not clear , however , how a first order transition from the 2sc phase to the crystalline phase at @xmath264 can be triggered by vanishingly small tachyonic masses of gluons .
we would like to emphasize here that the gluon type instability , indicated by negative values of the meissner screening masses squared , has nothing to do ( at least , directly ) with the so - called sarma instability @xcite .
it was shown for the first time in ref .
@xcite , and then confirmed in refs .
@xcite that the sarma instability in the effective potential for the order parameter is removed when the neutrality condition is imposed , or when the pairing interaction has a specific momentum structure and the ratio of the densities of the pairing fermions is kept fixed @xcite .
one may still argue that , despite the absence of the sarma instability in the effective potential , there exists another type of an instability , discussed in ref .
there it was suggested that paramagnetic currents should be spontaneously induced in the gapless phase .
in fact , we think that the instability connected with the @xmath195th gluon in the g2sc phase could indeed be of this type .
one may find some similarity between the chromomagnetic instability found in this paper and the instabilities that have been discussed in refs .
@xcite in application to a state of matter produced in heavy ion collisions
. it should be emphasized , however , that the quark distribution functions are completely isotropic in momentum space in the case of dense quark matter studied in this paper which is in contrast to a typical situation in ref .
this fact suggests that the mechanism of the instability seen here is more subtle . in passing
, we would like to mention that there might exist some analogy between the instability found in this paper and the so - called paramagnetic superconductivity discussed in condensed matter physics @xcite .
if these phenomena are related indeed , it would be natural to expect that the instability of neutral dense quark matter is resolved through a spontaneous chromo - magnetization .
again , the breakdown of the rotational symmetry in this case would seem inevitable .
in this paper , we calculated the debye and meissner screening masses for the gluons and the photon in the case of neutral , @xmath0-equilibrated two - flavor dense quark matter .
a general form of our result allows to use it in the gapped phase as well as in the gapless color superconducting phase by a simple change of the magnitude of the diquark pairing .
this latter determines the ratio of the gap parameter and the mismatch of the quark fermi momenta , i.e. , @xmath235 .
the qualitative dependence of this ratio on the diquark coupling constant in the case of neutral quark matter was studied in refs .
it was shown that the ground state corresponds to the normal phase at weak coupling , i.e. , @xmath244 . at some intermediate values of the diquark coupling
, the ground state is the g2sc phase .
the ratio @xmath235 is less than 1 in such a gapless phase .
finally , at large coupling , the ground state is the 2sc phase , and the ratio @xmath235 is larger than 1 .
one could also check that this ratio @xmath235 increases monotonically with increasing the coupling .
the results for the debye and the meissner screening masses in this paper give a natural interpolation between the known values in the normal phase @xcite and in the ideal 2sc phase @xcite .
the most important result of our calculation is that the expressions for the meissner screening masses of the five gluons , corresponding to the five broken generators of the @xmath251 group , are _ imaginary_. this is interpreted as an indication of a chromomagnetic instability in the corresponding phases of quark matter .
it remains to be clarified the consequences of such an instability , and the nature of the true ground state in neutral two - flavor quark matter . in the future
, it would be very interesting to investigate whether a chromomagnetic instability also develops in the gcfl phase @xcite , where the low - energy quasiparticle spectrum resembles the spectrum in the g2sc phase .
the authors thank m. buballa , m. forbes , t. hatsuda , c. kouvaris , j. lenaghan , m.s .
li , v. miransky , s. mrwczyski , r. pisarski , k. rajagopal , a. rebhan , s. reddy , p. reuter , d. rischke , p. romatschke , t. schfer , a. schmitt , d. son , m. strickland , m. tachibana , d. n. voskresensky , and q. wang for interesting discussions .
i.a.s . is grateful to the int at the university of washington in seattle for its hospitality .
the work of m.h . was supported by the alexander von humboldt - foundation , and the nsfc under grant nos .
10105005 , 10135030 . the work of i.a.s .
was supported by gesellschaft fr schwerionenforschung ( gsi ) and by bundesministerium fr bildung und forschung ( bmbf ) .
here we calculate the traces defined in eq .
( [ d - traces ] ) . by neglecting corrections of order @xmath265
, we derive the following results : @xmath266 & \simeq & 2\left(u^{\mu}u^{\nu } \mp \frac{u^{\mu } { \mathbf k}^{\nu}+{\mathbf k}^{\mu}u^{\nu}}{k } + \frac{{\mathbf k}^{\mu}{\mathbf k}^{\nu}}{k^{2}}\right ) , \label{tr2++ } \\ { \cal t}^{\mu\nu}_{\pm\mp } = \mbox{tr}\left[\gamma^{\mu } \gamma^0 \lambda^{(\pm)}_{k } \gamma^{\nu } \gamma^0 \lambda^{(\mp)}_{k'}\right ] & \simeq & -2\left(g^{\mu\nu}-u^{\mu}u^{\nu } + \frac{{\mathbf k}^{\mu}{\mathbf k}^{\nu}}{k^{2 } } \right ) , \label{tr2 + -}\\ { \cal u}^{\mu\nu}_{\pm\pm } = \mbox{tr}\left[\gamma^{\mu } \gamma^{5}\lambda^{(\pm)}_{k } \gamma^{\nu } \gamma^{5 } \lambda^{(\pm)}_{k'}\right ] & \simeq & - 2\left ( g^{\mu\nu } - u^{\mu}u^{\nu } + \frac{{\mathbf k}^{\mu}{\mathbf k}^{\nu}}{k^{2 } } \right ) , \label{tr1++}\\ { \cal u}^{\mu\nu}_{\pm\mp } = \mbox{tr}\left[\gamma^{\mu } \gamma^{5 } \lambda^{(\pm)}_{k } \gamma^{\nu } \gamma^{5 } \lambda^{(\mp)}_{k'}\right ] & \simeq & - 2 \left ( u^{\mu}u^{\nu } -\frac{{\mathbf k}^{\mu}{\mathbf k}^{\nu}}{k^{2 } } \pm \frac{u^{\mu } { \mathbf k}^{\nu}-{\mathbf k}^{\mu}u^{\nu}}{k } \right ) , \label{tr1 + -}\end{aligned}\ ] ] where @xmath267 and @xmath268 . in calculation of the polarization tensor , we also encounter the following type angular integrals : @xmath269 , \label{phi - int2}\end{aligned}\ ] ] where @xmath126 is the cosine of the angle between the vectors @xmath127 and @xmath128 .
these relations are used in eq .
( [ traces - xi ] ) .
as we saw in the main part of the paper , see eqs .
( [ pi11-n ] ) , ( [ pi99-n ] ) , and ( [ pi89-n ] ) , all components of the polarization tensor @xmath270 with @xmath271 were given in terms of the following coefficient functions ( here and below , @xmath272 for @xmath273 and @xmath274 , while @xmath275 for @xmath276 and @xmath277 , and @xmath278 ) : @xmath279,\end{aligned}\ ] ] @xmath280 , \end{aligned}\ ] ] @xmath281 , \end{aligned}\ ] ] @xmath282 , \end{aligned}\ ] ] @xmath283 , \end{aligned}\ ] ] @xmath284,\end{aligned}\ ] ] where @xmath285^{-1}$ ] is the fermi distribution function .
note that @xmath286 at zero temperature , these functions take the following form ( here and below , the upper sign corresponds to the first case , e.g. , @xmath287 , and the lower sign corresponds to the second case , e.g. , @xmath288 ) : @xmath289 } - \frac{(e_{\delta , k'}^- + e_{k'}^-)(e_{\delta , k'}^- + e_{k}^- - p_0 ) } { 2 e_{\delta , k'}^- \left[(p_0 - e_{\delta , k'}^-)^2 - ( e_{\delta , k}^-)^2\right ] } \nonumber \\ & - & \frac{(e_{\delta , k}^+ + e_k^+)(e_{\delta , k}^+ + e_{k'}^+ + p_0 ) } { 2 e_{\delta , k}^+ \left[(p_0 + e_{\delta , k}^+)^2 - ( e_{\delta , k'}^+)^2\right ] } - \frac{(e_{\delta , k'}^+ + e_{k'}^+)(e_{\delta , k'}^+ + e_{k}^+ - p_0 ) } { 2 e_{\delta , k'}^+ \left[(p_0 - e_{\delta , k'}^+)^2 - ( e_{\delta , k}^+)^2\right ] } \nonumber \\ & + & \theta(-e_{\delta , k}^-+\delta\mu ) \frac{(e_{\delta , k}^- \mp e_k^- ) ( e_{\delta , k}^- \mp e_{k'}^- \mp p_0 ) } { 2 e_{\delta , k}^- \left[(p_0 \mp e_{\delta , k}^-)^2 - ( e_{\delta , k'}^-)^2\right ] } \nonumber \\ & + & \theta(-e_{\delta , k'}^-+\delta\mu ) \frac{(e_{\delta , k'}^-
\mp e_{k'}^- ) ( e_{\delta , k'}^- \mp e_{k}^- \pm p_0 ) } { 2 e_{\delta , k'}^- \left[(p_0 \pm e_{\delta , k'}^-)^2 - ( e_{\delta , k}^-)^2\right]},\end{aligned}\ ] ] @xmath290 } - \frac{(e_{\delta , k'}^- - e_{k'}^-)(e_{\delta , k'}^- - e_{k}^- - p_0 ) } { 2 e_{\delta , k'}^- \left[(p_0 - e_{\delta , k'}^-)^2 - ( e_{\delta , k}^-)^2\right ] } \nonumber \\ & - & \frac{(e_{\delta , k}^+ - e_k^+)(e_{\delta , k}^+ - e_{k'}^+ + p_0 ) } { 2 e_{\delta , k}^+ \left[(p_0 + e_{\delta , k}^+)^2 - ( e_{\delta , k'}^+)^2\right ] } - \frac{(e_{\delta , k'}^+ - e_{k'}^+)(e_{\delta , k'}^+ - e_{k}^+ - p_0 ) } { 2 e_{\delta , k'}^+ \left[(p_0 - e_{\delta , k'}^+)^2 - ( e_{\delta , k}^+)^2\right ] } \nonumber \\ & + & \theta(-e_{\delta , k}^-+\delta\mu ) \frac{(e_{\delta ,
k}^- \mp e_k^- ) ( e_{\delta , k}^- \mp e_{k'}^- \pm p_0 ) } { 2 e_{\delta , k}^- \left[(p_0 \pm e_{\delta , k}^-)^2 - ( e_{\delta , k'}^-)^2\right ] } \nonumber \\ & + & \theta(-e_{\delta , k'}^-+\delta\mu ) \frac{(e_{\delta ,
k'}^- \mp e_{k'}^- ) ( e_{\delta , k'}^- \mp e_{k}^- \mp p_0 ) } { 2 e_{\delta , k'}^- \left[(p_0 \mp e_{\delta , k'}^-)^2 - ( e_{\delta , k}^-)^2\right ] } , \end{aligned}\ ] ] @xmath291 } - \frac{(e_{\delta , k'}^- - e_{k'}^-)(e_{\delta , k'}^- + e_{k}^+ - p_0 ) } { 2 e_{\delta , k'}^- \left[(p_0 - e_{\delta , k'}^-)^2 - ( e_{\delta , k}^+)^2\right ] } \nonumber \\ & - & \frac{(e_{\delta , k}^+ + e_k^+)(e_{\delta , k}^+ - e_{k'}^- + p_0 ) } { 2 e_{\delta , k}^+ \left[(p_0 + e_{\delta , k}^+)^2 - ( e_{\delta , k'}^-)^2\right ] } - \frac{(e_{\delta , k'}^+ - e_{k'}^+)(e_{\delta , k'}^+ + e_{k}^- - p_0 ) } { 2 e_{\delta , k'}^+ \left[(p_0 - e_{\delta , k'}^+)^2 - ( e_{\delta , k}^-)^2\right ] } \nonumber \\ & + & \theta(-e_{\delta , k}^-+\delta\mu ) \frac{(e_{\delta ,
k}^- \mp e_k^- ) ( e_{\delta , k}^- \pm e_{k'}^+ \mp p_0 ) } { 2 e_{\delta , k}^- \left[(p_0 \mp e_{\delta , k}^-)^2 - ( e_{\delta , k'}^+)^2\right ] } \nonumber \\ & + & \theta(-e_{\delta , k'}^-+\delta\mu ) \frac{(e_{\delta ,
k'}^- \mp e_{k'}^- ) ( e_{\delta , k'}^- \pm e_{k}^+ \mp p_0 ) } { 2 e_{\delta , k'}^- \left[(p_0 \mp e_{\delta , k'}^-)^2 - ( e_{\delta , k}^+)^2\right ] } , \end{aligned}\ ] ] @xmath292 } - \frac{(e_{\delta , k'}^- + e_{k'}^-)(e_{\delta , k'}^- - e_{k}^+ - p_0 ) } { 2 e_{\delta , k'}^- \left[(p_0 - e_{\delta , k'}^-)^2 - ( e_{\delta , k}^+)^2\right ] } \nonumber \\ & - & \frac{(e_{\delta , k}^+ - e_k^+)(e_{\delta , k}^+ + e_{k'}^- + p_0 ) } { 2 e_{\delta , k}^+ \left[(p_0 + e_{\delta , k}^+)^2 - ( e_{\delta , k'}^-)^2\right ] } - \frac{(e_{\delta , k'}^+ + e_{k'}^+)(e_{\delta , k'}^+ - e_{k}^- - p_0 ) } { 2 e_{\delta , k'}^+ \left[(p_0 - e_{\delta , k'}^+)^2 - ( e_{\delta , k}^-)^2\right ] } \nonumber \\ & + & \theta(-e_{\delta , k}^-+\delta\mu ) \frac{(e_{\delta ,
k}^- \mp e_k^- ) ( e_{\delta , k}^- \pm e_{k'}^+ \pm p_0 ) } { 2 e_{\delta , k}^- \left[(p_0 \pm e_{\delta , k}^-)^2 - ( e_{\delta , k'}^+)^2\right ] } \nonumber \\ & + & \theta(-e_{\delta ,
k'}^-+\delta\mu ) \frac{(e_{\delta , k'}^- \mp e_{k'}^- ) ( e_{\delta , k'}^- \pm e_{k}^+ \pm p_0 ) } { 2 e_{\delta , k'}^- \left[(p_0 \pm e_{\delta , k'}^-)^2 - ( e_{\delta , k}^+)^2\right ] } , \end{aligned}\ ] ] @xmath293 } - \frac{1}{2e_{\delta , k'}^-\left[(p_0-e_{\delta , k'}^-)^2 - ( e_{\delta , k}^+)^2\right ] } \nonumber \\ & - & \frac{1}{2e_{\delta , k}^+\left[(p_0+e_{\delta , k}^+)^2 - ( e_{\delta , k'}^-)^2\right ] } - \frac{1}{2e_{\delta , k'}^+\left[(p_0-e_{\delta , k'}^+)^2 - ( e_{\delta , k}^-)^2\right ] } \nonumber \\ & + & \frac{\theta(-e_{\delta , k}^-+\delta\mu)}{2e_{\delta , k}^- } \frac{1}{(p_0 \pm e_{\delta , k}^-)^2-(e_{\delta , k'}^+)^2 } + \frac{\theta(-e_{\delta , k'}^-+\delta\mu)}{2e_{\delta , k'}^- } \frac{1}{(p_0 \pm e_{\delta , k'}^-)^2-(e_{\delta , k}^+)^2}\bigg],\end{aligned}\ ] ] @xmath294 } - \frac{1}{2e_{\delta , k'}^-\left[(p_0-e_{\delta , k'}^-)^2 - ( e_{\delta , k}^-)^2\right ] } \nonumber \\ & - & \frac{1}{2e_{\delta , k}^+\left[(p_0+e_{\delta , k}^+)^2 - ( e_{\delta , k'}^+)^2\right ] } - \frac{1}{2e_{\delta , k'}^+\left[(p_0-e_{\delta , k'}^+)^2 - ( e_{\delta , k}^+)^2\right ] } \nonumber \\ & + & \frac{\theta(-e_{\delta , k}^-+\delta\mu)}{2e_{\delta , k}^- } \frac{1}{(p_0 \pm e_{\delta , k}^-)^2-(e_{\delta , k'}^-)^2 } + \frac{\theta(-e_{\delta , k'}^-+\delta\mu)}{2e_{\delta , k'}^- } \frac{1}{(p_0 \pm e_{\delta , k'}^-)^2-(e_{\delta , k}^-)^2}\bigg].\end{aligned}\ ] ] the components of the polarization tensor @xmath270 with @xmath4 are given in terms of the following coefficient functions : @xmath295\nonumber \\ & + & \left[{\tilde f}(- e_{bu , k}^+ ) + { \tilde f}(- e_{bd , k}^+)\right]\frac{-p_0- e_{b , k}^+ - e_{k'}^+ } { ( p_0 + e_{b , k}^+)^2-({e_{\delta , k'}^+})^2 } \nonumber \\ & + & \left[{\tilde f}(- e_{bu , k'}^- ) + { \tilde f}(- e_{bd , k'}^-)\right]\frac{p_0- e_{b , k'}^- - e_{k}^- } { ( p_0- e_{b , k'}^-)^2-(e^-_{\delta , k})^2},\end{aligned}\ ] ] @xmath296\nonumber \\ & + & \left[{\tilde f}(e_{bu , k}^- ) + { \tilde f}(e_{bd , k}^-)\right]\frac{-p_0 + e_{b , k}^- + e_{k'}^- } { ( p_0- e_{b , k}^-)^2-(e_{\delta , k'}^-)^2 } \nonumber \\ & + & \left[{\tilde f}(e_{bu , k'}^+ ) + { \tilde f}(- e_{bd , k'}^+)\right]\frac{p_0 + e_{b , k'}^+ + e_{k}^+ } { ( p_0 + e_{b , k'}^+)^2-(e^+_{\delta , k})^2},\end{aligned}\ ] ] @xmath297\nonumber \\ & + & \left[{\tilde f}(-e_{bu , k}^+ ) + { \tilde f}(-e_{bd , k}^+)\right]\frac{-p_0- e_{b , k}^+ + e_{k'}^- } { ( p_0 + e_{b , k}^+)^2-(e_{\delta , k'}^-)^2 } \nonumber \\ & + & \left[{\tilde f}(e_{bu , k'}^+ ) + { \tilde f } ( e_{bd , k'}^+)\right]\frac{p_0 + e_{b , k'}^+ - e_{k}^- } { ( p_0 + e_{b , k'}^+)^2-(e^-_{\delta , k})^2 } , \end{aligned}\ ] ] @xmath298\nonumber \\ & + & \left[{\tilde f}(e_{bu , k}^- ) + { \tilde f}(e_{bd , k}^-)\right]\frac{-p_0 + e_{b , k}^- - e_{k'}^+ } { ( p_0- e_{b , k}^-)^2-(e_{\delta , k'}^+)^2 } \nonumber \\ & + & \left[{\tilde f}(-e_{bu , k'}^- ) + { \tilde f}(- e_{bd , k'}^-)\right]\frac{p_0- e_{b , k'}^- + e_{k}^+ } { ( p_0- e_{b , k'}^-)^2-(e^+_{\delta , k})^2},\end{aligned}\ ] ] @xmath299\nonumber \\ & + & \left[{\tilde f}(- e_{bu , k'}^+ ) + { \tilde f}(- e_{bd , k'}^+)\right ] \frac{p_0- e_{b , k'}^+ - e_{k}^+ } { ( p_0- e_{b , k'}^+)^2-({e_{\delta , k}^+})^2 } \nonumber \\ & + & \left[{\tilde f}(- e_{bu , k}^- ) + { \tilde f}(- e_{bd , k}^-)\right ] \frac{-p_0- e_{b , k}^- - e_{k'}^- } { ( p_0 + e_{b , k}^-)^2-(e^-_{\delta , k'})^2},\end{aligned}\ ] ] @xmath300\nonumber \\ & + & \left[{\tilde f}(e_{bu , k'}^- ) + { \tilde f}(e_{bd , k'}^-)\right]\frac{p_0 + e_{b , k'}^- + e_{k}^- } { ( p_0 + e_{b , k'}^-)^2-(e_{\delta , k}^-)^2 } \nonumber \\ & + & \left[{\tilde f}(e_{bu , k}^+ ) + { \tilde f}(- e_{bd , k}^+)\right]\frac{-p_0 + e_{b , k}^+ + e_{k'}^+ } { ( p_0- e_{b , k}^+)^2-(e^+_{\delta , k'})^2 } , \end{aligned}\ ] ] @xmath301\nonumber \\ & + & \left[{\tilde f}(-e_{bu , k}^- ) + { \tilde f}(-e_{bd , k}^-)\right]\frac{-p_0- e_{b , k}^- + e_{k'}^+ } { ( p_0 + e_{b , k}^-)^2-(e_{\delta , k'}^+)^2 } \nonumber \\ & + & \left[{\tilde f}(e_{bu , k'}^- ) + { \tilde f}(e_{bd , k'}^-)\right]\frac{p_0 + e_{b , k'}^- - e_{k}^+ } { ( p_0 + e_{b , k'}^-)^2-(e^+_{\delta , k})^2 } , \end{aligned}\ ] ] @xmath302\nonumber \\ & + & \left[{\tilde f}(e_{bu , k}^+ ) + { \tilde f}(e_{bd , k}^+)\right]\frac{-p_0 + e_{b , k}^+ - e_{k'}^- } { ( p_0- e_{b , k}^+)^2-(e_{\delta , k'}^-)^2 } \nonumber \\ & + & \left[{\tilde f}(-e_{bu , k'}^+ ) + { \tilde f}(-e_{bd , k'}^+)\right]\frac{p_0- e_{b , k'}^+ + e_{k}^- } { ( p_0- e_{b , k'}^+)^2-(e^-_{\delta , k})^2}. \end{aligned}\ ] ] at zero temperature , the expressions for these coefficients become @xmath303 \frac { e_{b ,
k'}^- + e_{k}^- -p_0}{(e_{\delta , k}^-)^2- ( e_{b , k'}^- -p_0)^2},\end{aligned}\ ] ] @xmath304 \frac { e_{b , k}^- + e_{k'}^- -p_0}{(e_{\delta , k'}^-)^2- ( e_{b , k'}^- -p_0)^2},\end{aligned}\ ] ] @xmath305 @xmath306 \frac { e_{b , k}^- - e_{k'}^+ -p_0}{(e_{\delta , k'}^+)^2- ( e_{b , k}^- -p_0)^2 } \nonumber \\ & -&\left [ \theta(-e_{bu , k'}^- ) + \theta(-e_{bd , k'}^-)\right ] \frac { e_{b , k'}^- - e_{k}^+ -p_0}{(e^+_{\delta , k})^2- ( e_{b , k'}^- -p_0)^2},\end{aligned}\ ] ] @xmath307 \frac { e_{b , k}^- + e_{k'}^- + p_0}{(e_{\delta , k'}^-)^2- ( e_{b , k}^- + p_0)^2},\end{aligned}\ ] ] @xmath308 \frac { e_{b , k'}^- + e_{k}^- + p_0}{(e_{\delta , k}^-)^2- ( e_{b , k'}^- + p_0)^2},\end{aligned}\ ] ] @xmath309 \frac { e_{b , k}^- - e_{k'}^+ + p_0}{(e_{\delta , k'}^+)^2- ( e_{b , k}^- + p_0)^2 } \nonumber \\ & -&\left [ \theta(-e_{bu , k'}^- ) + \theta(-e_{bd , k'}^-)\right ] \frac { e_{b , k'}^- - e_{k}^+ + p_0}{(e^+_{\delta , k})^2- ( e_{b , k'}^- + p_0)^2},\end{aligned}\ ] ] @xmath310
in momentum integrals that appear in the expressions for the polarization tensor @xmath270 with @xmath271 , one can use some useful approximations that simplify the calculations considerably .
for example , in all integrals that come from particle - hole loops , the main contribution comes from a close vicinity of the average quark fermi momentum @xmath311 . in these integrals , therefore , it is justified to make the following replacement : @xmath312 the corrections to such momentum integrals are suppressed by inverse powers of @xmath313 . to be consistent with this approximation
, the antiparticle - antiparticle loops should be omitted , and the dependence of the particle - antiparticle loops on @xmath90 , @xmath245 and @xmath59 may be neglected . as a result , the only type of the particle - antiparticle contribution , that appears in the calculation , has the form @xmath314 the particle - hole loops give rise to the integrals of the following type : the calculation of the polarization tensor @xmath270 with @xmath4 reduces to the calculation of the integral in eq .
( [ c4pp - integral ] ) .
this can be written as a sum of two expressions .
one of them comes from integrating the first line in eq .
( [ cccc ] ) .
the result is @xmath316 } \nonumber\\ & & + \int_{\mu_{db}}^{\infty } k^2dk \frac{e_{\delta , k}^{-}-k+\bar\mu } { e_{\delta , k}^{-}(e_{\delta , k}^{-}+k-\bar\mu_b ) } \simeq \bar\mu^2 \left ( \int_{-\bar\mu}^{-\delta\mu-\mu_8 } \frac{dx(\sqrt{x^2+\delta^2}+x)}{\sqrt{x^2+\delta^2 } ( \sqrt{x^2+\delta^2}-x-\mu_8)}\right .
\nonumber\\ & & \left .
+ \int_{-\delta\mu-\mu_8}^{\delta\mu-\mu_8 } \frac{dx(2x^2+\delta^2+x\mu_8)}{\sqrt{x^2+\delta^2 } ( \delta^2 - 2 x \mu_8 - \mu_8 ^ 2 ) } + \int_{\delta\mu-\mu_8}^{\infty } \frac{dx(\sqrt{x^2+\delta^2}-x)}{\sqrt{x^2+\delta^2 } ( \sqrt{x^2+\delta^2}+x+\mu_8)}\right)\nonumber\\ & \simeq & 2\bar\mu^2\left [ 1-\frac{\delta^2}{4\mu_8 ^ 2}\ln \frac{\left(\delta^2+\mu_8 ^ 2\right)^2-\left(\mu_e\mu_8\right)^2 } { \delta^4}\right],\end{aligned}\ ] ] where @xmath222 and @xmath317 . in this calculation , we used the following table integrals : @xmath318 the other expression comes from integrating the second line in eq .
( [ cccc ] ) . the result is @xmath319 } \nonumber\\ & = & \bar\mu^2 \theta(\delta\mu-\delta ) \int_{-\sqrt{\delta\mu^2-\delta^2}}^{\sqrt{\delta\mu^2-\delta^2 } } \frac{dx(2x^2+\delta^2+x\mu_8)}{\sqrt{x^2+\delta^2 } ( \delta^2 - 2x\mu_8-\mu_8 ^ 2 ) } \nonumber\\ & = & \frac{\bar\mu^2 \delta^2 } { 2\mu_8 ^ 2 } \ln \frac{\delta^4-\mu_8 ^ 2(\delta\mu-\sqrt{\delta\mu^2-\delta^2})^2 } { \delta^4-\mu_8 ^ 2(\delta\mu+\sqrt{\delta\mu^2-\delta^2})^2}\theta(\delta\mu-\delta).\end{aligned}\ ] ] m. gell - mann , phys .
* 8 * , 214 ( 1964 ) .
collins and m.j .
perry , phys . rev
. lett .
* 34 * , 1353 ( 1975 ) .
shuryak , phys .
b * 78 * , 150 ( 1978 ) .
olive , nucl .
b * 190 * , 483 ( 1981 ) .
for a recent review see d.j .
schwarz , annalen phys .
* 12 * , 220 ( 2003 ) .
r. rapp , t. schfer , e.v .
shuryak and m. velkovsky , phys .
lett . * 81 * , 53 ( 1998 ) ; m. alford , k. rajagopal , and f. wilczek , phys .
b * 422 * , 247 ( 1998 ) .
alford , k. rajagopal and f. wilczek , nucl .
* b537 * , 443 ( 1999 ) .
son , phys . rev .
d * 59 * , 094019 ( 1999 ) ; d.k .
hong , v.a .
miransky , i.a .
shovkovy , and l.c.r .
wijewardhana , phys .
d * 61 * , 056001 ( 2000 ) ; d.k .
hong , nucl .
b * 582 * , 451 ( 2000 ) ; t. schfer and f. wilczek , phys .
d * 60 * , 114033 ( 1999 ) ; r.d .
pisarski and d.h .
rischke , phys .
d * 61 * , 051501 ( 2000 ) ; s.d.h .
hsu and m. schwetz , nucl .
* b572 * , 211 ( 2000 ) ; w.e .
brown , j.t .
liu , and h .- c .
ren , phys .
d * 61 * , 114012 ( 2000 ) .
i.a . shovkovy and l.c.r .
wijewardhana , phys .
b * 470 * , 189 ( 1999 ) ; t. schfer , nucl .
phys . * b575 * , 269 ( 2000 ) . m. alford and k. rajagopal , jhep * 0206 * , 031 ( 2002 ) .
steiner , s. reddy and m. prakash , phys .
d * 66 * , 094007 ( 2002 ) .
i. shovkovy and m. huang , phys .
b * 564 * , 205 ( 2003 ) .
e. gubankova , w.v .
liu and f. wilczek , phys .
lett . * 91 * , 032001 ( 2003 ) .
a. mishra and h. mishra , phys .
d * 69 * , 014014 ( 2004 ) . m. huang and i. shovkovy , nucl . phys .
a * 729 * ( 2003 ) 835 .
liao and p.f .
zhuang , phys .
d * 68 * , 114016 ( 2003 ) .
m. alford , c. kouvaris and k. rajagopal , phys .
rev . lett . * 92 * , 222001 ( 2004 ) .
m. alford , c. kouvaris and k. rajagopal , hep - ph/0406137 .
k. iida , t. matsuura , m. tachibana and t. hatsuda , hep - ph/0312363 .
rster , i.a .
shovkovy , and d.h .
rischke , nucl . phys .
a * 743 * , 127 ( 2004 ) .
s. reddy and g. rupak , nucl - th/0405054 .
i. shovkovy , m. hanauske and m. huang , phys .
d * 67 * , 103004 ( 2003 ) .
liu and f. wilczek , phys .
* 90 * , 047002 ( 2003 ) .
b. deb , a. mishra , h. mishra and p.k .
panigrahi , phys . rev .
a * 70 * , 011604(r ) ( 2004 ) . w.v
. liu , f. wilczek and p. zoller , phys .
a * 70 * , 033603 ( 2004 ) .
forbes , e. gubankova , w.v .
liu and f. wilczek , hep - ph/0405059 .
f. neumann , m. buballa , and m. oertel , nucl .
a * 714 * , 481 ( 2003 ) .
bedaque , h. caldas and g. rupak , phys .
lett . *
91 * , 247002 ( 2003 ) ; h. caldas , phys .
a * 69 * , 063602 ( 2004 ) .
m. huang and i.a .
shovkovy , phys .
d * 70 * , 051501(r ) ( 2004 ) .
rischke , phys .
d * 62 * , 034007 ( 2000 ) ; d.h .
rischke and i.a .
shovkovy , phys .
d * 66 * , 054019 ( 2002 ) .
a. schmitt , q. wang and d.h .
rischke , phys . rev .
d * 69 * , 094017 ( 2004 ) .
i. shovkovy and m. huang , in structure and dynamics of elementary matter " , nato scientific series in mathematics , physics and chemistry vol .
166 , edited by w. greiner et al .
( kluwer , dordrecht , 2004 ) pp .
329 - 336 .
taylor , nucl .
b * 33 * , 436 ( 1971 ) ; a.a .
slavnov , theor .
* 10 * , 99 ( 1972 ) .
s. elitzur , phys .
d * 12 * , 3978 ( 1975 ) .
y. nambu , phys .
* 117 * , 648 ( 1960 ) .
gusynin and i.a .
shovkovy , nucl .
a * 700 * , 577 ( 2002 ) .
faddeev and v.n .
popov , phys .
b * 25 * , 29 ( 1967 ) .
j. kapusta , _ finite - temperature field theory _ , ( cambridge , cambridge university press , 1989 ) .
m. huang , p.f .
zhuang and w.q .
chao , phys .
d * 67 * , 065015 ( 2003 ) s. rster , diploma thesis , j. w. goethe university , 2003 ; s.b .
rster and d.h .
rischke , phys .
d * 69 * , 045011 ( 2004 ) .
d. blaschke , s. fredriksson , h. grigorian and a.m. oztas , nucl .
a * 736 * , 203 ( 2004 ) .
a. gerhold and a. rebhan , phys .
d * 68 * , 011502 ( 2003 ) ; d.d .
dietrich and d.h .
rischke , prog .
phys . * 53 * , 305 ( 2004 ) .
m. huang , p.f .
zhuang and w.q .
chao , phys .
d * 65 * , 076012 ( 2002 ) .
h. vija and m.h .
thoma , phys .
b * 342 * , 212 ( 1995 ) . c. manuel , phys .
d * 53 * , 5866 ( 1996 ) .
m. g. alford , j. berges and k. rajagopal , nucl .
b * 571 * , 269 ( 2000 ) ; e.v .
gorbar , phys .
d * 62 * , 014007 ( 2000 ) .
litim and c. manuel , phys .
d * 64 * , 094013 ( 2001 ) .
gusynin , v.a .
miransky and i.a .
shovkovy , phys .
b * 581 * , 82 ( 2004 ) ; mod .
a * 19 * , 1341 ( 2004 ) .
we thank v. miransky for bringing this posibility to our attention .
h. mther and a. sedrakian , phys .
d * 67 * , 085024 ( 2003 ) .
m. g. alford , j. a. bowers and k. rajagopal , phys .
d * 63 * , 074016 ( 2001 ) ; d. k. hong and y. j. sohn , hep - ph/0107003 ; i. giannakis , j. t. liu and h. c. ren , phys . rev .
d * 66 * , 031501 ( 2002 ) ; r. casalbuoni and g. nardulli , rev .
* 76 * , 263 ( 2004 ) ; j. a. bowers , hep - ph/0305301 .
abrikosov , sov . phys .
jetp * 5 * , 1174 ( 1957 ) [ zh .
* 32 * , 1442 ( 1957 ) ] .
a. i. larkin and yu .
n. ovchinnikov , sov .
jetp * 20 * , 762 ( 1965 ) ; p. fulde and r. a. ferrell , phys . rev . * 135 * , a550 ( 1964 ) .
g. sarma , j. phys .
solids * 24 * , 1029 ( 1963 ) . s .-
wu and s. yip , phys .
a * 67 * , 053603 ( 2003 ) ; s. mrwczyski , phys .
b * 214 * , 587 ( 1988 ) ; phys .
b * 314 * , 118 ( 1993 ) ; phys .
c * 49 * , 2191 ( 1994 ) ; phys .
b * 393 * , 26 ( 1997 ) ; s. mrwczyski and m.h .
thoma , phys .
d * 62 * , 036011 ( 2000 ) .
p. romatschke and m. strickland , phys .
d * 68 * , 036004 ( 2003 ) ; phys .
d * 69 * , 065005 ( 2004 ) .
birse , c.w .
kao and g.c .
nayak , phys .
b * 570 * , 171 ( 2003 ) .
p. arnold , j. lenaghan and g.d .
moore , jhep * 0308 * , 002 ( 2003 ) .
s. mrwczyski , a. rebhan and m. strickland , phys .
d * 70 * , 025004 ( 2004 ) .
geim , s.v .
dubonos , j.g.s .
lok , m. henini , and j.c .
maan , nature * 396 * , 144 ( 1998 ) ; m. sigrist , t.m .
rice , rev .
67 * , 503 ( 1995 ) ; m.s .
li , phys .
rept . * 376 * , 133 ( 2003 ) . | the debye and meissner screening masses of the gluons and the photon in neutral and @xmath0-equilibrated dense two - flavor quark matter are calculated .
the results are presented in a general form that can be used in gapped as well as gapless color superconducting phases .
the results for the magnetic screening masses indicate that the system develops a chromomagnetic instability .
possible consequences of the instability are discussed . |
the emerging field of metamaterials and the realization of materials exhibiting a negative index of refraction ( nim ) in the past ten years opened the way for various fascinating applications . with these novel materials physical applications like cloaking devices @xcite and optical imaging beyond the diffraction limit
@xcite have become possible .
a metamaterial consists of artificial structures which are much smaller than the wavelength of the operating incident radiation .
nims have been realized by metal - dielectric structures e.g. split - ring resonators deposited on dielectrics @xcite or fishnet structures @xcite .
however , due to the electron - beam lithography fabrication process of these structures , the realization of three - dimensional metamaterials requires a sequentially stacking of single layers @xcite .
three - dimensional metamaterials consisting of multistacks of metal - dielectric layers can also be obtained by rolling - up strained layers into microrolls @xcite . that opens up the application of such microrolls for example as hyperlenses @xcite .
+ the efficiency of metamaterials suffers from the absorption of radiation in the metal , caused by the finite imaginary part of the metals dielectric function .
this causes strong energy dissipation and a rapid field decay .
a possible solution to overcome this problem would be the incorporation of gain materials .
different approaches in the recent literature include dyes @xcite , quantum dots @xcite and optically - pumped quantum wells that have been brought in close proximity to split - ring resonators @xcite .
semiconductor quantum wells show no bleaching and electrical pumping is well established for these structures .
recently it was demonstrated that the self - rolling process offers the possibility to easily include active quantum - wells in the semiconductor layer of three - dimensional metamaterials .
it was found that such a novel structure shows enhanced light transmission upon optical pumping @xcite . in the next step
the concept of rolled - up nanotech allows the realization of three - dimensional metamaterials with sophisticated metallic nanostructures instead of planar metal layers which enables us to profit from plasmonic resonances .
+ in this letter we investigate alternating layers of amplifying quantum wells and metal gratings supporting surface plasmon polaritions ( spps ) .
such structures can be realized in the walls of rolled - up metamaterials as sketched in fig . 1 .
experimentally one can prepare different type of layered systems rolling along or perpendicular to the grating stripes .
here we are interested in the fundamental effect of grating coupler enhanced gain and concentrate on a microroll with rolling direction along the grating stripes where we can precisely align the stripes on top of each other .
whereas in the case of a perpendicular rolling direction commensurability effects between the microroll perimeter and the grating period can occur .
we present finite - difference time - domain ( fdtd ) simulations which show a transmission enhancement in the structures with metallic gratings much larger than observed for samples with flat silver surfaces .
in fig . 1 a sketch of a microroll is shown .
the preparation of this structures is based on strain relaxation of pseudomorphically grown ( alin)gaas heterostructures , which roll up into a tube when released from the substrate by etching away an alas sacrificial layer ( for details see @xcite ) .
the important ingredients for our investigation here is that in addition ( i ) an ingaas gain layer and ( ii ) a grating is etched into the semiconductor layer and covered with ag before the rolling - up process . as a result one obtains alternating layers of quantum wells and ag gratings .
+ is the lattice constant , @xmath0 denotes the width of the etched region and @xmath1 is the grating depth . ]
we performed finite - difference time - domain simulations on the metamaterials by using the commercial software lumerical fdtd solutions @xcite .
for the simulations we neglected the curvature of the actual structure and investigate flat structures in two dimensions .
we simulated a grating unit cell ( lattice constant @xmath2 and filling factor @xmath3 where @xmath0 denotes the width of the etched region ) with three layers where we used periodic boundary conditions in x direction and perfectly matched layer boundary conditions in y direction ( fig . 2 ) .
we irradiated the investigated structures with a plane wave with zero incident angle . to fulfill the criterion for spp excitation
we assume that this plane wave has the h field @xmath4 and the e field @xmath5 . due to the periodic corrugation
also y components of the electric field are induced . in the frequency spectrum
the plane wave pulse has a gaussian - shaped distribution with a maximum at @xmath6thz and a full - width - half - maximum ( fwhm ) of @xmath7thz .
this corresponds to broadband excitation with a center wavelength of @xmath8 nm and a fwhm of @xmath9 nm . inside the simulation volume we used an appropriate non - uniform mesh which has been optimized by convergence testing .
the transmission spectrum was recorded in a distance of @xmath10 m in y direction behind the structure . for the dielectric functions we take polynomial fits on experimental data from palik @xcite .
the dielectric functions of the algaas , the ingaas and the alingaas used in the simulations are based on the dielectric function of gaas and were adjusted as follows : the percentage of in and al in the semiconductor layer decreases and increases the bandgap of the composite semiconductor which we assume to be al@xmath11ga@xmath12as , in@xmath13ga@xmath14as and al@xmath15in@xmath13ga@xmath16as . in first approximation this shift of the bandgap is linear causing the dielectric function to change slightly .
this effect was taken into account by dislocating the dielectric function of gaas linearly , @xmath17 where @xmath18 and @xmath19 are the percentages of in and al in the semiconductor , respectively .
the difference of the bandgap of gaas and in@xmath20ga@xmath21as @xcite is @xmath22ev and the difference of the bandgap of al@xmath23ga@xmath24as and gaas @xcite is @xmath25ev .
+ the amplification effect of the ingaas quantum well can be described by a negative imaginary part of the refractive index . in the simulations
the dielectric function @xmath26 of this gain layer was modelled qualitatively as a lorentz oscillator according to @xcite : @xmath27 with @xmath28 as the lorentz oscillator strength , @xmath29 the damping frequency and @xmath30 as the resonant frequency of the quantum well .
the beneficial exploitation of the spp excitations on the transmission requires that the lattice constant of the metallic grating is chosen appropriately to adjust the energy of the spp resonance to the active wavelength regime of the of the quantum well .
furthermore it is favorable to tune the spp resonance energy to the fabry - perot transmission peak of the system arising from the layered semiconductor / metal system with its high index of refraction .
the optically active wavelength regime of the quantum well can in fact be tailored by the amount of indium used during sample preparation . in the simulations presented in this letter
we consider a quantum well emitting at @xmath31 nm . the fabry - perot resonance peaks can be shifted by varying the individual layer thicknesses . the black line in fig .
3(a ) shows the simulated transmission spectrum for a three - layer structure with each layer consisting of @xmath32 nm ag , @xmath33 nm alingaas , @xmath34 nm ingaas and @xmath35 nm algaas .
we see that , for this dimensions , the structure exhibits a pronounced fabry - perot peak at the desired wavelength of @xmath31 nm . the appropriate grating was examined by performing simulations with various lattice constants . in fig .
3(a ) we display the transmission spectra for different gratings in comparison with the spectrum of unstructured layers .
we first choose a grating filling factor @xmath36 , a depth @xmath37 nm and an ag thickness @xmath38 nm .
one can see that for an @xmath39 nm grating a deep spp resonance is located at the desired wavelength of @xmath31 nm . for smaller grating periods the spp resonance shifts to shorter wavelengths ( @xmath40 nm for an @xmath41 nm grating and @xmath42 nm for an @xmath43 nm grating ) whereas for larger grating periods the spp - resonance wavelength is longer ( @xmath44 nm for an @xmath45 nm grating and @xmath46 nm for an @xmath47 nm grating ) .
variations in the lattice depth do not affect the position of the spp resonance but a smaller lattice depth causes less pronounced resonances .
moreover , several other spp excitations can be observed , however , they are considerably less pronounced . from the observed resonance positions we can reconstruct a dispersion relation @xmath48 where @xmath49 is the diffraction order .
we find that the pronounced minima , starting for @xmath47 nm at @xmath46 nm , follow a dispersion @xmath50 with @xmath51 .
the minima in the regime between @xmath52 to @xmath53 nm follow the same dispersion with @xmath54 , thus they represent a higher diffraction order .
this dispersion @xmath50 lies inbetween the spp dispersion of a flat - ag / vacuum interface and a flat - ag / gaas interface .
this is what we actually expect for our complex semiconductor / metal structures including vacuum gaps .
we further expect even different spp branches due to the asymmetric arrangement and coupling between the three metal layers .
indeed , the resonances starting for @xmath47 nm at @xmath55 nm follow a dispersion @xmath56 that lies energetically below @xmath57 .
our studies showed that the variation of the lattice filling factor ( we simulated filling factors between @xmath58 and @xmath59 ) does effect the position of the spp resonance only slightly by about @xmath60 nm . in fig .
3(b ) we independently show the transmission spectra for the above structure without and with an @xmath39 nm grating .
it can be seen clearly that the spp resonance and the fabry - perot resonance peak are matched to the active wavelength regime of the quantum well ( @xmath31 nm ) . in fig .
3(c ) we plot the two - dimensional electric - field distribution of the @xmath31 nm spp mode ( adjacent we show the one - dimensional intensity profile corresponding to the white dashed line ) .
we recognize that the intensity profile exhibits a large intensity inside the structure in particular at the quantum well locations .
we note that in two - dimensional metamaterials the gain material has to be brought in very close proximity to the metallic structure ( see , e.g. , reference @xcite ) which can cause manufacturing problems . in our three - dimensional case
the mode pattern allows us to use thick barrier layers which are necessary to maintain the optical quality of the quantum well and still have a strong field - coupling to the quantum well .
nm for illumination with a p - polarized wave .
the ingaas gain layer has a thickness of @xmath34 nm . ( d ) also shows the imaginary part of the refractive index of the gain layer .
( e ) transmission spectrum without gain ( compare with fig .
3(b ) ) for illumination with a p - polarized wave .
] nm for illumination with a s - polarized wave .
the quantum well has a thickness of @xmath34 nm . ( e ) transmission spectrum without gain for illumination with a s - polarized wave .
a pronounced spp resonance at @xmath31 nm wavelength is not observed ( compare with fig .
the lorentz oscillator strength in the gain layer was adjusted by investigating the transmission enhancement on a single layer : in previous experiments without grating it was shown , that for single layers with and without a gain layer , which exhibit transmissions @xmath61 and @xmath62 , respectively , a transmission enhancement of about @xmath63 is easily realizable @xcite .
we performed simulations where we adjusted the lorentz - oscillator strength @xmath28 to @xmath64 .
this corresponds to single layer transmission enhancements of @xmath65 and imaginary parts of the lorentz - oscillators refractive index of @xmath66 and @xmath67 at @xmath31 nm , respectively .
the results are plotted in fig .
figure 4(e ) shows the transmission spectrum with no gain as a reference .
figures 4(a)-(d ) depicts the transmission versus wavelength spectra for the above mentioned values of @xmath28 .
it is observed that the transmission at the spp resonance is drastically enhanced with increasing gain . for @xmath68 ( corresponding to @xmath69 and @xmath63 for an unstructured single layer film ) we already obtain a transmission of more than @xmath70 at the wavelength of @xmath71 nm . with increasing
gain the transmission at this wavelength increases up to 10 for @xmath72 ( @xmath73 , @xmath74 ) .
figure 3(c ) shows that there is a high intensity in all three layers .
this indicates that the multilayered systems that we can realize with the microroll technique is indeed quite helpful .
we have actually performed simulations , where we selectively turn on the gain in different layers and it turned out that the effects on the transmission are considerably smaller .
it is therefore advantageous to use a structure with three or even more layers .
we like to note that , as observed in fig . 4
, the transmission enhancement has actually a fano - type shape with a minimum at the low wavelength side .
this indicates that the negative imaginary part of the gain material shifts the spp resonance position , i.e. there is an increasing interaction between the spp and the gain layer .
+ while the excitation of spps is linked to illumination with p - polarized light we also performed for comparison simulations in s polarization .
the transmission versus wavelength spectra of this run are presented in fig .
figures 5(a)-(d ) show the spectra for lorentz - oscillator strengths @xmath64 corresponding to @xmath75 and @xmath76 , respectively .
figure 5(e ) depicts the transmission spectrum without gain .
as expected , there are no spp resonances . turning on the gain leads to an increasing transmission at the resonant frequency of the quantum well .
however , in the s - polarized case a strong transmission enhancement as in fig .
4(a)-(d ) is not observed .
our simulations show that the increased transmission for s polarization is comparable to simulations with flat layers instead of a grating .
this observation supports that the spp resonances are crucial for the extraordinary transmission .
further simulations show that the strength of the transmission enhancement as seen in fig .
4 is strongly dependent on the interplay of the fabry - perot resonance , the spp resonance and the active wavelength regime of the quantum well .
in conclusion we demonstrated that spp resonances on metallic gratings embedded into three - dimensional metamaterials containing quantum structures can lead to a strong transmission enhancement .
the dimensions of the structures and the grating period have to be chosen such that the wavelength of the spp resonance spectrally and spatially matches the emission of the quantum well .
moreover our simulations show that it is desirable to also shift the fabry - perot resonance peak to the spp resonance .
+ the authors thank markus brll and jens ehlermann for fruitful discussions and gratefully acknowledge financial support of the deutsche forschungsgemeinschaft via the graduiertenkolleg 1286 `` functional metal - semiconductor hybrid systems '' .
26ifxundefined [ 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 * * , ( ) http://dx.doi.org/10.1038/nphoton.2007.28 [ * * , ( ) ] link:\doibase 10.1126/science.1186351 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) http://dx.doi.org/10.1038/nphoton.2009.111 [ * * , ( ) ] @noop * * , ( ) http://ol.osa.org/abstract.cfm?uri=ol-30-11-1348 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.95.137404 [ * * , ( ) ] @noop * * , ( ) http://ol.osa.org/abstract.cfm?uri=ol-34-22-3478 [ * * , ( ) ] http://ol.osa.org/abstract.cfm?uri=ol-32-5-551 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevlett.102.163903 [ * * , ( ) ] http://dx.doi.org/10.1038/nature09278 [ * * , ( ) ] http://www.opticsexpress.org/abstract.cfm?uri=oe-17-10-8548 [ * * , ( ) ] link:\doibase 10.1364/oe.18.024140 [ * * , ( ) ] @noop http://www.lumerical.com/fdtd.php[__ ] , @noop _
_ ( , ) link:\doibase 10.1063/1.332655 [ * * , ( ) ] link:\doibase 10.1063/1.322979 [ * * , ( ) ] link:\doibase 10.1063/1.2800309 [ * * , ( ) ] | we investigate three - dimensional rolled - up metamaterials containing optically active quantum wells and metal gratings supporting surface plasmon polarition resonances .
finite - difference time - domain simulations show that by matching the surface plasmon polarition resonance with the active wavelength regime of the quantum well a strong transmission enhancement is observed when illuminating the sample with p - polarized radiation .
this transmission enhancement is further increased by taking advantage of the fabry - perot resonances of the structure . |
a thermally driven flow , or ludwig - soret effect , is observed when applying a thermal gradient to a complex fluid @xcite .
the corresponding mass transport is relevant for natural and technological processes , such as the global circulation of sea water @xcite and the phase behavior of eutectic systems at solidification @xcite . in recent years
, detailed experimental studies on macromolecular solutions and colloidal suspensions have revealed unambiguous and often surprising dependencies of the soret effect on system parameters such as salinity , surface coating , solute concentration , and molecular weight rau02,rau05,wie04,kit04,bra02,zha99,deg03,iac03,pia02,dem04,duh05,duh06 .
although the analogy to electrophoresis indicates the relevance of surface or marangoni forces and suggests a hydrodynamic treatment @xcite , the physcial mechanisms that drive thermophoresis in liquids are poorly understood and differ from those in gaseous phases @xcite .
diffusion theory accounts for the particle current due to density gradient in terms of fick s law @xmath0 , where @xmath1 is the number of suspended particles per unit volume . for a closed system
the stationary state is characterized by @xmath2 and , in the absence of external forces , corresponds to a homogeneous density .
thermophoresis describes an additional flow that is driven by a temperature gradient , resulting in the total particle current @xmath3with the thermal diffusion coefficient @xmath4 .
experimentally , @xmath4 is determined by applying a temperature gradient to a uniform suspension ( @xmath5 and by recording the initial current @xmath6 , or by measuring the density modulation @xmath7 induced by a temperature inhomogeneity @xmath8 in the steady state @xmath2 @xcite .
the latter method gives the soret coefficient @xmath9 eq .
( [ 2 ] ) provides a macroscopic description for the particle current in terms of the thermal forces @xcite . in a microscopic picture ,
one wishes to relate the kinetic coefficients @xmath10 and @xmath4 to the properties of solute and solvent . for this purpose
we rewrite the particle current as the sum of two terms @xcite , @xmath11the phoretic velocity @xmath12 arises from the interactions at the solute - solvent interface and thus is a single - particle effect , whereas the remainder is due to the mutual forces exerted by the suspended particles in terms of the gradient of the osmotic pressure @xmath13 . the bare mobility @xmath14 depends on the solvent viscosity @xmath15 and on the particle size @xmath16 .
( [ 4 ] ) describes a non - uniform system and can not be obtained from equilibrium statistical mechanics @xcite .
the stationary state @xmath2 provides the condition of mechanical equilibrium , i.e. , when all forces acting on a given particle cancel .
as an illustration , let us briefly discuss the case of an ideal gas . because of the absence of interactions , one has @xmath17 and the osmotic pressure reads @xmath18 .
the resulting particle current is then @xmath19 . comparing to eq .
( 2 ) gives the einstein relation @xmath20 as well as @xmath21 , with the soret coefficient @xmath22 .
this simply means that , at constant pressure , the stationary density is inversely proportional to the non - uniform temperature and that the particles accumulate in regions of lower temperature .
yet most colloidal suspensions show a considerably stronger , positive or negative , soret effect , i.e. , the thermally driven current by far exceeds that of an ideal gas and may be directed toward colder or warmer regions .
these deviations express the failure of the ideal - gas picture for the solute and emphasize the importance of particle - solvent interactions . in the present paper
we discuss the soret effect in charged colloidal suspensions ; we complete the theoretical framework given in our previous work @xcite and extend the electrostatic part to strongly charged particles . in sec .
[ sec2 ] we derive the single - particle velocity from the stokes equation for the surrounding fluid , with slip boundary condition that is related to the anisotropy of the maxwell tensor in the diffuse double layer . sec .
[ sec3 ] gives the thermodiffusion coefficient @xmath23 in terms of the surface stress , or marangoni force , and the osmotic pressure . in sec .
[ sec4 ] we obtain the electrostatic properties of the boundary layer and , in particular , the anisotropy parameter .
[ sec5 ] compare our predictions with experimental findings and previous work , and we summarize our main results in sec .
the thermal forces acting on a colloidal particle and the resulting velocity can not be obtained from a mechanical or thermodynamic potential .
thermophoresis is a hydrodynamic problem , and the particle velocity has to be derived from a fluid - mechanical treatment @xcite .
the thermally driven motion of micron or nanometer sized particles in a viscous liquid involves small reynolds numbers , i.e. , inertia effects are negligible . then the stationary velocity is given by stokes equation @xcite @xmath24 where @xmath25 is the solvent viscosity .
an incompressible fluid satisfies moreover @xmath26 and stick boundary conditions at the particle surface , @xmath27 .
besides the pressure gradient @xmath28 , the fluid is subject to the maxwell stress tensor @xmath29 which , contrary to the hydrostatic pressure @xmath30 , is anisotropic and thus takes different values parallel and perpendicular to the particle surface . the force field induced by surface charges , @xmath31 ,
is finite only within a boundary layer thickness @xmath32 , which is given by the debye screening length @xmath33 . throughout this paper
we suppose that the thickness of the boundary layer @xmath32 is much smaller than the particle radius @xmath16 , @xmath34the characteristic length scales of the normal and parallel derivatives in eq .
( [ 1 ] ) are given by @xmath32 and @xmath16 , respectively .
thus the forces vary rapidly in the normal direction , and much more slowly along the interface .
this separation of length scales permits us to calculate the particle velocity in two steps , as schematically drawn in fig .
first , resorting to a 1d approximation that is valid at distances much shorter than the particle size , we discuss the boundary layer and derive the surface force induced by the temperature gradient . in a second step
we obtain the fluid velocity field at distances well beyond @xmath32 , where the maxwell tensor vanishes .
.,width=312 ] close to particle , the surface may be considered as flat , with normal and parallel coordinates @xmath35 and @xmath36 . since the electric field @xmath37 of a uniformly charged spherical
particle is normal to the surface , the maxwell tensor is diagonal with elements @xmath38 where @xmath39 .
then stokes equation reads @xmath40 with @xmath41 , etc . the normal velocity vanishes close to the interface , @xmath42 ; thus @xmath43 so that @xmath44 is a constant .
inserting this relation in the equation for @xmath45 and noting that , because of the thermal gradient , @xmath45 is slowly varying along the surface , we find @xmath46 the derivative on the right - hand side gives the lateral force per unit volume exerted on the fluid .
since @xmath47 is finite within the boundary layer only , the force density @xmath48 vanishes beyond a distance @xmath49 from the interface .
integrating over the normal coordinate gives the total shear stress on the fluid accumulated through the boundary layer , @xmath50 the right - hand side gives the lateral force @xmath51 acting on the surface element @xmath52 of the particle .
well beyond the boundary layer , the surface can no longer be taken as flat and the above approximation ceases to be valid . at distances comparable to the particle radius @xmath16
, one has to deal with the 3d stokes equation , yet with modified boundary conditions .
the electric stress is absorbed in the boundary layer , and eq .
( [ 1 ] ) simplifies to @xmath53 .
it turns out convenient to define the anisotropy parameter characterizing the boundary layer @xmath54where @xmath55 and @xmath56 are the normal and parallel components of the maxwell tensor , and where we have used @xmath57
. then the oriented surface force reads @xmath58its direction is parallel to the particle surface and opposite to the 2d gradient @xmath59 . in the case of a non - uniform temperature one
has @xmath60where we have assumed that spatial variations arises from the thermal gradient only and that the elements of the maxwell tensor depend on temperature .
since heat propagation is much faster than particle migration , the temperature field may be taken as stationary .
starting from the overall thermal gradient of the fluid @xmath61with constant @xmath62 , the heat conduction equation for a spherical particle is readily solved , and the gradient at the particle surface reads lan87
@xmath63where the polar angle @xmath64 is given by @xmath65 see fig .
[ fig2 ] for a definition of the unitary vectors @xmath66 and @xmath67 . the parameter @xmath68is determined by the heat conductivities of solvent and particle . and @xmath67 .
the polar angle is noted @xmath69.,width=312 ] in a mesoscopic description , we may put @xmath70 and consider @xmath71 as a surface force acting at the particle - fluid interface .
thus eq .
( [ 1 ] ) reduces to the force - free stokes equation @xmath72 albeit with modified boundary conditions that account for the surface force accumulated through the boundary layer .
the stress tensor @xmath73 comprises the dissipative term or viscous force density @xmath74 , and the hydrostatic pressure @xmath30 @xcite .
the stick boundary conditions of hydrodynamics , imposing a continuous velocity at the interface , @xmath75 , cease to be valid in the presence of surface forces .
the fluid being incompressible , the normal component is continuous , @xmath76 whereas the parallel velocity in general exhibits a jump .
a second condition is obtained by noting that there is no net external force acting on the system consisting of the particle and the boundary layer .
thus the integrated normal force outside the boundary layer vanishes @xcite , @xmath77 the third condition involves the surface force ( [ 10 ] ) , which plays the role of the source field and imposes a finite shear stress @xcite , @xmath78 it turns out that this condition results in a finite slip velocity .
the source term ( [ 17 ] ) differs from that of electrophoresis , where the stokes equation in the boundary layer reads @xmath79 , with the external electric field @xmath80 and the local potential @xmath81 . integrating this equation twice , with
stick boundary conditions , gives the mesoscopic slip velocity @xmath82 as a function of the surface potential @xmath83 , resulting in @xmath84 @xcite . in the case of thermophoresis , integration of eq .
( [ 5 ] ) results in the surface stress @xmath71 ; the slip velocity is obtained only after solving eqs .
( [ 12])([17 ] ) . in the laboratory frame the particle moves at a velocity @xmath85 .
transforming to the reference frame in which the particle is at rest , we have @xmath86 and @xmath87 .
the solution of stokes equation at small reynolds numbers in spherical coordinates @xmath88 reads @xcite @xmath89 where @xmath64 is the polar angle with respect to the @xmath36 axis and the radial and tangential unit vectors @xmath90 and @xmath91 satisfy @xmath92 .
this flow field is related to a non - uniform hydrodynamic pressure @xmath93 the parameters @xmath94 are determined from the solulution of stokes with the boundary conditions ( [ 15])([17 ] ) .
the first two of these conditions involve the fluid velocity and stress only . in the particle - fixed frame
the normal velocity vanishes , @xmath95 , resulting in @xmath96 . the total stress at the interface can be written as @xmath97 , with the entries of the dissipative part in shperical coordinates @xcite @xmath98integrating ( [ 16 ] ) over a sphere just outside the boundary layer gives the relation @xmath99 .
one readily obtains the amplitudes of the velocity field varying with distance as @xmath100 and @xmath101 , respectively , @xmath102 the remaining condition ( [ 17 ] ) relates the stress @xmath103 and the surface force and determines the particle velocity @xmath104finally , taking the back transformation @xmath105 yields the fluid velocity in the laboratory frame @xmath106 the slip velocity , i.e. , the velocity change through the boundary layer , is given by the tangential component @xmath107 , @xmath108its maximum value @xmath109 at @xmath110 exceeds the particle velocity , which means that the particle and the fluid move in opposite directions .
( [ 18 ] ) provides the relation the phoretic velocity @xmath12 and the applied temperature gradient @xmath111 , @xmath112with the transport coefficient @xmath113this result displays how the phoretic velocity arises from the competition of particle - solvent interactions that drive the particle , and the fluid viscosity .
( [ 9 ] ) accounts for the balance of the stress on the fluid and the force @xmath114 acting on a surface element @xmath115 of the particle .
inserting the expression of @xmath116 and separating the dependence on the angle @xmath69 , we have @xmath117 the integral takes the value @xmath118 , resulting in the total force @xmath119 a phoretic mobility may be defined as the ratio of the one - particle velocity and the force , @xmath120 . comparing with ( 22 )
gives @xmath121 this expression differs from the mobility @xmath122 that describes a particle subject to an external force @xmath123 or the brownian force leading to diffusive motion .
the solution of the stokes equation with such an external force satisfies stick boundary conditions @xmath124 and the force balance @xmath125 , resulting in @xcite @xmath126the excess hydrodynamic pressure @xmath127 , and a velocity field @xmath128 that decays with distance as @xmath129 .
the results of the previous section are now applied to derive general expressions for the transport coefficient @xmath10 and @xmath130 .
we consider a suspension of interacting particles in an inhomogeneous temperature field @xmath131 , where both the density @xmath1 and the osmotic pressure @xmath132 are , in general , spatially varying functions .
such a system can not be dealt with in terms of equilibrium statistical mechanics ; its steady state and the relaxational kinetics are described in terms of the particle current eq .
( [ 4 ] ) . after expressing both terms by microscopic quantities proportional to @xmath133 and @xmath134 , we compare to eq .
( [ 2 ] ) .
the particle current contains , besides the single - particle velocity @xmath135 , a term proportional to the pressure gradient @xmath136 .
the latter depends on both density and temperature modulations , @xmath137 with the shorthand notation @xmath138 , etc . for a non - uniform system with a pair interaction @xmath139 ,
the pressure is calculated by explicitly writing the forces on a particle at position @xmath140 ; proceeding as with the clausius virial function one finds the sum of the ideal - gas pressure @xmath141 and the two - particle forces @xmath142 @xcite , @xmath143assuming the pair interaction @xmath144 to be isotropic and sufficiently short - ranged , one has with the density - density correlation function @xmath145 @xmath146this relation is a useful starting point for a perturbative evaluation of the diffusion current in terms of temperature and density gradients .
so far we have avoided the use of equilibrium thermodynamics ; the above relations involve mechanical and hydrodynamic forces only .
since @xmath10 and @xmath23 are linear - response coefficients , they may be evaluated in terms of equilibrium statistical mechanics , and @xmath147 may be considered as the canonical average with the gibbs distribution .
this means in particular that the density and temperature fields are replaced by their mean values @xmath1 and @xmath148 . for dilute and weakly interacting systems ,
a widely used approximation consists in expanding the pressure in powers of the particle density , @xmath149 and truncating at quadratic order , with a coefficient @xmath150 that still depends on temperature @xcite .
most relevant pair interactions are repulsive , such as the screened electrostatic potential in an electrolyte and the entropic short - range repulsion due to polymers grafted on colloidal particles .
then the virial coefficient is positive , @xmath151 and the interactions enhance the pressure . inserting @xmath12 and @xmath152 in eq .
( [ 4 ] ) and comparing to eq .
( [ 2 ] ) yields @xmath153together with the virial expansion eq .
( [ virial ] ) , the translational diffusion coefficient takes the form @xmath154which does not depend on the temperature gradient and is insensitive to surface forces .
a different picture arises for the remaining coefficient , @xmath155which depends explicitly on the temperature derivative @xmath156 and on the surface force parameter @xmath157 . for an ideal gas both the particle - solvent potential and the pair interaction
vanish , resulting in @xmath158 and @xmath159 .
one readily recovers the transport and soret coefficients @xmath20 , @xmath21 , and @xmath160 .
the above expression for the transport coefficient depends on the charged double layer through the parameter @xmath161 which characterizes the anisotropy of the maxwell stress tensor .
the electrostatic properties of the boundary layer are closely related to the distribution of the mobile co - ions and counter - ions of the electrolyte . in three dimensions , the resulting poisson - boltzmann
( pb ) equation can not be solved analytically .
approximate solutions exist for two particular situations .
first , for weak charges , the potential @xmath162 is small as compared to the thermal energy and the pb equation can be linearized .
second , for particles that are significantly larger than the debye length , the potential can be evaluated as an asymptotic series in terms of the small parameter @xmath163 since a small radius @xmath164 implies rather a weak charge , and large valencies are possible for sufficiently large particles only , the two approximations cover in fact most physical systems . in both cases ,
the resulting potential @xmath165 is an effective quantity that contains , besides the potential energy , the entropy of the mobile ions .
the limiting cases of weak and strong charges are expressed by the ratio of the bare valency @xmath166 and its effective value @xmath167 that is given in terms of the particle size @xmath16 , the debye screening parameter @xmath168 and the bjerrum length @xmath169 , with @xmath170 the dielectric constant of the solvent ( @xmath171 for water ) . in the mean - field approximation ,
the electrostatic potential of a charged particle satisfies the pb equation @xmath172this equation has to be solved together with the following boundary conditions .
first , the potential vanishes at infinity , @xmath173 , and second , the electric field at the surface of the colloid is related to the surface charge @xmath174 according to @xmath175however , since the pb equation is non - linear , one can only find an approximate solution in spherical geometry .
we discuss below the two relevant limits of weakly charged and highly charged colloids .
to proceed , it turns out convenient to define the reduced potential @xmath176that obeys the differential equation @xmath177 we first recall the electrostatic properties for weakly charged systems . in this case , the reduced potential is small everywhere , @xmath178 . then eq . ( [ s10 ] )
can be linearized , @xmath179 , and solved by the well - known debye - hckel potential @xmath180 this solution ceases to be valid if the surface potential @xmath181 approaches unity , i.e. , if the valency @xmath166 is of the order of the effective value @xmath182 .
for highly charged particles the nonlinearities in the pb equation are essential and result in a much more intricate problem .
still , an asymptotic expansion @xmath183 can be derived explicitly if the parameter @xmath184 is small , i.e. , if the debye length is smaller than the particle radius .
note that the coefficients @xmath185 in ( [ s16 ] ) still depend on @xmath184 , i.e. , we have not yet defined how to construct the series .
the usual boundary conditions of electrostatics impose that the potential vanishes at infinity , @xmath186 , and satisfies ( [ s6 ] ) at the particle surface .
since eq .
( [ s6 ] ) is independent of @xmath187 , the derivative involves the term @xmath188 only @xmath189 and vanishes for @xmath190 @xmath191 the multiple scale approach @xcite allows to systematically match the known 1d solution close to the particle surface to the screened potential at distances well beyond the debye - hckel length .
this is achieved by introducing @xmath192 where the `` fast '' variable @xmath193 accurately describes the problem in the vicinity of the surface , in addition to the `` slow '' variable @xmath194 which accounts for the spherical symmetry of the problem far from the particle .
the derivative thus becomes @xmath195 , with @xmath196 , and the laplace operator reads in terms of this new set of variables @xmath197 finally we expand the @xmath198 according to @xmath199 with @xmath200 and @xmath201 . inserting these relations in the pb equation ( [ s10 ] ) , we obtain a set of coupled equation for @xmath185 that can be solved by identifying terms of the same power of @xmath184 . we give explicitly the equation at order @xmath202 , @xmath203 and that at order @xmath204 , @xmath205 at lowest order in @xmath184 the differential equation ( [ s28 ] ) can be integrated as @xcite @xmath207 where we have used the condition @xmath208 and @xmath209 as @xmath210 this equation can be integrated once more and one finds @xcite @xmath211 where the unknown integration function @xmath212 remains to be determined .
the point is now that we artificially introduced an extra variable in the problem .
the equation of order @xmath213 and the boundary conditions at @xmath214 and @xmath215 are not sufficient to obtain the function of two variables @xmath216 .
following @xcite , we thus require that the non homogeneous part of eq .
( [ s30 ] ) for the _ first order _ solution decays faster than @xmath217 $ ] .
explicitly , this requirement reads @xmath218 so that @xmath219 , with @xmath220 constant . coming back to the original variable
, the lowest order term of the expansion reads @xmath221 with @xmath222 \ .\ ] ] the integration constant @xmath220 is finally obtained from the boundary condition ( [ s18 ] ) @xmath223 to leading order in @xmath224 , the parameter @xmath225 may be written as @xmath226 , i.e. , the ratio of debye and gouy - chapman lengths @xmath227 and @xmath228 .
if @xmath230 , one has @xmath231 and the solution ( [ s36 ] ) reduces to @xmath232 as expected , we recover the debye - hckel solution for weakly charged colloids . in the limit @xmath234 , the solution @xmath206 exhibits quite different behaviors at short and long distances . in the latter case ,
one recovers well beyond the debye length , @xmath235 , the screened potential @xmath236 albeit with the effective valency @xmath182 .
close to the surface of the particle , @xmath237 , the potential simplifies to a logarithmic law @xmath238 higher order terms in the expansion can be evaluated following the same method .
we give below only the first order correction , which is obtained by solving eq .
( [ s30 ] ) together with the requirement that the non homogeneous part of the equation for @xmath239 does not contains terms proportional to @xmath217 $ ] .
we obtain @xmath240 \ .\ ] ] the constant @xmath241 is obtained by enforcing the boundary condition at the surface of the particle @xmath242 .
explicitly , we find @xmath243 expanding this expression at lowest order in @xmath184 , one recovers the result of ref . @xcite .
incidentally , we note that the effective charge obtained from the behavior of the potential in the limit @xmath234 is now corrected to become @xmath244 , as pointed out by trizac and collaborators @xcite .
this correction is , however , irrelevant in the limit @xmath245 considered here .
contributions of quadratic and higher order in @xmath184 can be obtained by repeating the same procedure . in view of eq.([8 ] ) , we now evaluate the maxwell stress tensor which is given by the components of the electric field @xmath246 , @xmath247 since the electric field of a spherical particle is normal to the surface , the off - diagonal or shear components vanish ; the diagonal elements are finite , the normal one being positive and the two parallel to the surface negative , @xmath248 in the limit @xmath245 the particle surface may be considered as flat , and the anisotropy parameter ( [ 8 ] ) reads @xmath249 inserting @xmath206 one obtains @xmath250 . replacing the valencies by the charge number densities
@xmath251 we find @xmath252 in the two limiting cases with respect to the ratio @xmath253 one has @xmath254 the effective charge density @xmath255 depends on temperature , whereas the bare value @xmath256 is constant . in an inhomogeneous system
, @xmath257 is slowly varying along the interface , and @xmath255 and @xmath161 are well defined local quantities .
the derivative @xmath258 is calculated by rewriting the effective charge density as @xmath259 and by noting the dependencies of debye and bjerrum lengths , @xmath260 and @xmath261 , and of the dielectric constant @xmath262 @xcite @xmath263 thus one finds the relation @xmath264 , and finally obtains @xmath265 at room temperature the parameter @xmath266 takes the value @xmath267 .
in the limit of weak charges one has @xmath268 and thus recovers the expression given previously in wue07 @xmath269whereas in the opposite case we find @xmath270these expressions for the parameter @xmath271 constitute a main result of the present paper ; with eqs .
( [ 23 ] ) and ( [ 40 ] ) they provides the explicit form of the thermodiffusion coefficient @xmath4 .
the preceding sections provide an explicit expression for the thermally driven particle current @xmath272which according to eq .
( [ 40 ] ) consists of two contributions ; the first one , @xmath273 , describes a diffusive current due to the non - uniform stationary state in a temperature gradient , whereas the second one , @xmath274 , arises from surface forces .
the latter contribution is proportional to the parameter @xmath271 calculated above . in this section
we discuss the relative magnitude of these contributions , their dependencies on physical parameters of the solute and the solvent , and compare our results with experimental findings and previous theoretical work .
we mostly consider the dilute limit , where @xmath275 interaction effects that occur at higher densities are addressed separately .
it is instructive to estimate the relative magnitude of the two contributions to @xmath4 , that is the ratio @xmath276 this quantity exceeds unity for most experimentally relevant systems .
thus for micron - sized polystyrene beads with typical parameters @xmath277 @xmath278 m , @xmath279 e / nm@xmath280 , @xmath281 nm , and the viscosity and permittivity of water @xcite , one finds the value @xmath282 . even for solutes in the nanometer range ,
this ratio may be large ; for instance in the case of sodium dodecyl sulphate ( sds ) micelles of valency @xmath283 and radius @xmath284 nm pia02 , the resulting ratio @xmath285 lies between @xmath286 and @xmath287 .
note that we have used here @xmath288 , which holds for most colloidal suspensions .
similar arguments hold for polymer solutions , where the monomer - solvent interaction is at least of the order @xmath289 . the phoretic coefficient
@xmath290 scales with the inverse monomer size @xmath16 , whereas the mobility @xmath291 involves the gyration radius @xmath292 , resulting in @xmath293 @xcite .
thus we are led to the important conclusion that for solutes larger than a few nanometers , thermal diffusion is in general driven by surface forces , i.e. , @xmath294 this statement is confirmed by the experimental observation that hardly any system satisfies the ratio @xmath295 expected for diffusive thermal transport . in physical terms
this means that the ideal - gas or entropic contribution @xmath296 is much smaller than the surface force term @xmath157 that arises from solute - solvent interactions .
note also that @xmath258 is always positive for the electric double layer , which means @xmath297 and a thermophoretic current toward colder regions . in the range of parameters considered in this paper , @xmath298 ,
the parameter @xmath299 is independent of the particle size , and so are the surface energy @xmath300 and its derivative @xmath271 . as a consequence ,
the phoretic coefficient @xmath157 varies linearly with the particle size , @xmath301the stationary velocity @xmath302 is thus proportional to @xmath303 , whereas the diffusive contribution @xmath304 varies as @xmath305 . as a function of the reduced charge density @xmath306
the behavior of the phoretic coefficient switches from quadratic at low charge density to linear at high charge density.,width=312 ] for the case of weakly charged surfaces @xmath307 , one finds from eq .
( [ t4 ] ) that the phoretic coefficient @xmath157 varies quadratically with the charge density @xmath308moreover , @xmath157 is proportional to the particle size @xmath16 and to the debye screening length @xmath309 . both dependencies on @xmath16 and @xmath310
are confirmed by experiment , as discussed below for the soret coefficient . in the opposite limit of strongly charged particles , the phoretic coefficient @xmath311varies linearly with the surface charge density and with the particle size , but
is independent of the debye length and the dielectric constant of the electrolyte .
the dependence with the reduced charge density @xmath312 is shown on fig .
[ fig3 ] . for the case of small valencies
@xmath313 , the soret coefficient reads @xmath314 as discussed below , the ideal - gas value @xmath315 is small for particles larger than nanometers and may be safely neglected .
the soret coefficient is given by the remaining charge term .
the power laws @xmath316 and @xmath317 , as first predicted in @xcite with @xmath318 , have been confirmed recently in experiments on polystyrene beads in the parameter range @xmath319 nm and @xmath320 nm @xcite .
these dependencies may be combined to an interaction volume @xmath321 , i.e. , a spherical shell of thickness @xmath32 and mean radius @xmath16 , thus giving clear evidence for surface forces operating in a boundary layer at the particle surface . in the limit @xmath322
, the soret coefficient reduces to @xmath323 for large valencies one has the particularly simple form @xmath324 .
now we turn to higher particle densities where the first correction of the virial expansion is no longer small .
although the surface - force driven coefficient @xmath157 does not depend on density , the diffusion coefficient varies as @xmath325 .
thus for a repulsive pair interaction the soret coefficient decreases as a function of the density @xmath326 as shown in appendix [ appa ] , the virial coefficient @xmath327 depends on the debye length @xmath32 , which in turn depends on the electrolyte strength as @xmath328 . at high salinity , one has @xmath329 , resulting in strong screening of the electrostatic interaction and the excluded volume of hard spheres , i.e. , @xmath330 . in the dilute limit or at high salinity , the soret coefficient is proportional to @xmath331 and thus varies as @xmath332as salinity decreases , the debye length increases , and so does @xmath327 . if the screening length becomes comparable to the particle size , @xmath333 , the virial coefficient varies as @xmath334 . as a consequence , for a dense suspension in a weak electrolyte ,
the above behavior is modified by the virial coefficient , and the soret coefficient increases with salinity , @xmath335 both dependencies on @xmath1 and @xmath336 have been , at least qualitatively , observed for sds micelles in aqueous solution @xcite . the inverse coefficient
@xmath337 coefficient has been shown to vary linearly with density , and its slope to decrease as salinity augments , like the factor @xmath338 above .
moreover , fig . 1 of ref .
@xcite reveals that in the dilute limit , increasing the salinity reduces the soret effect , whereas for the dense micelle suspension @xmath339 increases when adding salt .
these experimental findings agree almost quantitatively with our theoretical expressions .
a more detailed comparison is not an easy undertaking , since both the size and the charge of the micelles slightly vary with the electrolyte strength .
also , part of the data are in the range @xmath340 whereas the present work relies on @xmath341 .
the charged double layer is predicted to result in a normal soret effect @xmath342 , that is , the particles drift in the direction opposite to the thermal gradient and thus accumulate in colder regions of the sample .
there are , however , systems that show a more complex behavior .
aqueous suspensions of maghemite particles present an inverse soret effect @xmath343 , whereas the normal behavior has been observed in organic solvents @xcite .
also , the soret coefficient of macromolecular solutions of lysozyme @xcite or dna @xcite changes sign as a function of temperature and salinity .
especially these latter examples strongly suggest that the observed transport coefficient @xmath23 comprises contributions of different physical origin and of opposite sign .
3a of ref .
@xcite displays particularly instructive data for the soret coefficient of polystyrene beads as a function of the debye length @xmath344 at different temperatures .
@xmath339 varies linearly with @xmath49 , as expected from ( [ d18 ] ) , yet does not vanish at @xmath345 ; the finite intercept is negative at @xmath346c , increases with temperature , and changes sign at about 20@xmath347c .
these observations give evidence for an additional contribution @xmath348 that is insensitive to charge effects but strongly temperature - dependent .
a similar behavior has been reported for macromolecular suspensions @xcite .
these experimental findings indicate that , besides the maxwell stress tensor of the charged double layer , additional forces operate at the particle surface @xcite . for micron
sized solutes , solid - fluid interface tension and grafted surfactant molecules contribute to the anisotropy parameter @xmath161 .
a very rough estimate for the tension may be obtained from the phenomenological law @xmath349 for its temperature dependence . with parameters
@xmath350 mn / m and @xmath351 k , the force parameter @xmath352 would by far exceed the double - layer contribution .
yet one should be aware that the actual situation at solid - fluid interfaces , especially at polar ones , is significantly more complex than suggested by the simple relation @xmath353 .
moreover , because of the short range of dispersion forces , the resulting stress anisotropy occurs in a narrow layer of about one nanometer thickness .
macroscopic hydrodynamics with constant viscosity may well not apply at such short distances .
finally , we want to compare our analysis with previous work . in the spirit of refs .
@xcite , our description of thermophoresis relies on hydrodynamics at low reynolds numbers . in this approach ,
the temperature dependence of the maxwell tensor results in a shear stress parallel to the surface , and the force acting on the particle is given by the stress accumulated through the boundary layer .
solving stokes equation with the appropriate boundary condition , one obtains the velocity of the particle and the flow of the surrounding fluid .
a different route has been taken by several authors in the past few years @xcite .
in those works , the driving force is related to the gradient of the effective energy @xmath354 of the charged double layer , @xmath355this expression relies on the assumption that @xmath354 may be treated like an external potential .
the phoretic velocity is obtained from the steady state where the external force @xmath356 and the stokes drag @xmath357 cancel , resulting in @xmath358 and in the thermodiffusion coefficient @xmath359comparison with eqs .
( [ d18 ] ) and ( [ d19 ] ) reveals several discrepancies which we discuss briefly . at small valencies
the double - layer free energy and the integrated anisotropy parameter @xmath300 turn out to be identical , @xmath360eq .
( [ c10 ] ) of appendix [ appb ] leads to the soret coefficient @xmath361 . comparing with eq .
( [ d18 ] ) shows that the charge terms obtained from hydrodynamics and from the external - force picture differ by a factor @xmath362the parameter @xmath363 accounts for the local modification of the temperature gradient due to the different heat conductivities of solvent and particle .
for @xmath364 one has @xmath365 and a constant gradient @xmath366 , whereas @xmath367 flattens the temperature profile around the particle and thus significantly reduces thermal transport .
the latter case is relevant for suspensions of crystalline particles , where the large heat conductivity indeed suppresses thermophoresis .
the remaining factor @xmath368 occurring in the hydrodynamic treatment is related to the shape of the particle . for a prolate particle oriented along the thermal gradient ,
this factor increases , whereas it is reduced for an oblate shape .
this dependency could be relevant for suspended rods , such as dna with a persistence length exceeding the size of the molecule .
although it misses the numerical prefactor @xmath369 , the picture of an external force provides , for weak charges , qualitatively correct results .
an additional discrepancy arises in the limit of high valencies . comparing the charging energy @xmath354 and the anisotropy parameter @xmath370 reveals , that they differ by a logarithmic factor @xmath371 . thus using @xmath354 for
strongly charged particles results in overestimating the soret coefficient by a factor @xmath372 . from a more formal point of view , eq .
( [ d24 ] ) suffers from two major shortcomings .
first , since the entropy @xmath373 of a classical system is defined with respect to an arbitrary reference value @xmath374 only , the free energy @xmath375 comprises an arbitrary constant @xmath374 , and its the absolute value of @xmath376 does not carry physical information .
this problem persists for the charging energy @xmath354 . since @xmath354 is defined with respect to the state where the excess charges are at infinity ; yet this reference state comprises the entropy of the mobile ions , which results in an arbitrary contribution to the derivative @xmath377 .
a second shortcoming occurs when applying ( [ d24 ] ) to polymer solutions and dense colloidal suspensions . since the thermally induced surface stress ( [ 9 ] ) is treated as an external force @xcite , the resulting flow field @xmath378
comprises , contrary to eq .
( [ 20 ] ) , a term @xmath379 that would result in long - range hydrodynamic interactions @xcite . as a consequence , for a particle suspension of volume fraction @xmath380 ,
one would find a reduced value @xmath381 with @xmath382 @xcite , which is , however , an artefact of the external - force picture .
an even more serious discrepancy arises for polymers , where ( [ d26 ] ) leads to underestimate the thermophoretic current by a factor @xmath383 , with the number of monomers @xmath384 and the exponent @xmath385 @xcite .
the correct treatment of the surface stress gives @xmath386 as in ( [ 20 ] ) , i.e. , there are no long - range hydrodynamic interactions . still , the debye - hckel approximation ( [ c10 ] ) provides a correct description for dilute suspensions of weakly charged particles fay05,duh06,dho07 .
given the uncertainty of the surface charge density , the missing numerical prefactor @xmath387 is of little practical consequence .
starting from stokes equation with a force term arising from the electric stress in diffuse double layer , we have obtained the thermodiffusion coefficient @xmath4 for charged colloidal particles .
we briefly summarize our main results .
\(i ) in the limit of small valencies we confirm previous work obtained in debye - hckel approximation .
novel results arise for strong charges , where the soret coefficient @xmath324 is proportional to the ratio of thermal conductivities @xmath363 and the valency @xmath166 only .
\(ii ) the coefficient @xmath388 is proportional to the particle size , for weak charges it is moreover linear in the debye length @xmath49 . for strong charges , however , @xmath4 does not vary with @xmath49 ; in this range the soret effect does not depend on the salinity nor on the permittivity of the electrolyte .
\(iii ) comparing the charging energy @xmath354 and the integrated anisotropy parameter @xmath389 of the maxwell stress tensor , we find that these quantities coincide for weak valencies , which explains the qualitatively correct results obtained from the simple external - force picture of eq .
( [ d24 ] ) . yet
significant discrepancies arise in the opposite case of high valencies , where @xmath390 results in linear behavior @xmath391 . in this range
@xmath354 carries a logarithmic factor @xmath372 .
the resulting ambiguities would alter the soret coefficient obtained from ( [ d26 ] ) , and illustrate the need of a rigorous treatment .
the authors thank e. trizac for helpful remarks .
acknowledges stimulating discussions with w. khler and s. semenov .
in view of the expressions for the diffusion coefficients @xmath10 we evaluate the virial coefficient @xmath327 for charged particles in a simple approximation . a `` particle '' consists of one colloidal macro - ion and its diffuse cloud of counterions and co - ions . for repulsive potentials ,
the coefficient @xmath327 accounts for an excluded volume and is expressed in terms of an effective particle radius @xmath392 that is larger than @xmath16 . at distances
@xmath393 well beyond the debye length , the pair interaction @xmath394 of two colloidal particles is given by the electrostatic energy of the first one in the debye - hckel potential of the second , @xmath395 here @xmath396 indicates the bare valency if @xmath229 , and @xmath397 in the opposite case .
since the virial coefficient @xmath327 can not be evaluated as it stands , we resort to an approximation that consists in replacing the integral in ( [ 36 ] ) by an excluded volume @xmath398where @xmath392 is the effective particle radius . since the particle may be considered as a rigid body , @xmath392 is always larger than @xmath16 , thus accounting for the repulsive interaction .
we define the effective radius as the distance where the pair potential is equal to the thermal energy , @xmath399indeed , the neighbors hardly penetrate in a region of size @xmath392 about a given particle , whereas they move rather freely in the remaining space . from the expression of the pair potential @xmath394 one finds
@xmath400where the quantity @xmath401closes an implicit equation for @xmath392 . in the absence of interactions , or at very high salinity ( @xmath402 ) , the excluded - volume radius is twice the particle size , @xmath403 .
the enhancement due to the electrostatic repulsion is of the order of the debye screening length . with typical parameters for sds micelles , @xmath404 , @xmath405 nm , @xmath406 nm
, one finds a logarithmic prefactor of the order @xmath407 .
the mean - field free energy of the charged double layer is usually given in terms of the charging energy , i.e. , the work required to assemble the charge @xmath408 on the particle from infinity , @xmath409 in terms of the reduced potential @xmath206 given in ( [ s36 ] ) , we find with @xmath410 @xmath411elementary integration gives@xmath412with the shorthand notation @xmath413 . in the limiting cases with respect to the ratio of bare and effective valencies this expression simplifies to @xmath414 | we present a microscopic theory for thermally driven transport in colloidal suspension of charged particles . for small valency ,
we recover known results from debye - hckel approximation , which are in quantitative agreement with recent experiments on polystyrene particles and sodium dodecyl sulphate micelles .
significant modifications occur for strong charges , where the thermodiffusion coefficient is independent of the debye length and the permittivity of the electrolyte , but is proportional to the valency of the macroion . |
safety line is a company that offers innovative solutions ( software and statistical analysis ) for risk management in the field of air transport ( airlines , maintenance organizations , airports ... ) .
the main expertise of safety line relies in hazard identification and risk assessment , assurance and safety promotion .
+ the objective of safety line for this cemracs project is to improve its technical modeling of plane systems . to monitor the proper functioning of a given system ,
their overall approach is to follow the time evolution of an indicator of the state of the system and detect deviations from the expected behaviour , which could be indicative of a malfunction . for this ,
the main challenge is to estimate as precisely as possible and at any time the value of this indicator in normal operating conditions .
in this project , we are interested in evaluating the state of the plane braking system . to this aim , we chose to focus on the indicator given by the deceleration force of the plane during landing , from the moment when the plane wheels touch the ground up to the moment when the plane leaves the track .
indeed , the deceleration force of the plane is the sum of an aerodynamic component and a component related to the brakes .
there are no wear on the aerodynamic component , unlike that of the brakes which varies according to the state of the system .
+ to this aim , we have at our disposal a database containing the measurements recorded over @xmath1 landings . for each landing ,
different time - dependent quantities are available such as the deceleration profile , the angle of the brake level manipulated by the pilot , the speed of the plane ... + the goal of this project is to build a purely statistical model from the available data , which can be used to reconstruct the deceleration force profile of the plane given a new set of input quantities . to achieve this task , we chose a bayesian approach based on gaussian processes , inspired from ideas of @xcite and present the results we obtained with this approach .
we also compared our strategy with other regression models which were already used by safety line , such as linear regression methods or random forests .
the approach using gaussian processes seems to perform significantly better than all these other methods .
+ in section [ sec 1 ] , we detail the structure of the database we trained our model on , before moving to the presentation of the bayesian approach with gaussian processes in section [ sec 2 ] .
the numerical results we obtained are commented in section [ sec 3 ] . in section [ disc ] , we provide some discussions about other methods that take into account time series influence in the prediction model .
as announced in the introduction , we are interested in reconstructing the time - dependent profile in the deceleration force of the plane during landing , which is a good indicator of the state of the braking system , from given input quantities
. + the objective of building this statistical model is to detect malfunctions of the plane from the deviation of the recorded deceleration force profile to the predicted one . in order to do so ,
ideally , our statistical model should be trained on data recorded for planes whose braking systems are new , or at least in a good state .
unfortunately this is a piece of information we did not have access to .
thus , we trained our model on the data we had at hand , the main objective being to demonstrate the feasibility and potential of our approach .
+ however , the state of the braking system of most planes recorded being hopefully good for most planes in average , it is reasonable to think that this data will nevertheless be sufficient to provide information on the mean behaviour of deceleration profiles of planes during landing , and that it should give useful indications in order to detect malfunctions of their braking system .
+ at this point , a first difficulty was to clearly define the set of inputs our statistical model will be built on .
indeed , for each landing , a huge amount of data is recorded and we have to select the quantities that are meaningful for the reconstruction of the deceleration force profile .
we chose to select input quantities that enable us to evaluate the different forces which act on the plane during its landing . +
all time - dependent data are provided discretely with one measurement per second during a period of @xmath2 seconds .
thus , for any quantity @xmath3 , its profile during landing will be characterized by a vector of size @xmath4 , @xmath5 , where @xmath6 denotes the measurement of the quantity @xmath3 at time @xmath7 .
we consider the following input data : + * the weight of the plane @xmath8 ; + * the initial kinetic energy of the plane @xmath9 , where @xmath10 is the initial speed of the plane ; + * the speed of the plane during the landing @xmath11 ; + * the thrust force @xmath12 , where for all time @xmath13 , @xmath14 is evaluated as the product of the square of the speed of the plane @xmath15 times the level of the reverse throttle ; + * the vector @xmath16 ; at any time @xmath13 , @xmath17 is equal to the product of the speed times the angle of the brake level , thus @xmath18 gives an indication of the braking force during landing ; * the drag force @xmath19 , which is a function of @xmath15 at any time @xmath20 .
all these quantities , except the weight and the initial kinetic energy , are time - dependent functions .
note that rebuilding the deceleration from only the speed of the plane is not that obvious .
in fact , the sensors are not optimal ( because of the noise ) , so that the derivative of the speed signal is not equal to the measured deceleration profile .
this leads to additional difficulties .
the output quantity we wish to reconstruct is the deceleration force of the plane @xmath21 , given as a vector @xmath22 , where for all @xmath23 , @xmath24 is equal to the product of the deceleration times the mass of the plane .
+ we consider a training data set which contains the recordings related to @xmath25 different landings . in all the rest of the document
, the superscript @xmath26 @xmath27 refers to the label of the landing recorded in the database ; the index @xmath28 refers to the type of input quantity among the @xmath29 considered and presented in section [ sec 1.2 ] ( weight , initial kinetic energy , speed , thrust force , braking force and drag force ) ; lastly , the superscript @xmath30 refers to the instant of the measurement ( @xmath23 ) .
+ more precisely , the data recorded for the @xmath31 landing at a time @xmath32 consists of : + @xmath33 where we use the same quantities as those introduced in sections [ sec 1.2 ] and [ sec 1.3 ] .
+ we also denote by @xmath34 and @xmath35 .
thus , for all @xmath36 , @xmath37 and @xmath38 . the training data set can then be written as @xmath39 .
we also denote by @xmath40 ( respectively @xmath41 ) the set of input ( respectively output ) quantities of the database . for all @xmath23
, we also define @xmath42 .
we use here a bayesian approach presented in @xcite .
+ let @xmath43 be a probability space .
all input and output quantities are considered as random vectors .
+ let @xmath13 be a fixed time .
we assume that , for all @xmath44 ( random ) value of the input quantities , the associated observed value of the output quantity at time @xmath30 , denoted by @xmath45 , can be written as : + @xmath46 where @xmath47 is a ( random ) function @xmath48 such that @xmath49 is equal to the true value of the output quantity at time @xmath30 for the input @xmath50 , and where @xmath51 is a random variable modeling the error due to the noise made on the measurement of the output quantity a time @xmath30 .
more generally , for a set of @xmath52 random inputs , @xmath53 , if we denote by @xmath54 the set of measured associated output values , @xmath55 and @xmath56 the set of noises made on the measurement of the output quantities , we have + @xmath57 in a bayesian framework , the law of @xmath58 is called the _ likelihood _ and is chosen a priori .
it is directly related to the choice of the law of the random vector @xmath59 modeling the noise made on the measurements of the output quantity . in the sequel
, we will assume that the variables @xmath60 are independent , identically distributed , centered gaussian variables with variance @xmath61 , so that @xmath59 is a random gaussian vector of law @xmath62 where @xmath63 denotes the identity matrix of @xmath64 .
thus , @xmath65 besides , the law of the function value @xmath66 conditioned to the knowledge of the value of the input @xmath50 is called the _ prior distribution _ and is also chosen a priori .
we use here a model where @xmath67 is assumed to be a gaussian process characterized by its mean @xmath68 and covariance function @xmath69 , i.e. @xmath70 in particular , this implies that @xmath71 is a random gaussian vector of size @xmath72 , of mean @xmath73 and covariance matrix @xmath74 where @xmath75 thus , @xmath76 . in the sequel
, we assume that @xmath77 and that the covariance function ( or kernel ) @xmath78 is chosen as a squared exponential covariance function defined by : + for all @xmath79 , @xmath80 , + @xmath81\ ] ] where @xmath82 denotes the frobenius norm on vectors of dimension @xmath4 .
+ the parameters @xmath83 @xmath84 , @xmath85 , which the prior distribution and likelihood depend on , are positive real numbers called _ hyperparameters , and their values depend a priori on the time @xmath20 . they have to be chosen in an appropriate way which is detailed in section [ sec 2.3 ] . _ at this point
, we would like to comment on the simplistic choice we made assuming that the mean function @xmath86 should be zero , and in the particular form of the kernel function @xmath78 we introduced above .
of course , this choice is not the only possibility and one could think for instance to borrow ideas from kriging methods in order to obtain a better guess of this mean function .
the universal kriging technique is an example of such a method which could be used to evaluate the function @xmath86 .
indeed , in this case , the function @xmath86 is approximated by @xmath87 for some @xmath88 , where @xmath89 is a set of a priori fixed basis functions , and @xmath90 are real coefficients which can be viewed as an additional set of hyperparameters and which are to be determined from the database we have at hand . however , in our case , because of the high - dimensional character of the input quantities of our database ( @xmath91 variates ) , the choice of a meaningful set of basis functions and the identification of associated parameters is a quite intricate task .
indeed , even if we chose a simplistic linear regression model , we would have to fit @xmath92 additional hyperparameters .
it would be interesting though to test if the choice of a better mean function @xmath86 could help in improving the results we obtained .
as announced above , the numerical results presented below were obtained in the simple case where the mean function @xmath86 is assumed to be @xmath93 .
assume for now that the values of the hyperparameters @xmath83 @xmath84 , @xmath94 ( @xmath95 ) , have been chosen for the time @xmath30 .
+ let @xmath96 and @xmath97 be a set of @xmath98 new input vectors such that a priori @xmath99 is not included in the set @xmath100 of input values of the database .
we present in this section how the set of the values of the deceleration of the plane at the time @xmath30 for each input vector , @xmath101 , can be reconstructed using the regression model based on gaussian processes .
+ let us consider a test random vector of input quantities @xmath102 and denote by @xmath103 where for all @xmath104 , @xmath105 is the `` true '' output value for the input vector @xmath106 . + the joint distribution of the previously observed target values @xmath107 and the randon vector @xmath108 can be written as ( using the gaussian process model introduced in the preceding section ):
@xmath109 \sim \mathcal{n } \left ( 0 , \left [ \begin{array}{cc } k_{\textbf{x},\textbf{x}}+ \sigma^2i_{n_{ob } } & k_{\textbf{x},\textbf{x}_*}\\ k_{\textbf{x}_*,\textbf{x } } & k_{\textbf{x}_*,\textbf{x}_*}\\ \end{array}\right ] \right),\ ] ] where @xmath110 @xmath111 @xmath112 we thus obtain the law of @xmath113 which reads @xmath114 where @xmath115 , \\ & = k_{\textbf{x}_*,\textbf{x}}\big ( k_{\textbf{x},\textbf{x}}+ \sigma^2i_{n_{ob } } \big ) ^{-1}\textbf{y}^t,\\ \textbf{s}_*^t & = k_{\textbf{x}_*,\textbf{x}_*}-k_{\textbf{x}_*,\textbf{x } } \big [ k_{\textbf{x},\textbf{x}}+\sigma^2i_{n_{ob}}\big ] ^{-1 } k_{\textbf{x},\textbf{x } _ * } \end{aligned}\ ] ] for all @xmath104 , the vector of the reconstructed values of the deceleration of the plane at time @xmath30 , @xmath116 associated to the set of input quantities @xmath117 is then given by @xmath118= k_{x_*,x_d}\big [ k_{x_d , x_d}+ \sigma^2i_{n_{ob } } \big ] ^{-1}y_d^t,\\\end{aligned}\ ] ] thus , for all @xmath119 , @xmath120 can be seen as a particular linear combination of the output values @xmath121 belonging to the database .
the full time - dependent evolution of the deceleration force profile is then given by @xmath122 .
we thus have built a purely statistical regression model from the database @xmath123 we have at our disposal : + @xmath124 from a training database @xmath123 , and a set of new random onput vectors @xmath125 , this regression model enables to reconstruct the profile of the deceleration force of the plane during the landing @xmath126 for the value of the input quantities @xmath127 ( @xmath104 ) .
let us denote by @xmath128 a set of hyperparameters for the bayesian gaussian process model introduced in section [ sec 2.1 ] , and let us denote by @xmath129 the random matrix defined by using the kernel function @xmath78 defined by with this set of hyperparameters .
+ in this section , we present how we choose the value of these hyperparameters ( which a priori depends on the time @xmath30 considered ) , @xmath130 , which we use in order to build the regression model for the reconstruction of the deceleration force profile of the plane , as described in section [ sec 2.2 ] .
+ the probability density functions of the random variables @xmath131 and @xmath71 are functions which depend on the value of these hyperparameters and we denote then respectively by @xmath132 and @xmath133 . the probability density function of the variable @xmath134 is called the _ marginal likelihood _ , depends also on the value of the hyperparameters @xmath135 and can be expressed as a function of the prior and likelihood distributions @xmath136 using the gaussian process model described in the preceding section , we can derive an explicit expression of the _ log marginal likelihood _ @xmath137 as follows ( see @xcite ) : @xmath138 a classical approach to set the values of the hyperparameters for a given time @xmath139 in an optimal way is to maximize the marginal likelihood of the database we have at our disposal , in other words , @xmath140 is chosen to be solution of @xmath141 where @xmath142 thus , this set @xmath130 of hyperparameters is chosen to be the one which makes the database we have `` as likely as possible '' . + in principle , the values of the hyperparameters @xmath130 should be computed for each time @xmath139 , which would lead to the resolution of @xmath143 optimization problems depending on @xmath144 parameters each .
this ideal approach is too costly from a computational point of view , so we adopted a simplified approach which requires however to make additional assumptions on the law of the process @xmath145 . + for @xmath146 ,
let us introduce @xmath147 such that @xmath148 instead of computing different sets of hyperparameters @xmath130 for all times @xmath139 , we only compute for all @xmath149 , one set of hyperparameters @xmath150 which will be the same for all times @xmath30 belonging to the time subinterval @xmath151 $ ] . to compute the optimal value @xmath150 of this set of hyperparameters , we make an additional assumption on the law of @xmath145 : we assume that for all @xmath152 , the random vectors @xmath134 are independent from one another for all @xmath153 $ ] .
this implies that we assume that there is no correlation between the values of the observed output quantities at two different times belonging to the same time subinterval @xmath151 $ ] , which is not true in general of course .
however , this very crude assumption enables us to significantly simplify the calculations of the hyperparameters while the produced regression model gives very reasonable results as will be seen in section [ sec 3 ] .
+ for all @xmath152 , the optimal value of the hyperparameters @xmath150 is then chosen as the solution of the following optimization problem @xmath154 where @xmath155 in the numerical results presented in section [ sec 3 ] , we illustrate two different ways to choose the times @xmath156 : * a first choice consists in taking @xmath157 , and thus @xmath158 and @xmath159 ; in this case , we only compute one set of hyperparameters @xmath160 which are valid for the reconstruction of the deceleration profile of the plane at all times @xmath139 ; * a second choice , consists in taking @xmath161 and for all @xmath152 , @xmath162 ; the true interval @xmath163 $ ] is partitioned into @xmath164 time subintervals and we compute @xmath164 sets of optimal hyperparameters for each of these subintervals . in section [ sec 3 ] , we compare the results obtained with the first and second strategy .
the optimization problems are solved in practice using a standard gradient algorithm .
as mentioned before , we compared the approach we detailed in section [ sec 2 ] . to this aim
, we consider the following different strategies : * linear regression ( lr ) ; * generalized additive model ( gam ) ; * multivariate adaptative regression splines ( mars ) ; * random forests ( rf ) .
we denote respectively by @xmath165 , @xmath166 , @xmath167 and @xmath168 the obtained regression models with the training database @xmath123 , which are all applications from @xmath169 to @xmath170 .
let us present the general idea of each of those models except for the well - known linear regression , using the notation of the preceding section .
for the sake of brievity , we do not give all implementation details here .
+ the generalized additive model ( see @xcite ) is an extension of the generalized linear regression approach to non - linear relationships proposed by hastie and tibshirani .
this model is constructed as a sum of smooth functions of each of the covariates .
the interest of such a method is that each smooth function is able to reproduce any shapes .
the smooth functions were estimated by cubic regression splines .
in other words , for all @xmath104 , if @xmath171 , then the associated deceleration profile @xmath172 is reconstructed as follows @xmath173 where @xmath174 is a smooth function for all @xmath175 and @xmath176 . + mars ( see @xcite ) is an automatic procedure for modeling the output using the most significant non - linear relationships and interactions between covariates .
the model is a sum of basis functions which are either a single hinge function or either a product of one or more hinge functions .
a hinge function is a piecewise function with two pieces on both sides of a knot .
one piece is set at zero and the other piece corresponds to a linear function . for all @xmath177 ,
then @xmath172 is reconstructed as follows @xmath178 where @xmath179 are the intercept and the slope parameters and @xmath180 being @xmath8 smooth real - valued functions defined on @xmath181 .
mars automatically selects the most significant basis functions by applying a procedure with two steps : a forward pass and a pruning pass .
the forward pass delivers a model with too many basis functions that overfit the data while the pruning pass removes the least significant basis functions to obtain the most accurate sub - model .
the forward pass selects iteratively the best reflected pair of hinge function among all possible functions .
the set of possible functions is built taking all observed covariates values as a knot .
then , the pruning pass removes the least significant basis function one by one until it finds the most accurate subset of basis functions .
+ finally , random forests ( see @xcite ) is a machine learning algorithm .
such a model is composed by an ensemble of decision tree models built on random learning datasets .
briefly , a tree model is a recursive partitioning of the observations according to their similarities in covariates and output .
tree models are built using the classification and regression tree ( cart ) algorithm .
the initial dataset is split into two clusters according to a threshold value for one covariate : one cluster have higher value or the other cluster lower value .
the algorithm evaluates all possible thresholds and selects the one which minimizes the total sum of squared errors .
the partitioning is repeated for each cluster until there are less than @xmath182 observations per cluster .
the random forest algorithm creates @xmath183 tree models from @xmath183 random samples of the original dataset .
the output estimated by random forest is an average of the individual estimation of all tree models . to assess the validity of a regression model @xmath184 such as the ones presented in section [ sec 2 ] or in section [ sec 3.1 ]
, we perform _ cross - validation _ tests .
the principle is the following : the full database @xmath123 we have at our disposal is composed of @xmath25 different landings , so that @xmath185 using the notation of section [ sec 1.4 ] . a regression model @xmath186 ( which can be built for instance using the bayesian approach with gaussian processes we detailed in section [ sec 2 ] ) depends of course of the information contained in @xmath123
. cross - validation tests consist in training a regression model from a smaller database than the one we have access to , and , for all the recordings which were drawn from the training database , to compare the profile of the output quantities reconstructed from the statistical regression model and the measured output profile .
more precisely , let @xmath187 be @xmath188 disjoint sets of recording indices .
for all @xmath189 , we denote by @xmath190 the regression model built from the database @xmath191 using one of the strategies presented in sections [ sec 2 ] or [ sec 3.2 ] , where @xmath192 let us assume for the sake of simplicity that for all @xmath189 , @xmath193 . assessing the validation of a regression model @xmath184 amounts to comparing the error between the measured output profiles @xmath194 and the reconstructed output profile @xmath195 taking @xmath196 as a test input set .
we performed cross - validation tests with these five different models using @xmath197 training data sets such that for all @xmath198 , @xmath199 .
for the gaussian process model , we tried two different strategies for the splitting of the time interval for the fitting of hyperparameters as mentioned in section [ sec 2.3 ] . in section [ sec 3.3.1 ] , we compared the different regression strategies using only the gaussian process model reconstructed with @xmath157 time subinterval . in section
[ sec 3.3.2 ] , we compare the results obtained for @xmath200 with @xmath157 or @xmath161 subintervals .
the curve legen is the following : * the black dots curve refers to the trye measured data ; * the black line curve refers to the gaussian process model ; * the black dashed line curve refers to the linear regression , * the black dotted line curve refers to the generalized additive model ; * the grey line curve refers to the random forest strategies ; * the grey dashed line curve refers to the multivariate adaptative regression splines ( mars ) .
we apply the five regression strategies on this problem and we draw the curve of each estimated deceleration profile on two different landings , the @xmath201and the @xmath202 landing for example ( see figure 1 ) , but only on the first @xmath203 seconds .
this is the time period when the influence of the braking system is the most important during the landing .
cc & [ 2lan ] + we can clearly see on these two examples that the gaussian process approach is the one that fits best the measured deceleration profiles .
this was observed for the majority of the landing profiles considered . in figure 2 , we plot the median absolute percentage error ( mape ) for all models , which is defined by the following formula ( using the notation of section [ sec 3.2 ] ) : + @xmath204 from this criterion , it can be seen that the gaussian process approach we propose significantly improves the predictive quality of the regression models previously used by safety line . [ eror ]
the histogram of the number of errors plotted in figure 3 provides another criterion to compare the different regression models .
more precisely , for all @xmath205 , the histogram plots the cardinal of the set @xmath206 [ hist ] of course , a high - quality model will produce a large number of errors for small values of the error threshold @xmath207 and a small number of errors for large values of @xmath207 .
this is indeed the case for the gaussian process model , and we can see that it also performs significantly better than the other regression models from this point of view .
+ let us now compare the results we obtained with the gaussian process model presented in section [ sec 2 ] , where we used only @xmath157 time interval , or @xmath161 different time subintervals .
+ in this figure , we compare the error associated to these two different strategies , for each subintervals of @xmath164 seconds .
in other words , for each @xmath208 , we compute the mape error @xmath209 for the two strategies , where @xmath210 and @xmath211 . + as expected , using different time subintervals improves the results , especially in the last part of landing .
in this section , we wish to comment the way time correlations are taken into account in our model described in section [ sec 2 ] through the very simplistic choice of the kernel @xmath78 we used here ( [ ker ] ) .
let us first note that it is quite usual in time - dependent regression models that the output quantity to be reconstructed at time @xmath30 , @xmath212 , usually depends only on the value at time @xmath30 of an input quantity @xmath213 ( see @xcite for instance ) . however , this is clearly not the case here since the deceleration of the place at an instant @xmath30 depends a priori on all the set of past input values @xmath214 .
the form of the kernel function @xmath78 enables to take into account in some way the fact that the value of the ouput quantity @xmath212 at time @xmath30 depends on the whole trajectory @xmath214 .
the dependence of the value of @xmath212 on the different values @xmath215 is somehow aggregated through the use of the frobenius norm @xmath216 , which is of course a very naive approach .
one shortcoming of this model in particular is that the value @xmath212 then depends on the _ future values of the input quantities , which is of course unrealistic from a physical point of view .
_ this very simplistic approach is sufficient to yield very satisfactory numerical results though and seems to capture somehow some features of the time correlations between input and output data
. a possible explanation of the significant improvement of the results using our gaussian process based approach compared with the other methods we presented in this proceeding ( which do not take time correlation into account as well ) may be the following .
+ in our approach , a new output signal is reconstructed as a linear combination of other signals that are already present in the database .
however , other methods reconstruct the output signal as a linear combination of input signals , which are not of the same nature as the output signal , and may present different behaviour of time correlation effects .
this particularity of the gaussian process based approach may account for the fact that the time dependencies of the output signal are qualitatively well reproduced in our case , even if we use such a naive way to incorporate time correlation effects in our statistical model . in the rest of the section ,
let us comment on different strategies which could have been used to incorporate time correlation effects using gaussian processes approach in our regression model .
a first strategy could have been to consider the time @xmath217 as a random variable , in the same way as the input quantity @xmath50 .
this would require to modify the mean function @xmath86 and kernel function @xmath78 so that they do not only depend on values of the input quantities @xmath50 but also on time .
more precisely , following ideas of @xcite , the output quantity @xmath218 could be modeled by @xmath219 where @xmath220 would be a random function to be determined by the regression model and @xmath221 a white noise random process for instance . using similar notation as those used above
, the random value @xmath222 could be modeled as a gaussian process characterized by a mean @xmath223 and a covariance kernel @xmath224 so that the law of @xmath225 would be given by @xmath226 such an approach would enable to take into account correlations between values of the output and input quantities at different times in a natural way .
the difficulty we encounter with such an approach is that the standard reconstruction procedure derived from this gaussian process approach requires the inversion of a matrix of size @xmath227 where @xmath228 in our case .
the large size of this matrix makes its inversion very difficult from a practical point of view .
however , in @xcite , the authors proposed to convert such a _ spatio - temporal gaussian process into an infinite - dimensional kalmann filtering .
the interest of such an approach is that the complexity of such an approach is linear ( instead of cubic ) in the number of time steps , thus avoiding the numerical difficulties mentioned above .
we did not test this strategy in our case though .
_ however , a second ( more tractable ) way to improve the choice of this kernel function could rely in the modification of the kernel function @xmath78 in order to take into account time correlatiosn as follows . indeed , using the squared exponential kernel function , in order to reconstruct the value of the deceleration of the plane at a time @xmath30 , all the values of the input quantities at all times have the same importance , which is of course unrealistic .
one would reasonably expect that that only the values of the input quantities at times anterior to @xmath30 would affect the value of the output quantity at a time @xmath30 . to take this into account
, one could think of using at each time @xmath30 a modified kernel function @xmath229 defined as follows : + for all @xmath230 @xmath231 , @xmath232\ ] ] where the standard frobenius norm @xmath82 would be replaced by a modified semi - norm @xmath233 , which would depend on @xmath30 and could be written as follows : for all @xmath234 , @xmath235 with a weight function @xmath236 , satisfying @xmath237 for all @xmath238 ( the standard frobenius norm used in corresponds to @xmath239 for all @xmath13 ) .
it would be interesting to test if these modifications could improve the quality of our regression model .
this strategy would lead to an additional computational cost though : it woud require the storage of @xmath4 matrices of size @xmath240 corresponding to all the matrices @xmath241 even if such a procedure could be more easly implementable than the first approach we described , we did not test this strategy here . it would be interesting though to check if one of these two possible strategies could help improving the numerical results presented in section [ sec 3.3 ] .
the authors thank tony lelivre and nicolas chopin for very helpful discussions .
labex amies is granted for financial support .
this work was done in a summer school `` cemracs 2013 '' .
we would like to thank the organizers of this school .
finally , houssam alrachid would like to thank `` enpc '' and `` cnrs libanais '' for supporting his phd thesis .
l. breiman , _ random forests _ , machine learning 45 ( 2001 ) , 5 - 32 . p. hall , h. muller and f. yao , _ modelling sparse generalized longitudinal observations with latent gaussian processes _ , j. r. statist .
b ( 2008 ) . j.h .
friedman , _ multivariate adaptive regression splines _ , the annals of statistics 19 ( 1991 ) .
t. hastie and r.tibshirani , _ generalized additive model _ , chapman and hall ( 1990 ) .
s. sarkka , s. member , ieee , a. solin , and j. hartikainen , _ spatio - temporal learning via infinite - dimensional bayesian filtering and smoothing _ , ieee signal processing magazine .
rasmussen and c.k.i .
williams , _
gaussian processes for machine learning _ ,
massachusetts institute of technology ( 2006 ) .
b. xu , z. shi , _ universal kriging control of hypersonic aircraft model using predictor model without back - stepping _ , iet control theory and applications ( 2012 ) . | we present numerical results obtained on the cemracs project predictive sms proposed by safety line .
the goal of this work was to elaborate a purely statistical method in order to reconstruct the deceleration profile of a plane during landing under normal operating conditions , from a database containing around @xmath0 recordings .
the aim of safety line is to use this model to detect malfunctions of the braking system of the plane from deviations of the measured deceleration profile of the plane to the one predicted by the model .
this yields to a multivariate nonparametric regression problem , which we chose to tackle using a bayesian approach based on the use of gaussian processes similar to the one presented in @xcite .
we also compare this approach with other statistical methods .
nous prsentons des rsultats numriques obtenus sur le projet cemracs predictive sms propos par safety line .
lobjectif de ce travail tait dlaborer une mthode purement statistique afin de reconstruire le profil de dclration dun avion durant son atterissage , partir dune base de donnes contenant peu prs @xmath0 enregistrements .
le but de safety line est dutiliser ce modle pour dtecter des anomalies du systme de freinage de lavion partir de lcart entre le profil de dclration de lavion mesur et celui prdit par le modle .
ceci mne un problme de rgression multivari non paramtrique que nous avons choisi de traiter via une approche baysienne utilisant des processus gaussiens similaire celle prsente dans @xcite .
nous comparons galement cette approche avec dautres mthodes statistiques classiques . |
damped ly@xmath0 ( dla ) absorbers in quasar spectra are the class of absorbers with the highest column density of neutral hydrogen ( @xmath2 ) .
they are important as it can be shown that they contain the bulk of the neutral hydrogen content of the universe ( e.g. storrie - lombardi & wolfe 2000 ) .
the similarity of the column densities of dlas to those through the disks of spiral galaxies have lead some authors to consider them as the direct progenitors of present - day spirals ( e.g. prochaska & wolfe 1998 ) .
however , dwarf galaxies and low surface brightness galaxies ( lsbgs ) also can have similar columns ( boissier , peroux & pettini 2003 ) , and metallicity arguments support dwarf galaxies ( pettini et al .
theorists have interpreted them in the context of hierarchical galaxy formation models as sub - units which will eventually merge to form @xmath1 galaxies today ( haehnelt , steinmetz & rauch 1998 ; maller et al . 2001 ) or as dwarf galaxies in which star formation is supressed by supernova feedback ( efstathiou 2000 ) .
the reason for the proliferation of theories for the nature of dlas is that reliably identifying the galaxies responsible for the absorption , even at @xmath3 , has proven to be very difficult .
imaging studies of @xmath3 absorbers , e.g. le brun et al .
( 1997 ) ; rao et al .
( 2002 ) indicate that a range of galaxy types might be responsible for producing dlas , but few candidate dlas have measured redshifts , so reliable constraints on the nature of the dla population remain poor .
paradoxically , mgii absorbers , with much lower column densities on average , have been easier to identify .
they seem to be sampling the normal galaxy population , with a mean luminosity of @xmath4 at @xmath5 ( steidel , dickinson & persson 1994 ; steidel et al .
possible reasons for the difficulty of identifying the galaxies causing dla systems include small impact parameters , low stellar luminosities or low surface brightnesses .
so far only six @xmath3 dlas have been spectroscopically - confirmed ( and all of these are at @xmath6 ) .
several are dwarf or low surface brightness galaxies , the best - studied example of which is sbs 1543 + 5921 ( bowen , tripp & jenkins 2001 ) , but the dla absorber of ao 0827 + 24 is a fairly normal , luminous spiral ( steidel et al.2002 ) .
many more suggested identifications have been made on the basis of proximity to the quasar and consistent photometry , but the reliability of these is unclear .
dlas are easy to identify in the spectra of high redshift ( @xmath7 ) quasars as they show up clearly in observed optical spectra , but they are hard to identify at lower redshift . by adopting the strategy of taking uv spectra with the _ hubble space telescope _ ( @xmath8 ) of the highest rest - frame equivalent width ( @xmath9 ) mgii absorbers in the literature , rao & turnshek ( 2001 ) were able to identify a sample of lower redshift ( @xmath10 ) dlas ( see also boisse & bergeron 1998 ) .
high @xmath9 mgii absorbers also typically show high @xmath9 feii absorption lines , and these have been hypothesised to be better tracers of cool gas than mgii ( bergeron & stasiska 1986 ) .
we have therefore used the first bright quasar survey ( fbqs ; white et al .
2000 ; becker et al . 2001 ) to select a similar sample with @xmath9 ( feii 2600)@xmath11 .
we have taken spectra of 26 of the sample with stis on the _ hst _ to look for dla systems .
we have followed up this sample using the keck telescope , taking near - infrared images in sub - arcsecond seeing , medium - resolution spectra of the quasars , and low - resolution spectra of the candidate absorbers .
we are mid - way through spectroscopic follow - up of the fbqs sample , and have identified four galaxies with redshifts the same as that of the feii absorbers . of these four , however , only two are associated with absorbers which are definitely damped .
these two galaxies , near the quasars fbqs 0051 + 0041 ( r.a .
@xmath12 dec .
@xmath13 ( j2000 ) ; absorber redshift , @xmath14 ; quasar redshift , @xmath15 ; apparent magnitude @xmath16 ) and fbqs 1137 + 3907 ( r.a .
@xmath17 dec .
@xmath18 ( j2000 ) ; @xmath19 ; @xmath20 ; @xmath21 ) , form the subject of this paper .
we assume a cosmology with @xmath22 and @xmath23 throughout .
where we have referenced galaxy luminosities , @xmath24 , to the present - day values of the break in the luminosity function , @xmath1 we have used @xmath25 values of norberg et al .
( 2002 ) in @xmath26 , lin et al .
( 1996 ) in @xmath27 and loveday ( 2000 ) in @xmath28 , adjusted to our passbands and cosmology as necessary ( @xmath29 , @xmath30 , @xmath31 ) .
the @xmath32-corrections have been derived using a model 5gyr old sd galaxy from the pegase library ( rocca - volmerange & fioc 1997 ) .
we picked this model as representative of a moderately young , star - forming galaxy , a decision motivated by the discovery of [ oii ] emission from the candidate dla galaxies .
fbqs 0051 + 0041 was imaged with the echellette spectrograph and imager ( esi ) on keck - ii and fbqs 1137 + 3907 with the near - infrared camera ( nirc ) on keck - i .
details of the observations are given in in table 1 .
a sequence of short ( 3 - 5min ) dithered exposures of each object were taken .
after dark subtraction and flat - fielding with dome flats the data were analysed using dimsum and xdimsum in iraf , respectively .
the dimsum package performs two passes on the data . in the first pass ,
a background image is made from a median of the dithered frames and the data are combined with integer pixel shifts to make masks which are used to mask out objects when constructing an improved sky background image for the second pass . in the second pass , the improved background is subtracted from the images before they are shifted and coadded , and a magnification factor may be applied to the data .
the pixel scale of the esi image was sufficiently small compared to the seeing that no magnification was used in the second pass for fbqs 0051 + 0041 .
the operation of the dimsum package is explained in more detail in stanford , eisenhardt & dickinson ( 1995 ) .
the xdimsum package is a modification of dimsum by l. davis at noao . amongst other improvements it runs significantly faster and allows non - integer magnification factors in the second pass .
we used a factor of 2.3 for the fbqs 1137 + 3907 image .
photometry for fbqs 0051 + 0041 was obtained using observations of the landolt ( 1992 ) standard 107 - 215 , and that for fbqs 1137 + 3907 from observations of the standard star p266-c from persson et al .
both quasar fields were imaged in photometric conditions .
the @xmath33 limiting magnitudes in 2-arcsec diameter apertures are @xmath34 for the fbqs 0051 + 0041 field and @xmath35 for the fbqs 1137 + 3907 field .
we performed approximate subtractions of the quasars using point spread functions derived from either from a star in the same image ( fbqs 0051 + 0041 ) or from other quasars observed on the same night ( fbqs 1137 + 3907 ) .
the psf was fitted with a least - squares fit using code developed for fitting quasar host galaxies ( lacy et al . 2002 ) .
the particular algorithm employed here subtracted the psf such that the mean value within a radius @xmath36 from the center of the quasar was set to zero .
the values of @xmath36 used were 06 for fbqs 0051 + 0041 and 03 for fbqs 1137 + 3907 .
neither psf subtraction was very satisfactory , as we made no serious attempt to obtain good , unsaturated psfs for the quasars .
they have successfully removed the wings of the psfs , however .
the resulting images are shown in figures 1 and 2 .
the quasars were observed with the stis with the mama detectors and the g230l grating as part of program 9051 .
the 02 slit was used , giving a spectral resolving power of 330 at the redshifted wavelength of ly@xmath0 , corresponding to a velocity resolution of @xmath37 .
each quasar was observed for two 1100s exposures , the second exposure being dithered @xmath38 along the slit .
the uv spectrum of fbqs 1137 + 3907 shows it to be a high - ionization broad - absorption line ( bal ) quasar , with a bal trough blueshifted by @xmath39 .
fortunately , the lyman-@xmath0 absorption line is clear of the bal troughs , falling just to the red of the ovi trough in the quasar spectrum .
the spectra are shown in figures 3 and 4 . a detailed discussion of the hst spectra and measurements of the damped absorbers are provided in a companion paper ( storrie - lombardi , et .
2003 , in preparation ) .
the column densities of the damped absorbers were initially estimated from measurements of the equivalent widths of the absorption lines with the redshift determined from the low ionization metal lines from our ground - based spectra .
profiles were then fit iteratively to determine the best fit column density and redshift for hi absorption feature .
the column densities should be accurate to @xmath40 and the redshifts to @xmath41 . though higher resolution spectra covering multiple lines in the lyman series would be ideal for this purpose , the substantial hst time that would be involved in obtaining them makes such observations impractical for large samples of objects . at higher redshifts ,
we have shown via observation and simulation that moderate resolution spectra provide an excellent estimate for the measured hi column densities ( e.g. storrie - lombardi et .
al . 1996 ; storrie - lombardi & wolfe 2000 ) , which gives us high confidence that the features in these spectra are damped and are not blends of lower column density lines .
the candidate absorbers were observed with the low resolution imaging spectrometer ( lris ) on keck - i using the 300 line mm@xmath42 , 5000@xmath43blaze grating in the red arm and a 1-arcsecond slit ( see table 2 ) .
this combination results in a spectral resolving power of @xmath44 , corresponding to a velocity resolution of @xmath45 .
the wavelength range covered was from @xmath46 .
in addition , observations of the quasars were made with esi on keck - ii to obtain medium - resolution spectra of the absorption lines .
the feii 2586 and 2600@xmath43lines fell in a noisy region near an order overlap in the esi spectra , but the mgii 2796 and 2803@xmath43lines were detected at high signal - to - noise in both quasars .
fbqs 0051 + 0041 was observed with esi on 2000 december 31 with a @xmath47 slit , and fbqs 1137 + 3907 on 2001 march 20 with a @xmath48 slit , yielding resolving powers of 5400 and 4100 respectively .
the esi spectra of the quasars in the region around mgii are shown in figures 5 and 6 , and the lris spectra of the galaxies at the absorber redshifts in figures 7 and 8 .
the _ hst _ spectrum shows that this absorber is a borderline dla , with a column density just above the division between lyman - limit and dla systems ( table 3 ) .
similarly , the @xmath9 of the feii absorption only just satisfies our selection criteria .
the candidate absorber galaxy , galaxy ` a ' in figure 1 , seems , at first sight , to be a fairly normal , star forming galaxy , offset by 24kpc from the quasar ( figures 1 and 7 ) .
the spectrum , however , shows a small mis - match between the [ oii]3727 redshift of 0.740 and the absorption line redshift from the ca ii h+k lines ( 0.738 ) , though the signal - to - noise ratio of the absorption lines is low .
two distinct velocity components are also seen in the low - ionization mgii , mgi and feii quasar absorption lines in our esi spectrum , one at @xmath49 and one at @xmath50 ( i.e. a @xmath51 velocity difference ) .
whether these two components correspond to two galaxies , the candidate and a dwarf companion nearer the quasar , or reflect internal gas motions in the candidate absorber is unclear .
this is a much higher column density absorber than that of fbqs 0051 + 0041 .
the _ hst _
spectrum shows several strong uv metal absorption lines in addition to the damped ly@xmath0 line ( figure 4 ) .
the lris slit was placed to cover both candidate absorbing galaxies , @xmath52 and @xmath53 , and line emission is seen from both of them ( barely resolved from each other ) at the wavelength corresponding to [ oii ] 3727 at @xmath54 ( figure 9 ) .
there is also a marginal detection of h@xmath55 at the same redshift ( figure 8) .
although we see no other features from @xmath52 or @xmath53 , it seems unlikely that the strong emission line can be anything other than [ oii ] at the absorber redshift , given its high equivalent width and the lack of lines nearby in the spectrum .
we also took a spectrum with the slit aligned with object ` b ' .
this showed that ` b ' had blue optical continuum emission , but no strong features from which to obtain a redshift . as it is further from the quasar than ` c ' and has no features at wavelengths corresponding to the absorption redshift
we assume that it is unrelated to the dla system . lccccc quasar field & telescope / instrument & filter & observation & total integration & seeing + & & & date ( ut ) & time ( minutes ) & ( arcsec ) + fbqs 0051 + 0041 & keck - ii /
esi & @xmath56 & 2001 june 22 & 20&0.8 + fbqs 1137 + 3907 & keck - i / nirc & @xmath57 & 2002 may 25 & 27&0.4 + lcccccccccc quasar field & @xmath58&mag . & impact & impact & abs.&@xmath59&@xmath60&ut date & slit pa & exposure + & & & parameter & parameter & mag .
& & & ( deg ) & & time ( s ) + & & & ( arcsec ) & ( kpc ) & & & & & & + fbqs 0051 + 0041a & 0.740 & @xmath61 & 3.3 & 24 & @xmath62&-0.11&0.8&2001 sept 25&26 & @xmath63 + fbqs 0051 + 0041b & ? & @xmath64 & 4.6 & ? & - & - & & & - & - + fbqs 0051 + 0041c & 0.070 & @xmath65 & 7.2 & 9.6 & @xmath66&-0.15&0.001&2001 sept 25&26 & @xmath63 + fbqs 1137 + 3907a & 0.7185 & @xmath67&2.5 & 18 & @xmath68&-0.08&0.3&2002 dec 31 & 126&@xmath69 + fbqs 1137 + 3907b & ? & @xmath70&1.8 & ? & - & -&&2002 dec 31 & 228&@xmath69 + fbqs 1137 + 3907c & 0.7185 & @xmath71&1.5 & 11 & @xmath72&-0.08&0.2&2002 dec 31 & 126&@xmath69 + notes : @xmath59 is the amount that was added to the absolute magnitudes to account for the difference in the flux density of the galaxy between the observed band and the band the absolute magnitude was calculated in .
this correction depends only on the assumed galaxy sed and reduces to the usual @xmath32-correction when the two bands are the same .
llccccc quasar & @xmath73 & mgii @xmath9 & feii @xmath9 & lg(@xmath74 + & & ( 2796 ) & ( 2600 ) & @xmath75 ) + fbqs 0051 + 0041 & 0.740 & 2.4 & 1.0 & 20.4 @xmath40 + fbqs 1137 + 3907 & 0.719 & 3.0 & 2.5 & 21.1 @xmath40 + llccccccll absorber & @xmath58&@xmath76&impact & observed&absolute & @xmath77&@xmath60&type&refs .
+ candidate & & @xmath75)¶meter&mag . & mag . & & & & + & & & ( kpc ) & & & & & & + sbs 1543 + 5921 & 0.009&20.4 & 0.4 & @xmath78 & @xmath79 & - & 0.02 & lsbg & 9,10 + oi 363 ( b0738 + 313 ) g1 & 0.221&20.9 & 20 & @xmath80 & @xmath81&0.2 & 0.2 & de & 3,4 + fbqs 1137 + 3907c & 0.720&21.1 & 11 & @xmath71 & @xmath72&-0.1 & 0.2 & spiral ? & 1 + pks 1127 - 145 & 0.312&21.7 & @xmath827 & @xmath83 & @xmath84&-0.2 & 0.4 & lsbg & 7,8 + pks 1243 - 072a & 0.437&@xmath85&12&@xmath86 & @xmath87&0.2 & 0.4 & starburst / seyfert 2&8 + fbqs
0051 + 0041a & 0.740&20.4 & 24 & @xmath61 & @xmath62&-0.11 & 0.8&spiral ? & 1 + b2 0827 + 243 g1 & 0.526&20.3 & 36 & @xmath88 & @xmath89&0.2 & 1.6 & spiral & 5,6 + ao 0235 + 164 a@xmath90 & 0.524&21.7@xmath91&13 & @xmath92&@xmath93&0.3&@xmath94&seyfert 1&2 + notes : @xmath90 object a1 in ref . 2 is closer to the quasar but has no spectrum , if it is at the absorber redshift it has @xmath95 , a has a bright nuclear component , and the luminosity of the underlying galaxy is uncertain ; we have quoted upper limits based on the sum of the host and agn fluxes . @xmath91 column density from 21 cm line absorption , dependent on the assumed spin temperature
. @xmath96 sum of components 1 - 4 of rao et al .
see table 3 for the definition of @xmath77 .
references : ( 1 ) this paper ; ( 2 ) burbidge et al . 1996 ;
( 3 ) turnshek et al.2001 ; ( 4 ) cohen 2001 ; ( 5 ) steidel et al .
2002 ; ( 6 ) rao et al .
2003 ; ( 7 ) lane et al .
1998 ; ( 8) kanekar et al . 2002 ; ( 9 ) reimers & hagen 1998 ; ( 10 ) bowen et al .
2001 the esi spectrum shows a complicated velocity profile , with two to four distinct low - ionization absorption components seen , depending on which line is examined .
the two deepest features , visible in all the lines , are at @xmath97 and @xmath98 ( i.e. a @xmath51 velocity difference ) .
thus the [ oii ] emission is most likely to be associated with the deeper , less redshifted absorption component .
the dla absorber candidate close to fbqs 0051 + 0041 has a luminosity of approximately @xmath1 , whereas that associated with fbqs 1137 + 3907 is about @xmath99 .
the galaxy types are unclear .
both are apparently forming stars , possessing both blue continua and [ oii ] emission lines , and the complicated velocity structures of the mgii absorption lines seen in the esi spectra , might be indicative of starburst - induced winds ( e.g. ellison et al .
further evidence for this is the offset between the emission and absorption redshifts in fbqs 0051 + 0041a , if real .
thus there is some evidence that both may be starburst galaxies .
however , the continuum slope of fbqs 0051 + 0041a is a little too red unless the starburst is reddened by dust , and fbqs 1137 + 3907c has a high axial ratio , suggesting a spiral rather than an irregular galaxy ( though the shape of ` c ' may also be due to it being composed of a chain of knots ) .
higher spatial resolution multicolor imaging will be required to determine the nature of these galaxies .
table 4 summarizes the properties of the eight known candidate dla galaxies at @xmath3 which spectroscopy has shown to lie at the absorber redshift .
selection effects will mean that the absorber population for which spectroscopic identification is possible will be biased .
observational requirements for successful spectroscopy , namely an impact parameter @xmath100 arcsec and an apparent magnitude brighter than @xmath101 , will naturally favor massive galaxies . for our fbqs sample discussed in the introduction , and
from which fbqs 0051 + 0041 and fbqs 1137 + 3907 were drawn , we have so far only successfully identified candidates and obtained spectra with redshifts consistent with those of the absorber system for four out of thirteen of our high @xmath9 absorption systems .
in addition , there is another subtle selection effect from the way most ( though not all ) of the dlas in table 4 have been selected , namely on the basis of high @xmath9 metal absorption lines .
this will tend to bias the sample towards higher metallicity , more massive , galaxies .
working against these two effects is the possibility that many of the more massive dla galaxies are dusty enough to significantly extinguish the light from the background quasar , and cause the quasar to drop out of optically - selected samples .
this may tend to produce a bias towards relatively dust - free , less massive hosts , and may explain why the mgii absorbers , with their generally higher impact parameters and lower hi column densities , are observed to have luminosities around @xmath25 . however , the results of the survey of radio - loud quasars of ellison et al .
( 2001 ) suggest that this is not an important effect .
any such effect will also be much reduced for fbqs quasars as they are selected on the palomar @xmath102-plates with a relatively red ( @xmath103 ) color selection criterion .
thus , despite most of the selection effects favouring successful spectroscopic identification of the most luminous absorbing dla galaxies , it is clear from table 4 that many are sub-@xmath1 , and this suggests that the majority of the dla population is probably sub-@xmath1 .
both our dla candidates appear to be actively forming stars , consistent with them being relatively gas - rich galaxies .
a large fraction of the known spectroscopically - confirmed dla systems at low redshift listed in table 4 appear to be merging or interacting in some way , for example , fbqs 1137 + 3907c , pks 1127 - 145 and pks 1243 - 072 a ( though obviously the statistics are poor at present , and few apparent pairs have been spectroscopically confirmed ) .
this is intriguing in the light of theories in which dla galaxies are dwarf galaxies which merge to become massive galaxies today ( haehnelt et al .
1998 ; maller et al . 2001 ) .
though these theories are more applicable to the high redshift systems , the lower redshift dlas may represent the tail end of this activity .
we thank an anonymous referee for helpful comments .
this work was partly carried out at the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration ( nasa ) , and partly under the auspices of the u.s .
department of energy , national nuclear security administration by the university of california , llnl , under contract no .
w-7405-eng-48 , with additional support from nasa grant hst - go-09051.01-a and nsf grant ast-00 - 98355 ( university of california , davis ) .
some of the data presented herein were obtained at the w.m .
keck observatory , which is operated as a scientific partnership among the california institute of technology , the university of california and nasa .
the observatory was made possible by the generous financial support of the w.m .
keck foundation .
the authors wish to acknowledge the significant cultural role and reverence that the summit of mauna kea has always had within the indigenous hawaiian community .
this research has also made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with nasa .
becker , r.h .
2001 , apjs , 135 , 227 bergeron , j. & stasiska , g. 1986 , a&a , 169 , 1 boisse , p. , le brun , v. , bergeron , j. & deharveng , j .-
1998 , a&a , 333 , 841 boissier , s. , peroux , c. & pettini , m. 2003 , mnras , 338 , 131 bowen , d.v . ,
tripp , t.m .
& jenkins , e.b .
2001 , apj , 121 , 1456 burbidge , e.m . , beaver , e.a .
, cohen r.d . ,
junkkarinen , v.t . & lyons , r.w .
1996 , aj , 112 , 2533 cohen , j.g .
2001 , aj , 121 , 1275 efstathiou , g. 2000 , mnras , 317 , 697 ellison , s.l .
, yan , l. , hook , i.m . , pettini , m. , wall , j.v . & shaver , p. 2001
, a&a , 379 , 393 ellison , s.l .
, malln - ornelas , g. & sawicki , m. 2003 , apj , in press ( astro - ph/0302147 ) haehnelt , m.g . ,
steinmetz , m. & rauch , m. 1998 , apj , 495 , 647 kanekar , n. , athreya , r.m .
& chengalur , j.n . , 2002 ,
a&a , 382 , 838 kulkarni , v. , et al .
2001 , apj , 551 , 37 landolt , a.u .
1992 , aj , 104 , 340 lane , w. , smette , a. , briggs , f. , rao , s. , turnshek , d. & meylan , g. , 1998 , aj , 116 , 26 le brun , v. , bergeron , j. , boiss , p. & deharveng , j.m .
1997 , a&a , 321 , 733 lin , h. , kirshner , r.p . ,
shectman , s.a , landy , s.d .
, oemler , a. , tucker , d.l . & schechter , p.l .
1996 , apj , 464 , 60 loveday , j. 2000 , mnras , 312 , 557 maller a.h . ,
prochaska , j.x . ,
somerville , r.s .
& primak , j.r .
2001 , mnras , 326 , 1475 norberg , p. , et al .
2002 , mnras , 336 , 907 persson , s.e . ,
murphey , d.c . ,
krezeminski , w. , roth , m. & rieke , m.j .
1998 , aj , 116 , 2475 pettini , m. , ellison , s.l .
, steidel , c.c . & bowen , d.v . 1999 ,
apj , 510 , 576 prochaska , j.x . & wolfe , a.m. 1998 , apj , 507 , 113 rao , s. & turnshek , d.a . 2000 , apjs , 130 , 1 rao , s. , nestor , d.b . ,
turnshek , d.a . , lane , w.m . ,
monier , e.m . & bergeron , j. apj , submitted ( astro - ph/0211297 ) ridgway , s.e .
& stockton , a.n .
1997 , aj , 114 , 511 rocca - volmerange , b. & fioc , m. 1997 , a&a , 326 , 950 reimers , d. & hagen , h .- j .
1998 , a&a , 329 , l25 stanford , s.a . , eisenhardt , p.r.m . & dickinson , m. 1995 , apj , 450 , 512 steidel , c. , dickinson , m. & persson , e. 1994 , apj , 437 , l75 steidel , c. , kollmeier , j.a . , shapley , a. , churchill , c.w . ,
dickinson , m. & pettini , m. 2002 , apj , 570 , 526 songaila a. , cowie l.l . , hu e.m .
, gardner j.p . , 1994 , apjs , 94 , 461 storrie - lombardi , l.j , mcmahon , r.g . ,
irwin , m.j . & hazard , c. 1996 , apj , 468 , 121 storrie - lombardi , l.j . & wolfe , a.m. 2000 , apj , 543 , 552 turnshek d.a . , rao , s. , nestor , d. , lane , w. , monier , e. , bergeron , j. & smette , a. 2001 , apj , 553 , 288 warren , s.j . ,
moller , p. , fall , s.m .
& jakobsen , p. 2001 , mnras , 326 , 759 white , r.l .
2000 , apjs , 126 , 133 zwaan m. , briggs , f.h . &
verheijen , m. 2002 , in extragalactic gas at low redshift , eds j.s . mulchaey & j. stocke , asp vol .
254 , p. 169 | we have identified galaxies near two quasars which are at the redshift of damped lyman-@xmath0 ( dla ) systems in the uv spectra of the quasars .
both galaxies are actively forming stars .
one galaxy has a luminosity close to the break in the local galaxy luminosity function , @xmath1 , the other is significantly fainter than @xmath1 and appears to be interacting with a nearby companion . despite the strong selection effects favoring spectroscopic identification of the most luminous dla galaxies , many of the spectroscopically - identified dla galaxies in the literature are sub-@xmath1 , suggesting that the majority of the dla population is probably sub-@xmath1 , in contrast to mgii absorbers at similar redshifts whose mean luminosity is close to @xmath1 . |
recent studies of the lattice qcd in the maximally abelian ( ma ) gauge suggest the remarkable properties of the qcd vacuum , such as abelian dominance@xcite and monopole condensation@xcite , which provide the dual superconductor picture of the qcd vacuum as is described by the dual ginzburg - landau ( dgl ) theory@xcite . in the ma gauge
, qcd is reduced into an abelian gauge theory including color - magnetic monopoles . according to the lattice
qcd results , the nonperturbative quantities as the string tension and the chiral condensate are almost reproduced only by the diagonal gluon part , while the off - diagonal gluon does not contribute to such the long - range physics , namely , abelian dominance .
furthermore , the world - line of the color - magnetic monopole in the confinement phase appears as the global network , which indicates monopole condensation .
then , the dgl theory can be constructed by extracting the diagonal gluon as the relevant degrees of freedom and taking into account monopole condensation .
based on the dgl theory , the quark confinement is explained by the flux - tube formation through the dual meissner effect , and chiral symmetry breaking is described as the function of monopole condensate@xcite . in this paper , we focus such the dual superconductor picture of the qcd vacuum in the ma gauge , and confirm the connection between nonperturbative qcd and the dgl theory . then , we would like to apply the dgl theory to hadron physics , especially , to the analysis of the scalar glueball properties .
abelian dominance and monopole condensation in the ma gauge are the keywords to connect the qcd with the dgl theory , and the recent lattice qcd simulations show the former on the string tension and the chiral condensate , and the latter as the large clustering of the monopole world - line . in such situation , we still have interest in this subject , since the physical essence of abelian dominance is not understood yet . furthermore , we must not jump to a conclusion that the global network of the monopole world - line is really the evidence of monopole condensation , which also should be evaluated quantitatively . to answer these questions , we first study the gluon propagator in the ma gauge and evaluate the mass of the off - diagonal gluon field using the su(2 ) lattice qcd simulation@xcite .
this study is based on the following idea .
if the off - diagonal gluon has a mass such as the massive vector boson , its propagator @xmath0 would be described by the yukawa - type function @xmath1 , and if we find the linear behavior for @xmath2 , the mass @xmath3 can be extracted from its slope . as a result , we find that the off - diagonal gluon has the large mass @xmath4 gev as shown in fig .
1 . that is to say , the interaction range of the off - diagonal gluon is limited within the short distance corresponding to its inverse mass @xmath5 fm .
thus , the off - diagonal gluon does not contribute to the long - range physics , which predicts general infrared abelian dominance in the ma gauge .
as for monopole condensation , we study the inter - monopole potential and evaluate the dual gluon mass using the su(2 ) lattice qcd simulation@xcite .
the dual gluon field @xmath6 is introduced to satisfy @xmath7 and @xmath8 . here , @xmath9 is the color - magnetic monopole current .
the idea used here is quite similar to the evaluation of the off - diagonal gluon mass .
if monopole condensation is occurred , the dual gluon becomes massive due to the dual higgs mechanism .
then , its mass @xmath10 can be extracted by fitting the yukawa potential @xmath11 , since the dual gluon behaves as the massive vector boson . from this analysis
, we find that the dual gluon acquires the mass @xmath12 gev as shown in fig .
2 , which is just the quantitative evidence of monopole condensation . as an interesting application of abelian dominance for the inter - quark potential
, we can calculate the quark single - particle potential @xmath13 for the low - lying hadron ( @xmath14=300 mev ) . here ,
@xmath13 is defined by the superposition of the inter - quark potential @xmath15 ( @xmath16 gev / fm , @xmath17 ) with the weight of the color charge distribution @xmath18 as @xmath19 . solving the self - consistent equations between the quark wave function and the quark potential
, we obtain the color charge distribution and the quark single - particle potential as shown in figs . 3 and 4 . the color charge distribution is spread over a intermediate region @xmath20 fm .
the quark single - particle potential is found to be flat at the short distance , which can be connected with the bag model .
fig . 1 .
the gluon propagator as a function of 4-dimensional distance @xmath21 in the ma gauge .
= 7.5 cm fig . 2 . the inter - monopole potential as a function of 4-dimensional distance @xmath21 in the ma gauge . fig .
the color charge distribution for the low - lying hadron .
= 6.2 cm fig .
the quark single particle potential in the low - lying hadrons .
the dgl theory can be constructed by taking into account abelian dominance and monopole condensation in the ma gauge in qcd .
the dgl lagrangian is given as @xmath22,\ ] ] where @xmath23 and @xmath24 denote the dual gluon field with two components @xmath25 and the complex scalar monopole field , respectively .
the quark field is included in the current @xmath26 . here
, @xmath27 is the root vector of su(3 ) algebra , and @xmath28 denotes an arbitrary constant 4-vector , which corresponds to the direction of the dirac string .
the gauge coupling @xmath29 and the dual gauge coupling @xmath30 hold the relation @xmath31 .
monopole condensation is characterized by @xmath32=@xmath33 , and the dual gluon field acquires the mass @xmath34 0.5 gev through the dual higgs mechanism .
then , the dgl theory describes the qcd vacuum as the dual superconductor .
the quark confinement is explained by the dual meissner effect , which forces the color - electric field between the quarks to form the flux - tube configuration , and leads the linear inter - quark potential .
this flux - tube also provides intuitive picture of hadrons .
if we apply this flux - tube picture to the glueball , it would be identified with the flux - tube ring , since the glueball is considered to have no valence quarks , and the lowest state is the scalar glueball .
from the flux - tube ring solution in the dgl theory , we find the mass and the size of the scalar glueball as 1.6 gev and 0.5 fm , respectively@xcite .
it is interesting to note that this mass spectrum is consistent with the recent lattice qcd results @xmath35 = 1.50 - 1.75 gev@xcite . here
, we find another aspect of the scalar glueball in the dgl theory , which is closely related to the dual higgs mechanism .
taking monopole condensation into account , the monopole field can be defined as @xmath36 , where @xmath37 and @xmath38 are real variables denoting the magnitude of the vacuum fluctuation and the phase , respectively . here , @xmath39=1 , 2 , 3 labels the color - magnetic charge of the monopole field , dual - red , dual - blue and dual - green . since the origin of the monopole field is the off - diagonal gluon field in the ma gauge in qcd , this field @xmath37 would present the scalar gluonic excitation corresponding to the dual higgs particle . in particular
, the weyl symmetric monopole field defined by @xmath40 is the color - singlet field@xcite so that it can be regarded as the scalar glueball with the mass @xmath41 gev .
although the relation between the flux - tube ring is not clear , it can be considered as another feature of the scalar glueball . here , we concentrate on the calculation of the @xmath42 vertex function , which plays an important role to understand how the scalar glueball interacts with the quarks . the lowest diagram is shown in fig .
the scalar glueball interacts with the dual gluon at first , and then , the dual gluon interacts with the quarks . we show the typical behavior of the vertex function in the scalar channel , as a function of the coupled quark momentum in fig . 6 . here , we have set @xmath43=0 for simplicity .
we find that the heavy quark ( @xmath44 gev ) interacts with the scalar glueball about four times stronger than the light quarks ( @xmath45 gev ) .
it seems to indicate the flavor dependence of the interaction of the scalar glueball .
it is interesting to study how this interaction property reflects on the scalar glueball decay into the two pseudo - scalar mesons and the glueball - quarkonium mixing states , which are now investigating .
k. amemiya and h. suganuma , ; in _ proceedings of international symposium on innovative computational methods in nuclear many - body problems ( innocom97 ) _ , osaka , nov .
1997 , ( world scientific , 1998 ) p.284 . | we study abelian dominance and monopole condensation for the quark confinement physics using the lattice qcd simulations in the ma gauge .
these phenomena are closely related to the dual superconductor picture of the qcd vacuum , and enable us to construct the dual ginzburg - landau ( dgl ) theory as an useful effective theory of nonperturbative qcd .
we then apply the dgl theory to the studies of the low - lying hadron structure and the scalar glueball properties .
# 1#2#3#4#1 * # 2 * , # 3 ( # 4 ) |
microtubules are filamentous protein aggregates that appear in all eukaryotic cells .
they have an inherent polarity that results in different dynamics at their two ends .
the so - called plus end is highly dynamic , alternating between prolonged periods of polymerization ( growth ) and depolymerization ( shrinkage ) @xcite . on the other end of the filament ,
the minus end often remains connected to the locus of nucleation @xcite or is found to exhibit relatively steady depolymerization @xcite .
the combination of slow depolymerization at the minus end and prolonged growth at the plus end leads to a phenomenon known as treadmilling , whereby the individual tubulin dimers appear to move from the plus to the minus end @xcite . as the stiffest of the cytoskeletal filaments , microtubules
are widely used in intracellular transport and for structural support .
therefore , the dynamic properties of the microtubules are tightly regulated by the cell , often through the use of microtubule - associated proteins ( maps ) @xcite . in this work
we investigate one particular type of map that causes the severing of microtubules .
it was noted by vale @xcite that otherwise stable microtubules could be severed in mitotic extracts of xenopus eggs .
this activity was traced back to a protein that is able to use atp hydrolysis to sever microtubules .
the protein was identified only later and given the name katanin after the katana , the japanese samurai sword @xcite .
katanin is a heterodimer , consisting of the p60 and p80 subunits : the p80 subunit is thought to be responsible for the targeting of the protein , whereas p60 is involved in the actual severing as part of a hexameric ring @xcite .
the hexameric form of katanin appears to remove individual dimers from the microtubule lattice , thereby compromising the structural integrity of the microtubule .
it is not currently clear whether katanin acts uniformly along the microtubule , or whether it is attracted by pre - existing lattice defects @xcite .
katanin homologs have since been discovered across the animal and plant kingdoms @xcite . another severing protein by the name of spastin has also been identified . like katanin
, it also assembles in a hexameric ring , suggesting a severing mechanism similar to that of katanin @xcite .
both katanin and spastin are capable of severing microtubules at seemingly random locations , and are used by the cell for the regulation of the cytoskeleton @xcite , for example in the mitotic and meiotic spindles @xcite or the formation of the transverse cortical array in plant cells @xcite .
generally , the activity of severing proteins leads to a decrease in the average microtubule length , but an increase in their number @xcite .
surprisingly , this increase in number can sometimes more than offset the loss of microtubule length due to the average length decrease @xcite .
in this work , we investigate theoretically how the occurrence of microtubule severing at random positions affects the length distribution of microtubules .
previous studies have assessed the effect of severing on actin filaments @xcite . however , whereas actin has a single growth mode that is described well by constant polymerization and depolymerization rates , microtubules show richer dynamics because the plus end switches between growing ( polymerizing ) and shrinking ( depolymerizing ) states .
we base our model on the basic dynamic instability model that was introduced by dogterom and leibler @xcite . in this model , microtubules exist in either the growing or the shrinking state , and switch between these states with a ` catastrophe ' rate @xmath0 ( growth @xmath1 shrinkage ) and a ` rescue ' rate @xmath2 ( shrinkage @xmath1 growth ) . in the growing state , the microtubule extends with an effective speed @xmath3 , and in the shrinking state it retreats with an effective speed @xmath4 .
new microtubules are nucleated with a steady nucleation rate @xmath5 .
this set of constraints gives rise to a steady state distribution of microtubule lengths , provided that @xmath6 .
we model microtubule severing by a constant severing rate per unit length .
this is a valid approximation if we assume that microtubule severing process occurs on a time scale that is much shorter than the time in which a microtubule grows significantly . by taking a constant severing rate
, we also implicitly assume that severing is limited by the availability of microtubules , i.e. the severing protein is available in abundance .
however , because we focus on steady - state results , where the total amount of microtubules is constant , the possible invalidity of this assumption will not qualitatively affect the results .
following the approach in @xcite , we construct a set of evolution equations for the length distributions of growing and shrinking microtubules . denoting the growing and shrinking microtubule length distributions by @xmath7 and @xmath8 , respectively
, the equations can be written as [ eq : seveqnall ] @xmath9 where the derivatives with respect to @xmath10 reflect the translation of the distributions due to growth and shrinkage of microtubules . the severing contribution @xmath11
will be constructed below .
these equations are supplemented by the boundary condition @xmath12 specifying the nucleation of new microtubules with rate @xmath5 , and by the physically motivated constraint that both distributions tend to zero for large lengths ( there are no infinitely long microtubules ) .
[ c]llmodel parameters & + @xmath13 & growth speed + @xmath14 & shrinkage speed + @xmath15 & catastrophe rate + @xmath16 & rescue rate + @xmath17 & nucleation rate + @xmath18 & severing rate + dimensionless parameters & + @xmath19 & speed ratio + @xmath20 & rescue rate + @xmath21 & severing rate + in the following we derive the form of the severing terms in the set of equations .
the inclusion of severing events leads to two types of contributions to the evolution of the length distributions : the _ disappearance _ flux @xmath22 of microtubules of a certain length @xmath10 and the _ appearance _ fluxes @xmath23 and @xmath24 of two new microtubules with a total length @xmath25 .
in addition , the newly created microtubules can be created from microtubules that were initially growing ( + ) or shrinking ( - ) .
symbolically , we write @xmath26 where @xmath27 stands for the removal of microtubules in state @xmath28 , and @xmath29 represents the appearance of new microtubules of length @xmath10 in state @xmath30 as a result of the severing of microtubules in state @xmath31 . the contributions will be discussed individually below .
the process of severing is controlled by the severing rate @xmath32 that is given as a rate per unit of length .
thus we find that the fluxes of disappearing microtubules @xmath33 ( growing ) and @xmath34 ( shrinking ) are given by @xmath35 in order to specify the influx terms @xmath36 it is necessary to specify the process of severing in more detail .
we will assume that the action of the severing protein is local , thus having no effect on the remote plus and minus ends of the microtubule it severs .
this implies that the plus end fragment of a severed microtubule remains in the same state .
however , we must make an explicit assumption regarding the state of the newly created plus end . in line with biological observations @xcite
, we assume that this plus end always starts out in the shrinking state . a severing event thus shortens an existing microtubule without affecting its growth state and creates an additional microtubule that is in the shrinking state . if necessary
, the model can easily be extended to handle ( a fraction of ) severing - created plus ends that start out in the growing state .
let us define the uniform probability density of selecting a severing location @xmath10 on a microtubule of length @xmath37 as @xmath38 , with the value @xmath39 for @xmath40 $ ] , and @xmath41 otherwise .
we then derive the influx @xmath42 of growing microtubules of length @xmath10 that results from the severing of growing microtubules .
@xmath43 where @xmath44 is the heaviside step function . in a similar way , and taking account the fact that the minus end fragment is always in a shrinking state , we derive @xmath45 \,\mathrm{d}l'}\nonumber\\ & = 2 r_s { \int_{l}^{\infty } m^-(l ' ) \,\mathrm{d}l'}.\end{aligned}\ ] ] and , because the severing of shrinking microtubules can not produce growing microtubules , we have @xmath46 .
note that we do not need to keep track of the correlations between the lengths of the _ individual _ microtubules that are created from a single cutting event , because we are looking only at ensemble - averaged length distributions .
the contributions derived above are similar in form to those introduced by edelstein - keshet and ermentrout @xcite as the continuous limit of a discrete monomer addition and severing model for actin filaments .
however , we use distinct asymmetric terms for growing and shrinking microtubules .
the model as described does not explicitly take into account the often observed depolymerization of microtubules at their minus ends that leads to treadmilling .
however , this case is easily addressed through a renormalization of the growth and shrinkage speeds . denoting the shrinkage speed at the minus end by @xmath47 ,
this is achieved by the substitutions @xmath48 and @xmath49 , leaving the results qualitatively unchanged by the incorporation of treadmilling .
inserting the severing terms into equations we obtain [ eq : seveqnfull ] @xmath50 \,\mathrm{d}l ' } \mathrm{,}\end{aligned}\ ] ] we note that in the special case when @xmath51 , the equation for the length distribution of the growing microtubules is equivalent to the model of microtubule gtp - cap dynamics by flyvbjerg _
@xcite with a vanishing diffusion constant .
the number of parameters in these equations can be reduced by scaling the parameters and variables using natural units for length , time and microtubule number . as a unit of time , we take the mean time to catastrophe ( @xmath52 ) ; the unit of length is the microtubule growth in that time ( @xmath53 ) and unit of microtubule number is the number of microtubules nucleated in this time ( @xmath54 ) . in terms of the dimensionless length unit @xmath55
, we define @xmath56 note that the definition of @xmath57 is inverted with respect to its natural conversion into dimensionless parameters , but this choice simplifies the notation in what follows . in the steady state ,
the equations in dimensionless form become [ eq : sever - ssall ] @xmath58 \,\mathrm{d}x ' } \mathrm { , } \label{eq : sseqn2}\end{aligned}\ ] ] with the boundary conditions @xmath59 in the absence of severing ( @xmath60 ) these equations are solved by the exponential functions @xmath61 and @xmath62 @xcite .
note that these solutions are only valid for @xmath63 , when the average microtubule length is bounded .
we proceed to analyze the steady state equations in a number of steps .
first , we determine global properties of the length distributions .
subsequently , we derive an explicit expression for the microtubule length distributions in the special case in which rescues are absent ( @xmath64 ) . we conclude with a numerical method using which the distribution can be calculated for arbitrary parameter values .
multiplying equations by @xmath65 , integrating from @xmath41 to @xmath66 and subtracting the results yields @xmath67 also , integrating over yields @xmath68 \,\mathrm{d}x},\ ] ] which can be combined with and to give the following expression for the total number of microtubules .
@xmath69 surprisingly , the total number of microtubules in the system does _ not _ depend on the rate of microtubule severing .
specifically , we note that severing can not prevent the diverging microtubule count for @xmath70 . in the special case @xmath71 ( no rescue events ) ,
equation for @xmath72 decouples from and can be solved analytically .
this solution can then be used to obtain an expression for @xmath73 as well .
we introduce the functions @xmath74 , allowing us to rewrite equations as @xmath75 from equations and we find that @xmath76 ( because @xmath71 ) , and using the boundary condition @xmath77 we solve to obtain @xmath78 inserting this solution into equation , it can be solved using the boundary conditions @xmath79 ( from equation ) and @xmath80 to give @xmath81,\label{eq : fminr0}\end{aligned}\ ] ] where @xmath82 is the complementary error function @xmath83 the expressions for @xmath72 and @xmath73 follow by differentiation of . the resulting distribution for @xmath84 and various values of @xmath85 is shown in figure [ fig : severingr0 ] . , @xmath71 ) .
the choice @xmath84 is an approximate value based on the _ in - vivo _ measurements in interphase plant cells reported by vos _
_ @xcite . *
a ) * length distributions for three different values of the severing rate : @xmath60 ( black ) , @xmath86 ( dark gray ) and @xmath87 ( light gray ) .
the inset shows the same on a logarithmic scale . *
b ) * average length as a function of the severing parameter @xmath85 .
the squares indicate the parameter values of the distributions in ( a ) .
the dashed line is the small-@xmath85 approximation .
length distributions have been calculated from and ; the average length from . ] looking at the derivative of the combined distribution @xmath88 , we find that for @xmath89 @xmath90 the perhaps surprising implication is that if @xmath91 , sufficiently high severing rates can lead to a positive slope of the distribution at @xmath89 .
in other words , in that case the length distribution no longer decreases monotonically with length .
finally , we compute the average microtubule length as @xmath92 with @xmath93 .
this confirms that the average length decreases with increasing severing activity , a fact that is reflected in figure [ fig : severingr0 ] . for the more general case @xmath94 we use a numerical method to solve equations .
in order to do so , we reparameterize the problem . inspired by the results and we define @xmath95 hence , @xmath96 is a ( dimensionless ) measure of the total microtubule density . in terms of these variables ,
the steady state equations can be written as @xmath97 \label{eq : divqdef}\\ \frac{\mathrm{d}}{\mathrm{d}x } p(x ) & = & -(1-r v)p(x)-(1+r)q(x)\nonumber\\ & & -s(1-v)x p(x)-s x q(x)+\nonumber\\ & & s \int_x^{\infty } \left[q(x')+(1 - 2v)p(x ' ) \right ] \mathrm{d}x ' .
\label{eq : divpdef}\end{aligned}\ ] ] equation can be formally solved to give @xmath98 \mathrm{d}x ' \\ & = s v \int_x^{\infty}\left [ ( 2x - x ' ) p(x ' ) \right ] \mathrm{d}x ' \mathrm{. } \label{eq : qprelation}\end{aligned}\ ] ] inserting this expression into equation we transform the problem into a single integral equation .
we can subsequently remove the integrals by differentiating three times and obtain the linear fourth - order ode @xmath99 where @xmath100 stands for @xmath101 . to derive the boundary conditions for this problem we use , and to derive @xmath102 by repeated application of these equalities and differentiation of equations and we obtain the boundary conditions @xmath103p(0 ) + ( 1+v)\times\nonumber\\ & \left[(1+r)(1-r v)-s\left ( \frac{3-r v+2 r v^2}{1-r v } \right ) \right]\\ p^{(3)}(0 ) = & r(1+v)\left [ ( 1-r v)^2 + s r(-3 + 7v ) \right]p(0 ) \nonumber\\ & - ( 1+v)(1+r)(1-r v)^2 \nonumber\\ & + s(1+v)\left [ 6 + 3r-4v-5r v+2r v^2 \right]\nonumber\\ & - s^2\left ( \frac{1+v}{1-r v } \right)(3 - 4v+8v^2).\end{aligned}\ ] ] we note that these boundary conditions are of the form @xmath104 with a single undetermined parameter @xmath105 . because equation is a homogeneous linear ode , we can evaluate it twice , using both @xmath106 and @xmath107 as boundary conditions .
we generally find that both solutions diverge with opposing signs .
the value of @xmath105 can therefore be determined from the constraint @xmath108 . and
@xmath109 have been used . *
a ) * length distributions for three different values of the severing rate : @xmath60 ( black ) , @xmath86 ( dark gray ) and @xmath87 ( light gray ) .
the inset shows the same on a logarithmic scale .
also shown for @xmath86 is data obtained by a stochastic simulation of microtubule severing using a gillespie algorithm , matching the predicted distribution .
simulation parameters were [ @xmath110 m s@xmath111 , @xmath112 m s@xmath111 , @xmath113s@xmath111 , @xmath114s@xmath111 , @xmath115s@xmath111 , @xmath116m@xmath111s@xmath111 ] .
length data was distributed into 500 bins and sampled 1000 times at 50 second intervals after an initial equilibration period of 50,000 seconds .
* b ) * average length as a function of the severing parameter @xmath85 .
the squares indicate the parameter values of the distributions in ( a ) .
the dashed line is the small-@xmath85 approximation the light gray curve , which converges for large @xmath85 , is the result for @xmath71 ( figure [ fig : severingr0 ] ) .
length distributions have been calculated from . ] to this end , both solutions should be evaluated over a range that is as large as possible , whilst maintaining a very high numerical accuracy , because the final result is obtained by subtracting two diverging functions . in our numerical calculations
( using wolfram mathematica 6.0 ) , we have evaluated the differential equation with sufficient precision to achieve an accuracy of 13 significant digits and limited the integration range to 10 decay lengths @xmath118 _ or _ the point at which the first solution exceeded the value @xmath119 , depending on which occurred first .
figure [ fig : severingr1 ] shows the numerically computed distributions for @xmath84 , @xmath109 and various values of @xmath85 .
it is interesting to note that the total distribution is no longer monotonically decreasing for @xmath86 and @xmath87 .
figure [ fig : fplusminus ] shows that this is solely due to the contribution from the growing microtubules .
we also note that the average length decreases rapidly for relatively small severing rate .
as the severing rate increases , the average length converges to that of the system without rescue events ( @xmath71 ) . in other words , if severing events occur very frequently , rescue events are no longer significant , presumably because the microtubules become so short that they disappear before they can be rescued .
this statement is summarized by the condition @xmath120 , where @xmath121 is the average microtubule length , or in dimensionless units @xmath122 . ,
@xmath109 , @xmath86 ) .
the total distribution function @xmath123 has been computed numerically using ; @xmath72 and @xmath73 have been computed using and . ]
we have constructed a model that describes the dynamic instability of microtubules in combination with the severing of microtubules .
this model takes the form of two coupled integro - differential equations that are a function of three parameters : @xmath57 , the ratio of polymerization and depolymerization speeds , @xmath124 , the ratio of the rescue and catastrophe rates , and the dimensionless severing rate @xmath85 .
we have analyzed the steady state solutions and their properties : notably the number of microtubules and their average length .
for the special cases of no rescue events ( @xmath71 ) and small severing rates ( see appendix [ sec : smallsev ] ) , we have presented analytical solutions .
the general case has been addressed by transforming the coupled integro - differential equations into a single fourth order differential equation that is solved numerically .
the resulting microtubule length distributions show a number of interesting properties .
as expected , an increase in the severing rate always leads to a decrease in the average length of the microtubules .
in addition , because the rate of severing is proportional to the microtubule length , the number of very long microtubules is strongly reduced , leading to a distribution that is more compact ( see appendix [ sec : smallsev ] for an explicit expression ) .
furthermore , in contrast to the dynamic instability model without severing , the length distributions are no longer guaranteed to decrease monotonically with increasing length .
for some parameters , a ` bump ' is observed in the length distribution , caused by the continuous creation of short ( but not vanishingly small ) microtubules through the severing process .
we have also demonstrated that the total number of microtubules does not depend on the severing rate , even though the average length of each microtubule decreases . to understand this counterintuitive result
, we consider a system without severing , in which the following steady - state relation holds : [ population size ] @xmath125 [ nucleation rate ] @xmath126 [ average lifetime ] .
suppose that a single microtubule is severed ( consistent with an infinitesimally small severing rate ) .
the expected lifetime of the created segments is obviously shorter than that of the original segment .
however , the fact that the total number of microtubules is unaffected by the presence of severing implies that this decrease in lifetime is compensated _ exactly _ by the fact that every severing event ` nucleates ' an extra microtubule . for this cancellation to occur
, the sum of the lifetimes of the two fragments of a severed microtubule should equal the expected lifetime of the microtubule if severing had not occurred .
that this is indeed the case can be shown using a simple argument , based on the lifetime of a microtubule in the absence of severing . in this case
the dynamics of the microtubule tip are independent of its length .
this implies that the average time it takes for a growing microtubule of length @xmath10 to later return ( in a shrinking state ) to the _ same _ length is independent of the value of @xmath10 , and therefore equal to the average microtubule lifetime @xmath127 ( obtained by starting from @xmath128 ) . the remaining contribution to the lifetime
is given by the time it takes for the shrinking microtubule of length @xmath10 to disappear . in the absence of rescue events , this would take a time @xmath129 . however , each rescue event switches the microtubule back to the growing state , extending the microtubule s expected lifetime by @xmath127 before it returns to the same position in the shrinking state .
the expected number of such rescue events is equal to @xmath130 .
this argument is illustrated in figure [ fig : shrinkdynamics ] .
collecting the terms described above , we obtain the following expression for the expected microtubule lifetime @xmath131 of a microtubule of length @xmath10 and growth state @xmath28 : @xmath132 the same relation , including an expression for @xmath127 , has been derived analytically by and bicout @xcite and an equivalent expression for microtubules that grow in discrete units was produced by rubin @xcite . each severing event conserves the total length @xmath10 of the severed microtubule and the state @xmath30 of the existing plus end , and the newly created plus end immediately undergoes a catastrophe ( @xmath133 ) . in terms of life times
we find that @xmath134 , so that a single severing event indeed preserves the total lifetime . by induction
it follows that _ any _ number of severing events will leave the expected number of microtubules in the system intact . .
the return times @xmath135 have an expectation value @xmath127 . ]
we stress that this argument holds regardless of the frequency and location of severing events .
this implies that any distribution of severing events provided they lead to a catastrophe of the trailing end
conserves the total number of microtubules in steady state .
specifically , this also applies to severing at positions where microtubules of different orientations cross @xcite , a process that may be relevant to the formation of the cortical array in plant cells .
the conservation of microtubule number in the presence of severing is in apparent contradiction with the experimentally reported increase in microtubule numbers @xcite .
however , this can be readily explained by the fact that our results are based on the assumption of constant parameter values . in a living cell ,
the decrease in average microtubule length that is the result of severing will lead to an increased availability of free tubulin dimers . in turn , this is likely to increase the polymerization rate ( growth speed ) and nucleation rate , which would indeed cause an increase in the number of microtubules .
st thanks frank jlicher and benjamin lindner for very helpful discussions and is grateful for the hospitality that was provided by the max planck institute for the physics of complex systems .
the authors thank nils becker for critical comments on the manuscript .
st was supported by a grant from the nwo `` computational life sciences '' program ( contract cls 635.100.003 ) and a visiting stipend from the max planck society .
this work is part of the research program of the `` stichting voor fundamenteel onderzoek der materie ( fom ) '' , which is financially supported by the `` nederlandse organisatie voor wetenschappelijk onderzoek ( nwo ) '' .
in the absence of severing , equations are solved by @xmath61 and @xmath136 . to investigate the changes that occur when a small severing rate is included , we perturb these solutions as follows @xmath137 inserting these expressions into equations and dropping all higher order terms gives a set of equations that can be solved to yield @xmath138 \right)\nonumber\\ & + o(s^2)\\ f^-(x ) = & v e^{-(1-r v)x } \biggr(1 + s \biggr [ \frac{1+v}{(1-rv)^2 } + \frac{-v+r v+2 r v^2}{(1-r v)^2}x \nonumber\\ & -\frac{1+r v^2}{2(1-r v)}x^2 \biggr ] \biggr ) + o(s^2)\end{aligned}\ ] ] we note that both solutions will become negative for large values of @xmath65
. however , even though this is decidedly unphysical , the effect on measurable parameters such as the average microtubule length is small ( for small @xmath85 ) , because of the rapid decay of @xmath139 and @xmath140 .
the properties of the resulting distributions , such as the average length , will therefore still remain valid , subject to the bounds on @xmath85 computed below . the first - order distributions in @xmath85
satisfy the total microtubule number constraint @xmath141 \,\mathrm{d}x } = ( 1+v)/(1- r v ) + o(s^2)$ ] , consistent with the general result . for the average length we obtain ( to first order in @xmath85 ) @xmath142 for @xmath71 ,
this is consistent with the exact result .
the resulting ( linear ) predictions are indicated in figures [ fig : severingr0 ] and [ fig : severingr1 ] . finally , we determine the length variation @xmath143 and , from that , the coefficient of variation ( @xmath144 ) @xmath145 this number provides a measure for the relative width of the distribution . the results and confirm that severing decreases both the weighted average and the relative width of the length distribution .
the results above have been obtained under the assumption that @xmath85 is very small .
to make an _ a priori _ estimate for the validity range of @xmath85 , we determine the relative importance of the terms on the right - hand side of equations and . using the results in the absence of severing as a benchmark , the terms _ not _ involving @xmath85 give contributions of the order @xmath146 .
comparing the integral term ( evaluated for the @xmath60 situation ) with this term gives @xmath147 .
the term that is proportional to @xmath148 will dominate the other terms for large @xmath65 , but this does not significantly affect the results if it only occurs for lengths that are much longer than the average length .
evaluating the terms at @xmath149 , where @xmath150 is the number of average lengths , we obtain the constraint @xmath151 .
because @xmath150 is of the order 1 , we simply state that the approximation is accurate for @xmath152 .
j. roland , j. berro , a. michelot , l. blanchoin , and j .- l . martiel .
stochastic severing of actin filaments by actin depolymerizing factor / cofilin controls the emergence of a steady dynamical regime .
, 94(6):2082 2094 , 2008 . j.w .
vos , m. dogterom , and a.m.c .
icrotubules become more dynamic but not shorter during preprophase band formation : a possible `` search - and - capture '' mechanism for microtubule translocation . , 57(4):246258 , 2004 . | microtubules are highly regulated dynamic elements of the cytoskeleton of eukaryotic cells .
one of the regulation mechanisms observed in living cells is the severing by the proteins katanin and spastin .
we introduce a model for the dynamics of microtubules in the presence of randomly occurring severing events . under the biologically motivated assumption that the newly created plus end undergoes a catastrophe
, we investigate the steady state length distribution .
we show that the presence of severing does not affect the number of microtubules , regardless of the distribution of severing events . in the special case in which the microtubules can not recover from the depolymerizing state ( no rescue events ) we derive an analytical expression for the length distribution . in the general case
we transform the problem into a single ode that is solved numerically . |
several passenger boarding schemes are used by the airline industry in effort to quickly load passengers and their luggage onto an airplane .
since the passenger boarding time often takes longer than refueling and restocking the airplane its reduction could constitute a significant savings to a particular carrier , especially for airplanes which make several trips in a day .
conventional wisdom would suggest that boarding from the front to the back is the worst case but that boarding from the back to the front is optimal or nearly so . indeed , this is the strategy that is often employed , boarding passengers in blocks from the rear of the plane to the front . in this case ,
conventional wisdom only provides an answer that is half right .
the worst boarding method is , indeed , to board the plane from front to back .
as i will show however , boarding the airplane from the back to the front is very likely the _ second worst _ method .
previous studies of the airplane boarding used a variety of approaches .
pioneering work done by @xcite used computer simulations of several boarding schemes and found there is much room for improvement over traditional back - to - front boarding .
of particular interest is that they point out that random boarding is superior to many traditional methods .
a later computer simulation study @xcite confirmed that traditional methods are not optimal , even in light of different assumptions regarding the primary cause of delay in the boarding process .
analytic work by @xcite , which modeled optimal airplane boarding as an extremal path in a two - dimensional , lorentzian geometry , broadly confirmed the findings of @xcite and was able to interpret those findings in the context of their model thus providing an explanation for why the different boarding strategies perform as they do relative to each other .
each of these studies , particularly the computational works , focused on different components of delay . consequently , while all found traditional methods lacking , none agree on the best overall approach .
a portion of their disagreement arises from differing assumptions .
another point of discrepancy stems from the fact that their focus was largely on finding the best practical approach or the best of a set of approaches .
this limitation on the scope of previous works meant that they did nt always study the same boarding methods .
thus , the question remains , `` what method , practical or otherwise , gives the fastest boarding time ? '' it is this question that i address here . the answer to this question , whether directly applicable or not , is valuable as it can inform decisions regarding the worth changing an existing policy since it indicates how much room there is for improvement .
moreover , once the optimal boarding method is identified one can explore the reasons why it is successful and incorporate those characteristics into a more practical scheme . in this work
i infer from the results of a computer simulation and an associated optimization algorithm the optimal passenger boarding method .
the fundamental assumption that i make is that the bulk of the time to load the airplane is consumed by time that it takes the passengers to load their luggage .
all other effects , such as the time used by passengers who stand up to retrieve an item , who sit in the wrong seat , or who must pass by someone that is already seated in their row are initially treated as negligible .
a more sophisticated model could include such detail ; or , many of their effects could be folded into the model via some of the available parameters or components ( such as the distribution from which the passenger luggage loading times are selected ) . with the stated assumption
, i find that the boarding time for the optimal scheme can be significantly faster than the boarding time of the worst case between a factor of 4 and 10 faster depending upon the length of the airplane and other model parameters . in this article
i describe the techniques used to find the optimal method , i interpret the results and use that interpretation to discuss the merits of some schemes that are employed by the industry , finally i give some concluding remarks .
note that i generally use the term `` boarding '' to refer to the boarding process itself and the term `` loading '' to refer to the passengers loading their luggage .
thus , boarding time and loading time are the times required to fill the aircraft and the time required to load one s luggage respectively .
the nominal airplane model that i use seats 120 passengers with six passengers per row and 20 rows .
since the focus is on the general boarding procedures , there is no first - class cabin , no priority seating , and each flight is completely full .
i discuss the effects of changes to this airplane model in section [ results ] .
the passengers are each assigned a seat and the number of time steps that they need to load their luggage , a random number between 0 and 100 unless otherwise stated .
other , human nature assumptions include : 1 ) that a person will not move unless there is enough space between them and the person in front of them two steps in this case , 2 ) that if they are moving , then they will occupy any empty space in front of them prior to stopping ( this results in passengers bunching - up again as they come to a halt ) , 3 ) that they require one space either in front of or behind them in order to load their luggage ( this is related to the congestion parameter @xmath0 introduced by @xcite ) , and 4 ) that they only load their luggage into the bins above their assigned row . in section [ results ]
i discuss the effects of changing any of these parameters or the distribution from which the luggage loading times are assigned .
this model does not include the effects of aisle vs. window seats , the clustering of passengers into companions or families , and other effects of human nature . while adding these features
might improve the accuracy of the results , these effects are not likely to be the primary issue and consequently should not be the fundamental concern when finding the general strategy for a passenger boarding scheme .
moreover , many of their effects can be accounted for once the optimal boarding method , based upon the stated assumptions , is identified .
i discuss these issues in sections [ robust ] and [ practical ] .
the algorithm that i use to find the optimal loading order is based upon a markov chain monte carlo ( mcmc ) algorithm and is similar to the metropolis algorithm @xcite . starting with an initial passenger order i load the airplane and
record the loading time . then , starting with that initial order , the positions of two random passengers are exchanged and the airplane is loaded again . if the airplane boards as fast or faster than the previous iteration , then i accept the current passenger order , swap the positions of two additional random passengers , and repeat the process .
if the current configuration loads more slowly than the previous , then i reject the change , return to the previous configuration , and repeat the process beginning there .
i stop after @xmath1 iterations since adding additional steps does not significantly change the results . unlike a traditional mcmc
, i do not allow any configuration which loads more slowly to be accepted .
technically , this means that the stated algorithm only finds a local minimum .
however , when i include this aspect the results are unchanged while the time needed to converge increases .
thus , the choice to neglect that aspect of an mcmc analysis should not affect the results stated here .
that being said , it is possible for many configurations to board in the same time or to be near enough that the differences in boading time are not important .
this shows that a class of configurations that are effectively equivalent is more important than a single , optimum order .
for example , there is no difference in the loading time if the two aisle - seat passengers in the same row are swapped , and there is little difference in swapping two random passengers . to identify the class of optimal configurations
i tabulate the differences in seat number between adjacent passengers .
it is this distribution of seat number differences , the `` seating distribution '' , that remains effectively constant from one optimization run to the next .
this fact illustrates the important point that the actual seat numbers of two adjacent passengers is less important than how far apart they sit from each other .
the results of applying the above analysis gives the seating distribution shown in figure [ seatdist ] .
these results are from 100 realizations of the boarding scenario where the passengers are reassigned their luggage loading time for each of the realizations .
the largest feature is the peak near a seat difference of 12 .
that difference corresponds to two rows , or the distance that i assume neighboring passengers require to load their luggage .
other features of the peak , aside from its location , are its shape , its height relative to the rest of the distribution , and its width .
all four of these aspects depend upon the passenger and airplane model parameters and each of them could be calibrated with data as i describe below .
as stated , the location of the peak corresponds to the distance required by a passenger to load his luggage .
if passengers need only the space that corresponds to their assigned row , then the location of the peak would shift to a value of 6 or one row of difference . if passengers require three rows of loading space ( including their assigned row ) , then the peak shifts to 18 .
this effect can be seen in figure [ personal ] where i make these changes while leaving all other model parameters fixed .
the peak of the seating distribution is symmetrical near its apex .
this is because the passengers can load their luggage using either the space in front of them or the space behind them .
the width of the peak is related to the number of seats per row .
if there are only four seats per row , then the width of the peak is more narrow .
if there are eight seats per row ( still with only one aisle ) , then it is more broad .
the height of the peak depends upon the time that the passengers take to load their luggage .
if passengers load their luggage instantaneously , then the peak disappears altogether . as the luggage loading time increases the penalty for having someone out of order increases and the algorithm forces more passengers to be separated by amounts nearer the minimum row separation required to load their luggage ( here between 7 and 18 seats ) .
the height of the peak ultimately saturates when the average luggage loading time approaches the time to walk the length of the airplane .
this means that if it takes significantly longer for a typical passenger to walk the length of the airplane than it does for him to load his luggage , then the passenger ordering is much less important .
the optimal boarding scheme is found by extrapolating from the results given above , using the insight that they provide .
the reason that these seating distributions load faster than the worst case is that they allow multiple passengers to load their luggage at the same time .
the peak occurs at a distance that corresponds exactly to the space needed by adjacent passengers to do so . taking this to an extreme
, we wish to find the configuration that allows the maximum number of passengers to load their luggage at all times during the boarding process . that is the case where all adjacent passengers are assigned seats that are separated by exactly two rows such that the person at the front of the line is assigned a seat on the back row .
the second - order effect of windows vs. aisle seats can be incorporated here by having the passengers in the window seats enter the airplane first , then the middle seats , then the aisle seats .
the resulting seat ordering could be one of those shown in figure [ seatdiagram ] though there are many equivalent orderings .
this ordering scheme provides nearly a five - fold reduction in the time that it takes over the worst case an improvement that gets larger with a larger aircraft .
@xmath2 a look at the optimal boarding method shows why loading from the back of the plane to the front does not provide any benefit .
if the back two rows of passengers were to board the airplane first , they would occupy roughly 12 rows of the aisle .
all but the first few would be putting their luggage away while the others waited their turn ; the passengers load their luggage serially .
the optimal boarding strategy uses this aisle space more efficiently since each member of the first group of passengers who enter the airplane ( 10 passengers in the fiducial model ) can put their luggage away ; they load their luggage in parallel . in this manner
the aisle is not used as a passive extension of the waiting area , but rather as a place for passengers to actively situate themselves . ideally ,
all of the passengers that are inside the aircraft should either be seated or be loading their luggage ; with none waiting .
one question that arises from this is whether or not it is practical to implement the optimal boarding scheme , where each passenger enters the airplane in a particular order .
such a scenario may well be possible since southwest air has recently implemented a similar policy , at least to some extent .
given that , however , there will always be some fraction of the passengers who are out of order ; there will always be families or other groups who board together regardless of their assignments .
these same issues will affect other boarding methods in similar ways .
in the next section i test the robustness of the optimal boarding method under these circumstances .
section [ practical ] is a comparison of a few practical boarding methods where the passengers board in groups but are ordered randomly within those groups .
to test the robustness of the optimal boarding scheme i conducted two experiments .
the first is to change the distribution from which i select the passenger s loading time .
the second test is to make random changes to the passenger ordering .
these changes include swapping the locations of several random pairs of passengers and shifting the entire line by some random number ( moving people at the end of the line to the front ) .
to test the effect of a different distribution of luggage loading time , i ran my minimization software on 100 realizations of each of several distributions .
these distributions include a uniform distribution with a given mean , a normal distribution with the same mean and with a variance equal to that mean ( essentially a poisson distribution ) , and an exponential distribution with the same mean . for each of these cases
the resulting passenger seating distributions were statistically indistinguishable as shown by a kolmogorov - smirnov test @xcite .
moreover , the time required to load the entire plane is not affected by these different distributions ; it depends primarily upon the mean luggage loading time .
the fact that the results of this analysis does not depend upon the distribution from which the luggage loading times were chosen indicates that the optimal boarding method does not depend upon the presence of non - gaussian outlier disturbances ( as it should not ) .
thus , if a particular passenger takes an unusually long time to load his luggage , the optimal method of passenger boarding is still , on average , optimal .
moreover , since one can not generally identify which passenger
if it is a passenger will be the cause of delay and since that passenger s seat will be at a random location within the aircraft , the best that one can robustly expect to do is to board optimally both before and after the disturbance . randomly shifting
the line does not significantly affect the boarding time .
this is because it only changes the starting point of the boarding process .
all of the passengers keep their 12 seat spacing from their neighbors and so the advantage of the optimal boarding scheme is preserved .
if pairs of passengers are swapped , which effectively randomizes portions of the line , then the time to board the airplane can change significantly .
indeed , a 20% increase in the boarding time results from randomly swapping only 10% of the passengers that is 6 pairs for the case of 120 passengers .
however , there is an upper limit to the adverse results of swapping passengers .
once they are completely randomized additional swaps preserve the random nature of the passenger ordering and it does nt get any worse . interestingly , random boarding takes much less than half the time of the worst case boarding ; indicating that randomization is not catastrophic , a finding also identified by @xcite .
indeed , we will see in the next section that random boarding compares favorably with traditional boarding techniques .
the effects of swapping pairs of passengers is shown in figure [ passengermix ] .
while the optimal scheme would produce the fastest boarding times , there are issues of practicality to consider .
it may be challenging to arrange all of the passengers in the proper order though , as mentioned , at least one airline has implemented this policy . regardless , most airlines board the airplane in groups , presumably out of convenience and in effort to reduce confusion .
in this section i introduce a few practical modifications to the optimal boarding scheme and choose one to compare with existing methods .
the advantage of the optimal boarding method comes from the fact that neighboring passengers do not sit near each other and consequently can load their luggage simultaneously .
one way to accomplish a similar effect while allowing blocks of passengers to board is to have each block contain passengers from widely separated rows . after trying several possibilities i chose to use , as a modification to the optimal method , blocks of three consecutive seats separated by 12 .
this scenario has four boarding groups and is equivalent to calling all passengers that are in even rows and from one side of the airplane .
the three remaining groups are for the other side of the airplane in the same row , then the two sides of the odd rows .
the loading time that results from this scheme is not as fast as the optimum , indeed it took about twice as long to board , but it was more than a factor of two faster than the worst case .
i call this boarding method the `` modified optimal '' method
. similar boarding methods , arrived at by very different means , are suggested by @xcite and @xcite and these methods also prove successful in the respective studies .
i selected two group boarding strategies to compare with the optimal , the modified optimal , the worst case , and the second worst case ( boarding from the back to the front with the passengers in order ) .
these are : 1 ) ordered blocks , where a fourth of the cabin is loaded at a time starting in the back and moving to the front and 2 ) a scheme where the windows are boarded first , then the middle , then the aisle seats . within each of these groups
the travelers were randomly distributed .
tests which used unordered blocks ( blocks of 5 rows but not in order from back to front ) gave similar results to the ordered blocks and were therefore not included .
the ordered block scenario reduced the boarding time to 74% of the worst case . since the main contributor to fast boarding is having multiple passengers load their luggage at once
, this method suffers from the fact that only passengers within a small portion of the airplane are boarding at a given time . within the boarding block
there is the possibility of multiple people loading their luggage at once , but it is relatively small since at most three passengers ( out of 30 ) could be simultaneously loading their luggage .
the windows - middle - aisle approach reduced the boarding time to 43% of the worst case .
the advantage here is that passengers from anywhere in the cabin are allowed to board .
thus , many people can load their luggage at once and the probability having two passengers from adjacent rows near each other is small .
moreover , this probability decreases as the length of the airplane increases .
another advantage is that the second - order effect of getting past the person in the aisle is eliminated .
one drawback of this approach is that many people travel in groups and would likely not adhere strictly to this particular boarding policy . in general , however , this approach did very well ( as also found by @xcite though for different reasons ) .
the modified optimal approach performed slightly better than , though almost identically to , the windows - middle - aisle approach .
this method does not depend as strongly on the length of the airplane .
so , for shorter airplanes it compares better while on longer planes it compares less well .
an advantage that the modified optimal approach has over boarding window - middle - aisle is that it allows passengers who are travelling together and sitting side - by - side to board at the same time without boarding out of order .
random boarding , where the passengers positions in line are completely uncorrelated with their seat assignment , shares the advantage of the windows - middle - aisle method of spreading passengers throughout the length of the airplane .
moreover , it is not disadvantaged , in implementation at least , by traveling groups .
the random method performed the similar to , but slightly worst than , the windows - middle - aisle method and the modified optimal method .
this demonstrates that the optimal approach , even with a significant fraction of people out of order can do at least as well as or better than the most efficient of the methods that employ boarding groups .
figure [ blockresults ] shows a histogram of the loading times for 100 realizations of seven different boarding schemes , each realization having different selections for the time it takes to load the luggage and different passenger ordering where applicable .
these include : 1 ) optimal boarding , 2 ) the modified optimal approach , 3 ) the window - middle - aisle method , 4 ) random boarding , 5 ) ordered blocks from back to front , 6 ) back to front with all passengers in order , and 7 ) the worst - case ( front to back with all passengers in order ) .
we see from this figure that optimal boarding has , by far , the best improvment in loading times , nearly a factor of five faster than the worst case , more than a factor of three better than the ordered blocks , and it is more than a factor of two faster than the modified optimal , random , and windows - middle - aisle methods .
this improvement grows with the length of the airplane such that an airplane that seats 240 passengers ( 40 rows ) will board over seven times faster than the worst case !
the results of this study , based upon the assumption that a passenger loading his luggage consumes the bulk of the time that it takes for him to be seated , identify the primary cause for delay in the boarding process as well as the best means to overcome these delays . by boading passengers in a manner that allows several passengers to load their luggage simultaneously the boarding time can be dramatically reduced .
this result contradicts conventional wisdom and practice that loads passengers from the back of the airplane to the front .
indeed , it shows that loading from the back to the front is hardly better than the worst case scenario .
the goal of an optimized boarding strategy should focus on spreading the passengers who are loading their luggage throughout the length of the airplane instead of concentrating them in a particular portion of the cabin . by boarding in groups where passengers whose seats
are separated by a particular number of rows , by boarding from the windows to the aisle , or by allowing passengers to board in random order one can reduce the time to board by better than half of the worst case and by a significant amount over conventional back - to - front blocks which , while better than the worst case performed worse than all other block loading schemes .
the primary drawbacks for any of these methods is likely to be psychological instead of practical .
groups of passengers who wish to board together would be an issue to investigate from both a customer satisfaction point of view and as a component in a more detailed model .
if a workable method to have passengers line up in an assigned order could be found and it likely may be employed already , then there is the potential for a substantial savings in time .
such a savings would most likely benefit flights between nearby cities where a particular airplane would make several trips in a given day since it might allow one or two additional flights .
or , it might allow an airline to reduce the number of gates that it requires to meet its obligations since each gate would be cleared more rapidly . while the generic features of this model are well understood , a real application of it would require some data so that it can be properly calibrated . in particular , the distributions of luggage loading times , the fraction of people traveling in groups , the queueing habits of passengers , and empirical measurements of `` personal space '' are all pieces of information that are necessary to state these results in terms of actual times and distances instead of arbitrary time steps and lengths that are used by a computer model .
regardless of the ultimate application of this technique , the establishement of a firm lower bound can be used to inform a decision maker of the worth of further improvements to a particular boarding strategy .
if an improvement could provide only a marginal gain while costing significant amounts money and time to implement then it is not likely to be worth the investment . on the other hand ,
if a particular strategy is clearly failing to meet the demands of competition and customer satisfaction , then knowing just how much room there is for improvement could expedite changes . in the end ,
the time that it takes to load passengers into the airplane affect not only the airline company and the airport , it also affects the passengers .
few enjoy standing in line longer than necessary and fewer still enjoy sitting in an airplane longer than needed .
faster boarding would be a significant improvement for all involved parties .
* acknowledgements : * j. steffen acknowledges the generous support of the brinson foundation .
fermilab is supported by the u.s .
department of energy under contract no . de - ac02 - 07ch11359 .
a record of invention of this software and its results are on file at fermilab . | using a markov chain monte carlo optimization algorithm and a computer simulation , i find the passenger ordering which minimizes the time required to board the passengers onto an airplane .
the model that i employ assumes that the time that a passenger requires to load his or her luggage is the dominant contribution to the time needed to completely fill the aircraft .
the optimal boarding strategy may reduce the time required to board and airplane by over a factor of four and possibly more depending upon the dimensions of the aircraft .
in addition , knowledge of the optimal boarding procedure can inform decisions regarding changes to methods that are employed by a particular carrier .
i explore some of the salient features of the optimal boarding method and discuss practical modifications to the optimal .
finally , i mention some of the benefits that could come from implementing an improved passenger boarding scheme . |
the direct detection of gravitational radiation presents one of the greatest scientific challenges of our day . with interferometers such as ligo , virgo , geo , and tama @xcite expected to be operating in the next few years , and
a new generation of spherical resonant mass detectors under study @xcite , the calculation of the signals expected from various astrophysical sources has a high priority .
accurate calculations of the waveforms are needed to enable both the detection and identification of sources @xcite .
in particular , short duration bursts are expected to be more difficult to detect than longer - lived signals .
one interesting class of sources includes rapidly rotating compact objects that develop the rotationally - induced `` bar instability '' .
this instability derives its name from the bar - like deformation it induces .
the resultant object is potentially a strong source of gravitational radiation because of its highly nonaxisymmetric structure .
examples of compact astrophysical objects that may rotate rapidly enough to encounter this instability include stellar cores that have expended their nuclear fuel and are prevented from undergoing further collapse by centrifugal forces @xcite ; a neutron star spun up by accretion from a binary companion @xcite ; and the remnant of a compact binary merger @xcite .
such global rotational instabilities in fluids arise from nonaxisymmetric modes @xmath0 , where @xmath1 is known as the `` bar mode '' .
it is convenient to parametrize them by @xmath2 where @xmath3 is the rotational kinetic energy and @xmath4 is the gravitational potential energy @xcite . in this paper , we focus on the _ dynamical _ bar instability , which is driven by newtonian hydrodynamics and gravity , and is expected to be the fastest growing mode .
it operates for fairly large values of the stability parameter @xmath5 and develops on a timescale on the order of the rotation period of the object . for
the uniform density , incompressible , uniformly rotating maclaurin spheroids , @xmath6 . in the case of differentially rotating fluids with a polytropic equation of state ,
the @xmath1 dynamical stability limit @xmath6 has been numerically determined to be valid for initial angular momentum distributions that are similar to those of maclaurin spheroids @xcite ; see also @xcite .
( we note that when @xmath7 is greater than some critical value @xmath8 , a _
secular _ instability can arise from dissipative processes such as gravitational radiation reaction and viscosity .
when this instability arises , it develops on the timescale of the relevant dissipative mechanism , which can be several rotation periods or longer @xcite . in recent years
, much work has also been carried out on various other modes in rotating relativistic stars as detectable sources of gravitational radiation ; see @xcite for a review and references . ) the first numerical simulations of the dynamical bar instability were carried out by tohline , durisen , and mccollough ( tdm ; @xcite ) in the context of star formation . using a polytropic equation of state , @xmath9 with polytropic index @xmath10 , they evolved differentially rotating axisymmetric models with a 3-d eulerian hydrodynamics code , or hydrocode , in newtonian gravity . in all models with initial @xmath11 ,
the @xmath1 mode grew to nonlinear amplitudes and a two - armed spiral pattern was produced as mass and angular momentum were shed from the ends of the bar .
numerous other simulations have confirmed this basic scenario for the evolution of the bar mode into the nonlinear regime ; see sec .
[ prev_stud ] for references and further discussion .
more recently , these techniques have been extended to the context of rapidly rotating , compact objects in the newtonian regime , with the gravitational waves calculated in the quadrupole limit .
this is a reasonable first approximation for an object , such as a centrifugally - hung stellar core with a density intermediate between that of white dwarfs and neutron stars , with initial mass @xmath12 and radius @xmath13 km , and hence @xmath14 .
centrella and collaborators used both smooth particle hydrodynamics ( sph ) and eulerian finite difference hydrodynamics to evolve the bar instability in a model with @xmath15 @xcite . in all of their runs
the gravitational wave signal was a relatively short duration burst lasting for several bar rotation periods , and the system evolved to a nearly axisymmetric central core surrounded by a flattened , disk - like halo .
new @xcite carried out a similar study , with an improved version of tohline s eulerian code @xcite .
her simulation employed a symmetry condition that only permitted the growth of even modes @xmath16 ; see sec .
[ prev_stud ] .
this simulation produced a final state with a persistent bar - like core , which yielded a gravitational wave signal of much longer duration than that found by centrella and collaborators .
given the requirements of reliable waveforms for the detection and identification of sources , it is important to resolve this issue of the late - time gravitational wave signal from the dynamical bar instability .
to this end , we have carried out a set of long - duration runs using the two eulerian codes employed by new and by centrella in their earlier work , and have made a detailed study of the resulting models . in section [ prev_stud ]
we review previous numerical studies of the dynamical bar instability , highlighting the various assumptions and restrictions used by different authors .
the numerical techniques we used are discussed in sec .
[ num_tech ] . in sec .
[ analysis ] we present our simulations and analyze the results .
a discussion of these results follows in sec .
[ discuss ] .
finally , the appendix contains additional information about the two hydrocodes used in this work .
as mentioned above , the work of tdm @xcite set the stage for subsequent numerical calculations of the dynamical bar instability .
their initial models consisted of differentially rotating , axisymmetric equilibrium spheroids with a maclaurin rotation law for the angular momentum distribution .
the maclaurin law produces rigid rotation when it is applied to an incompressible ( @xmath17 ) fluid ; when it is used in a polytrope , it produces differential rotation @xcite . after small amplitude
random perturbations were applied to the density , each model was evolved into the nonlinear regime using a 3-d eulerian hydrocode with newtonian gravity .
this hydrocode solved the mass continuity and euler equations in cylindrical coordinates @xmath18 ; the resulting evolutions were adiabatic and maintained the same polytropic equation of state , eq .
( [ poly ] ) .
tdm used equatorial symmetry and `` @xmath19-symmetry '' in their simulations .
equatorial symmetry is a reflection symmetry through the equatorial plane @xmath20 .
the @xmath19-symmetry condition imposes periodic boundary conditions at angles @xmath21 and @xmath22 ; thus , physical variables are the same in the interval @xmath23 as they are in @xmath24 .
it is computationally advantageous to impose an equatorial- and/or @xmath25 symmetry condition on such a simulation because , by doing so , half as many computational grid zones are required in order to achieve a given spatial resolution in the vertical and/or azimuthal coordinate directions , respectively .
it was also physically reasonable for tdm to adopt both of these symmetry conditions because the eigenfunction ( a pure , @xmath1 barmode ) to which their models were expected to be initially unstable had both equatorial and @xmath25symmetry @xcite .
as we discuss more fully below , ultimately one would like to remove these computational constraints in order to test whether or not the physical outcome is sensitive to them .
the first work to address the late - time development of the bar instability was published by durisen , gingold , tohline , and boss @xcite , who ran simulations with @xmath26 and @xmath27 for @xmath15 polytropes .
they used three different 3-d hydrocodes : tohline s eulerian code as used in tdm , another eulerian code developed in spherical coordinates by boss , and an sph code developed by gingold .
boss s code also enforced equatorial- and @xmath19-symmetries but , being gridless , gingold s sph code imposed neither of these symmetries .
however , the sph simulations were limited to a very small number of particles , @xmath28 .
the results produced by these three separate simulation codes were qualitatively similar .
for example , at early times all simulations showed evidence of the development of a bar - like pattern instability , consistent with the results of tdm and the predictions of linear perturbative analysis @xcite .
peturbative analysis says this instability is the result of the growth of a coherent bar - like wave that propagates around the system with a well - defined pattern speed , while material moves differentially through that pattern . at subsequent times in the simulations ,
the barmode distortion developed into a two - armed , trailing spiral pattern as described by tdm ; when the spiral pattern reached a nonlinear amplitude , some relatively high specific angular momentum material was expelled in the equatorial plane of each system ; and the primary structure that remained at the end of each simulation was a dynamically stable , centrally condensed object exhibiting a value of @xmath29 .
but there were significant quantitative differences among the various evolutions presented by durisen et al .
for example , the simulations produced central remnants that had different total masses and exhibited different degrees of nonaxisymmetric distortion .
this disagreement signified , in part , that the simulation techniques being used were rather primitive and , in part , that the available computing resources did not permit the simulations to be carried out with adequate spatial resolution .
williams and tohline subsequently carried out an investigation of the dynamical barmode instability in models with different polytropic indices . using the tdm code with @xmath19-symmetry and an improved azimuthal grid resolution , they first considered models with initial @xmath30 and @xmath31 , and @xmath32 , and focused their analysis on the measurement of barmode growth rates and pattern speeds in the linear - amplitude growth regime @xcite . the runs with @xmath33 and @xmath34 were then extended to later times through the development of nonlinear - amplitude nonaxisymmetric structures and yielded a rotating triaxial central remnant @xcite .
williams and tohline noted that such a configuration would be of interest when viewed in the context of compact stellar objects because `` its existence would presumably be discernable from the spectrum of any emitted gravity wave radiation , '' but they did not derive such a spectrum from their models .
houser , centrella , and smith @xcite were the next to carry out 3-d simulations of the dynamical bar instability for the case @xmath10 and initial @xmath35 , this time in the context of rapidly rotating stellar cores .
using both an sph and an eulerian code , they considered the matter to be a perfect fluid with equation of state @xmath36 where @xmath37 is the specific internal energy , and solved an equation for the internal energy . using artificial viscosity ,
they could account for the energy generation by shocks that occurs when the spiral arms form and deflect the streamlines of the supersonically moving fluid .
routines were added to calculate the gravitational waveforms and luminosities in the quadrupole approximation .
the sph code ( developed from treesph ; see @xcite ) imposed no symmetry restrictions and was run with up to @xmath38 particles .
their eulerian code , written in cylindrical coordinates , imposed symmetry through the equatorial plane but not @xmath19-symmetry @xcite .
overall , their simulations produced nearly axisymmetric central remnants at late times . houser and centrella @xcite carried out additional sph simulations with @xmath39 and @xmath40 , and initial @xmath41 using improved initial models with @xmath42 particles .
as before , the @xmath15 case resulted in an almost axisymmetric central remnant and a correspondingly short burst of gravitational radiation . the runs with @xmath43 and
@xmath44 underwent additional episodes of spiral arm ejection , with the number of episodes increasing as @xmath45 decreased ; such behavior was also observed by williams and tohline @xcite .
this resulted in longer - lived nonaxisymmetric structure in the central remnants , accompanied by longer duration gravitational waveforms as the models grew stiffer ( i.e. , as @xmath45 was decreased ) .
note that the relatively small number of particles present in the sph simulations of centrella and collaborators , accompanied by the velocity dispersion in their initial models , may make it difficult for models with softer equations of state ( larger @xmath45 ) to maintain long - lived nonaxisymmetric structures .
new @xcite used an improved version of tohline s code @xcite to study the @xmath15 , @xmath46 case .
this code solves an energy equation and incorporates artificial viscosity to handle the shocks .
she added a routine to calculate the gravitational radiation in the quadrupole limit .
her simulation , which imposed both equatorial and @xmath25symmetries , produced a persistent bar structure and a long - duration gravitational waveform .
all of the studies mentioned above in this section started from initially axisymmetric models with the same radial distribution of specific angular momentum as in a maclaurin spheroid .
pickett , durisen , and davis @xcite studied the instabilities that result in an @xmath15 polytrope , when the angular momentum distribution is varied .
they used a ( different ) updated version of tohline s code with neither equatorial plane symmetry nor @xmath19-symmetry imposed ; all their evolutions were adiabatic . using the maclaurin rotation law , they evolved a model with initial @xmath47 to late times , and obtained a bar - shaped central remnant .
recently imamura , durisen , and pickett @xcite have performed additional adiabatic simulations of dynamical instabilities in @xmath15 and @xmath48 polytropes with the maclaurin rotation law , using the same hydrodynamics code used in @xcite .
they focused on comparing the early phases of nonlinear mode growth in their runs with the predictions of quasi - linear approximations .
their high resolution simulations of @xmath15 polytropes with @xmath49 and @xmath50 both resulted in bar - like endstates .
the properties and outcomes of the long duration bar mode runs with @xmath10 and @xmath34 are summarized in table [ long_runs ] for convenience .
all of the times reported in table [ long_runs ] are given in units of @xmath51 , where @xmath51 is defined as one central initial rotation period ( cirp ) .
when surveying the information catalogued in table i , one should keep in mind that the identified `` final '' state has been reported at different evolutionary times in the various references . as this table emphasizes , over the fifteen years that have passed since the original durisen et al .
comparison paper @xcite , there remain significant quantitative differences among the results of various published simulations of the bar mode instability . in particular , as indicated by our comments under the `` remarks '' column , these previous simulations do not clearly indicate whether or not the end product of the evolution should be a central , steady - state structure that has a bar - like geometry .
the new simulations of the dynamical bar instability presented here begin with rotating spheroidal models above the maclaurin stability limit , @xmath52 , constructed in hydrostatic equilibrium . for fluids rotating about the @xmath53axis with angular velocity @xmath54 , where @xmath55 is the distance from the rotation axis , the equations of motion reduce to the equation of hydrostatic equilibrium , @xmath56 where @xmath57 is the centrifugal potential and
@xmath58 is a constant .
the gravitational potential @xmath59 is a solution to poisson s equation , @xmath60 the initial models for the runs discussed in this paper were constructed using hachisu s self - consistent field ( hscf ; @xcite ; see also @xcite ) technique , which is a grid - based iterative method . to facilitate treatment of the boundary conditions
, it uses the integral form of the hydrostatic equilibrium condition , eq .
( [ diff - hse ] ) .
this gives @xmath61 where @xmath62 is the enthalpy of the fluid and @xmath63 is a constant determined by the boundary conditions .
the models are computed on a uniformly - spaced @xmath64 grid .
the method requires an equation of state @xmath65 . for the polytropic relation in eq .
( [ poly ] ) , the enthalpy takes the form @xmath66 for purposes of comparison with earlier work , we follow bodenheimer and ostriker @xcite and adopt a specific angular momentum profile that is the same function of cylindrical mass as a maclaurin spheroid , namely , @xmath67\varpi^{-2 } , \label{omega}\ ] ] where @xmath68 is the total mass of the system , @xmath69 is the mass interior to cylindrical radius @xmath55 , the constant @xmath70 , and @xmath71 is the total angular momentum .
hence , the centrifugal potential is , @xmath72 ^ 2 \varpi^{-3 } d\varpi . \label{psi}\ ] ] because the angular velocity is assumed initially to be only a function of @xmath55 , lichtenstein s theorem implies that the configuration will have reflection symmetry through the equatorial plane @xcite .
the hscf method requires that two boundary points , @xmath73 and @xmath74 , on the surface of the model be selected @xcite . for spheroids ,
point @xmath73 is set along @xmath55 at the equatorial radius , @xmath75 , and point @xmath74 is set on the @xmath53axis at the polar radius , @xmath76 .
the axis ratio @xmath77 is given as an input parameter ; varying it produces equilibrium models with different values of @xmath7 .
points @xmath73 and @xmath74 set the boundary conditions for the solution of eq .
( [ hse ] ) . since @xmath78 , @xmath79 , and
therefore @xmath80 vanish on the surface of the polytropic fluid , we have [ all_bc ] @xmath81 @xmath82 once @xmath59 and @xmath83 are known , eqs .
( [ all_bc ] ) can be solved for the constants @xmath63 and @xmath84 .
the hscf iteration process begins with an initial guess for @xmath85 , which also specifies the mass enclosed within each cylindrical radius @xmath69 .
given @xmath78 , the gravitational potential @xmath86 is determined by solving poisson s equation , eq .
( [ poisson ] ) ; see ref.@xcite for details . given @xmath69 , the centrifugal potential @xmath87 is determined using eq .
( [ psi ] ) . then , @xmath63 and @xmath84 are found from the boundary conditions , eqs .
( [ all_bc ] ) , and the enthalpy @xmath80 is computed from eq .
( [ hse ] ) .
finally , a new density distribution is calculated from @xmath80 by inverting eq .
( [ enthalpy ] ) ; this is used as input for the next iteration cycle .
the process is repeated until fractional changes in @xmath63 and @xmath84 and the maximum fractional change in @xmath80 between two successive iteration steps are less than some threshold ( in this work , @xmath88 ) .
the virial error @xmath89 provides a measure of how well the energy is balanced , and thus is indicative of the quality of the resulting equilibrium configuration .
it is defined by @xcite @xmath90 where @xmath91 is the total kinetic energy , and @xmath92 is the volume of the model .
the @xmath89s for the models used here are @xmath93 .
the simulations presented in this paper were carried out using two hydrocodes that employ eulerian finite - differencing techniques to solve the equations of hydrodynamics coupled to newtonian gravity .
the @xmath94 ( drexel ) hydrocode is the same one that was used by smith , houser , and centrella @xcite in their studies of the bar instability , whereas the @xmath95 ( lsu ) hydrocode is the one that was used by new and tohline @xcite . in this section
we briefly describe these codes , highlighting differences between them that we believe to be most relevant to the analysis and discussion of our results .
further details on the @xmath94 and @xmath95 hydrocodes may be found in the appendix .
both 3-d hydrocodes are written on uniform grids in cylindrical coordinates @xmath18 .
the @xmath94 code assumes equatorial plane symmetry .
the @xmath95 hydrocode allows the use of both equatorial and @xmath19-symmetries , as discussed in sec .
[ prev_stud ] .
both codes handle the transport terms using similar monotonic advection schemes that are second - order accurate in space , and impose the same outflow boundary conditions on the edges of the grid .
the @xmath94 and @xmath95 codes both solve energy equations , using the perfect fluid relation of eq .
( [ gamma - law ] ) to calculate the gas pressure and artificial viscosity to handle shocks .
finally , both codes solve poisson s equation , eq .
( [ poisson ] ) , for the newtonian gravitational potential @xmath59 with boundary conditions on the edges of the grid specified in terms of spherical harmonics .
eulerian codes typically require that the mass density in a grid zone never be zero , and thus fill the `` vacuum '' regions with a fluid having some small density , @xmath96 . to facilitate the comparison of results in this paper , both codes impose essentially the same conditions in the `` vaccum '' regions .
the density is set to @xmath97 if the density drops below @xmath96 in a zone .
the specific internal energy is similarly limited by @xmath98 , where eqs .
( [ poly ] ) and ( [ gamma - law ] ) give @xmath99 and @xmath100 is the polytropic constant of the initial model . in addition
, the velocities in the low density zones must be limited to prevent them from becoming too large and thereby requiring very small timesteps through the courant criterion @xcite .
the velocities are limited when @xmath101 .
specifically , in cells where @xmath102 , @xmath103 and @xmath104 are set to the value @xmath105 , _ if _ they exceed @xmath106 .
here , @xmath106 is the globally maximum sound speed .
additionally , @xmath107 is set to zero in cells where @xmath102 and @xmath108 . here , @xmath109 , where @xmath110 is the central rotation speed of the initial model .
the codes do have a number of differences .
the most important of these is that , as discussed in the appendix , the hydrodynamical equations in the @xmath95 code are written in flux - conservative form whereas in the @xmath94 code they are not .
the accuracy of the @xmath95 code is second - order in both space and time @xcite .
however , while the @xmath94 code is spatially second - order accurate , the accuracy of its time evolution scheme is between first and second orders @xcite .
finally , the @xmath94 code was written in fortran 77 and optimized to run on cray vector computers such as the c90 and t90 in single - processor mode .
these machines use 64-bit words in single precision , which allows the use of very low densities in the vacuum region , typically @xmath111 .
the @xmath95 code was developed for the parallel maspar mp-1 computer , and was written in maspar fortran , which is maspar s version of fortran 90 .
the maspar computers use 32-bit words in single precision , and the @xmath95 code uses @xmath112 in the vacuum regions .
the initial axisymmetric equililbrium models used for the simulations presented in this paper were computed with the hscf method @xcite as described in sec .
[ init_mods ] . specifying an axis ratio @xmath113 , polytropic index @xmath15 , and the maclaurin rotation law , eq .
( [ psi ] ) , yields a model with @xmath114 .
models with two different resolutions were constructed for this work .
the lower resolution model was computed on a grid with @xmath115 radial and @xmath116 axial zones ; its equatorial radius extends out to zone @xmath117 and its polar radius to zone @xmath118 .
the higher resolution model was computed on a grid with @xmath119 ; its equatorial radius extending out to zone @xmath120 and its polar radius to zone @xmath121 .
( note that @xmath122 in radial zone @xmath123 and @xmath20 in vertical zone @xmath124 . )
the computations of both models required 21 iterations .
the @xmath125 model had a virial error @xmath126 , as defined in eq .
( [ ve ] ) ; for the @xmath127 model , @xmath128 .
density contours of this model in the @xmath129 plane are shown in fig .
[ init_den ] and plots of the angular velocity @xmath130 ( cf .
( [ omega ] ) ) and equatorial plane density profile @xmath131 are given in fig .
[ init_omega ] .
we have normalized the equatorial radius and central density to unity , @xmath132 and @xmath133 . to prepare the initial data for evolution with a hydrocode
, the 2-d model is swept around in the azimuthal direction to create a 3-d axisymmetric model .
random perturbations are then imposed on the density to trigger the instability when the evolutions are run .
following tdm @xcite , we set @xmath134 , \label{pert}\ ] ] where @xmath135 is the density calculated with the hscf code , @xmath136 , and @xmath137 is a random number .
the perturbed initial models were then evolved with the hydrocode .
note that the models were initially centered on the origin .
all calculations used equatorial plane symmetry .
two simulations were performed with the @xmath94 code .
model d1 had @xmath138 zones while model d2 had twice the angular resolution , @xmath139 ; neither model used @xmath19-symmetry .
models d1 and d2 both used the density @xmath111 in the `` vacuum '' regions .
three simulations were performed with the @xmath95 code .
model l1 was run without @xmath19-symmetry and used the same resolution as model d2 ; for comparison , model l2 was run with @xmath19-symmetry and @xmath140 and thus had the same angular zone size @xmath141 as model l1 .
finally , model l3 was run without @xmath19-symmetry and used twice the radial and axial resolution , @xmath142 .
the models run with the @xmath95 code all used @xmath112 .
some basic properties of these models are summarized in the first five columns of table [ model_props ] for convenience .
all the models give similar results for the initial growth of the instability and its development into the nonlinear regime .
an initially axisymmetric system develops bar - shaped structure as the @xmath1 mode grows to nonlinear amplitude ( cf .
[ modes ] , @xcite ) .
a pattern of trailing spiral arms is formed as mass and angular momentum are shed from the ends of the rotating bar .
after the bar reaches its maximum elongation , it recontracts somewhat towards a more axisymmetric shape .
all the runs displayed these basic characteristics , which are illustrated in fig .
[ contours ] using data from model l3 .
time is measured in units of @xmath143 , where @xmath144 is the dynamical time for a _ sphere _ of mass @xmath68 having the same radius as the initial equatorial radius of the model , @xmath145 .
significant differences in the models arise in the next stage of the evolution , shown in figs .
[ d - late ] and [ l - late ] .
since the models go unstable at slightly different times , they encounter the various phases of the instability at somewhat different intervals ; the frames in these figures are labeled with the time measured from the initial moment in each model
. frames ( a - c ) of fig .
[ d - late ] show that the central regions of model d1 appear to undergo a very slight re - expansion to a weak bar .
after another @xmath146 bar rotation periods this weak bar disappears , leaving behind a nearly axisymmetric remnant that is somewhat displaced from the center of the grid . in model d2 the central bar re - expands more strongly , although not to the extent of its previous maximum elongation .
after another @xmath146 bar rotation periods , the model evolves toward a nearly axisymmetric remnant , which is offset from the center of the grid ; see frames ( d - f ) of fig .
[ d - late ] .
similar behavior is seen in model l1 , although the second bar elongation phase is stronger and lasts somewhat longer ; representative density contours are shown in frames ( a - c ) of fig .
[ l - late ] .
the higher resolution model l3 , shown in frames ( d - f ) of fig .
[ l - late ] , also re - expanded to a fairly strong bar , which underwent two additional episodes of contraction and expansion .
the remnant in model l3 remained bar - like in shape for @xmath147 bar rotation periods , but eventually it also decayed and settled into a nearly axisymmetric remnant as it moved off the center of the grid .
in contrast , model l2 ( which was run with @xmath19-symmetry ) retained a fairly strong bar for many ( @xmath148 ) bar rotation periods .
this model was run for a longer time than any of the others and experienced @xmath149 episodes of contraction and expansion . at the end of the simulation , model
l2 still had a strong bar centered on the origin ; see frames ( g - i ) of fig .
[ l - late ] . as is illustrated in fig .
[ beta - plots ] , for all five model evolutions each episode of expansion and contraction of the bar is mirrored in the time - evolution of the stability parameter @xmath150 .
as the bar develops and expands , @xmath7 drops and reaches a local minimum when the bar is at its maximum elongation
. then @xmath7 rises to a local maximum as the central regions recontract .
this behavior occurs because @xmath151 .
hence , as the bar elongates its moment of inertia @xmath152 increases which , assuming its angular momentum is approximately conserved , reduces the rotational kinetic energy @xmath3 . each subsequent episode of bar re - expansion
can also be associated with a local minimum in the global parameter @xmath7 . as has been illustrated by figs .
[ d - late ] and [ l - late ] , in each simulation except model l2 , the `` final '' remnant appears to have moved off of the center of the grid . in each case
this displacement is associated with measurable motion of the center - of - mass of the system ( such center - of - mass motion was also seen in some of the simulations in @xcite ) .
[ cm ] shows the position of the system center - of - mass @xmath153^{1/2}$ ] as a function of time for each of the four affected models .
( note that @xmath154 at all times in all of our simulations since they all employ equatorial plane symmetry . ) in runs d1 and d2 , the center - of - mass begins to move rather abruptly at a time @xmath155 , then after moving to a location @xmath156 ( d1 ) and @xmath157 ( d2 ) , the center - of - mass motion slows considerably .
because both models have radial zones of size @xmath158 , this location corresponds to approximately @xmath159 and @xmath160 radial grid zones , respectively , or in terms of the initial equatorial radius of the model , @xmath161 and @xmath162 , respectively . for models l1 and l3
, the center - of - mass motion does not appear to level off , as seen in fig .
[ cm](b ) .
notice that the center - of - mass moves farthest from the center of the grid in model l1 , which has the same resolution as d2 , and that the onset of this motion is significantly delayed when the radial resolution is doubled in model l3 .
since these simulations all have newtonian gravitational fields and the systems are assumed to be isolated from the external environment , in each case the center - of - mass _ should _ remain fixed at the origin if the total mass of the system remains confined within the boundaries of the computational grid ( _ i.e. , _ if the system remains isolated from its environment ) and if the equations governing the dynamics of an isolated system are being properly integrated forward in time . although some ( @xmath163 ) mass ( and associated linear and angular momentum ) does flow radially off the grid , this mass loss does not appear to be large or asymmetric enough to account for the center - of - mass motion .
we have verified this by running a simulation ( not presented in detail in this manuscript ) with the same resolution as model l1 but with an expanded grid ( @xmath164 ) .
the center of mass motion of the model evolved on the expanded grid was not significantly different from that present in the run on the smaller grid ( @xmath165 ) ; the mass lost from the expanded grid was less than @xmath166 . instead
, we suspect that the center of mass motion arises from numerical errors . in `` flux - conservative '' finite - differencing schemes , such as the one employed here in the @xmath95 code
( see the appendix ) , the advection term is handled in such a way that , for example , momentum and mass are guaranteed to be globally conserved _ if _ the dynamical equations contain no source or sink terms .
however , such algorithms are not explicitly designed to preserve the position of the center - of - mass of the system and source terms due to gradients in the pressure and gravitational potential do naturally arise ( see , for example , the right - hand - sides of equations ( b1)-(b3 ) ) .
so discrepancies that inevitably will arise between the finite - difference representation of the dynamical equations and their exact differential counterparts can lead to center - of - mass motion that is unphysical .
( an exception to this is the case where @xmath19-symmetry is explicitly enforced .
with @xmath19-symmetry imposed , odd azimuthal modes can not grow and , in particular , no center - of - mass motion can develop . )
once such motion begins in either the @xmath94 or @xmath95 code , it evidently has a tendency to amplify rather than to damp .
it is also instructive to examine the conservation of total angular momentum @xmath71 .
all models show some degree of angular momentum loss as the bar expands into the `` vacuum '' regions of the grid . to quantify this
, we consider the loss of @xmath71 in each model at the time that @xmath7 reaches its first local minimum , and thus at the time the bar reaches its point of maximum expansion ; see , for example , fig .
[ beta - plots ] .
( shortly after this time , the models typically lose mass as the spiral arms expand and material flows off the grid . ) in general , table [ model_props ] shows that models run with the @xmath94 code lose more angular momentum during this stage than those run with the @xmath95 code .
additonal tests with the @xmath94 code showed that this loss of angular momentum increases as @xmath96 increases @xcite .
overall , we attribute the better angular momentum conservation obtained with the @xmath95 code to the fact that it is written in an explicitly flux - conservative form whereas the @xmath94 code is not ; see the appendix .
we can quantify the development of the dynamical instability shown in the previous section by examining various fourier components in the density distribution .
to this end , the density in a ring of fixed @xmath55 and @xmath167 can be written using the complex azimuthal fourier decomposition @xmath168 where the amplitudes @xmath169 of the various components @xmath16 are defined by @xcite @xmath170 we shall also find it useful to define the normalized amplitude @xmath171 where @xmath172 is the mean density in the ring .
the phase angle of the @xmath173 component is defined by @xmath174 .
\label{phim_def}\ ] ] when nonaxisymmetric structure propagating in the @xmath175-direction develops into a global mode we can write the phase angle as @xmath176 where @xmath177 is called the eigenfrequency of the @xmath173 mode .
the pattern speed is then @xmath178 and the pattern period is @xmath179 @xcite . for the @xmath1 mode , @xmath180 is twice the angular velocity of the bar and , hence , the rotation period of the bar is @xmath181 . in practice ,
we implement eq .
( [ cm ] ) in the codes by summing up the contributions over the azimuthal coordinate ( @xmath182 ) in rings of width @xmath183 centered on the origin at various values of @xmath55 in the equatorial plane @xmath20 . by examining the amplitudes of the fourier components @xmath184 at various distances from the rotation axis
, we can determine when global modes arise in the models .
since the rings are always centered on the origin , care must be taken when interpreting the results once the center - of - mass motion becomes significant .
figure [ modes - graph ] shows the growth of the amplitudes @xmath184 for the first four fourier components , @xmath185 .
these amplitudes were calculated in the equatorial plane in a ring of radius @xmath186 , which corresponds to radial zone @xmath187 , for models d1 , d2 , l1 , and l2 . for model l3 ,
the amplitudes were calculated in a ring of radius @xmath188 , which is radial zone @xmath189 .
similar plots were obtained for rings at other values of @xmath55 within the central regions , indicating that these are all global modes .
notice that the exponential growth of the @xmath1 bar mode ( thick solid line ) dominates all evolutions , as expected from visual inspection of the density contours shown in figs .
[ contours ] - [ l - late ] .
in addition , all models show a clear @xmath190 mode ( thin solid line ) that begins growing exponentially once the bar mode is well into its exponential growth regime .
the @xmath1 and @xmath190 modes both reach their peak amplitudes at about the same time in all runs .
the odd modes appear at later times in all models except l2 , in which @xmath19-symmetry was imposed ( which prevents the development of odd modes ; see sec .
[ prev_stud ] ) .
the @xmath191 , or translational , mode ( dotted line ) begins growing after the @xmath190 mode . comparing figs .
[ cm ] and [ modes - graph ] shows that the @xmath191 mode grows as the center of mass motion increases , as expected .
the @xmath192 , or pear , mode is shown by the dashed line . in models l1 and l3 , the @xmath192 mode is the last one to grow . in models d1 and d2 , the @xmath192 mode grows at early times , but then lags behind the @xmath191 mode . as mentioned in sec .
[ prev_stud ] , analysis of maclaurin spheriods suggests that models near the dynamical stability limit are only unstable to the @xmath1 mode ( higher order , even harmonics of the @xmath1 mode may arise in the subsequent evolution ) .
models near this stability limit are not physically susceptible to the growth of odd modes .
this has been confirmed again recently by the perturbative analysis of toman et al .
( tipd ; @xcite ) .
in fact , tipd demonstrate that @xmath15 polytropes with the maclaurin rotation law [ eq . ( [ psi ] ]
are only unstable to the @xmath192 mode when @xmath7 is @xmath193 . in general all our models
( for which @xmath194 , initially ) conform to this expectation , with the odd modes ( which here are numerical artifacts ) growing earlier in the models with lower resolution .
in the l3 simulation , which has the highest resolution , the @xmath191 and @xmath192 modes develop at late times and are cleanly separated from the @xmath1 and @xmath190 modes .
the growth rate @xmath195 for the @xmath1 and @xmath190 modes can be determined by fitting a straight line through the data points in fig .
[ modes - graph ] during the time intervals in which @xmath196 is growing linearly with time .
we find that all models have approximately the same growth rate for these modes , as seen in table [ bar_props ] .
to find @xmath177 for these modes we plot @xmath197 versus time and use a trigonometric fitting routine to get @xmath198 ; cf .
the function @xmath197 was used because @xmath198 itself is multi - valued due to the @xmath199 in eq .
( [ phim_def ] ) .
the fit was performed over the same time interval , chosen `` by eye '' , used to calculate the corresponding mode growth rate .
table [ bar_props ] shows that models d1 and d2 have smaller eigenfrequencies and pattern speeds than those run on the @xmath95 code .
however , in all models we find that the pattern speeds for the @xmath1 and @xmath190 modes are nearly the same , @xmath200 .
this confirms that the @xmath190 mode is a harmonic of the bar mode and not an independent mode , as mentioned above and first pointed out by williams and tohline @xcite .
it is interesting to compare the data from our runs with the results of tipd , who used a perturbative , linearized eulerian scheme to calculate mode growth rates and eigenfrequencies of differentially rotating polytropes .
as discussed by tipd , this method produces much more precise results than traditional lagrangian normal mode analysis @xcite ( including the tensor virial approach @xcite , which has been demonstrated to be inappropriate for differentially rotating polytropes ) .
tipd calculated the bar mode growth rate and eigenfrequency for an axisymmetric @xmath15 polytrope with the maclaurin rotation law , eq .
( [ psi ] ) , and @xmath201 using their perturbative eulerian method .
they found @xmath202 and @xmath203 , where we have converted from their units .
tipd state that the uncertainties in their results are on the order of a few percent ( excluding any systematic errors that may be present ) .
comparison with the data in table [ bar_props ] shows that the numerical models all have growth rates in good agreement with that of tipd .
the eigenfrequencies @xmath180 for the models run with the @xmath95 code are also in very good agreement with the tipd results , while those for d1 and d2 are @xmath204 smaller .
the reason for the discrepancies in the eigenfrequencies of the @xmath205 code runs is unknown .
we use the quadrupole approximation , which is valid for nearly newtonian sources @xcite , to calculate the gravitational radiation produced in our models .
since the gravitational field in both codes is strictly newtonian , we compute only the gravitational radiation produced and do not include the effects of radiation reaction .
the reduced or traceless quadrupole moment of the source is @xmath206 where @xmath207 are spatial indices and @xmath208 is the distance to the source . for an observer located on the axis at @xmath209 of a spherical coordinate system with its origin located at the center of mass of the source , the amplitude of the gravitational waves for the two polarization states becomes simply @xcite @xmath210 where an overdot indicates a time derivative @xmath211 . a straightforward application of eqs .
( [ iij ] ) - ( [ hplus - axis ] ) in an eulerian code involves calculating @xmath212 directly by summing over the grid , and then taking the time derivatives numerically .
however , such successive application of numerical time derivatives generally introduces spurious noise into the waveforms , especially when the timestep is allowed to change from cycle to cycle . to reduce this problem ,
we have used the partially integrated versions of the standard quadrupole formula developed by finn and evans @xcite .
the @xmath94 code uses the first moment of momentum formula , qf1 , which allows the calculation of @xmath213 directly from quantities available in the code @xcite .
this gives @xmath214d^3r , \label{qf1}\ ] ] where @xmath215 of course , another time derivative is still required to obtain @xmath216 .
when this is taken numerically , the resulting waveform amplitudes @xmath217 and @xmath218 can still be dominated by noise . to cure this problem , we pass @xmath216 through a filter to smooth it before calculating the waveforms ; see ref .
@xcite for details .
the @xmath95 code does not use numerical time derivatives to compute @xmath216 . instead
, the equations of motion are used in conjuction with eq .
( [ qf1 ] ) to compute @xmath216 directly from quantities available in the code .
this gives @xmath219 d^3x .
\label{stress - gw}\end{aligned}\ ] ] here , the summation convention is implied on repeated up and down indices and the `` a. v. terms '' contain contributions to the stress tensor from the artificial viscosity ; see refs . @xcite and @xcite for details .
the expression for @xmath220 in eq .
( [ stress - gw ] ) yields smooth waveforms which do not require filtering .
note that both the @xmath95 and @xmath94 codes compute @xmath216 with respect to the origin of the coordinate system .
hence when the center - of - mass motion becomes significant , the waveforms computed are no longer precisely correct .
however , the center - of - mass motion itself is a numerical artifact ; thus , upon its development the simulation as a whole becomes physically unreliable .
figure [ h+_fig ] shows the gravitational waveform @xmath221 given in eq .
( [ hplus - axis ] ) as a function of time for all models . at early times , the waveforms are all similar , as the initial expansion of the bar gives rise to gravitational radiation . comparing fig . [ h+_fig ] with fig .
[ beta - plots ] , we see that the maximum amplitude of @xmath221 occurs at the same time as the first local minimim in @xmath7 , which marks the maximum expansion of the bar .
the amplitude of @xmath221 then dips as the central bar recontracts and @xmath7 rises again to a local maximum . when the bar re - expands , the amplitude of @xmath221 rises again ; cf.figs .
[ d - late ] and [ l - late ] . in models d1 ,
d2 , and l1 , the central remnant soon becomes nearly axisymmetric and @xmath221 decays rapidly . in l3 ,
the bar persists for @xmath222 bar rotation periods and undergoes two additional episodes of expansion and contraction , producing a longer - lived gravitational wave signal .
model l2 maintains a strong bar with multiple expansions and contractions , and thus a strong gravitational wave signal , throughout its evolution . in fig .
[ hnorm_fig ] we plot the quantity @xmath223 note that the absolute scale of the @xmath224 axis in this figure is not labeled because the curves for the three runs have been shifted horizontally in order to line up their initial peaks .
all models show a strong initial peak in @xmath225 that coincides with the maximum expansion of the bar .
the secondary peaks in @xmath225 correspond to the secondary expansions of the bar and the additional local minima in @xmath7 shown in fig .
[ beta - plots ] .
a plot of @xmath225 is instructive because its time variation is not complicated by the periodic rotation of the bar .
thus @xmath226 reflects how the `` mean '' properties of the system ( e.g. , the moment of inertia ) change with time .
the @xmath225 curve should be perfectly flat once ( and _ if _ ) the remnant settles down into a steady - state structure ; any slight downward slope will provide a measure of long - term secular changes .
ultimately , every hydrodynamics code should produce the same , correct , plot of @xmath226 for this simulation .
that is , the amplitude and frequencies present in @xmath226 should be quantitatively reproduceable . fig .
[ hnorm_fig ] thus shows to what degree our ( three best ) simulations are converging toward the same answer and thereby begins to establish the nature of the `` correct '' result .
we have carried out simulations of dynamical instability in a rapidly rotating @xmath15 polytrope using two different eulerian hydrocodes and several different resolutions .
the rapidly rotating polytropic initial models used were constructed with the maclaurin rotation law and had a ratio of kinetic to gravitational energy @xmath227 .
all models evolved by both codes agree on the following basic properties of the early nonlinear development of the instability . _
( i ) _ the @xmath1 mode dominates the evolutions , producing a central rotating bar which sheds mass and angular momentum at its ends to produce a spiral arm pattern .
once the bar reaches its point of maximum elongation , it contracts and then re - expands .
_ ( ii ) _ the @xmath190 mode is the next one to reach nonlinear amplitudes . _
( iii ) _ the growth rates for the @xmath1 and @xmath190 modes are @xmath228 and @xmath229 , respectively .
the pattern speeds @xmath200 , indicating that the @xmath190 mode is a harmonic of the bar mode . _
( iv ) _ the instability produces a gravitational wave signal with maximum amplitude [ @xmath230 $ ] @xmath231 for an observer on the axis at @xmath232 , @xmath233 of a spherical coordinate system centered on the source
. the models also exhibit some differences .
in particular , simulations run with the @xmath94 code have smaller values of the eigenfrequencies @xmath180 and @xmath234 , show weaker secondary bars , and lose more angular momentum during the initial bar expansion than those run with the @xmath95 code .
it appears that most of these differences can be attributed to the lower order time differencing and the lack of consistent flux - conservative differencing in the @xmath94 code .
overall the simulations presented here , and those carried out by previous authors , agree on the qualitative nature and many quantitative aspects of the initial development of the bar structure .
however , as detailed in
i and ii , such agreement has not been universally present among simulations that follow the long duration evolution of this instability .
such lengthy evolutions are nontrivial for hydrodynamics codes as they may allow any numerical inaccuracies present to grow to the point where they signficantly influence the simulations . as described below , the late growth of odd modes in our bar mode evolutions is an example of the numerical difficulties that can arise in extended simulations .
linear analysis indicates that the @xmath15 , @xmath235 models used in our simulations are initially unstable to the @xmath1 bar mode only , and not to any odd modes @xcite ( even harmonics of the m=2 mode may subsequently develop ) .
thus the odd modes that develop in all but one ( l2 ) of the simulations presented here are numerical artifacts arising from shortcomings in the finite - difference techniques utilized in the @xmath94 and @xmath95 hydrocodes ( see sec .
iv.b and also @xcite ) .
once these artificial odd modes reach nonlinear amplitudes , the physical reliability of the simulations is degraded .
in particular , the growth of the @xmath191 mode is tied to the development of center - of - mass motion in the simulations we ran without @xmath19-symmetry .
as the center - of - mass moves away from the origin of the cylindrical grid , the finite - differencing of the curvilinear form of the hydrodynamics equations becomes asymmetric .
this causes the accuracy of the evolution to deteriorate .
the growth of the center - of - mass motion appears directly related to the decay of the bar - like structure of the system ; the bar decays when the center - of - mass motion becomes significant .
because the center - of - mass motion is unphysical , we believe the decay of the bar structure is unphysical as well .
this conclusion is substantiated by run l3 , which had twice the radial ( and axial ) resolution of run l1 .
the onset of the spurious center of mass motion was significantly delayed in l3 and the central remnant maintained its bar - like structure for a correspondingly longer time .
the growth of odd modes was also delayed .
thus as the resolution is increased , the l code evolutions converge towards the predictions of linear analysis .
( unfortunately , we could not repeat this experiment with the @xmath94 code due to a lack of computational resources . )
thus it is our belief that a simulation of this instability that did not develop the nonphysical center of mass motion ( e.g. , one performed with very fine radial resolution ) , would produce a long - lived nonaxisymmetric structure . recall that this is the result of the model l2 simulation , which was run with @xmath19-symmetry .
that symmetry condition prevents the growth of odd modes and thus prohibits the development of center of mass motion .
hence , overall l2 is the most physically relevant of the simulations presented here .
we believe our results demonstrate that the physically acurate outcome of the rotational instability in the object studied here , is a persistent bar with an accompanying long - lived gravitational wave signal . this dynamically stable configuration can be viewed as a compressible analog of a riemann ellipsoid ; efforts to understand
the detailed structural properties of such configurations are underway @xcite .
the nonaxisymmetric structure of the remnant will decay on a secular timescale due to viscous dissipation and/or gravitational radiation emission .
the gravitational radiation timescale is likely to be shorter than the viscous timescale for sufficiently compact objects @xcite .
note that @xmath7 will also decrease as a result of this secular evolution . the system will continue to evolve until it reaches a configuration that is secularily stable . a number of factors including the presence of an envelope surrounding the rotating object , the variation of the rotation law and the equation of state , and the influence of general relativity could potentially affect the outcome of the instability .
such effects should be the subject of future study .
we thank l. lowe for her considerable help with data analysis and producing the figures for this paper .
jmc also thanks l. lowe , h. luo , and c. hempel for assistance with the @xmath94 hydrocode .
we appreciate stimulating conversations with john cazes , patrick motl , and brian pickett .
this work was supported in part by nsf grants phy-9208914 and phy-9722109 at drexel , and nasa grant nag5 - 8497 and nsf grant ast-9528424 at lsu .
this research was also supported in part by grant number phu6php from the pittsburgh supercomputing center ( psc ) , which is supported by several federal agencies , the commonwealth of pennsylvania and private industry ; and by nsf cooperative agreement aci-9619020 through computing resources provided by the national partnership for advanced computational infrastructure ( npaci ) at the san diego supercomputer center ( sdsc ) .
the numerical simulations using the @xmath94 code were run at sdsc and psc , and those using the @xmath95 code were run at nasa s hpcc / gsfc .
this work performed under auspices of the u.s .
department of energy by los alamos national laboratory under contract w-7405-eng-36 .
the @xmath94 and @xmath95 codes solve the equations of hydrodynamics , which govern the structure and evolution of a fluid , in cylindrical coordinates .
these equations include the continuity equation , @xmath236 the three components of euler s equation , @xmath237 @xmath238 @xmath239 and poisson s equation , eq .
( [ poisson ] ) .
here , @xmath240 is the fluid velocity with components @xmath241 in the @xmath18 directions .
the quantities @xmath242 , @xmath243 , @xmath244 are the radial , vertical , and angular momentum densities , respectively .
the codes solve slightly different forms of the energy equation .
the @xmath94 code evolves the specific internal energy @xmath37 : @xmath245 the @xmath95 code evolves the internal energy density @xmath246 : @xmath247 in both codes , the pressure @xmath79 is obtained from the perfect fluid equation of state , eq .
( [ gamma - law ] ) .
the @xmath94 hydrocode was developed by clancy , smith , and centrella @xcite .
it is written in cylindrical coordinates @xmath18 with reflection symmetry through the equatorial plane @xmath20 .
the original version allowed nonuniform radial and axial grids and was used to carry out the eulerian runs in ref .
@xcite ; for the simulations described in this paper , the code was restricted to uniform grids .
this code was written in fortran 77 and optimized for cray vector computers ; it currently runs on the cray t90 .
the actual form of the hydrodynamics equations ( [ cont])-([eq_a ] ) used in the @xmath94 code is given in ref .
@xcite , with the exception that eq .
( [ eq_a ] ) takes the form @xmath248 where @xmath249 is the specific angular momentum .
note that the equations the @xmath94 code solves are not written in flux - conservative form . in the discrete form of these equations ,
the scalar quantities @xmath78 , @xmath37 , @xmath59 , and @xmath79 are defined at cell centers and at integral timesteps .
the @xmath94 code actually evolves the velocities , which are defined on the faces between cells and at half - integral timesteps , located halfway between the integral timesteps .
the velocities are face - centered in the coordinate along which they are directed ; for example , @xmath104 is defined at the center of the grid zone faces normal to the @xmath53axis @xcite .
the @xmath94 code uses operator splitting to evolve the discrete versions of the hydrodynamical equations , eqs .
( [ cont ] ) - ( [ d_energy ] ) , forward in time @xcite .
the accuracy of this time integration method is between first and second orders .
the source step is carried out first .
this begins by holding @xmath78 constant and updating the velocities due to the pressure gradient , gravitational force , and centrifugal force terms in eqs .
( [ eq_s ] ) - ( [ eq_a ] ) using centered differences ; note that in the source step we advance the azimuthal velocity component @xmath107 instead of the specific angular momentum @xmath250 in eq .
( [ euler-3 ] ) . using these updated values ,
the artificial viscosity terms are applied to advance the velocities and @xmath37 .
these new values are then used to update the energy due to the compressional or `` pdv '' terms .
we next carry out the transport step to evolve @xmath78 , @xmath251 , @xmath103 , @xmath104 , and @xmath250 due to the advection of fluid from one cell to the next .
the transport step consists of three advection sweeps , one in each of the three coordinate directions .
we use a monotonic advection scheme developed by leblanc @xcite that is second - order accurate in space to calculate the fluxes in each direction . on each cycle
, we vary the order in which the advection sweeps are carried out to avoid setting up a preference for any one direction ; the order changes on each successive cycle as all six permutations are exhausted . on each sweep ,
the same mass flux used to advect the density in eq .
( [ cont ] ) is employed to advect @xmath103 , @xmath104 , and @xmath250 in eqs .
( [ eq_s ] ) - ( [ eq_a ] ) @xcite . during the transport step ,
the density is held constant ; thus , @xmath252 is written as @xmath253 in eq .
( [ eq_s ] ) , and similarly for eqs .
( [ eq_t ] ) - ( [ d_energy ] ) .
after updating the advection terms on each cycle , a momentum conservation is applied with the new density to update the velocities .
the equation of state , eq .
( [ gamma - law ] ) , is then used to calculate a new value of the pressure .
once the hydrodynamical equations have been advanced , the newtonian gravitational potential @xmath59 is calculated by solving poisson s equation , eq .
( [ poisson ] ) , using the updated density .
the boundary conditions at the edge of the grid are specified using a spherical multipole expansion .
the discretization yields a large , sparse , banded matrix equation which we solve using a preconditioned conjugate gradient method with diagonal scaling @xcite .
the @xmath254 hydrocode was originally developed by tohline @xcite , and has been refined and updated with collaborators and students .
the modern version of the code is fully second order accurate in both space and time @xcite . the parallel version of the code that we use here was originally developed for the maspar
mp-1 computer and was written in maspar fortran , which is maspar s version of fortran 90 ; see @xcite .
the @xmath254 code uses uniformly spaced grids in cylindrical coordinates @xmath18 .
the code allows the use of reflection symmetry through the equatorial plane @xmath20 and @xmath25symmetry in the azimuthal direction ; cf . sec .
[ prev_stud ] .
the fluid equations , eqs .
( [ cont])-([eq_a ] ) and ( [ l_energy ] ) , are written in flux - conservative form @xcite .
when they are discretized on the uniform cylindrical grid , the density @xmath78 , the angular momentum density @xmath255 , and the gravitational potential @xmath59 are defined at cell centers .
the radial and vertical velocities @xmath256 and
momentum densities @xmath257 are defined at cell vertices or nodes .
the source terms on the right - hand sides of eqs .
( [ eq_s ] ) - ( [ eq_a ] ) are approximated using standard second - order centered differences .
the flux or divergence terms are written as a summation over the six faces of a cylindrical grid zone @xcite , @xmath258 here @xmath92 is the volume of the cylindrical grid cell , @xmath259 is the area of a particular face , and @xmath260 is the product of the quantity @xmath261 and the corresponding velocity component at the face @xmath262 ( i.e. , @xmath263 is the velocity normal to the @xmath264 face ) .
these terms are updated using a monotonic interpolation scheme developed by van leer @xcite that is second - order accurate in space . when the system is evolved forward in time , the physical variables @xmath261 are updated by applying the source terms and the flux terms in different steps .
second - order accuracy in time is obtained via a lax - wendroff scheme that uses velocity values in the flux terms ( [ flux - terms ] ) that are centered at time @xmath265 @xcite . to accomplish this ,
the source terms are applied for a half timestep @xmath266 and the updated values @xmath267 are saved .
the flux terms are then applied for @xmath266 with the updated values @xmath268 to obtain velocities at time @xmath265 . with these new velocities
, fluxing is performed for a full timestep on the saved quantities @xmath267 .
an additional half timestep of sourcing is then performed .
currently at los alamos national laboratory , los alamos , nm 87545 see , e.g. , articles in _ gravitational wave detection _ , eds . k. tsubono , m .- k .
fugimoto , and k. kuroda ( universal academic press , tokyo , 1996 ) for recent articles describing these interferometers .
.properties of long - duration simulations of the bar mode instability .
@xmath270 is the time at which the bar reaches its maximum elongation and @xmath271 is the end of the simulation .
the length of time the bar persists is @xmath272 . in this table ,
time is measured in units of @xmath51 , where @xmath273 = 1 central initial rotation period ( cirp ) . [ cols="^,^,^,^,^,^,^,^,^ " , ] | compact astrophysical objects that rotate rapidly may encounter the dynamical `` bar instability . ''
the bar - like deformation induced by this rotational instability causes the object to become a potentially strong source of gravitational radiation .
we have carried out a set of long - duration simulations of the bar instability with two eulerian hydrodynamics codes .
our results indicate that the remnant of this instability is a persistent bar - like structure that emits a long - lived gravitational radiation signal . |
codes , proposed by arikan @xcite , are proved to achieve the symmetric capacities of the binary - input discrete memoryless channels ( b - dmcs ) .
this capacity - achieving code family is based on a technique called channel polarization . by performing the channel splitting and channel combining operations on independent copies of a given b - dmc ,
a set of synthesized binary - input channels can be obtained .
let @xmath0 denote the symmetric capacity of a b - dmc @xmath1 .
it is proved in @xcite that : with @xmath2 uses of @xmath1 , @xmath3 , when @xmath4 is large enough , it is possible to construct @xmath4 synthesized channels such that @xmath5 of them are completely unreliable and @xmath6 of them are noiseless . by transmitting free bits ( called information bits ) over the noiseless channels and transmitting a sequence of fixed bits ( called frozen bits ) over the others , polar codes can achieve the symmetric capacity under a successive cancellation ( sc ) decoder with both encoding and decoding complexity @xmath7 . in @xcite , it is proved that the block error probability of polar code under sc decoding satisfies @xmath8 for any @xmath9 when code length @xmath4 is large enough and code rate @xmath10 .
furthermore , it was shown by korada et al .
@xcite that the error exponent @xmath11 can be arbitrarily close to 1 for large @xmath4 with a general construction using larger kernel matrices than the @xmath12 matrix proposed by arikan .
to construct polar codes , the channel reliabilities can be calculated efficiently using bhattacharyya parameters for binary - input erasure channels ( becs ) @xcite .
but for channels other than becs , density evolution is required @xcite .
more practical methods for calculating the channel reliabilities are discussed in @xcite and @xcite , and these techniques are extended to input channels @xcite .
the channel polarization phenomenon is believed to be universal in many other applications , such as parallel communications @xcite @xcite , coded modulation systems @xcite , multiple access communications @xcite @xcite , source coding @xcite @xcite , information secrecy @xcite @xcite and other settings .
although polar codes have astonishing asymptotic performance , the finite - length performance of polar code under sc decoding is not satisfying . with the factor graph representation of polar codes , a belief propagation
( bp ) decoder is introduced by arian in @xcite . and in @xcite , hussami et .
al . show that bp decoder significantly can outperform sc decoder , and point out that , for channels other than bec , the schedule of message passing in bp plays an important role .
and they also show that the performance of bp decoder can be further improved by utilization of overcomplete factor graph representations over bec . unfortunately ,
due to the sensitivity of bp decoder to message - passing schedule , this is not realized on other channels . in @xcite a linear programming ( lp ) decoder is introduced without any schedule , and also , by using the overcomplete representations can improve the performance of lp decoder . but lp decoder can not work on channels other than bec .
maximum likelihood ( ml ) decoders are implemented via viterbi and bcjr algorithms on the codeword trellis of polar codes @xcite , but because of their high complexity , they can only work on very short code blocks .
successive cancellation ( sc ) decoding of polar codes essentially shares the same idea with the recursive decoding of rm codes @xcite . like the recursive decoders
can be improved by using a list @xcite or a stack @xcite , sc can also be enhanced in the same way . as an improved version of sc , successive cancellation list ( scl ) decoding algorithm
is introduced to approach the performance of maximum likelihood ( ml ) decoder with an acceptable complexity @xcite , @xcite . and
later , an other improved decoding algorithm based on sc named successive cancellation stack ( scs ) decoding algorithm is proposed whose computational complexity will decrease with the increasing of signal - to - noise ratio ( snr ) and can be very close to that of the sc decoding in the high snr regime @xcite .
compared with scl , scs will have a much lower computational complexity .
but it comes at the price of much larger space complexity and it will fail to work when the stack is too small . combining the ideas of scl and scs ,
a new decoding algorithm named successive cancellation hybrid ( sch ) is proposed in this paper , and it can achieve a better trade - off between computational complexity and space complexity . in this paper , all the three improved sc decoding algorithms , scl , scs and sch ,
are described under a unified manner of a path searching procedure on the code tree .
further , to reduce the complexity , a pruning technique is proposed to avoid unnecessary path searching operations .
the remainder of the paper is organized as follows .
section [ section_preliminaries ] reviews the basics of polar coding and describes the sc decoding algorithm as a path searching procedure on a code tree using _ a posteriori _ probabilities ( apps ) as metrics .
then the three improved successive cancellation ( isc ) decoding algorithms and the pruning technique are introduced in section [ section_isc ] .
section [ section_simulations ] provides the performance and complexity analysis based on the simulation results of polar codes under isc decoders with different parameters .
finally , section [ section_conclusion ] concludes the paper .
in this paper , we use blackboard bold letters , such as @xmath13 and @xmath14 , to denote sets , and use @xmath15 to denote the number of elements in @xmath13 .
we write the cartesian product of @xmath13 and @xmath14 as @xmath16 , and write the @xmath17-th cartesian power of @xmath13 as @xmath18 .
we use calligraphic characters , such as @xmath19 to denote a event . and let @xmath20 denote the event that @xmath19 is not happened .
we use notation @xmath21 to denote a @xmath4-dimension vector @xmath22 and @xmath23 to denote a subvector @xmath24 of @xmath21 , @xmath25 .
particularly when @xmath26 , @xmath23 is a vector with no elements in it and the empty vector is denoted by @xmath27 .
we write @xmath28 to denote the subvector of @xmath21 with odd indices ( @xmath29 ; @xmath30 is odd ) .
similarly , we write @xmath31 to denote the subvector of @xmath21 with even indices ( @xmath29 ; @xmath30 is even ) . for example , for @xmath32 , @xmath33 , @xmath34 and @xmath35 .
further , given a index set @xmath36 , @xmath37 denote the subvector of @xmath21 which consists of @xmath38s with @xmath39 .
only square matrices are involved in this paper , and they are denoted by bold letters .
the subscript of a matrix indicates its size , e.g. @xmath40 represents a @xmath41 matrix @xmath42 .
we write the kronecker product of two matrices @xmath42 and @xmath43 as @xmath44 , and write the @xmath17-th kronecker power of @xmath42 as @xmath45 .
let @xmath46 denote a b - dmc with input alphabet @xmath13 and output alphabet @xmath47 . since the input is binary , @xmath48 .
the channel transition probabilities are @xmath49 , @xmath50 , @xmath51 . for code length @xmath52 , @xmath53 , and
information length @xmath54 , i.e. code rate @xmath55 , the polar coding over @xmath1 proposed by arikan can be described as follows : after channel combining and splitting operations on @xmath4 independent uses of @xmath1 , we get @xmath4 successive uses of synthesized binary input channels @xmath56 , @xmath57 , with transition probabilities @xmath58 where @xmath59 and the source block @xmath60 are supposed to be uniformly distributed in @xmath61 .
let @xmath62 denote the probability of maximum - likelihood ( ml ) decision error of one transmission on @xmath63 , @xmath64 where @xmath65 , @xmath66 and the indicator function @xmath67 and @xmath68 is the module-@xmath69 addition .
the reliabilities of polarized channels @xmath70 are usually measured by ( [ equ_pe ] ) , and can be evaluated using bhattacharyya parameters @xcite for binary erasure channels ( becs ) or density evolution @xcite for other channels . to transmit a binary message block of @xmath54 bits , the @xmath54 most reliable polarized channels @xmath70 with indices
@xmath39 are picked out for carrying these information bits ; and transmit a fixed bit sequence called frozen bits over the others .
the index set @xmath71 is called information set and @xmath72 . and the complement set of @xmath36 is called frozen set and is denoted by @xmath73 . alternatively , the polar coding can be described as follow : a binary source block @xmath60 which consists of @xmath54 information bits and @xmath74 frozen bits is mapped to a code block @xmath75 via @xmath76 . the matrix @xmath77 , where @xmath78 $ ] and @xmath79 is the bit - reversal permutation matrix .
the binary channel @xmath80 are then sent into channels which are obtained by @xmath4 independent uses of @xmath1 .
as mentioned in @xcite , polar codes can be decoded by successive cancellation ( sc ) decoding algorithm .
let @xmath81 denote the estimate of the source block @xmath60 .
after receiving @xmath82 , the bits @xmath83 are determined successively with index @xmath84 from @xmath85 to @xmath4 in the following way : @xmath86 where @xmath87 the block error rate ( bler ) of this sc decoding is upper bounded by @xmath88 this successive decoding can be represented as a path searching process on a code tree . for a polar code with code length @xmath4 , the corresponding code tree
@xmath89 is a full binary tree .
more specifically , @xmath89 can be represented as a 2-tuple @xmath90 where @xmath91 and @xmath92 denote the set of nodes and the set of edges respectively , @xmath93 , @xmath94 .
the depth of a node @xmath95 is the length of the path from the root to the node .
the set of all nodes at a given depth @xmath96 is denoted by @xmath97 , @xmath98 .
the root node has a depth of zero .
all the edges @xmath99 are partitioned into @xmath4 levels @xmath100 , @xmath101 , such that the edges in @xmath100 incident with the nodes at depth @xmath102 and the nodes at depth @xmath103 . except the nodes at the @xmath4-th depth @xmath104 , each @xmath105 has two descendants which belong to @xmath106 , and the two corresponding edges are label as @xmath107 and @xmath85 respectively .
the nodes @xmath108 are called leaf nodes . fig .
[ fig_tree ] gives a simple example of code tree with @xmath109 .
a @xmath84-length decoding path @xmath110 consists of @xmath84 edges , with @xmath111 , @xmath112 .
a vector @xmath113 is used to depict the above decoding path , where @xmath38 is corresponding to the binary label of edge @xmath114 .
the reliability of a decoding path @xmath113 can be measured using _ a posteriori _
probability @xmath115 the apps can be regarded as normalized versions of the channel transition probabilities defined in ( [ equ_polarized_channels ] ) .
the two kinds of probabilities are related by a multiplicative factor @xmath116 . by eliminating the factor ,
the apps take values in a more stable range , and all the decoding paths with the same lengths have the sum probability equals to one , i.e. @xmath117 this property will help in understanding the path searching procedure in the code tree and is more suitable for hardware implementation .
similar to the recursive expressions of ( [ equ_polarized_channels ] ) given in @xcite , the apps can also be calculated recursively . for any @xmath118 , @xmath2 , @xmath119 , @xmath120 @xmath121 sc decoding
can be seen as a greedy search algorithm on the code tree . in each level , only the one of two edges with larger probability is selected for further processing . .
the bold branches show a decoding path of sc with @xmath122 . ]
the red bold edges in fig .
[ fig_tree ] shows the sc decoding path .
the number written next to each of the nodes provides the app metric of the decoding path from the root to that node .
the nodes which are extended during the sc decoding procedure are represented by the numbered circles , and the corresponding numbers indicate the processing order .
the black circles represent the nodes which are visited ( whose app metric is calculated ) but failed in competition for further exploring . and the gray ones are those which are not visited during the searching process . in the example , four times of calculations of equation ( [ equ_app ] ) are required , one for each level .
however , the decoding path is not guaranteed to be the most probable one . as shown in the example
, the one labeled @xmath123 has the largest probability of all the @xmath4-length paths , but it failed in the competition at the first level . for further practical considerations , we use the logarithmic apps as the path metrics : @xmath124 for @xmath125 , the path metric can be recursively calculated as @xmath126 and @xmath127 where function @xmath128 is the jacobian logarithm and @xmath129 , @xmath130 .
then , the decision function of sc in ( [ equ_sc_h ] ) is rewritten as @xmath131 using the space - efficient structure @xcite to implement a sc decoder , the time and space complexity are @xmath132 and @xmath133 respectively .
the performance of sc is limited by the bit - by - bit decoding strategy . since whenever a bit is wrongly determined , there is no chance to correct it in the future decoding procedure .
theoretically , the performance of the maximum _ a posteriori _ probability ( map ) decoding ( or equivalently ml decoding , since the inputs are assumed to be uniformly distributed ) can be achieved by traversing all the @xmath4-length decoding paths in the code tree @xmath89 . but this brute - force traverse takes exponential complexity and is difficult to be implemented .
two improved decoding algorithms called successive cancellation list ( scl ) decoding and successive cancellation stack ( scs ) are proposed in @xcite @xcite and @xcite .
both of these two algorithms allow more than one edge to be explored in each level of the code tree . during the scl(scs )
decoding , a bunch of candidate paths will be obtained and stored in a list(stack ) . since for every single candidate path ,
the metric calculations and bit determinations are still performed bit - by - bit successively , scl and scs can be regarded as two improved versions of conventional sc decoding . in this section
, we will restate scl and scs under a unified framework with the help of app metrics and the code tree representations . then to overcome the own shortages of scl and scs , a new hybrid decoding algorithm named successive cancellation hybrid ( sch ) decoding is proposed .
furthermore , to reducing the computational complexities , we propose a pruning technique to eliminate the unnecessary calculations during the path searching procedure on the code tree . .
] as an enhanced version of sc , the successive cancellation list ( scl ) decoder @xcite @xcite searches level - by - level on the code tree , which is just the same with sc .
however , unlike sc where only one path is reserved after processing at each level , scl allows at most @xmath134 candidate paths to be further explored at the next level .
scl can be regarded as a breadth - first searching on the code tree @xmath135 with a searching width @xmath134 . at each level
, scl doubles the number of candidates by appending a bit @xmath107 or a bit @xmath85 to each of the candidate paths , and then selects at most @xmath134 ones with largest metrics and stores them in a list for further processing at the next level .
finally , when reaching the leaf nodes , the binary labels @xmath21 corresponding to the edges in path @xmath136 which has the largest metric in the list , are assigned to the estimated source vector @xmath81 .
let @xmath137 denotes the set of candidate paths corresponding to the level-@xmath84 of code tree in a scl decoder .
the @xmath137s are stored and updated in a list structure .
the scl decoding algorithm with searching width @xmath134 , denoted by scl(@xmath134 ) , can be described as follows : ( a.1 ) initialization .
a null path is included in the initial list and its metric is set to zero , i.e. @xmath138 , @xmath139 .
( a.2 ) expansion . at the @xmath84-th level of the code tree ,
the number of candidate paths in the list are doubled by concatenating new bits @xmath140 taking values of @xmath107 and @xmath85 respectively , that is , @xmath141 for each @xmath142 , the corresponding path metric(s ) are updated according to ( [ equ_metric_info_or_froz ] ) , ( [ equ_metric_odd ] ) and ( [ equ_metric_even ] ) . ( a.3 ) competition .
if the number of paths in the list after ( a.2 ) is no more than @xmath134 , just skip this step ; otherwise , reserve the @xmath134 paths with the largest metrics and delete the others .
( a.4 ) determination .
repeat ( a.2 ) and ( a.3 ) until level-@xmath4 is reached .
then , the decoder outputs the estimated source vector @xmath143 , where @xmath21 is the binary labels of the path with the largest metric in the list . fig .
[ fig_scl ] gives a simple example of the tree searching under scl decoding with @xmath144 .
compare with sc in fig .
[ fig_tree ] , scl find the most probable path @xmath123 .
but the times of metric computations is increased from four to seven .
scl maintains @xmath134 decoding paths simultaneously , each path consumes a @xmath133 space , the space complexity of scl then is @xmath145 . during the decoding process at each level ,
each of the @xmath134 candidates is copied once and extended to two new paths , these copy operations require @xmath145 computations .
moreover , since the code tree has @xmath4 levels , a direct implementation of scl decoder will take @xmath146 computations . in @xcite , a so called `` lazy copy '' technique based on the memory sharing structure among the candidate paths
is introduced to reduce this copy complexity .
therefore , the scl decoder can be implemented with computational complexity @xmath147 .
note that , the path metric ( [ equ_metric_info_or_froz ] ) of a certain decoding path with binary label vector @xmath113 will not be smaller than that of any of its descendants , i.e. for any @xmath148 and @xmath149 , @xmath150 hence , if the metric of a @xmath4-length decoding path is larger than that of another path with length @xmath151 , it must also be larger than the metric of any of the @xmath4-length descendant path of the latter .
so rather than waiting after processing at each level , we can keep on searching along the single candidate path until its metric is no longer the largest . once a @xmath4-length path is found with the largest metric among all the candidate paths ,
its binary label vector is simply output as the final estimation , the unnecessary computations for extending other paths are then saved
. the scs decoder @xcite uses a ordered stack @xmath152 to store the candidate paths and tries to find the optimal estimation by searching along the best candidate in the stack .
whenever the top path in the stack which has the largest path metric reaches length @xmath4 , the decoding process stops and outputs this path . unlike the candidate paths in the list of scl which always have the same length , the candidates in the stack of scs have difference lengths
let @xmath153 denote the maximal the stack @xmath152 in scs decoder .
a little different from the original scs in @xcite , an additional parameter @xmath134 is introduced to limit the number of extending paths with certain length in decoding process .
a counting vector @xmath154 is used to record the number of the popping paths with specific length , i.e. @xmath155 means the number of popped paths with length-@xmath84 during the decoding process . the scs decoding algorithm with the searching width @xmath134 and
the maximal stack depth @xmath153 , denoted by scs @xmath156 , is summarized as follows : ( b.1 ) initialization : push the null path into stack and set the corresponding metric @xmath139 .
initialize the counting vector @xmath157 with all - zeros , and the instantaneous stack depth @xmath158 .
( b.2 ) popping : pop the path @xmath159 from the top of stack , and if the path is not null , set @xmath160 .
( b.3 ) expansion : if @xmath140 is a frozen bit , i.e. @xmath161 , simply extend the path to @xmath162 ; otherwise , if @xmath140 is an information bit , extend current path to @xmath163 and @xmath164
. then calculate path metric(s ) by ( [ equ_metric_info_or_froz ] ) , ( [ equ_metric_odd ] ) and ( [ equ_metric_even ] ) .
( b.4 ) pushing : for information bit @xmath165 , if @xmath166 , delete the path from the bottom of the stack .
then push the two extended paths into the stack .
otherwise , for frozen bit @xmath140 , push the path @xmath167 into stack directly .
( b.5 ) competition : if @xmath168 , delete all the paths with length less than or equal to @xmath169 from the stack @xmath152 .
( b.6 ) sorting : resort paths in the stack from top to bottom in descending metrics .
( b.7 ) determination : if the top path in the stack reaches to the leaf node of the code tree , pop it from the stack .
the decoding algorithm stops and outputs @xmath143 as the decision sequence .
otherwise go back and execute step ( b.2 ) .
[ fig_scs ] gives a simple example of the tree searching under scs .
compare with scl in fig .
[ fig_scl ] , scs can also find the most probable path @xmath123 with two fewer metric computations .
similar to sc and scl , the space efficient structure and `` lazy copy '' technique are applied in the implementation of scs decoders .
the time and space complexity of scs are @xmath147 and @xmath170 respectively . however , under the same searching width @xmath134 , the actual computations of scs@xmath171 will be much fewer than that of scl@xmath171 when workding in the moderate or high snr regime .
compared with scl , scs decoding can save a lot of unnecessary computations especially when working in the high signal - to - noise ( snr ) regime @xcite .
however , the stack used in scs consumes a much larger space than scl . theoretically , to prevent performance deterioration
, the stack depth @xmath153 needs to be as large as @xmath172 , thus the space complexity will becomes @xmath173 .
fortunately , as shown in @xcite , a much smaller stack - depth @xmath153 is enough for moderate and high snr regimes .
but the most appropriate value of @xmath153 is relied on the specific snr and is hard to determine . in this paper ,
a new hybrid decoding algorithm called successive cancellation hybrid ( sch ) is proposed .
sch , as the name suggests , is a hybrid of scl and scs .
sch has two working modes called _ on - going _ and _ waiting_. at first , sch decoder works on the on - going mode , it searches along the best candidate path using a ordered stack just the same as that scs does .
but when the stack is about to be full , sch stops searching forward and switches to the waiting mode . under the waiting mode , sch turns to extend the shortest path in the stack until all the candidate paths in the stack have the same length .
the processing under waiting mode is somewhat similar to scl and it decreases the number of paths in stack .
then , sch switches back to the on - going mode again . fig . [ fig_sch ] gives a graphic illustration .
this decoding procedure goes on until an @xmath4-length path appears at the top of the stack .
the sch algorithm with the searching width @xmath134 , the maximal stack depth @xmath153 , denoted by sch @xmath174 , is summarized as follows : ( c.1 ) initialization : push the null path into stack @xmath152 and set the corresponding metric @xmath139 . initialize the counting vector @xmath175 with all - zeros , and the instantaneous stack depth @xmath158 .
the working mode flag @xmath176 is set to @xmath107 , where @xmath107 denote the on - going mode and @xmath85 denote the waiting mode .
( c.2 ) popping : when @xmath177 , pop the path @xmath159 from the top of stack ; else when @xmath178 , pop the path @xmath159 with the shortest path length in the stack . then
, if the popped path is not null , i.e. @xmath179 , set @xmath160 .
( c.3 ) expansion : if @xmath140 is a frozen bit , i.e. @xmath180 , simply extend the path to @xmath181 ; otherwise , if @xmath140 is an information bit , i.e. @xmath39 , extend current path to @xmath163 and @xmath164
. then calculate path metric(s ) by ( [ equ_metric_info_or_froz ] ) , ( [ equ_metric_odd ] ) and ( [ equ_metric_even ] ) .
( c.4 ) pushing : for information ( frozen ) bit @xmath140 , push the new two paths ( one path ) into the stack .
( c.5 ) competition : if @xmath168 , delete all the paths with length less than or equal to @xmath169 from the stack @xmath152 .
( c.6 ) mode switching : when @xmath177 and @xmath182 , switch @xmath178 ; when @xmath178 and all the candidate pathes in the stack have equal lengths , @xmath178 ; ( c.7 ) sorting : resort paths in the stack from top to bottom in descending metrics .
( c.8 ) determination : if the top path in the stack reaches to the leaf node of the code tree , pop it from the stack .
the decoding algorithm stops and outputs @xmath143 as the decision sequence .
otherwise go back and execute step ( c.2 ) .
the time and space complexity of sch are @xmath147 and @xmath170 respectively .
the actual computations of sch decoding is less than that of scl but is usually more than that of scs . for sch decoding ,
since no path is dropped when the stack is about to be full , the performance will not affected by @xmath153 . however ,
when decoder stays in the waiting mode , unnecessary computations will be taken . and the smaller the maximum stack depth is @xmath153 , the more likely the decoder will switch to the waiting mode .
so , the computational complexity grows with the decreasing of @xmath153 . to have enough space for waiting mode
, the minimum value of @xmath153 is @xmath183 . particularly , when @xmath184 , sch(@xmath134,@xmath153 ) is equivalent to scl(@xmath134 ) ; and when @xmath185 , sch(@xmath134,@xmath153 ) is equivalent to scs(@xmath134,@xmath153 ) .
candidates divide the probability space into @xmath134 partitions . ] during the path searching on the code tree , the candidate paths with too small metrics and their descendants will hardly have the chance to be reserved in the future process . in this subsection
, we propose a pruning technique to reduce the computational complexity of the improved successive cancellation decoding algorithms .
an additional vector @xmath186 is used to record the pruning reference for each level , where @xmath187 is the largest metric of all the traversed @xmath84-length decoding paths on the code tree .
more specifically , for scl decoding , @xmath188 and equivalently , for scs and sch , @xmath187 is set to the metric of the first @xmath84-length path popped off the stack .
we introduce a new parameter called probability ratio threshold @xmath189 . during the processing at level-@xmath84 on the code tree
, a @xmath84-length path with metric smaller than @xmath190 is dropped directly .
recall that the path metrics are defined as the logarithmic apps ( [ equ_metric_info_or_froz ] ) .
therefore , the pruned paths are those whose apps @xmath191 intuitively , the correct path will possibly be dropped in this pruning operation . in the following part of this subsection ,
an upperbound of the additional performance deterioration brought by @xmath189 is derived and a conservative configuration of @xmath189 is given .
hereafter , scl , scs and sch are collectively referred to as improved successive cancellation ( isc ) decoding algorithms .
the block error event of polar code with information set @xmath36 under isc decoding is defined as @xmath192 by introducing pruning operations , the error events can be classified into two kinds .
the first kind is the correct path is not lost until the final decision phase , i.e. the correct path is contained in the final list(or stack ) but does not have the largest metric .
the second kind is the correct path is lost before the decision step .
so , the block error rate ( bler ) of isc can be decomposed as @xmath193 where @xmath194 means the correct path loss .
the event @xmath195 can be further decomposed as @xmath196 where @xmath197 is the event that the correct path is not lost until the processing at the @xmath84-th level .
there are three kinds of event which will lead to path loss at the @xmath84-th level .
the first is brought by the searching width limitation , i.e. the correct path is excluded from the @xmath134 best paths in @xmath84-th decoding step , and is denoted by @xmath198 .
the second is brought by the maximum probability ratio limitation , i.e. the metric of the correct path is much smaller than that of the best one , and is denoted by @xmath199 .
the third is brought by the maximum stack depth limitation , which only exist in the scs decoding that the correct path is abandoned when the path length equals @xmath84 and the metric is much smaller than the maximum one at that moment , and this event is denoted by @xmath200 .
then @xmath201 for scl , sch decoding or scs with a large enough stack depth @xmath153 , @xmath202 .
the additional bler performance deterioration brought by pruning is @xmath203 during the processing on the code tree @xmath89 , we will have at most @xmath134 paths at level-@xmath84 with apps @xmath204 which is calculated by ( [ equ_app ] ) , and @xmath205 without loss of generality , we assume that @xmath206 . by the assumption that the one of these paths is the correct path , the @xmath134 probability divided the whole probability space into @xmath134 parts as shown in fig . [ fig_prob_space ] .
the event of correct path loss in the pruning processing at the @xmath84-th level has a probability @xmath207 for each of these eliminated paths , the corresponding probability @xmath208 where @xmath209 .
so we have @xmath210 the additional error probability brought by @xmath189 is upper bounder by @xmath211 given a tolerable performance deterioration @xmath212 , the value of @xmath189 can be determined as @xmath213 in most cases , since the upperbound in ( [ equ_ptau_upperbound ] ) is very loose , the accrual performance deterioration is usually far less than @xmath212 .
the configuration of @xmath189 in ( [ equ_tau_value ] ) is very conservative .
in this section , the performance and complexity of the improved successive cancellation ( isc ) decoding algorithms will be discussed . to simplify the complexity evaluation of polar decoding
, we measure the average computational complexity in terms of the number of metric recursive operations , which are defined in ( [ equ_metric_odd ] ) or ( [ equ_metric_even ] ) .
for example , the computational complexity of sc decoder is @xmath214 .
[ fig_diff_len_performance ] gives the simulation results with code length @xmath4 set as @xmath215 and @xmath216 , and the code rate @xmath217 . and
[ fig_diff_rate_performance ] shows the bler performances with code rate @xmath218 set as @xmath219 and @xmath220 , and the code length @xmath4 is fixed to @xmath215 .
the lowerbounds of bler performance under maximum - likelihood ( ml ) decoding are obtained by performing scl(@xmath221 ) decoding and counting the number of times the decoded codeword is more likely than the transmitted one .
the probability ratio threshold @xmath189 for pruning operation is set by ( [ equ_tau_value ] ) with @xmath222 . as shown in the figures , under proper configurations ,
all the three decoding algorithms can achieve the performance very close to that of ml decoding .
the average computational complexities under different decoding algorithms with code length @xmath223 and code rate @xmath217 are shown in fig .
[ fig_diff_dec_complexity ] .
we can see that the complexity of sch is not monotonically decreasing with the increasing of snr .
this is because the switching between the two working modes is relied on the certain code construction and searching procedures .
however , sch always has a much lower computational complexity than that of scl .
although it needs more computations than scs , sch occupies less memory space without any deterioration in performance .
in fact , under some specific configurations , sch can be equivalent to the other two decoding algorithms : when @xmath184 , sch(@xmath134 , @xmath183 ) is equivalent to scl(@xmath134 ) ; and when @xmath153 is very large , sch(@xmath134 , @xmath153 ) is equivalent to scs(@xmath134 , @xmath153 ) ; therefore , sch can achieve a better trade - off between computational complexity and space complexity .
furthermore , by applying the pruning technique introduced in section [ subsec_prune ] , the computational complexity can be significantly reduced and very close to that of sc in the moderate and high snr regime .
compared with sc , isc decoding algorithms introduce three more parameters : the searching width @xmath134 , the maximum stack depth @xmath153 and probability ratio threshold for pruning @xmath189 . in the following part of this section
, we will analysis the impacts on performance and complexity of this three parameters one - by - one . ] ] fig .
[ fig_diff_l_performance ] gives the performance comparisons under scl decoding with different @xmath134 .
the code length and code rate are set as @xmath223 and @xmath224 respectively .
the searching width @xmath134 varies from @xmath85 ( equivalent to sc ) to @xmath225 .
note that , scl(@xmath134 ) is equivalent to sch(@xmath134 , @xmath183 ) and scs(@xmath134 , @xmath153 ) with a large enough @xmath153 .
the affects brought by different @xmath134 in scl decoding are the same with that in scs and sch .
the larger the searching width is , the less probable to lose the correct path , i.e. @xmath226 in ( [ equ_event_ci ] ) is a decreasing function of @xmath134 . but according to the results depicted in fig .
[ fig_diff_l_complexity ] , the computational complexity is approximately proportional to @xmath134 . as shown in fig .
[ fig_diff_l_performance ] , @xmath227 is good enough for @xmath223 and @xmath224 . ] for polar codes under scs decoding , a too small value of the maximum stack depth @xmath153 will lead to significant deterioration on performance . as shown in fig .
[ fig_diff_d_performance ] , @xmath153 need to be larger than @xmath215 for scs decoding . but for sch , the different configurations of @xmath153 no longer affect the bler performance but the computational complexity . as shown in fig .
[ fig_diff_d_complexity ] , the computational complexity of sch is decreasing with the increasing of @xmath153 .
although it needs more computations than scs , sch occupies less memory space without any deterioration in performance .
compared with scl , sch has much lower computational complexity and only require a little more memory space .
in fact , under some specific configurations , sch can be equivalent to the other two decoding algorithms : when @xmath184 , sch(@xmath134 , @xmath183 ) is equivalent to scl(@xmath134 ) ; and when @xmath153 is very large , sch(@xmath134 , @xmath153 ) is equivalent to scs(@xmath134 , @xmath153 ) ; hence , sch can achieve a better trade - off between computational complexity and space complexity . ] ] ] fig .
[ fig_diff_tau_performance ] and fig .
[ fig_diff_tau_complexity ] give simulations of polar codes with code length @xmath223 code rate @xmath224 over binary - input additive gaussian noise channels ( bawgncs ) .
the codes are decoded by sch decoders with @xmath227 , @xmath228 and @xmath189 varies from @xmath85 to @xmath229 . as shown in the figures ,
the computational complexity will be reduced when the increasing of @xmath189 , while the bler performance will be deteriorated with a too small @xmath189 .
larger values of @xmath189 such as @xmath230 will introduce little deterioration in performance , but will lead to larger complexities .
however , when the codes work in a moderate signal - to - noise ratio ( snr ) regime such as @xmath231db where the bler is less than @xmath232 , the computational complexity differences of sc and sch decoding under different @xmath189 in the simulated regime tends to negligible as shown in fig .
[ fig_diff_tau_complexity ] .
the successive cancellation ( sc ) decoding algorithm of polar codes and its improved versions , successive cancellation list ( scl ) and successive cancellation stack ( scs ) are restated as path searching procedures on the code tree of polar codes . combining the ideas of scl and scs ,
a new decoding algorithm named successive cancellation hybrid ( sch ) is proposed , which can achieve a better trade - off between computational complexity and space complexity . to avoid unnecessary path searching ,
a pruning technique which is suitable for all improved successive cancellation ( isc ) decoders is proposed .
performance and complexity analysis based on simulations show that , with the help of the pruning technique , all the isc decoders can have a performance very close to that of maximum - likelihood ( ml ) decoding , and the computational complexities can be very close to that of sc in the moderate and high signal - to - noise ratio ( snr ) regime .
e. arikan , `` channel polarization : a method for constructing capacity - achieving codes for symmetric binary - input memoryless channels , '' _ ieee trans .
inf . theory _
55 , no . 7 , pp . 3051 - 3073 , jul . | as improved versions of successive cancellation ( sc ) decoding algorithm , successive cancellation list ( scl ) decoding and successive cancellation stack ( scs ) decoding are used to improve the finite - length performance of polar codes .
unified descriptions of sc , scl and scs decoding algorithms are given as path searching procedures on the code tree of polar codes .
combining the ideas of scl and scs , a new decoding algorithm named successive cancellation hybrid ( sch ) is proposed , which can achieve a better trade - off between computational complexity and space complexity .
further , to reduce the complexity , a pruning technique is proposed to avoid unnecessary path searching operations .
performance and complexity analysis based on simulations show that , with proper configurations , all the three improved successive cancellation ( isc ) decoding algorithms can have a performance very close to that of maximum - likelihood ( ml ) decoding with acceptable complexity .
moreover , with the help of the proposed pruning technique , the complexities of isc decoders can be very close to that of sc decoder in the moderate and high signal - to - noise ratio ( snr ) regime .
polar codes , successive cancellation decoding , code tree , tree pruning . |
quantum bits ( _ qubits _ ) are the elementary units of information that are used to represent quantum data @xcite .
thus , the idea of a qubit underlies all investigations in the rapidly growing science of quantum information including quantum information theory , quantum communication , quantum computation , quantum complexity , and quantum game theory @xcite .
in particular , qubits are the basic building blocks for defining the standard model of quantum computation as introduced by deutsch @xcite , which has so far provided the appropriate representation for identifying and understanding _ efficient _ ways of processing information using quantum mechanics .
its investigation resulted in feasible algorithms for factoring large integers @xcite and for simulating many - particle quantum systems @xcite , two problems not known to be efficiently solvable with classical computers .
a qubit can be thought of as the extension of a classical bit obtained by applying the superposition principle .
when quantum superposition states of many qubits are constructed in the tensor product state space that quantum mechanics prescribes for a composite system , quantum entanglement arises as an additional information resource with no classical counterpart .
however , qubits share with classical bits the fundamental property of being a _ fungible _ information resource @xcite : while both classical and quantum information is intended , in fact required , to be physically realized , it is _ abstractly _ defined and therefore independent of the details of the underlying physical realization .
the fungibility property is essential to quantum information in two respects .
first , by defining quantum information independent of the details of specific physical devices and their complex physics , it has been possible to study qubit properties at the abstract level and thus to obtain a deeper understanding of the distinctive features that qubits inherit from their intrinsic quantum - mechanical nature . examples of fundamental results following from the basic properties of superpositions and of the randomness associated with quantum measurements include the fact that qubits in an unknown quantum state can not be perfectly copied ( no - cloning theorem @xcite ) and they can not be broadcast ( no - broadcasting theorem @xcite ) .
on the other hand they can be reliably communicated by means of the quantum teleportation protocol @xcite , an extremely useful protocol with many applications @xcite .
second , from a practical standpoint , the arbitrariness of the physical realization implied by fungibility allows for greater flexibility in the identification and design of quantum information processors .
this is reflected in the amazing variety of representations of qubits that have appeared in recent proposals for physical realizations of quantum computers @xcite . while such variety greatly increases the number of physical systems with the potential for quantum information processing , it can be intimidating when considering a new prospective system for implementing qubits .
as recently observed by divincenzo @xcite , `` recognizing a qubit can be trickier than one might think '' .
thus to determine the suitability of a physical system for quantum information processing we need to answer the following basic questions : what is a qubit , and how do we look for them ? in this paper
, we identify essential criteria to be met by a system to contain physical embodiments of qubits .
our primary goal is to develop the intuition needed for `` distilling '' qubits from physical systems , in the hope that such intuition will serve as a guide for exploring the full range of possibilities available for physical implementation .
we do so by revisiting a few prototypical examples to obtain a general operational characterization of qubits .
the examples are examined in detail in section [ sect : fungibility ] .
the first example shows how the state space of a collection of bosonic modes can be exploited for representing qubits based on the requirement of obtaining realizable control over the qubit observables .
the second and third example are essentially motivated by reconsidering the notion of what a qubit is in the light of our knowledge from quantum error - correcting methods .
in particular , we analyze two situations where a qubit can be embedded into the larger state space of three physical two - state systems in a way which makes it robust against noise to first - order in time and to arbitrary orders in time , respectively . as the level of abstraction required for constructing our example qubits grows , the qubit states are less easily related to states of the underlying physical degrees of freedom , and more powerful mathematical tools are required to describe the qubits .
the emerging qubit criteria are summarized in section [ sect : criteria ] .
our analysis points to the crucial role played by _
control resources _ in determining a preferred qubit realization , and to an operator picture in terms of _ quantum observables _ , rather than states , as the appropriate framework for capturing the idea of a qubit in full generality .
the first proposals for quantum computers implemented qubits literally by exploiting explicit two level systems
. examples of qubits in these proposals include the two levels corresponding to absence and presence of a photon in a mode @xcite , spin @xmath0 nuclear spins @xcite , and the ground and one of the excited states of ions @xcite .
the proposals have in common the goal of implementing qubits at the fundamental level of a physical system .
the recognition of how fragile these qubits states are led to the development of quantum error correction , which necessarily involves maintaining the information by distributing it in a controlled way over a potentially large number of physical systems . as a result
we reconsidered the notion of a qubit . to illustrate the different facets of the notion of a qubit we determine how an entity equivalent to an abstractly defined qubit can be constructed in three prototypical situations .
an abstractly defined qubit can be identified with an ideal two - state quantum system q whose associated state space is a two - dimensional complex hilbert space @xmath1 .
an orthonormal basis @xmath2 and we make the identification @xmath3 , @xmath4 with vectors in @xmath5 . operations on states are conveniently expressed in terms of the pauli operators @xmath6 , which obey commutation and anti - commutation relationships of the form @xmath7 & = & 2i \,\sum_\gamma \varepsilon_{\alpha\beta\gamma } \sigma_\gamma \ : , \label{comm } \\ \
{ \sigma_\alpha , \sigma_\beta \ } & = & 2 \,\delta_{\alpha\beta } { \leavevmode\hbox{\small1\normalsize\kern-.33em1}}\ : , \label{anticomm}\end{aligned}\ ] ] with @xmath8 , and @xmath9 and @xmath10 being the completely antisymmetric symbol and the identity on @xmath11 , respectively . to simplify notations ,
we define @xmath12 .
as we will see , the abstract state space of a qubit is not necessarily directly related to the state space of the physical systems .
however , the operators are naturally embedded in the operator algebras associated with the physical observables .
bosonic qubits are the basic building blocks in a recent proposal for linear optics quantum computation ( loqc ) @xcite . the relevant physical system is a collection of @xmath13 distinguishable modes which describe the elementary excitations of the quantized electro - magnetic field @xcite .
each mode is characterized by annihilation and creation operators @xmath14 , @xmath15 , which satisfy bosonic commutation rules @xmath16= [ { \bf a}_k^\dagger , { \bf a}_{k'}^\dagger]=0 $ ] , @xmath17=\delta_{k k'}$ ] .
in particular , number states of mode @xmath18 are defined as eigenstates of the number operator @xmath19 for mode @xmath18 , @xmath20 is the overall state space in which we have to look for qubits .
we use the abbreviation @xmath21 for product number states in @xmath22
. clearly , a variety of ( inequivalent ) prescriptions are conceivable for representing qubits in @xmath22 . in loqc ,
an encoding based on two modes and one boson is adopted via the mapping @xmath23 for a qubit supported by modes @xmath24 .
this choice is motivated by the potential for easy implementations of single qubit transformations via passive linear optical elements .
it is worth pointing out that encodings similar to the above , where the degree of freedom carrying quantum information is identified with the presence of a quasi - particle in one of two modes or sites , or the presence / absence of a quasi - particle in a single mode or site , can make sense for quantum statistics other than the bosonic one .
thus , the present discussion is relevant to situations where information is stored in anyonic @xcite or fermionic @xcite qubits . a different scheme for embedding a generic finite - dimensional quantum system ( or _ qudit _ ) in the state space of bosonic degrees of freedom
was recently proposed in @xcite . according to ( [ bq ] ) , the state space of the bosonic qubit , @xmath25 . by letting @xmath26
denote the eigenvalue of the joint number operator @xmath27 , one has formally @xmath28 where @xmath29 is the subspace of states in @xmath30 with @xmath31 bosons , and @xmath32 .
the state space of @xmath33 bosonic qubits , which is obtained via an effective tensor - product construction of one - qubit spaces , likewise relates to the overall state space @xmath22 in a non - trivial way . assuming a pairing between modes of the form @xmath34 for the @xmath35th encoded qubit , @xmath36
, we obtain @xmath37 \simeq \big [ { { \cal h}}_{{\sf q}_1 } \otimes { { \cal h}}_{{\sf q}_2 } \ldots \otimes { { \cal h}}_{{\sf q}_n } \big ] \oplus \tilde{{{\cal h}}}\ : , \label{mqss}\end{aligned}\ ] ] where for each two - mode state space @xmath38 a decomposition similar to ( [ sqss ] ) has been used , @xmath39 , and @xmath40 collects all the contributions with excitation different from one in at least one of the pairs of modes .
several remarks on the above state space structure are in order . in the one - qubit case
, the presence of the summand @xmath41 in ( [ sqss ] ) causes the bosonic qubit to be effectively embedded into a larger state space where infinitely many `` non - qubit '' levels are present .
one can formally rewrite this separation so that the additional levels form part of an effective environment @xcite .
the formal state space of the two resulting interacting subsystems is written as @xmath42 where @xmath43 and @xmath44 are adjoined `` vacuum '' states , and we make the identification @xmath45 .
the idea generalizes to @xmath33 qubits , in which case the role of the effective `` non - qubit '' environment is assumed by @xmath40 in ( [ mqss ] ) . whenever a non - zero amplitude is found in the space @xmath40 , one or more qubits have `` leaked - out '' from the intended logical space @xmath46 so that the state no longer maps to a state of qubits . to guarantee that the state belongs to the qubits for _ all _ times @xmath47
means restricting the overall evolution operator @xmath48 to the sub - manifold of unitary operators @xmath49 .
it is essential to realize that such a requirement is in general unnecessary as well as overly restrictive .
while care should be taken to ensure that no leakage allows the qubits to stray away from the logical space at the beginning and at the end of every quantum gate , one should not insist on retaining a mapping into qubits at every intermediate time .
in other words , qubits need only exist _ stroboscopically _ in time . in the loqc proposal , relaxing
the constraint of well - defined qubit states at intermediate steps is indispensable for achieving the required two - bit conditional dynamics .
for example , in @xcite , the most basic conditional sign - flip gate c-@xmath50 on two bosonic qubits , say @xmath51 and @xmath52 , is decomposed into a sequence of operations @xmath53 where both the beam - splitter gate @xmath54 and the non - deterministic gate @xmath55 on mode @xmath56 cause a temporary departure from the one - excitation space of each qubit .
this implementation is also non - deterministic , meaning that it requires post - selection , with a known success probability . in @xcite ,
the non - determinism can be removed by exploiting the quantum teleportation protocol @xcite . in spite of the complicated underlying state space structure , a bosonic qubit @xmath57 can be straightforwardly characterized in terms of its generating observables . on the @xmath13-mode state space @xmath22 ,
define the operator @xmath58 on @xmath30 , and @xmath10 elsewhere .
@xmath59 satisfies @xmath60 , and @xmath61 hence , the restriction @xmath62 of @xmath59 to @xmath30 , @xmath63 , and the product @xmath64 : = @xmath65 , @xmath66 , are the projectors onto the one - qubit and the @xmath33-qubit space , respectively . clearly , the action of @xmath62 is the identity operation when further restricted to the qubit space , _ i.e. _ @xmath67 , where @xmath68 means equality over @xmath69 . for fixed @xmath70 ,
one finds that @xmath71 from this one can readily check that the following operators act as encoded generating observables for the bosonic qubit @xmath57 , with @xmath72 : @xmath73 and @xmath74/2i =- i z_{{\sf q}(k , k')}x_{{\sf q}(k , k')}$ ] .
these observables obey commutation / anti - commutation rules identical to ( [ comm])-([anticomm ] ) .
evolutions generated by the hamiltonians given in ( [ bqobs ] ) can be readily constructed from the action of optical phase shifters and beam splitters respectively , which allows for an easy implementation of arbitrary one qubit ( @xmath75 ) gates via passive linear optics .
the final ingredients that are required to make the bosonic qubit useful include the ability to initialize the qubit in the intended logical space @xmath69 , and the ability to read - out the qubit observables . in loqc
, the state preparation of qubit @xmath57 can be accomplished by using a single - photon source to prepare mode @xmath18 in @xmath76 and mode @xmath77 in @xmath78 . to measure the bosonic qubit it suffices to use a photo - detector on the mode @xmath18 , which destructively determines whether one or more photons were present in the mode .
further details are in @xcite .
our second example comes from quantum error correction using stabilizer codes .
consider the simplest possible situation , where we wish to protect one quantum bit against single bit - flip errors by encoding it into three physical qubits .
a quantum code can be specified by a two - dimensional subspace of the overall state space @xmath22 , @xmath79 .
to protect against single bit - flip errors , a repetition code @xmath80 can be defined by the encoding @xmath81 it is easily checked that @xmath80 satisfies the necessary and sufficient conditions for error - recovery with respect to the error set @xmath82 @xcite .
note that @xmath83 for errors in @xmath84 .
let @xmath85 denote the subspace spanned by @xmath86 , for @xmath87 , and let us choose as orthonormal bases in the @xmath85 s the result of applying the errors to the logical states ( [ code2 ] ) , _ i.e. _ @xmath88 with @xmath89 , @xmath87 , @xmath90
. then @xmath91 and a recovery super - operator can be explicitly constructed by defining , for each error @xmath92 , @xmath93 the fact that the quantum operation @xmath94 defined by @xmath95 for density operators @xmath96 actually restores the state of the encoded qubit after an error in @xmath84 happens is due to the property that every combination @xmath97 is a multiple of the identity operation on @xmath80 .
thus , the basic idea for using this code is to apply @xmath98 after the errors happened .
while this procedure successfully protects our bit of quantum information , it is conceptually dissatisfying , because it would appear that _
after _ the errors happened , but _ before
_ application of @xmath98 , the information is corrupted by noise .
is there a representation that clearly separates errors and information in such a way that it is clear that the quantum information is never affected ?
in other words , where does the protected qubit reside both before and after errors occurred ? the basic insight is to regard the error - correcting code as an appropriate _ subsystem _ @xcite .
this is possible by establishing the mapping @xmath99 for the basis vectors of the @xmath85 s ( hence @xmath22 ) introduced before . in the right hand - side of ( [ qmap ] ) ,
the @xmath100 vectors , @xmath87 , are taken as basis states of a two - dimensional complex space that will serve as the protected qubit state space , while the @xmath101 store the bit string that uniquely identifies the error syndrome .
essentially , the state @xmath101 is meant to fully encode the effect of the noise on the code .
the correspondence ( [ qmap ] ) is a prescription for decomposing the physical coding space @xmath22 into the tensor product of a qubit space @xmath102 and a syndrome space @xmath103 : @xmath104 if we choose to represent the syndrome corresponding to @xmath105 ( no error ) by @xmath106 , then clearly @xmath107 , we find @xmath108 \otimes |v_a^0 \rangle_{{{\cal e } } } = |\psi \rangle_{{{\cal q } } } \otimes the last equality the mapping ( [ code3 ] ) is made explicit . by construction ,
the vector @xmath101 depends on @xmath109 alone .
thus , information in @xmath102 is completely unaffected by errors in @xmath84 : the factor @xmath102 in ( [ code3 ] ) is the qubit subsystem where the protected quantum information resides .
it is worth comparing this picture with the familiar description of the error - correcting code based on the stabilizer formalism @xcite . in the stabilizer language ,
the code @xmath80 is characterized by its stabilizer group , @xmath110 s is generated by @xmath111 and @xmath112 , and the code @xmath80 is a joint eigenspace of the two stabilizer generators . the commutation pattern between each error @xmath109 and the stabilizer generators @xmath113 diagnoses the error syndrome completely . in our case
, it simply reads @xmath114 , @xmath115 , @xmath116 , @xmath117 , where @xmath118 encodes whether the error commutes or anti - commutes with the corresponding generator , respectively .
a basis of the state space of the three qubits can be built from joint eigenvectors of a sufficiently large ( maximal ) set of commuting operators .
such a set can be obtained by adding to @xmath119 and @xmath120 the operator @xmath121 ( for example ) .
the corresponding joint eigenvectors may be labeled @xmath122 , where , for convenience , we denote by @xmath123 not the @xmath124 eigenvalues of the corresponding operators @xmath125 , but the eigenvalues of the projectors onto the @xmath126 subspace _
e.g. _ , the label @xmath127 ( @xmath128 ) corresponds to a @xmath129 ( @xmath126 ) eigenvalue of the operator @xmath130 .
the code @xmath80 is then spanned by the vectors @xmath131 and @xmath132 .
the quantum number @xmath133 can be thought of as labeling a two - state degree of freedom .
this also translates into a prescription for decomposing the overall coding space , @xmath134 via a correspondence of the form @xmath135 note that this prescription is not unique , as it depends on the choice of phase of the different eigenvectors .
( it can be made unique by considering @xmath130 and the global flip @xmath136 as generators for the algebra of observables for the degree of freedom carried by @xmath137 . )
the difference between the representations ( [ qmap2 ] ) and ( [ qmap ] ) accounts for the different views of error - correcting codes that comes from the subsystem idea .
the difference can be explicitly appreciated by looking at the correspondences between the states in the two factorizations .
we have , for instance , @xmath138 but @xmath139 meaning that the degree of freedom @xmath137 has its values flipped by some error combination and is therefore not explicitly protected . specifically , @xmath140 \simeq [ |c_0 |1 \rangle_{\sf q ' } + c_1 |0 \rangle_{\sf q'}]\otimes @xmath137 conditional on the state of @xmath141 .
thus , while in this picture error detection and error recovery necessarily appear as two separate steps , the earlier factorization ( [ qmap ] ) makes sure that the need for an explicit error correction step is removed once information is properly stored : to first order in time , the effect of the noise is only to `` heat '' the ancillary degrees of freedom carried by the syndrome subsystem @xmath103 , and all that is required to recover from errors is to `` reset '' the latter subsystem to the state @xmath106 .
this is exactly what the recovery quantum operation @xmath98 accomplishes , as is clear by representing every operator @xmath142 in ( [ recovery ] ) using the mapping ( [ qmap ] ) , _ i.e. _ @xmath143 the best method for characterizing the protected qubit living in @xmath102 is in terms of its observables .
this is easily done by starting with the obvious observables restricted to the code subspace : @xmath144 can then be obtained by applying the errors in the following way : @xmath145 note that the required behavior of @xmath146 , @xmath147 on the subsystem @xmath102 follows from the property that each operator @xmath148 in ( [ obscode ] ) , @xmath149 or @xmath150 , has the intended action on states of the form @xmath151 , @xmath87 . the relevant commutation and anti - commutation rules ( [ comm])-([anticomm ] ) are readily verified .
in addition to belonging to the set of operators commuting with the stabilizer group ( the so - called normalizer @xmath152 @xcite ) , the above qubit observables have the property that they also commute with every operator that is a repeated combination of a reset operator followed by an error , _ e.g. _ @xmath153 .
this sheds light on the algebraic nature of our three - bit subsystem , which is characterized as a _
noiseless subsystem _ of the multiplicative algebra @xmath154 constructed from recovery operators followed by errors generated by a linear set of operators @xmath155 contains all the linear complex combinations of products of operators in @xmath156 . ] .
the fact that _ every _ quantum error - correcting code can be pictured as a noiseless subsystem of a suitable operator algebra is established in @xcite .
consider a system @xmath157 composed by three physical , distinguishable spin @xmath0 particles .
suppose that the interaction with the environment @xmath158 results in a particularly simple form of noise , _ collective _ noise , where @xmath158 couples in a symmetric way to each spin @xcite .
a relevant example of this situation occurs when the spins couple identically to a fluctuating magnetic field .
our task is to devise a scheme for embedding in @xmath157 a qubit that is protected to noise . for collective noise ,
the interaction hamiltonian @xmath159 can we written as @xmath160 where @xmath161 , @xmath162 , are the components of the total spin angular momentum and the @xmath163 s are linearly independent environment operators .
note that the operators @xmath164 are the generators of a lie group @xmath165 which corresponds to global spatial rotations of the spins , and can be identified with the familiar @xmath165 of angular momentum theory @xcite .
suppose that the self - hamiltonian of the spins can also be expressed in terms of the @xmath166 .
one possibility for constructing a protected qubit is to look for a pair of simultaneous degenerate eigenstates of the @xmath164 s @xcite _ i.e. _ , @xmath167 since the eigenvalues @xmath168 do not depend on @xmath169 , the two eigenstates can not be distinguished by the environment .
thus , if they can be found , they define a basis of a protected qubit s state space . in our case
, one finds that ( [ deg ] ) can be only satisfied with @xmath170 span a semi - simple lie algebra @xmath171 @xcite . ] , meaning that the states @xmath172 belong to the so - called @xmath173 singlet representation of @xmath171 _ i.e. _ they are invariant under @xmath165 rotations .
unfortunately , no state of three spin @xmath0 particles obeys this invariance condition , and a minimum number of four physical spin @xmath0 particles is required for the singlet representation to occur with degeneracy at least two @xcite . in spite of this impossibility to find a _
subspace _ of the three - spin state space @xmath22 which is immune against noise , it turns out that we are still able to construct a _
noiseless subsystem _ of @xmath22 @xcite by considering observables commuting with the @xmath164 s .
the idea , which we describe next , is to identify in @xmath22 a protected degree of freedom .
since the hamiltonians @xmath164 generate the spatial rotation group acting symmetrically on the three spins , it is natural to decompose the state space @xmath22 of the spins according to the total angular momentum @xmath174 : @xmath175 where the subspaces @xmath176 are the eigenspaces of @xmath177 corresponding to angular momentum @xmath178 , with @xmath179 .
they have dimension dim@xmath180 , dim@xmath181 , respectively .
let us focus on the @xmath182 component .
the fact that @xmath183 has dimension four implies that the two - dimensional spin-@xmath0 representation of the rotation group occurs twice , meaning that physically there are two distinct , equivalent routes for generating angular momentum @xmath182 which , for our three - spin system , reads @xmath184 @xcite . ] .
a basis of states for @xmath183 is constructed by considering joint @xmath177 , @xmath185-eigenvectors , @xmath186 , where @xmath187 is the @xmath185-eigenvalue and the quantum number @xmath188 identifies which of the distinct routes the corresponding eigenvector belongs to . because angular momentum operators @xmath164 are confined to act non - trivially and equivalently within each route , the @xmath164 have an identical , diagonal action on the degree of freedom supported by the quantum number @xmath188 .
this degenerate behavior of noise operators with respect to @xmath188 is exactly the two - fold degeneracy we are looking for .
a better grasp of the protected structure which is emerging in @xmath183 is obtained by establishing the mapping @xmath189 where now @xmath190 is an orthonormal basis in @xmath191 and @xmath192 is the @xmath185-eigenvector with eigenvalue @xmath187 .
under such an identification , the subspace @xmath183 can be represented as @xmath193 the actions of the noise operators @xmath164 on @xmath183 take a correspondingly simple form , @xmath194 where @xmath195 is a unit linear combination of pauli operators which depends on the choice of basis states in @xmath11 .
these algebraic identities provide the starting point for identifying @xmath183 as the state space of an effective bi - partite system , and for associating an abstract subsystem to each factor in the tensor product ( [ factors ] ) . in particular , by virtue of the identity action of noise operators in ( [ actions ] ) , the left factor @xmath11 provides a _ noiseless _ subsystem where a qubit can safely reside , protected from noise for ( ideally ) arbitrarily long times . because the @xmath191-basis vectors @xmath196 are left arbitrary in ( [ factors ] ) ,
various realizations are possible as basis states of our three - spin noiseless qubit .
two convenient choices are listed below : @xmath197 which can be easily built up from singlet / triplet states of spins 1 and 2 @xcite , or @xmath198 with @xmath199 , which connects directly with standard basis states for the two - dimensional irreducible representation * d*@xmath200 of the permutation group @xmath201 acting on the spins @xcite . in ( [ basis1])-[basis2 ] ) , the notations @xmath202 and @xmath203 have been used to account for the different choices of the basis states @xmath204 in @xmath11 .
it is important to realize that a possibly mixed state of @xmath22 of the form @xmath205 , where @xmath206 and @xmath207 is an arbitrary density operator on @xmath208 , is a pure state of the qubit living in @xmath11 .
the bases given above establish an equivalence between the state spaces of a pair of two - state systems and the subspace @xmath183 of the three spins state space @xmath22 .
the method can be systematized and the introduction of a different state space can be avoided by directly considering the set of physical observables available for the three spins .
observables that are not affected by the interaction operators @xmath164 are the scalars under spatial rotations , which are given by @xmath209 since these scalars commute with the interactions , their expectations are protected from noise .
if we can combine them so that their expectations define states of a qubit , a noiseless qubit has been found .
it is sufficient to form combinations that algebraically behave like the pauli operators .
let @xmath210 denote the projector over @xmath11 , @xmath211 @xmath10 denoting the identity operator over @xmath22 and @xmath212 satisfying @xmath213 when restricted to @xmath11 .
then the following choice of observables works and corresponds to the @xmath214-basis introduced above : @xmath215 where @xmath216 is the unitary operator swapping the first two spins .
@xmath146 is given explicitly as @xmath217/2i= \sqrt{3}/6\ , \tau_{123}$ ] , @xmath218 being the totally antisymmetric operator @xmath219 , @xmath220 .
similar expressions hold for the singlet - triplet basis ( [ basis1 ] ) , _ e.g. _
@xmath221 , and @xmath222 .
since the commutation and anti - commutation rules obeyed by the observables are identical to ( [ comm])-([anticomm ] ) , the operator algebra they generate by multiplication and hermitian conjugation is the algebra of a two - state quantum system . an important simplification to the above discussion is worth mentioning .
this occurs when not only @xmath177 but also @xmath185 are good ( _ i.e. _ conserved ) quantum numbers for the dynamics of the three physical spins . in this case , using appropriate encoded states rather than the physical three - spin states is motivated by the possibility of achieving universal control by means of purely angular momentum preserving quantum gates .
note that three is the minimal number of physical qubits allowing for simultaneous eigenvectors of @xmath223 with degeneracy at least equal to two .
the resulting qubits are the basic ingredients in a recent proposal for universal quantum computation via exchange interactions @xcite .
the mathematical description of these qubits is simpler than the one of the noiseless three - spin case , as the full subsystem structure ( [ factors ] ) is not required .
the relevant observation is that if no mixing between the sets of states in @xmath183 corresponding to @xmath224 occurs , a further decomposition applies within the @xmath183 subspace : @xmath225 this means that the two subspaces spanned by @xmath226 and @xmath227 , @xmath228 ( the first and second pair of states in either ( [ basis1 ] ) or ( [ basis2 ] ) ) are _ individually _ capable of encoding a qubit . of course
, none of these qubits will retain the robust behavior against all collective noise . over each @xmath191-summand of ( [ summands ] ) ,
the relevant qubit observables are still given by the combinations of scalars determined above .
note that , once a fixed angular momentum component is chosen , say @xmath229 , states belonging to @xmath230 as well as to @xmath231 behave as `` non - qubit '' levels similarly to what we encountered in the case of the bosonic qubit . in particular , this results in a many - qubit state space structure analogous to ( [ mqss ] ) , with a non - trivial summand @xmath40 accounting for all the possible `` non - qubit '' configurations .
explicit prescriptions are mentioned in @xcite for both initializing the logical qubit to the @xmath232 state , and for implementing the final qubit measurement .
initialization may be achieved by relying on the natural equilibration processes of the physical spins in the presence of a polarizing magnetic field , while measurement of the qubit in the @xmath233 criteria for a qubit -------------------- so what is a qubit ? from the mathematical point of view , the main lesson that emerges from the above examples , and from the many others that can be analyzed along similar lines , is that the algebraic notion of a _ subsystem _ provides the most general framework for capturing the variety of ways in which qubits may be constructed .
subsystems , intended as factors ( in the tensor product sense ) of subspaces of a possibly larger state space , are the structures where we have identified qubits throughout our analysis . because the information - carrying qubits can be far removed from the `` natural '' physical systems that a device is based on , states of qubit subsystems often look complicated when expressed in a basis associated with the physical degrees of freedom , making it difficult to recognize their properties and dynamical behavior .
the situation is much simpler , and the description more compact , if the qubit is realized in an _ operator _ sense through its _
observables_. the idea of describing quantum systems in terms of operators forming a complex associative algebra , whose hermitian elements provide the system s observables , underlies the operator approach to quantum and quantum - statistical mechanics @xcite .
similarly , the general definition of a subsystem is motivated @xcite by a fundamental representation theorem for finite dimensional associative operator algebras closed under hermitian conjugation , stating that for any such algebra @xmath154 a direct sum representation of the overall state space @xmath22 exists , @xmath234 in such a way that @xmath154 has identity action over each of the factors @xmath235 s , @xmath236 .
thus , if @xmath154 is the algebra constructed from noise operators ( _ interaction algebra _
@xcite ) , a noiseless factor ( subsystem ) @xmath235 is naturally characterized by an irreducible representation of the so - called _ commutant _ @xmath237 , which is formed from the operators commuting with everything in @xmath154 . motivated by this general perspective
, a necessary condition for having a qubit is that it is a subsystem whose associative operator algebra is identical with ( isomorphic to ) the `` right '' operator algebra of a two - state quantum system , _
i.e. _ one whose generators satisfy the set of composition rules specified in ( [ comm])-([anticomm ] ) . notice that this requirement is stronger than the one based on the @xmath171 commutation rules ( [ comm ] ) alone . while the latter are crucial in determining the appropriate _ lie_-algebraic structure ( thereby obtaining the correct symmetry properties ) of our qubit , ( [ comm ] ) and ( [ anticomm ] ) together are necessary ( and sufficient ) for ensuring the correct _ associative _ structure of the relevant algebra .
in particular , the condition @xmath238 should hold for the generating observables @xmath239 to ensure that the appropriate representation in terms of an abstract spin-1/2 particle is realized . how sensible is this definition from a physical standpoint ?
looking back at our examples once more , one observes that identifying the correct algebraic structure does not guarantee by itself the ability of using the associated subsystem as a qubit . in a sense , this is only the first step toward constructing a qubit .
suppose , however , that we did actually succeed at determining a set of observables which generate the correct operator algebra , and suppose , in addition , that we have the capabilities for implementing the following manipulations : * _ unitary control_@xmath240 apply the observables as hamiltonians to effect universal control operations on the subsystem . *
_ initialization_@xmath240 apply suitable non - unitary control , _
i.e. _ a quantum operation , so as to leave the subsystem in a state whose expectation on the observables matches that of @xmath241 for some choice of the observable @xmath242 . *
_ read - out_@xmath240 strong version : perform von neumann projective measurements of the subsystem observables , which together with unitary control implies the ability to initialize .
weak version : perform weak ensemble measurements of the subsystem observables .
then what we have constructed is , for all practical purposes , a qubit .
we have provided an operational guideline for constructing qubits in physical systems .
our analysis emphasizes the role of operator algebras and observables as the most powerful and comprehensive language to be used for defining qubits in full generality .
one obvious implication is that a qubit ends up being a much more versatile and general object than one might at first conceive of . in a broader context , it was recently argued by steane @xcite that a picture in terms of operators rather than state vectors could provide a more insightful perspective for understanding various aspects of quantum information processing , including the origin of the apparent improvements in efficiency over classical information processing . in a sense , a definition of the qubit in general operator terms
can then be regarded as the first necessary step of this program .
this work was supported by the nsa and by the doe under contract w-7405-eng-36 . | the notion of a qubit is ubiquitous in quantum information processing . in spite of the simple abstract definition of qubits as two - state quantum systems ,
identifying qubits in physical systems is often unexpectedly difficult .
there are an astonishing variety of ways in which qubits can emerge from devices .
what essential features are required for an implementation to properly instantiate a qubit ?
we give three typical examples and propose an operational characterization of qubits based on quantum observables and subsystems . |
to verify that the evolution operator ( eq . [ u ] ) is exact , we provide an example associated with the hamiltonian of spin-1/2 particle in an effective magnetic field having a known solution : @xcite @xmath98 the energy eigenvalues of ( [ ht-2 ] ) are @xmath99 where @xmath100^{1/2}$ ] .
we construct a normalized orthogonal set of eigenspinors of hamiltonian ( [ ht-2 ] ) as : @xmath101 where @xmath102 since the hamiltonian ( [ ht-2 ] ) is time dependent , the general time dependent schrdinger equations can be written as @xmath103 the exact solution of time dependent schrdinger eqs .
( [ pdt-1 ] ) and ( [ pdt-2 ] ) can be written as : @xcite ( color online ) transition probability vs time .
here we chose @xmath111 and @xmath112 .
transition probabilities obtained from eqs .
( [ probability-1 ] ) and ( [ probability-2 ] ) ( solid and dashed lines , respectively ) are seen to be in excellent agreement to the ones obtained from the feynman disentangling operator scheme .
the transition probabilities are given by @xmath113 ( circles ) and @xmath114 ( triangles ) .
, width=321,height=188 ] next , we apply the feynman disentangling operator scheme and find the exact evolution operator of the hamiltonian ( [ ht-2 ] ) which initially takes the form of ( eq.9 , see the paper ) but with three different coupled riccati equations ( evolved during the disentangling operator scheme of the hamiltonian ( [ ht-2 ] ) ) .
these are : @xmath115 usually , an exact solution of such riccati differential equations does not exist .
however , in this case , it is possible to find the exact solution of eqs .
( [ alpha-2 ] ) , ( [ beta-2 ] ) and ( [ gamma-2 ] ) as : @xmath116 where @xmath117 , @xmath118 , @xmath119^{1/2}}{\omega},\label{n1 } \\ n_2= \frac{\delta-\omega}{\omega}-\frac{\left[\left(\delta-\omega\right)^2+\omega^2\right]^{1/2}}{\omega}.\label{n2}\end{aligned}\ ] ] in fig
. [ fig8 ] , the transition probability obtained from eqs .
( [ probability-1 ] ) and ( [ probability-2 ] ) ( solid and dashed lines ) is seen to be in excellent agreement to the one obtained from the disentangling scheme ( circles and triangles ) .
thus , we have demonstrated that the evolution operator obtained from the disentangling operator scheme is exact .
finding an exact unitary operator is one of the requirements for quantum computing and is one of the motivations of the present work .
35ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) link:\doibase 10.1038/ncomms2003 [ ( ) ] link:\doibase 10.1103/physrevlett.109.060401 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.109.237601 [ * * , ( ) ] link:\doibase 10.1103/physrevb.87.195307 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.109.206602 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevlett.108.140503 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.96.026804 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevb.79.195307 [ * * , ( ) ] link:\doibase 10.1103/physrevb.84.155208 [ * * , ( ) ] link:\doibase 10.1063/1.3675620 [ * * , ( ) ] link:\doibase 10.1103/physrevb.87.235202 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1038/nnano.2013.63 [ * * , ( ) ] link:\doibase 10.1103/physrevb.74.165310 [ * * , ( ) ] link:\doibase 10.1103/physrevb.79.085407 [ * * , ( ) ] http://stacks.iop.org/0957-4484/21/i=30/a=302001 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevb.86.085301 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevb.86.075144 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.108.175303 [ * * , ( ) ] link:\doibase 10.1103/physrevb.85.165418 [ * * , ( ) ] @noop * * , ( ) @noop link:\doibase 10.1103/physrev.84.108 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1142/s0217984912500947 [ * * , ( ) ] link:\doibase 10.1103/physrev.80.580 [ * * , ( ) ] @noop _ _ ( , ) | we investigate the evolution of spin dynamics in graphene nanoribbon superlattices ( gnsls ) with armchair and zigzag edges in the presence of a drift field .
we determine the exact evolution operator and show that it exhibits spin echo phenomena due to rapid oscillations of the quantum states along the ribbon .
the evolution of the spin polarization is accompanied by strong beating patterns .
we also provide detailed analysis of the band structure of gnsls with armchair and zigzag edges .
manipulation of electron spins using gate potentials in low dimensional semiconductor nanostructures is of interest , among other things , in that it provides a promising approach for the practical realization of robust qubit operations .
@xcite in recent years , experimental and theoretical research has sought a better understanding of the underlying physics of electrostatically defined quantum dots formed in two - dimensional electron gases for applications to solid state based quantum computing.@xcite in these devices , the spin - orbit interaction gives rise to decoherence due to the coupling of the electron spins to lattice vibrations .
hyperfine interactions between electron and nuclear spins are also a factor in some systems .
much work has focused on iii - v systems although si quantum dots @xcite are also of interest because of their relatively long decoherence times due to weak spin - orbit and hyperfine interactions .
@xcite in another promising approach , experimentalists have succeeded in fabricating and testing a low operation voltage organic field effect transistor using graphene as the gate electrode placed over a thin polymer gate dielectric layer .
@xcite graphene is promising because it exhibits extremely weak spin orbit coupling and hyperfine interactions .
@xcite in this paper , we present a theoretical investigation of the spin echo phenomena in gnsls under an externally applied drift field . we find that the spin echo is accompanied by a strong beating pattern in the evolution of spin dynamics along the gnsls with armchair and zigzag edges .
we show that with a particular choice of the period and drift field , the spin polarization can be controlled to propagate on the surface of the bloch sphere in a desired fashion .
( color online ) schematic diagram of graphene sheet with armchair along x - axis and zigzag along y - axis .
spin orientation shown by arrow sign in fig .
[ fig1 ] ( a ) and ( b ) move rapidly between @xmath0 and @xmath1 for armchair and between @xmath2 and @xmath3 for zigzag gnsls that induce spin echo under an externally applied drift field . , width=321,height=264 ] ( color online ) band structure of gnsls with zigzag edge . localized and
edge states wave function squared are plotted in fig .
[ fig2 ] ( b ) and ( c ) at @xmath4 .
here we chose @xmath5 for electron - like state and @xmath6 for hole - like state .
also we chose @xmath7 , @xmath8 , @xmath9 , @xmath10 , @xmath11 , @xmath12 and @xmath13 . , width=321,height=151 ] ( color online ) quasi - eigen energy of gnsls with zigzag and armchair edges .
the material parameters are chosen to be the same as in fig .
[ fig2 ] but @xmath14 . ,
width=321,height=151 ] ( color online ) fidelity , @xmath15 , vs time in gnsls with zigzag edge .
here we chose @xmath10 , @xmath16 and @xmath17.,width=321,height=188 ] ( color online ) components of the evolution operator ( exact results ) for zigzag gnsls under an applied drift field @xmath18 .
the material constants are chosen to be the same as in fig .
[ fig4].,width=321,height=151 ] ( color online ) spin - flip transition probability vs time in both zigzag and armchair gnsls at @xmath19 ( solid lines with circles ) and @xmath16 ( solid lines ) .
the material constants are chosen to be the same as in fig .
[ fig4 ] .
, width=321,height=188 ] ( color online ) spin echo is observed in armchair and zigzag gnsls under an externally applied drift field .
spin waves propagate in the form of temporal constructive and destructive interferences .
here we chose @xmath20 , @xmath21 , @xmath16 and @xmath17 .
, width=321,height=188 ] the effective mass hamiltonian for single layer gnsls elongated along either armchair or zigzag direction ( see fig .
[ fig1 ] ) near the k point of the brillouin zone can be written as : @xmath22 . here
@xmath23 is the fermi velocity , @xmath24 is the miniband width , @xmath25 are the pauli spin matrices and @xmath26 is the confining potential . for gnsls with
zigzag edge , we consider @xmath27 and the wave function of the above hamiltonian @xmath28 is @xmath29 ( see ref . for details )
now , we define the relative coordinate @xmath30 and relative momentum @xmath31 and formulate the total hamiltonian in the form : @xmath32 .
here , @xmath33 where @xmath34 and @xmath35 . under an externally applied drift field @xmath36 , we write the equation of motion for @xmath37 as : @xmath38 ( see ref . for details ) .
thus we can write @xmath39 with @xmath40 . in ( [ hz - quasi ] ) , we have written the quasi momentum @xmath41 .
we have used the finite element method and solved the corresponding eigenvalue problems for ( [ hpy ] ) and ( [ hz - quasi ] ) and plotted the band structures of gnsls with zigzag edge in fig . [ fig2](a ) .
in addition to the localized states ( fig .
[ fig2 ] ( b ) ) , we also see the presence of edge states ( fig . [ fig2 ] ( c ) ) in gnsls .
the hamiltonian @xmath42 associated to the relative coordinates and relative momentum does not couple to the lowest spin states .
only the quasi - hamiltonian @xmath43 induces bloch oscillations in the evolution of spin dynamics in gnsls .
@xcite for convenience , we write eq .
( [ hz - quasi ] ) as : @xmath44 where @xmath45 , @xmath46 and @xmath47 .
the band structures of quasi electron - hole states described by the hamiltonian ( [ hz - quasi-1 ] ) can be written as @xmath48^{1/2}.\label{epm}\ ] ] the band structure of electron and hole states in the first brillouin zone @xmath49 $ ] with the specific choice of the parameters is shown in fig .
[ fig3 ] .
we construct a normalized orthogonal set of eigenspinors of the quasi - hamiltonian ( [ hz - quasi-1 ] ) as : @xmath50^{1/2 } } \left(\begin{array}{c } \frac{c}{\delta-\delta } \\ 1\\ \end{array}\right ) , \label{chi - p}\ ] ] @xmath51^{1/2 } } \left(\begin{array}{c } -1 \\ \frac{c^\star}{\delta-\delta } \\
\end{array}\right ) .
\label{chi - m}\ ] ] the above spinors ( [ chi - p ] ) and ( [ chi - m ] ) satisfy @xmath52 and @xmath53 . with the use of the feynman disentangling technique , the exact evolution operator of hamiltonian ( [ hz - quasi-1 ] ) can be written as ( see also supplementary material ): @xmath54 where @xmath55 is the time ordering operator . at present , the time dependent functions @xmath56 are unknown and can be found by the feynman disentangling scheme as discussed below .
@xcite for a spin-1/2 particle , the exact evolution operator ( [ u-2 ] ) can be written as : @xmath57 which can be seen to satisfy @xmath58 and @xmath59 due to the fact that @xmath60 is unitary .
at @xmath61 , we use the initial condition @xmath62 , where @xmath63 denotes transpose and write @xmath64 as @xmath65 next , we find the functional form of @xmath66 , @xmath67 , and @xmath68 of gnsls with zigzag edge by utilizing the feynman disentangling scheme .
@xcite in a similar way , for armchair gnsls , we write the total hamiltonian @xmath80 , where @xmath81 here @xmath82 and @xmath83 . we construct a normalized orthogonal set of eigenspinors of the quasi - hamiltonian ( [ ha - quasi ] ) as : @xmath84^{1/2 } } \left(\begin{array}{c } -1\\ \frac{-i{c'}^\star}{\delta-\delta } \\
\end{array}\right ) , \label{chi - pa}\ ] ] @xmath85^{1/2 } } \left(\begin{array}{c } \frac{ic'}{\delta-\delta } \\
1\\ \end{array}\right ) , \label{chi - ma}\ ] ] where @xmath86 .
at @xmath61 , we use the initial condition @xmath87^{1/2 } } \left(\begin{array}{c } \frac{c'_2 } { \delta-\delta } \\ 1\\ \end{array}\right ) , \label{chia}\ ] ] and find @xmath88 .
the three coupled riccatti equations for the quasi - hamiltonian of gnsls with armchair edge can be written as @xmath89 where @xmath90 .
we have solved the three coupled riccatti equations for zigzag and armchair gnsls numerically and found the exact evolution operator ( [ u ] ) .
the supplementary material provides an additional example of finding exact evolution operator for a spin-1/2 particle in presence of tilted magnetic field .
@xcite in fig .
[ fig4 ] , we have plotted the fidelity vs time for several different values of the applied drift field . as expected from the analytical solutions ( see eqs . [ chi - m ] and [ chi ] ) , with the increasing drift field , the fidelity becomes enhanced because the amplitude of the spin polarization increases as the magnitude of the applied drift field is increased while the spatial frequency of the propagating waves is reduced .
these effects cause dips in fidelity at an early stage with increasing magnitude of the drift field . in fig .
[ fig5 ] , we study the evolution operator in graphene under a drift field .
the components of @xmath91 are superposed leading to constructive and destructive interferences .
we observe hahn echo patterns in a constructive interference regime of the evolution of spin dynamics.@xcite this will be separately discussed with reference to fig .
[ fig7 ] .
spin flip occurs when the localized states in gnsls with armchair and zigzag edges move rapidly along the ribbon with the application of the time dependent gate controlled electric field .
such spin flip probabilities are plotted in fig .
[ fig6 ] for both the armchair and zigzag gnsls at different values of the period of the superlattices .
as can be seen , the spin - flip probabilities in both the armchair and zigzag gnsls are enhanced with increasing value of the period of the superlattice .
this is due to the fact that the enhancement in the amplitude of the spin waves occur with the increasing value of the period of the superlattice . in fig .
[ fig7 ] , we have plotted the expectation values of the pauli spin matrices vs time under an externally applied drift field in both gnsls with zigzag and armchair edges . for @xmath92 ,
the components of the spin waves fluctuate rapidly and thus we observed hahn echo patterns in gnsls . in this regime , where the spin echo can be observed , the spin waves are superposed which induce constructive interference .
when the superposed spin waves induce destructive interference , we find strong beating patterns .
the hahn echo accompanied by strong beating patterns can be seen at different intervals of time in gnsls with zigzag ( solid line with circles ) and armchair ( solid lines with diamond ) edges due to the fact that the spin waves in both kinds of superlattices travel with different velocities .
note that the hahn echo patterns in @xmath93 of zigzag edge resemble to the @xmath94 of armchair edge .
this is due to the fact that the coefficients of @xmath95 and @xmath96 of the quasi - hamiltonian of the gnsls with zigzag and armchair edges eq .
( [ hz - quasi ] ) and ( [ ha - quasi ] ) respectively interchanges during the evolution of spin dynamics . to conclude
, we have shown that the quasi - hamiltonian of the gnsls with zigzag and armchair edges can be used to investigate the evolution of spin dynamics under an externally applied drift field .
we have shown that the exact evolution operator can be found for such hamiltonian via the feynman disentangling technique . with a particular choice of the period and drift field ,
the spin polarization can be controlled to propagate on the surface of the bloch sphere in a desired fashion . during the transportation of the spin on the surface of the bloch sphere , the hahn echo , accompanied by strong beating patterns , is observed .
the authors in ref . have fabricated a field effect transistors ( fet ) using graphene where the gate electrode was placed over a thin polymer gate dielectric layer .
such devices show excellent output and transfer characteristics ( mobility @xmath97 ) for gate and drain operations ( see fig.(4b ) in ref . ) .
the present work was motivated in part by these experimental studies and our results ( e.g. , figs . [ fig1]-[fig7 ] ) might provide useful information for the design of quantum logic gates based on the application of the externally applied drift fields in gnsls with both the armchair and zigzag edges .
this work has been supported by nserc ( canada ) and canada research chair programs . |
the presence of dust seems to be a ubiquitous property of galaxies throughout the observable universe .
even the most distant quasars at @xmath10 show evidence for copious amounts of dust ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this indicates rapid metal enrichment of the interstellar medium within the first billion years after the big bang . about 30% of the known luminous @xmath10 quasars
are detected in the mm continuum with many of them also detected in co ( e.g. * ? ? ?
* and references therein ) .
such studies confirm the notion that most of the rest - frame far - infrared ( fir ) emission comes from massive star formation , possibly indicating the formation of early galactic bulges .
thus , these objects signify an important stage in the connection between the build - up of stellar mass and black hole growth .
most high - redshift objects lack full fir / sub - mm spectral energy distributions ( seds ) . l@xmath4 and m@xmath11 are commonly determined using single photometric measurements , typically obtained at 1.2 mm ( = 250ghz ) and applying standard values for the dust temperature as determined from lower redshift objects .
it is unknown whether this assumption is appropriate for high - redshift objects .
ground - based 350@xmath3 m observations of a few @xmath0 quasars tentatively support the assumed values for the dust temperatures when combined with further sub - mm and mm data @xcite . for a more comprehensive picture of the dust emission at high redshifts
we have obtained pacs ( 100 + 160@xmath3 m ) and spire ( 250 + 350 + 500@xmath3 m ) photometry of 69 quasars at @xmath0 as part of our _ herschel _ space observatory @xcite key project `` the dusty young universe '' . _
spitzer _ space telescope @xcite observations complement these data .
this enables the study of the full optical through infrared sed of these objects in the rest frame wavelength range @xmath12 m , which most importantly covers the fir peak of the sed .
while the photometry for the complete key project sample will be presented in a forthcoming paper , we here report on the _ herschel _ observations and sed analysis of the eleven objects in the sample which were previously detected at 1.2 mm ( tab.[sample ] ) .
currently , this sub - sample is best suited to explore the relative importance of fir and sub - mm / mm measurements for the interpretation of the total infrared sed and the contribution from the coolest dust components .
moreover , the increased wavelength coverage allows us to develop our fitting procedure with optimal constraints .
we outline the data reduction in section 2 . in section 3
we describe how we extract physical properties from our measurements , which are then discussed in section 4 .
we summarize and conclude in section 5 . throughout the paper we use a @xmath13cdm cosmology with @xmath14 km s@xmath15 mpc@xmath15 , @xmath16 , and @xmath17 .
lc [email protected] c r@ , l name & redshift & & @xmath18 & + sdss & & & mjy & + ( 1 ) & ( 2 ) & & ( 4 ) & + j020332.35 + 001228.6 & 5.72 & 20&94 & @xmath19 & 1 & 2 + j033829.30 + 002156.2 & 5.03 & 20&01 & @xmath20 & 4 & 5 + j075618.13 + 410408.5 & 5.11 & 20&15 & @xmath21 & 6 & 7 + j081827.40 + 172251.8 & 6.00 & 19&34 & @xmath22 & 8 & 3 + j084035.09 + 562419.9 & 5.84 & 20&04 & @xmath23 & 8 & 9 + j092721.82 + 200123.7 & 5.77 & 19&87 & @xmath24 & 8 & 3 + [email protected] & 5.78 & 19&21 & @xmath26 & 10 & 3 + j104845.05 + 463718.3 & 6.23 & 19&25 & @xmath27 & 11 & 12 + j114816.64 + 525150.2 & 6.42 & 19&03 & @xmath28 & 11 & 12 + j133550.80 + 353315.8 & 5.90 & 19&89 & @xmath29 & 8 & 9 + [email protected] & 6.04 & 20&60 & @xmath30 & 1 & 3 +
all objects were observed with the photodetector array camera and spectrometer ( pacs ; @xcite ) at 100 and 160@xmath3 m using the mini - scan map observing template . for each source we obtained two maps with different scan angles using observing parameters as recommended in the mini - scan map astronomical observation template ( aot ) release note . for each scan direction ,
five repetitions were executed .
this resulted in a total on - source integration time of @xmath2900s for each object .
data reduction was performed within the herschel interactive processing environment ( hipe , @xcite ) , version 8.0.1 .
we followed standard procedures for deep field data reduction , including source masking and high - pass filtering .
the maps of the two scan directions were processed individually and mosaicked at the end of the work flow . a first version of the combined map was used to create a source mask for high - pass filtering .
masking was performed by hand through visual inspection of the mosaicked maps .
this proved to be more reliable than a strict sigma cut , as it also allowed the masking of faint structures which can potentially influence the fluxes of faint targets if located close to the object of interest .
a second processing was then performed , including the mask .
the resulting maps ( individual and mosaic ) were inspected visually and if necessary the mask or width of the high - pass filter was adjusted .
images around the target positions at 100 and 160@xmath3 m are presented in fig.[images ] .
aperture photometry of the final mosaics was performed in idl .
we used relatively small apertures of 6- 10 radius to maximize the signal - to - noise ratio .
appropriate aperture corrections were determined from tabulated values of the encircled energy fraction of unresolved sources .
details on the properties of the pacs point spread function ( psf ) can be found on the pacs calibration web pages .
the uncertainty in the photometry of our maps can not be determined directly from the pixel - to - pixel variations because the final pacs maps suffer from correlated noise ( where the level of pixel noise correlation depends on the details of the data reduction and final map projection ) . in order to estimate
robust photometric uncertainties we therefore implemented the following procedure : for any given map a set of 500 apertures of the same size as used for the qso photometry was placed on random positions on the sky ( see also * ? ? ?
* ; * ? ? ?
the only constraint on the placement of these background apertures was that the central pixel of the aperture has to have at least 75% of the integration time as the position of the qso ( to exclude noisy areas at the edge of the map ) .
the distribution of the measured fluxes in these 500 apertures was then fitted by a gaussian .
the sigma value of this gaussian was taken as the 1@xmath31 photometric uncertainty of this map .
the resulting flux values for the quasars are given in table[photometry ] .
c r@@xmath32l r@@xmath32l r@@xmath32l r@@xmath32l r@@xmath32l r@@xmath32l r@@xmath32l r@@xmath32l r@@xmath32l r@@xmath32l r@@xmath32l name & & & & & & & & & & & + & & & & & & & & & & & + ( 1 ) & & & & & & & & & & & + & 80&1 & 88&1 & 106&6 & 106&7 & 353&111 & 680&44 & & & & & + & 81&2 & 71&2 & 82&7 & 158&9 & & 1187&52 & 10.7&1.0 & 18.5&2.0 & 19.6&5.9 & 18.5&6.2 & 12.6&6.5 + & 61&2 & 62&2 & 70&6 & 123&7 & & 698&36 & 6.2&0.8 & 9.0&1.0 & 11.4&5.3 & 19.0&4.8 & 19.9&5.0 + & 168&2 & 200&2 & 167&8 & 216&10 & 425&127 & 1004&30 & & & & & + & 58&1 & 80&1 & 61&7 & 62&6 & & 440&29 & & & & & + & 47&2 & 50&2 & 42&7 & 76&7 & & 639&40 & & & 13.1&5.3 & 15.3&5.0 & 19.5&5.8 + & 106&2 & 131&2 & 108&8 & 186&9 & & 1436&39 & 6.7&0.8 & 8.5&1.0 & & & + & 110&2 & 120&2 & 95&7 & 128&7 & & 818&41 & & & & & + & 137&3 & 146&2 & 143&8 & 214&8 & 304&100 & 1349&39 & 3.9&0.6 & 7.4&1.7 & 21.0&5.3 & 21.8&4.9 & 12.4&5.7 + & 66&1 & 69&1 & 55&4 & 57&6 & & 483&32 & & & & & + & & & & & & & & 9.8&1.3 & 15.2&5.4 & 12.0&4.9 & + all quasars in the sample were also observed with the spectral and photometric imaging receiver ( spire , @xcite ) at 250 , 350 , and 500@xmath3 m in small scan map mode for five repetitions and a total on - source integration time of @xmath2190s per source .
data reduction followed standard procedures in hipe as recommended by the spire instrument team .
source extraction was performed with the hipe build - in task sourceextractorsussextractor @xcite using information on the psf ( e.g. fwhm ) given in the spire observer s manual .
our observations are dominated by confusion noise which is on the order of @xmath33mjybeam@xmath15 in the spire photometric bands @xcite as determined from deep extragalactic fields . in order to estimate the uncertainties due to confusion noise specifically in our target fields
, we implemented the following procedure ( see also * ? ? ?
* ; * ? ? ?
* ) : first , the source extractor was run over the full calibrated maps
. detections within less than half the fwhm from the nominal target position were tentatively considered to belong to the quasar , pending further confirmation from our check for confusion with nearby fir bright sources .
we then created an artificial image which included all the sources found by the source extractor and subtracted this `` source image '' from the observed map . on this `` residual map ''
we determined the pixel - to - pixel rms in a box with a size of 8 times the fwhm ( fwhm size : 18.2 , 24.9 , and 36.3 for default map pixel sizes of 6 , 10 , and 14 at 250 , 350 , and 500@xmath3 m , respectively ) , centered on the nominal position of the qso .
the size of this box was chosen large enough to allow an appropriate sampling of the surroundings of the source , but small enough to avoid including the lower coverage areas at the edges of the map even for the longest wavelengths .
in addition , the number of pixels per fwhm is approximately constant for the three wavelengths in the final maps ( @xmath34px / fwhm ) which translates into a similar number of pixels used for determining the rms in the background box .
the resulting estimates for the noise ( limited by confusion ) are comparable to the average values given in @xcite , but have a tendency to be slightly lower .
the fluxes and uncertainties we determine are given in table[photometry ] and the 250@xmath3 m maps are presented in fig.[images ] .
we note that table[photometry ] lists a number of spire flux measurements which are nominally below the estimated 3@xmath31 value of the noise . in these cases ,
the inspection of the images revealed a clear excess of flux at the position of the quasar .
it has been shown that the use of positional priors can reduce the effect of confusion noise by 20 - 30% @xcite . while our strategy is somewhat different from that work , we benefit not only from accurate ( relative and absolute ) positional information , but also from information on the seds of the quasar and potential confusing sources in the field via our multi - wavelength data .
this leads us to include these flux measurements in this study , although they have to be treated with caution .
similarly , fluxes at 500@xmath3 m should be considered tentative because at this wavelength the beam is large ( @xmath236 fwhm ) , the confusion noise is high , and the significance of the detections is often low . for all _ herschel _ targets ( except j2054@xmath250005 )
we also have available mid - infrared ( mir ) imaging from _ spitzer _ at 3.6 , 4.5 , 5.8 , and 8.0@xmath3 m with the infrared array camera ( irac , @xcite ) as well as at 24@xmath3 m with the multiband imaging photometer for spitzer ( mips , @xcite ) .
for the redshifts of our sources these passbands cover the rest frame optical and near - infared ( nir ) wavelengths ( @xmath2@xmath35@xmath3 m ) .
the _ spitzer _
data have been processed using standard procedures within the mopex software package provided by the _
science center ( ssc ) .
aperture photometry was performed in idl using standard sets of aperture radii and appropriate aperture corrections as outlined in the respective instrument handbooks ( also available from the ssc website ) .
errors were estimated in a similar fashion as for pacs by measuring the flux in randomly placed apertures on empty parts of the background and determining the variations between these background flux measurements .
all objects in this paper with _ spitzer _
data are detected at high significance in the five bands .
the resulting photometry is summarized in table [ photometry ] and is usually consistent with measurements published previously , where available ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the 24@xmath3 m images are presented alongside the pacs and spire 250@xmath3 m images in fig.[images ] .
our multi - wavelength data set , and in particular the _ spitzer _ 24@xmath3 m images , provide a tool for determining the exact position of the quasar in the _ herschel _ bands . since we can often identify several sources per field that are visible both at _ spitzer _ and at _ herschel _ wavelengths , the exact location of the quasar in the fir maps can be determined from the relative positional information . with this procedure
we can robustly identify faint _ herschel _ detections with the quasars as well as avoid mis - identifications due to nearby objects . during this exercise
we observe absolute spatial offsets between _
spitzer _ and _ herschel _ of typically @xmath362 , in line with expectations from the absolute pointing accuracies .
+ + + + + + + + + the all - sky data release of the wide - field infrared survey explorer ( wise ; * ? ? ?
* ) was queried for photometry or upper limits in the 12@xmath3 m band which can fill the gap in the _ spitzer _ photometry between 8 and 24@xmath3 m . only three quasars in this paper
are detected at 12@xmath3 m and their significance is low ( @xmath363.5@xmath31 , tab.[photometry ] ) .
some objects ( e.g. j2054@xmath250005 ) are not detected in any wise band , and consequently no upper limits are available in the point - source catalogue . for such sources we performed aperture photometry on the wise atlas images to determine upper limits , following the guidelines in the explanatory supplements to the wise all - sky data release products @xcite .
additional data from the literature was compiled , mainly in the observed nir or mm regimes and often taken from the discovery papers .
the latest data release of the ukidss survey @xcite was also checked which yielded additional photometry in the nir for six objects .
in combination with other supplemental data from the literature ( section 2 ) , we compile seds covering a rest frame wavelength range of typically 0.1 - 400@xmath3 m ( see figs.[sed_fits_detected ] and [ sed_fits_nondetected ] ) .
these seds are then fitted with a combination of models to represent the different components contributing to the observed sed . for this purpose
we have divided our sample into two groups , depending on the amount of data available to constrain the fitted components .
objects detected in at least two _ herschel _ bands were subject to full sed fits ( five sources ) , except j2054@xmath250005 , for which the lack of strong photometric constraints at rest frame wavelengths @xmath3710@xmath3 m ( no _ spitzer _ observations and only wise upper limits ) prevented a full sed fit . in our fits we consider four distinct components : 1 . a power law in the uv / optical regime which represents the emission from the accretion disk .
we extend this component into the nir and introduce a break to a the rayleigh - jeans slope of @xmath38 at 3@xmath3 m in the rest frame @xcite . in the fitting , the power - law slope in the uv / optical and the overall normalization are free parameters . 2 .
a blackbody of typically 1300k temperature , thus peaking in the restframe nir .
empirically , such a component is often required to fit the optical through mir seds of luminous quasars ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and is generally interpreted as a signature of hot ( graphite ) dust close to the sublimation temperature .
3 . a clumpy torus model from the library of @xcite to account for the agn heated dust from the `` dusty torus '' in the central parts of the agn . this component dominates the mir and is important to disentangle , to first order , the contributions from the nuclear dust to the rest frame fir emission . besides the choice of a particular model ( see below ) , the absolute scaling of the model is the only free parameter for this component .
4 . a modified black body to account for possible fir excess emission ( over the agn heated torus ) which we here interpret as powered by star formation .
for this component we fix the emissivity index @xmath39 to a value of 1.6 ( e.g. * ? ? ?
* ; * ? ? ?
* but see section [ betafit ] ) .
the temperature of the modified black body and the normalization are free parameters .
the torus models we consider here are available for seven different inclinations ( starting at 0 and increasing in steps of 15 ) . for each inclination , @xcite provide models of various combinations of parameters ( opening angle , radial dust distribution , etc . ) .
our fitting procedure takes one of the torus models and fits a linear combination of the four components to the observed sed via chi - square minimization using mpfit @xcite in idl .
we then cycle through all torus models in the library and repeat the fitting for each of them .
however , for efficiency we limit the torus component to models with inclinations @xmath4045 degrees which seems reasonable given that we observe luminous and largely unreddened type-1 quasars .
this leaves a total of 959 different torus models .
we visually inspect the best - fitting 10% of the model combinations to confirm the fits . with this approach we do not intend to develop a highly accurate model for the full sed emission in these objects . for the current work we aim to construct a physically motivated approximation that yields a reasonable description of the observed seds which allows us to isolate excess fir emission and to account for ( to first order ) contributions of the agn heated nuclear dust to the fir photometry .
as outlined above , our fitting includes an additional , emprically motivated 1300k black body in the nir .
we have also performed the fits excluding this component , only fitting a power - law , a torus model and a fir modified black body .
the comparison between both cases shows that the fits including the nir black body generally represent the observed photometry better .
this is particularly apparent in wavelength regions dominated by very hot dust ( @xmath1@xmath21 - 3@xmath3 m ) , in the overlap region between the torus and the fir black body ( @xmath41@xmath220 - 30@xmath3 m ) and in the fit at @xmath41@xmath42100@xmath3 m .
the temperatures of the fir dust component also come out consistently lower ( by about 5 - 10k ) in fits including the additional nir black body .
recall that the additional nir component contributes significantly ( or dominantly ) to the short infrared wavelengths . in cases
where this component is absent , torus models with a strong emphasis on emission at @xmath43@xmath3610@xmath3 m are favored in the fits to accomodate ( in particular ) the mips photometry . by design
such torus models contribute less flux at longer wavelengths ( @xmath43@xmath4220@xmath3 m ) thus requiring a hotter fir component to match the _ herschel _ photometry .
this in turn negatively affects the fit in the observed sub - mm / mm regime . including the nir component , more power in the torus component can shift to slightly longer wavelengths , allowing a cooler fir component and providing a better overall fit to the data .
the need to add an additional hot component to torus models when fitting type-1 agn seds has also been noted by e.g. @xcite .
nir reverberation mapping observations ( e.g. * ? ? ?
* ) show that the size of the emitting region of this very hot dust is a factor of @xmath220 smaller than the torus as measured in the mir via interferometry ( burtscher et al .
, in preparation ) , supporting this additional complexity in the distribution of the agn - heated dust .
the following results and discussion will therefore be based on the fits including the additional nir black body .
m band for the redshifts of the sources in this plot ( @xmath44).[flux_ratios ] ]
in our new _ herschel _ observations we detect six out of eleven sources ( tab.[photometry ] ) .
typically , the quasars are either detected in all five bands or not at all with _ herschel_. this is an important point which will be revisited in section [ sec : stacking ] .
the exceptions are j0927 + 2001 which is detected with spire but not with pacs and j1044@xmath251025 for which the opposite is the case .
from the ten objects observed with _ spitzer _ ,
all are detected in all bands . with wise at 12@xmath3 m
only three objects are detected and at low significance .
more detailed information on the individual objects can be found in the appendix .
previous studies of star formation in high - redshift ( @xmath0 ) quasars often had to rely on single band mm emission as a tracer for the starburst heated dust ( e.g. * ? ? ?
* ; * ? ? ?
far - infrared luminosities were determined by fitting a single modified black body to the mm photometry and integrating under this component . since no knowledge about the temperature of the dust was at hand for most cases , typical values found for fir bright quasars at lower redshift ( @xmath45@xmath2@xmath46 ) were assumed ( e.g. t = @xmath47k , @xmath48 ; @xcite ) .
this approach has been tentatively supported for some high - redshift quasars by ground - based observations at 350@xmath3 m @xcite .
the new multi - wavelength fir photometry now allows us to estimate the temperature of the fir emitting dust directly while simultaneously accounting for the contributions from agn heated nuclear dust to the infrared . in figure
[ sed_fits_detected ] we present the seds of the five objects which are detected in at least two _ herschel _ bands and their accompanying best fits .
as explained above , details of the sed of the agn heated dust torus affects the shape ( and temperature ) of the fir component .
we have taken this into account when calculating the uncertainties of the dust temperature ( see section [ sec : error_estimates ] ) .
the values of t@xmath4 we obtain here for these five objects are reported in table [ results ] .
we clearly see a range of temperatures among these objects , spanning almost 20k .
one important result from the sed fits in figure [ sed_fits_detected ] is that the flux at @xmath43@xmath49@xmath4250@xmath3 m is usually dominated by the fir excess emission with only minor contributions from the agn heated torus .
this is further illustrated in fig.[flux_ratios ] where we show the fractional contribution of the hot dust plus dusty torus component ( both presumably powered by the agn ) compared with the contributions from the fir excess component , which may be powered by star formation . as indicated in this figure , the spire 350@xmath3 m band is typically dominated by emission from the fir black body for the redshifts considered here .
in fact , fitting only a single modified black body ( @xmath50 ) to the photometry at @xmath43@xmath51350@xmath3 m gives very similar dust temperatures compared to the full sed fits .
this result can immediately be utilized for an estimate on the dust temperature for objects which do not qualify for full sed fits ( see figure [ sed_fits_nondetected ] ) . in these cases
we use the 250ghz detection as an anchor for the ( modified ) black body while the fir upper limits at @xmath43@xmath51350@xmath3 m ( mostly from spire ) allow us to constrain the maximum permitted temperature of this component . because of our findings from the full sed fits ( fig.[flux_ratios ] ) we can assume that our upper limits to the dust temperature are reasonably robust when limiting the fits to @xmath43@xmath51350@xmath3 m , even without suitable constraints on the agn dust emission in these objects .
we determine dust temperatures of @xmath4057k , with j0818 + 1722 being the only exception ( t@xmath52@xmath4071k ) .
however , the latter value has to be taken with caution because the 250ghz photometry of this source could be contaminated by the emission from the nearby galaxy that is detected in most infrared bands ( see notes on individual objects in the appendix and figure [ images ] ) . in the previous fits ,
the emissivity index @xmath39 was fixed to a value of 1.6 to enable the comparison with earlier literature studies .
the good photometric coverage in the fir and sub - mm for some of our objects now allows us to explore in how far @xmath39 can be constrained using the quasar seds .
for such a study , additional photometry at lower frequencies ( typically around 90ghz in the observed frame ) is very important as it helps to further constrain the rayleigh - jeans tail of the fitted dust component .
in fact , considering only photometry at @xmath53@xmath54250ghz does not provide good constraints on @xmath39 ( or the fir dust temperature ) if both parameters are kept free during fitting .
photometry at @xmath290ghz in combination with _
fir detections is available for four objects in our sample .
re - fitting the seds as outlined previously but now keeping @xmath39 as a free parameter , we find relatively high @xmath39 values ( @xmath55 ) combined with relatively low temperatures ( @xmath233k , but still 54k for j1148 + 5251 ) .
the integrated luminosity of the fir dust when determined from these new fits remains virtually unchanged as compared to a fixed @xmath39 approach .
we caution , however , that a reliable measure of @xmath39 is hard to obtain in these objects because the peak of the dust emission is not well defined ( or isolated ) in the sed due to the strong nuclear dust emission from the torus .
therefore , the peak wavelength ( and temperature ) of the fir black body depends on the choice of the torus model which adds additional uncertainty in the determination of @xmath39 .
lcccccccc name & @xmath56@xmath57 & l@xmath57 & l@xmath58 & t@xmath4 & l@xmath4 & % of & sfr & m@xmath11 + & & @xmath59ergs@xmath15 & @xmath59ergs@xmath15 & k & @xmath60l@xmath7 & l@xmath4 & @xmath61m@xmath7yr@xmath15 & @xmath62m@xmath7 + ( 1 ) & ( 2 ) & ( 3 ) & ( 4 ) & ( 5 ) & ( 6 ) & ( 7 ) & ( 8) & ( 9 ) + j0203 + 0012 & @xmath63 & @xmath64 & @xmath3710.5 & @xmath3757 & @xmath371.3 & & @xmath372.2 & + j0338 + 0021 & @xmath65 & @xmath66 & @xmath67 & @xmath68 & @xmath69 & 24 & @xmath70 & @xmath71 + j0756 + 4104 & @xmath72 & @xmath73 & @xmath74 & @xmath75 & @xmath76 & 35 & @xmath77 & @xmath78 + j0818 + 1722 & @xmath79 & @xmath80 & @xmath378.3 & @xmath3771 & @xmath371.8 & & @xmath373.1 & + j0840 + 5624 & @xmath81 & @xmath82 & @xmath378.8 & @xmath3745 & @xmath370.9 & & @xmath371.6 & + j0927 + 2001 & @xmath83 & @xmath84 & @xmath377.3 & @xmath85 & @xmath86 & 62 & @xmath87 & @xmath88 + j1044@xmath250125 & @xmath89 & @xmath90 & @xmath91 & @xmath3753 & @xmath371.2 & @xmath3724 & @xmath372.1 & + j1048 + 4637 & @xmath92 & @xmath93 & @xmath3711.5 & @xmath3756 & @xmath371.6 & & @xmath372.7 & + j1148 + 5251 & @xmath94 & @xmath95 & @xmath96 & @xmath97 & @xmath98 & 54 & @xmath99 & @xmath100 + j1335 + 3533 & @xmath92 & @xmath101 & @xmath377.3 & @xmath3755 & @xmath371.4 & & @xmath372.4 & + j2054@xmath250005 & @xmath102 & @xmath103 & @xmath3722.5 & @xmath3754 & @xmath371.3 & & @xmath372.2 & + + integrating the emission of the fir excess component only ( i.e. the modified black body ) between 8 and 1000@xmath3 m in the rest frame allows us to determine its luminosity l@xmath4 .
the fir luminosities we derive are of the order of @xmath210@xmath6l@xmath7 ( see tab.[results ] ) , which corresponds to @xmath225 - 60% of the bolometric fir luminosity in these objects .
for sources with full sed fits , the values for l@xmath4 do not change significantly ( @xmath104% ) if we fit only a single modified black body to the photometry at @xmath105@xmath51350@xmath3 m . assuming that the fir emission is powered by star formation , we converted the fir luminosities into star - formation rates using the formula in @xcite .
the results are reported in table [ results ] . using our fits of the fir excess component we can also derive an estimate for the dust masses in the star forming regions : @xmath106 where @xmath107 is flux level at a rest frame wavelength of 250@xmath3 m as determined from the _ fit _
, @xmath108 is the luminosity distance , @xmath109 is the dust absorption coefficient at 250@xmath110 m as determined from the models of @xcite , and @xmath111 is the value of the planck function with temperature @xmath112 at a wavelength of 250@xmath110 m .
the results are also reported in table[results ] . in order to estimate uncertainties in the derived parameters we studied the distribution of their values resulting from all the fitted models ( n@xmath21000 ; as outlined in section [ sec : fitting ] we only include torus models with inclinations of @xmath4045 degrees ) .
this also allows us to account for the influence the choice of a particluar torus model has on these parameters . as a first step
, we calculated for all the fitted models the residuals between the global fit and the observed data in the infrared ( @xmath113 m ) and determined the error - weighted rms for these points . a typical distribution of the resulting rms values is presented in panel (a) of figure [ uncertainties ] ( we here use the quasar j0756 + 4104 as an example to demonstrate our approach ) . in this figure
we see a clear peak at low rms values representing a family of good fits , with an extended tail to large rms values corresponding to increasingly worse model representations of the observed sed .
we then fitted the right side of the rms peak with a gaussian ( dashed line ) .
all fits with an rms value within 3@xmath31 of the centroid value of the gaussian are identified as acceptable model fits .
the values corresponding to these fits are marked as blue and hashed regions in all panels of figure [ uncertainties ] and are used for estimating uncertainties on the derived values .
each of the n@xmath21000 model fits provides a value for the temperature of the modified black body in the fir ( t@xmath4 , panel (b) in fig.[uncertainties ] ) .
from the temperature and the normalization of this component we can then calculate ( see section [ sec : l_fir ] ) l@xmath4 ( panel (c) ) , a star - formation rate ( panel (d) ) , and a dust mass m@xmath11 ( panel (e) ) . in each histogram
we fit the parameter values obtained from the acceptable fits ( as determined from the residual rms distribution ; blue and hashed regions ) with a gaussian ( dashed line ) .
this gaussian fit provides us with a mean parameter value ( centroid ) and an uncertainty ( @xmath31 ) for each fitted object .
these results are reported in tab.[results ] . +
the average temperature of the modified black body used to model the fir emission is comparable to the @xmath247k measured for lower redshift fir bright quasars @xcite .
this finding is confirmed by the average seds presented in section [ sec : stacking ] below .
while this is true on average , we see a significant spread in dust temperature ( @xmath220k ) between individual objects , even for comparable fir luminosities . despite the low number of objects for which such fits can be performed
, this finding highlights that the choice of the dust temperature can add uncertainty to the estimate of @xmath114 and m@xmath11 , in particular for objects with only single photometric measurements and thus no individual constraints on @xmath115 .
a related issue is that of possible agn contributions to the heating of the fir dust which will be discussed briefly in the following section .
we also find that in our modeling strategy the fir component can be isolated from the torus component if only data at @xmath43@xmath116@xmath4250@xmath3 m are considered ( see section [ sec : t_fir ] and fig.[flux_ratios ] ) .
consequently , single - component fits to data at these wavelengths yield estimates of the fir luminosity , temperature and dust mass that match the values based on the full sed fits .
this result validates the approach in previous studies of high-@xmath45 quasars @xcite in which single - component fits to ( ground - based ) photometry at @xmath105@xmath51@xmath117 m was used to derive physical parameters ( with the caveat of unknown dust temperature in some of these studies ) .
it also adds further significance to the upper limits on @xmath112 , @xmath118 , and sfr we determine for the remainder of our sample where we are limited to single component fits at @xmath105@xmath51@xmath117 m .
the strong overlap between the torus and the fir components in our fits does not provide good constraints on the emissivity index @xmath39 of the latter component .
the difficulty of determining reliable @xmath39 estimates in objects with a strong agn is also apparent from the literature : @xcite find a high @xmath39 value of @xmath22 ( with a fir dust temperature of 41k ) by combining the available photometry for a number of @xmath45@xmath24 quasars into a single sed and using this global sed to constrain a modified black body fit . on the other hand , using a similar approach ( and much of the same data ) , @xcite find @[email protected] and t@xmath4@xmath247k for a sample of quasars with @xmath45=@xmath119 . in order to better constrain the fir emission of the spire non - detected objects , we stacked the individual spire observations at the nominal position of the quasar ( excluding j0818 + 1722 due to possible confusion issues)0125 from this stack .
while the source is also not detected with spire , its pacs detection would influence the average pacs flux of this subsample significantly . ] . even in the stacked images no significant detection was achieved ( see fig.[stacked_images ] ) . stacking the corresponding pacs data we recover a faint ( @xmath23@xmath31 ) average signal at 100 and 160@xmath3 m .
we iterated during the stacking , leaving a different source out of the stack for every iteration to verify that the result is not biased by any individual object .
differences between these stacks were usually smaller than the uncertainty on the photometry of the total stack .
the photometry was performed in an identical manner to the individual frames and as outlined in section2 . using the stacked _ herschel _ fluxes and combining them with stacked wise data as well as averaged _ spitzer _ and mm photometry
, we can produce an average sed for the fir non - detected objects which is presented in fig.[average_seds ] .
performing the same stacking / averaging procedure for the objects detected in the fir individually ( excluding j1148 + 5251 due to possible confusion issues ) also provides us with an average sed for these objects . fitting the average fir seds with a modified black body ( at @xmath105@xmath51350@xmath3 m , @xmath39=1.6 ) reveals dust temperatures of 47k for the stack of the individually detected objects , as expected considering the individual fir dust temperatures of the objects in this stack ( tab.[results ] ) . for the average sed of the objects individually undetected in spire we determine an upper limit of 49k which is lower than the upper limits determined individually ( tab.[results ] ) .
these temperatures are very similar to the values commonly adopted for mm detected quasars with only few or single photometric data points ( e.g. * ? ? ?
* ; * ? ? ?
they also agree well with the dust temperatures observed for sub - mm galaxies of comparable fir luminosity @xcite . in fig.[average_seds
] we also compare our average seds with the sdss quasar template of @xcite .
when normalized at 1@xmath3 m ( using a power - law fit to the rest frame uv / optical emission ) , we see that below @xmath230@xmath3 m the average sed of the spire non - detections ( blue points in fig.[average_seds ] ) looks similar to the typical sdss quasar template . at longer wavelengths we see an additional cool dust component for our sources , as indicated by the available mm observations .
not considering these longer wavelengths , we see agn powered emission with possibly increased emphasis on the hottest dust ( nir ) when compared to the template . ) to the observed photometry at @xmath43@xmath54350@xmath3 m .
the solid black line shows the mean sdss quasar template presented in @xcite , matched in the uv / optical and extended by a power law of @xmath120 at the longest wavelengths .
all seds shown here were normalized at a rest frame wavelength of 1@xmath3 m .
[ average_seds ] ] in contrast , for a similar scaled uv / optical emission , the average infrared sed of the spire detected quasars ( red points in fig.[average_seds ] ) exceeds the scaled template significantly at any wavelength @xmath1@xmath421@xmath3 m .
this includes the presumably agn powered dust emission at short infrared wavelengths ( @xmath41@xmath3620@xmath3 m ) as well as the fir emission possibly powered by additional star formation .
the observed discrepancy between the average seds in figure [ average_seds ] is somewhat puzzling .
both groups of objects have similar uv / optical properties ( on average ) , indicating that their black holes grow at comparable rates .
now , assuming that the nir and mir emission is powered by the agn , the objects with individual _ herschel _ detections ( red symbols ) convert a higher fraction of their accretion luminosity into re - processed dust emission .
this could , in principle , be caused by a different dust geometry ( or dust content ) in the inner parts of these objects .
this in turn could also lead to increased contributions of agn powered dust emission at fir wavelengths .
if agn - powered dust emission extends further into the fir than anticipated by the torus models we utilize here , the values for the inferred star - formation luminosity ( and star formation rates ) in these objects ( tab.[results ] ) would be overestimated . while a detailed discussion of this issue is beyond the scope of this paper , it is worth keeping in mind that the agn may be contributing to the heating of the fir dust depending on the dust distribution and geometry .
new _ herschel _ observations of eleven @xmath0 quasars with detections at 1.2 mm are combined with data at other wavelengths to analyze their full seds covering the rest frame wavelength range of @xmath20.1 - 400@xmath3 m .
our results can be summarized as follows : * six out of the eleven objects are detected in at least two of the five _ herschel _ bands .
five of them have sufficient data coverage and quality to allow full sed fits .
* in all cases where such fits could be performed , agn powered emission from a dusty torus is not sufficient to explain the observed fir fluxes .
an additional fir component is required to model the seds .
similar to other studies of luminous ( but lower redshift ) quasars , we notice the need for a hot ( @xmath21300k ) dust component to account for the strong rest frame nir emission . *
the additional fir component was modeled as a modified black body and shows temperatures of t@xmath240 - 60k .
we interpret this emission as being powered by star formation with luminosities of l@xmath5@xmath210@xmath6l@xmath7 which translate into star - formation rates of several thousand solar masses per year . *
our fits also allow us to estimate that the contributions of the agn powered dust to the infrared sed are small at wavelengths @xmath121@xmath4250@xmath3 m . for the redshifts of our objects
this implies that the star - formation powered fir component can be isolated and characterized adequately by single component fits if only photometric measurements at @xmath105@xmath42350@xmath3 m are considered .
this explains the good match for the temperature and luminosity of the fir emission with previous studies of such objects which relied on ( ground - based ) observations at @xmath105@xmath42350@xmath3 m .
* by stacking the _ herschel _ data of individually undetected sources we recover a signal in pacs but not in spire .
we use this stacking approach to construct average seds for objects with and without individual _ herschel _ detections .
we find that the high - redshift objects which are individually undetected with _
herschel _ show an sed similar to to quasar templates constructed from lower redshift and lower luminosity objects . * the average sed of the _ herschel _ detected objects , on the other hand , shows a surplus of nir and mir emission relative to the uv / optical when compared to the _ herschel _ nondetections or to quasar templates .
this may suggest a correlation between strong fir emission ( here modeled as star - formation powered ) and strong mir emission ( here modeled as agn powered ) , and possibly indicates significant agn contributions to the fir emission .
this publication makes use of data products from the wide - field infrared survey explorer , which is 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 .
this work is based in part on data obtained from the ukirt infrared deep sky survey ( ukidss ) .
cl acknowledges funding through dfg grant le 3042/1 - 1 .
xf acknowledges supports from nsf grants ast 08 - 06861 and 11 - 07682 .
mh is supported by the nordrhein - westflische akademie der wissenschaften und der knste .
we thank the anonymous referee for helpful comments on the paper . _
facilities : _ , .
j0203 + 0012 ( @xmath122)_.the combined optical and nir spectroscopy by @xcite reveals broad absorption line features in this quasar which could be one explanation for the abrupt change in sed slope observed at @xmath43@[email protected]@xmath3 m ( fig.[sed_fits_nondetected ] ) .
it is fairly radio bright at 1.4ghz compared to most @xmath123 quasar @xcite and is detected at 250ghz with a flux of @xmath19mjy @xcite .
these authors did not detect the quasar in co(6 - 5 ) nor in the corresponding continuum .
while clearly seen in all our _ spitzer _ observations , no detection was achieved with _
herschel_. several nearby galaxies can be identified in sdss and irac , some of which are prominent in most infrared channels .
the closest bright object is @xmath230 northeast of the quasar . from sdss spectroscopy of some galaxies in this area
it appears that they might belong to a foreground cluster at redshift of @xmath124 . _
j0338 + 0021 ( @xmath125)_.this source was detected at 250ghz ( @xmath20mjy ) , but remained undetected at 1.4ghz @xcite .
the 850@xmath3 m flux is measured to be @xmath126mjy @xcite .
@xcite report co(5 - 4 ) emission from the quasar , but no continuum at 95.6ghz is detected .
inspection of the field at optical through fir wavelengths reveals a nearby source @xmath215 to the west of the quasar .
this object is clearly visible in the sdss @xmath127 and @xmath128 bands , as well as with irac and mips . in all these filters ,
the quasar is typically the brighter source , a situation which reverses at 100 and 160@xmath3 m where the nearby object is @xmath21.4 and @xmath21.3 times brighter , respectively , than the quasar . in all spire bands ,
however , we only detect a single source .
interestingly , the quasar and the nearby object can both be seen in ground - based 350@xmath3 m observations @xcite and the flux of the quasar ( @xmath129mjy / beam ) is very comparable to the spire 350@xmath3 m flux determined for the single detection ( @xmath130mjy ) .
this suggests that the sed of the source close to the qso peaks somewhere around the pacs bands but does not contribute significantly at wavelengths @xmath131 m . _
j0756 + 4104 ( @xmath132)_.the 250ghz flux ( @xmath133mjy ) has been presented by @xcite who also detect the source at 1.4ghz and constrain the size of the radio emission to @xmath134 .
@xcite report detections at 850@xmath3 m ( @xmath135mjy ) and 450@xmath3 m ( @xmath136mjy ) where `` the source clearly appears elongated '' in the 850@xmath3 m map at a position angle of @xmath270 degrees .
we detect the quasar in all our five _ herschel _ bands , but the significance is often marginal .
our flux at 350@xmath3 m is consistent with the ground - based measurements of @xcite . _
j0818 + 1722 ( @xmath137)_.radio continuum emission at 1.4ghz is clearly detected @xcite and the 250ghz continuum is observed at the 3@xmath31 level ( @xmath22mjy ; @xcite ) .
the close inspection of our multi - wavelength images reveal a resolved foreground galaxy @xmath26 north - east of the quasar . both objects ,
the galaxy and the quasar , are individually detected by mips at 24@xmath3 m and in shorter wavelength bands .
however , only a single source is detected in pacs and spire ( note that the spatial resolution of _ herschel_/pacs at 100@xmath3 m is comparable to _
spitzer_/mips at 24 @xmath3 m ) . from the relative positions of other sources in the field we can identify the detection at 100 , 160 , and possibly at 250@xmath3 m with the foreground galaxy .
it is conceivable that the faint ( 3.5@xmath31 ) detection at 350@xmath3 m is also due to this source .
no detection is achieved at 500@xmath3 m . in the light of these results ,
higher resolution mm observation are clearly needed to determine the source of the 250ghz continuum emission . _
j0840 + 5624 ( @xmath138)_.we do not detect this source in our pacs nor in the spire data . the quasar has been detected at 250ghz ( @xmath139mjy ) , but not at 1.4ghz @xcite .
co emission is seen in this source ( but no continuum at either 85ghz or 101ghz ) and the 350@xmath3 m emission is `` marginally detected '' from the ground @xcite , which is consistent with our spire 350@xmath3 m upper limit .
@xcite also report the presence of another source visible at 350@xmath3 m as well as at 1.4ghz located @xmath230 north - west of the quasar .
this source is also detected in all our infrared bands .
inspection of the irac maps reveals this detection to coincide with two close objects which could be two slightly resolved galaxies separated by @xmath21.8as seen on an archival _ hst_/wfc3 image in the f105w filter .
_ j0927 + 2001 ( @xmath140)_.previously detected at 250ghz ( @xmath24mjy ; @xcite ) as well as in co and in the 90ghz continuum @xcite .
the 350@xmath3 m observations by @xcite show the quasar ( @xmath141mjybeam@xmath15 ) and a secondary source of equal brightness 15 to the southeast .
while the quasar detection was confirmed in @xcite with better sensitivity ( @xmath142mjybeam@xmath15 ) , the secondary source was not .
we detect the quasar with spire , but not with pacs . _
j1044@xmath250125 ( @xmath143)_.this well studied quasar shows a broad civ absorption feature in its spectrum ( e.g. * ? ? ? * ; * ? ? ? * ) . the continuum emission is detected at 850@xmath3 m ( @xmath144mjy ; @xcite ) and at 250ghz ( @xmath26mjy ; @xcite ) . @xcite report the detection of co ( 6 - 5 ) but can only give an upper limit on the continuum at 102ghz .
the quasar is not seen in ground - based observations at 350@xmath3 m @xcite and 1.4ghz @xcite .
we detect the quasar with pacs , but not with spire .
_ j1048 + 4637 ( @xmath145)_.while clearly detected in the available _ spitzer _ bands , this quasar remains undetected in our _ herschel _ photometry .
@xcite only give an upper limit on the 350@xmath110 m flux , just like @xcite at 450 and 850@xmath110 m .
these authors , however note that based on the published 1.2 mm detection ( @xmath27mjy ; @xcite ) , the scuba 850@xmath110 m observations are deep enough to enable the detection of the source with 4@xmath31 significance given a dust temperature of 40k . at even longer wavelengths ,
@xcite detect co(6 - 5 ) emission as well as the continuum at 96ghz .
observations at 1.4ghz only provide an upper limit on the source flux @xcite .
_ j1148 + 5251 ( @xmath146)_.this famous object was the highest redshift quasar known for half a decade @xcite and as such has been studied at many wavelengths , including the mm and sub - mm regime .
we also have observed this quasar previously with _ herschel_/pacs and reported detections at 100 and 160@xmath3 m @xcite .
surprisingly , we discovered a secondary object @xmath210 north - west of the quasar which is brighter at 160@xmath3 m but can still be identified at 100@xmath3 m .
ground - based data at 350@xmath3 m @xcite and 250ghz @xcite revealed an intriguing elongation of the quasar detection in the direction of the second source .
we argued in our previous paper that this could be an indication for the presence of the secondary source .
a possible counterpart is also seen in the 24@xmath3 m images . in the irac band ,
three sources are detected around the position of this secondary source , two of which can also be identified on deep _
hubble _ space telescope ( _ hst _ ) images with the acs camera in the f850lp filter .
recently , new deep _ hst _ images from wfc3 in the nir revealed also the third source seen in irac .
this object is clearly detected , but faint in f125w and gets significantly brighter in f160w .
this is our best candidate for a counterpart of the secondary source seen at 24 , 100 and 160@xmath3 m .
surprisingly , however , follow - up observations with the plateau de bure interferometer ( pdbi ) at 1.2 mm at @xmath21resolution did not yield a detection and the 3@xmath31 upper limit we derive is 0.9mjy . since our initial photometry @xcite we have re - observed the quasar with _
herschel _ and obtained new images at 70 , 160 , 250 , 350 , and 500@xmath3 m . while the quasar itself is faintly detected at 70@xmath3 m , there is no sign for a secondary source .
the new 160@xmath3 m observations confirm our earlier findings that the flux appears to come from two sources .
the source complex is also detected in all spire bands , but the spatial resolution is too low to identify a possible double source . combining the multi - wavelength photometry of the secondary source , we find that the sed is consistent with a star - forming galaxy at @xmath147 with ulirg - like luminosity ( l@xmath5@xmath2few times 10@xmath148l@xmath7 ) .
_ j1335 + 3533 ( @xmath149)_.the optical spectrum of this source is quite unusual as it shows a typical quasar continuum but virtually no ly@xmath56 emission @xcite . at longer wavelengths ,
the quasar is seen in the _ spitzer _ bands , but not in our _ herschel _ data .
@xcite report the detection of the co(6 - 5 ) transition and give upper limits on the continuum at 350@xmath3 m and 100ghz .
the 250ghz ( @xmath29mjy ) and 1.4ghz continuum was detected by @xcite . _
j2054@xmath250005 ( @xmath150)_.this is the only source in our sample for which we do not have _ spitzer _ observations .
the sdss imaging featured a sufficient number of sources that could also be identified on the pacs maps to determine the position of the quasar accurately .
the qso is detected at 250ghz ( @xmath30mjy ) , but not at 1.4ghz @xcite .
co observations revealed the ( 6 - 5 ) transition but no continuum at 98ghz @xcite .
we see the source at 160@xmath3 m with pacs , but can only give an upper limit on the 100@xmath3 m flux .
a faint 3 - 4@xmath31 source is visible at 250 and 350@xmath3 m . at 500@xmath3 m
we run into confusion issues with a source located @xmath230 north of the quasar s nominal position , which can also be identified ( separate from the quasars ) in the other spire bands and at 160@xmath3 m .
since our photometry indicates that the fir peak of the quasar in f@xmath151 falls close to the 250@xmath3 m band , we do not expect significant flux in the 500@xmath3 m channel . in a nir spectrum ,
@xcite see a very strong mgii absorber at @xmath152 . | we present _ herschel _ far - infrared ( fir ) photometry of eleven quasars at redshift @xmath0 that have previously been detected at 1.2 mm .
we perform full spectral energy distribution ( sed ) fits over the wavelength range @[email protected] - 400@xmath3 m for those objects with good _ herschel _ detections .
these fits reveal the need for an additional far - infrared ( fir ) component besides the emission from a dusty agn - powered torus .
this additional fir component has temperatures of t@xmath4@xmath240 - 60k with luminosities of l@xmath5@xmath210@xmath6l@xmath7 ( accounting for 25 - 60% of the bolometric fir luminosity ) .
if the fir dust emission is due to star formation it would suggest star formation rates in excess of 1000 solar masses per year .
we show that at long wavelengths ( @xmath8 m ) the contribution of the agn - powered torus emission is negligible .
this explains how previous fir studies of high - redshift quasars that relied on single component fits to ( ground - based ) observations at @xmath9 m reached t@xmath4 and l@xmath4 values similar to our complete sed fits .
stacking the _ herschel _ data of four individually undetected sources reveals a significant average signal in the pacs bands but not in spire .
the average sed of sources with individual _ herschel _ detections shows a striking surplus in near- and mid - infrared emission when compared to common agn templates .
the comparison between two average seds ( sources with and without individual _ herschel _ detections ) matched in the uv / optical indicates that for these objects the strength of the mir emission may correlate with the strength of the fir emission . |
the phase of a wave function plays an interesting role in quantum mechanics .
it plays a special and basic role in the quantum behaviour of a system . on the one hand , the introduction of an extra phase for a whole quantum state does not change the expectation value of an observable quantity . on the other hand
, the phase difference between two quantum states may lead to observable interference effect .
atom interferometry , one of the most promising technologies suggested by modern atomic physics , is based on the phase differences between two or more states via corresponding paths @xcite . in an atom interferometer ,
the kapitza - dirac ( kd ) scattering of matter waves from a standing wave is a very powerful and versatile tool for the coherent splitting and/or mixing of momentum modes @xcite .
the study on the kd scattering of atoms by light has been numerous , both in experiments @xcite and in theories @xcite . according to the intensity and duration of the atom - light interaction ,
it is usually divided into the bragg , the raman - nath and the channeling regimes @xcite .
the bragg scattering , which is limited to regime of low intensity and long pulse , relies severely on the energy - momentum conservation ( bragg condition ) @xcite . in the high intensity and short pulse , the so - called raman - nath ( rn ) regime @xcite ,
the atomic displacement along the standing wave is small which leads to the neglecting of the atomic kinetic energy term in the hamiltonian . in the channeling regime ,
numerical methods are usually needed @xcite .
nevertheless , there is still lack of an analytical expression for the diffraction amplitude over the whole regime , not even an approximate one . besides
, the diffraction phase is rarely studied . in ref .
@xcite , henkel _ et al .
_ discussed the diffraction phase near " the rn regime by expanding the action integral along the classical trajectory of atoms perturbatively . here
we study the diffraction of an atom from a standing - wave pulse by an evolution operator method @xcite , which is not limited by the intensity or duration of the pulse , compared with the usual method for bragg scattering , rn regime and channeling regime .
the evolution operator is calculated through the dyson expansion series .
the elements of the evolution matrix are found to be complex , giving rise to nonzero diffraction phases . applying the evolution operator to an atom initially in the zero momentum state , approximate analytical expressions for the diffraction amplitudes of different orders
are obtained , from which the corresponding diffraction phases can be calculated straightforwardly .
the analytical dependence of the amplitudes and phases on the pulse intensity and duration shows good consistency with the numerical results .
the diffraction phase has no observable effect in the relative population in the one - pulse scattering .
interferometry is one of the methods to observe the diffraction phase @xcite .
however , it is not easy to detect the above mentioned diffraction phase in interferometers realized in the rn regime or with the bragg scattering , partially because the diffraction phase is small in the rn regime or it may cancel in some interferometers . in order to explore the effect of the diffraction phases of different orders ,
a temporal talbot - lau ( tl ) interferometer @xcite is demonstrated experimentally with a bose - einstein condensate ( bec ) .
the temporal tl interferometer consists of two identical kd pulses separated by an interval , which is varied in the experiment .
the intensity and duration of each pulse are large enough , so that the diffraction is beyond the rn regime , as a result of which the amplitude of the second diffraction order is considerable .
the relative population of the bec around the zero momentum state after the tl interferometer is measured as a function of the interval . in the population oscillation
, an asymmetry induced by the diffraction phase differences is observed between the rising and the falling edges .
the experimental observation agrees well with our analysis including the high - order diffraction phases .
we further discuss the role of interatomic interaction in the tl interferometer .
the paper is organized as follows . in sec .
ii , we calculate the evolution operator for an atom scattered by a standing - wave pulse and apply it to an atom initially in the zero momentum state . in sec .
iii , the effect of the diffraction phases is studied experimentally and theoretically using a temporal tl interferometer with a bec .
a brief summary and discussion is given in sec .
iv . direction .
( b ) time sequences of one pulse and two pulses . the one pulse process , which is illustrated by solid curves , is turned on between @xmath0 and @xmath1 .
the strength and duration of the pulse is @xmath2 and @xmath3 , respectively .
a second identical pulse is applied after a variable interval @xmath4 ( dotted curve ) .
( c ) a typical tof image obtained in our experiment . as we can see , the fraction of @xmath5 order is considerable.,width=302 ]
we first consider the diffraction of an atom by a standing - wave pulse as shown in fig .
[ fig : scheme ] .
the standing wave pulse consists of a pair of linearly polarized , counter - propagating laser beams along the @xmath6 direction ( fig .
[ fig : scheme](a ) ) .
the duration of the pulse , which is turned on at @xmath7 and turned off at @xmath8 , is @xmath9 ( fig .
[ fig : scheme](b ) ) .
the detuning of the laser beams is far enough we can ignore the spontaneous emission .
the hamiltonian for a single atom diffracted by the standing wave light field is given by @xmath10 where the first term in the right hand side is the single particle kinetic hamiltonian @xmath11 with the atomic momentum operator @xmath12 and atomic mass @xmath13 .
the second term in the right hand side is the time - dependent scattering potential , with @xmath14 describing the shape of the pulse .
@xmath15 represents the effective interaction between the atom and the light field , which is a spatially periodic potential @xcite , with the position operator @xmath16 , the trap depth @xmath2 and the wave vector @xmath17 ( @xmath18 is the wavelength of the laser ) . due to the position operator @xmath16 ,
@xmath19 can be described as a momentum shift operator : @xmath20 @xcite . in the momentum representation ,
the periodical potential @xmath21 , apart from a constant term , can be written in an operator form as @xmath22 , which explicitly shows the momentum change @xmath23 of an atom scattered by the standing wave @xcite .
this process is accomplished by absorbing a photon from one of the laser beams and emitting it into the counter - propagating one .
the iteration of such a two - photon process results in an equally spaced distribution separated by @xmath23 in the momentum space . with a standard procedure @xcite
, the evolution operator for the one - pulse scattering process can be obtained via the dyson expansion series : @xmath24 where the zeroth order term @xmath25 is the unitary operator , and @xmath26 \end{split } \label{eq : un}\ ] ] is the @xmath27 order term with @xmath28 . here
@xmath29 is the hamiltonian in the interaction picture .
each @xmath30 in eq .
( [ eq : un ] ) stands for a single two - photon scattering process , and their product at different time represents all possible configuration of the @xmath27 order two - photon processes . with the expression for @xmath31 , it is straightforward to get @xmath32 where @xmath33 is the detuning of the two - photon process for an atom of momentum @xmath34 , which simply reflects the energy and momentum conservation for a free particle .
the second term of @xmath35 is the doppler shift @xcite , which is neglected in ref .
@xcite .
the evolution operator @xmath36 can be expressed as a matrix of infinite dimension , whose matrix element @xmath37 represents the transition amplitude from the initial state @xmath38 to the final state @xmath39 in the interaction picture , with @xmath40 the quasi - momentum . from eq .
( [ eq : alm ] ) we can see that , @xmath41 depends not only on the quasi - momentum @xmath42 , but also on the starting time @xmath7 and the ending time @xmath8 of the pulse .
additionally , some properties of @xmath41 can be drawn out without knowing its explicit expression .
firstly , the evolution matrix @xmath36 is symmetric with respect to the main diagonal , which means @xmath43 , because of the real hermite operator @xmath31 .
secondly , the transition probability should not depend on the starting time but the duration of the pulse , so one has @xmath44 .
the matrix element @xmath45 can be expanded as , @xmath46 where @xmath47 is the @xmath27 order expansion coming from the corresponding term in the dyson series .
substituting eq .
( [ eq : vi ] ) into eq .
( [ eq : un ] ) , we have @xmath48 [ eq : an_expansion ] the transition amplitude of the zeroth order approximation ( [ eq : alm:0 ] ) is independent of @xmath49 and equals unity only when the momentum state does not change , otherwise it equals zero . eq .
( [ eq : alm:1 ] ) and eq .
( [ eq : alm:2 ] ) reflect the first and second order of the interaction between the atom and the laser pulse , respectively .
the kronecker @xmath50 function is due to the special form eq .
( [ eq : vi ] ) of the periodical scattering potential , reflecting the momentum change @xmath23 in a single two - photon process .
however , cascade of two - photon processes may result in no momentum change , which can be seen by the expansion of the two kronecker @xmath50 functions .
as we have mentioned , @xmath41 depends on both @xmath7 and @xmath8 , which comes from the kinetic operator @xmath11 .
the dependence can be removed through the picture transformation by introducing the transition amplitude in the schrdinger s picture @xmath51 { a}_{l , m}\exp[ie_m(q)t_i/\hbar]$ ] @xcite . here
@xmath52 and @xmath53 are eigen - values of @xmath11 for states @xmath54 and @xmath55 , respectively
. then @xmath56 depends only on the quasi - momentum @xmath42 and the pulse duration @xmath57 .
it is noticed that the scattering amplitude @xmath56 is a complex quantity , characterized by a modulus and a phase @xmath58 the modulus determines the transition probability that the atom is scattered from state @xmath55 to state @xmath54 .
the phase @xmath59 , which stems from the detuning @xmath60 as we can see from eq .
( [ eq : an_expansion ] ) , is known as the diffraction phase @xcite .
the study on the diffraction phase is rare , even it is of crucial importance in the atom interferometry , as it may modify the interference signals .
the diffraction phase has no observable effect in the relative population in the one - pulse scattering , as it does not change the transition probability , which is what we can measure after the pulse when all orders are well separated .
however , it plays a special role in the interference of two - pulse scattering . as we will show in sec .
[ sec : exp ] , if another pulse is applied after some variable interval @xmath49 , the role of the diffraction phase can be clearly seen . as a special case
, we consider the diffraction of an atom by a square shaped standing - wave pulse as shown by the solid line in fig .
[ fig : scheme](b ) . in this case , the initial momentum @xmath61 , which means @xmath62 , @xmath63 , and @xmath64 with @xmath65 the heaviside step function .
we denotes @xmath66 and @xmath67 by @xmath68 and @xmath69 , respectively . as @xmath63 , @xmath70
, the diffraction should be symmetric with respect to the zeroth order , which means @xmath71 . substituting the value of @xmath72 and @xmath42 and the expression of @xmath14 into eq .
( [ eq : an_expansion ] ) , up to the second order , we can get approximate expressions for the scattering amplitude @xmath73 , which depend on the pulse depth @xmath74 in unit of the single photon recoil energy @xmath75 and duration @xmath57 explicitly : @xmath76 [ eq : phase ] with @xmath77 , which is a basis unit of the scattering process .
the diffraction phase obtained from @xmath78 for different order can be extracted directly from the corresponding expressions . when @xmath79 , @xmath80 reduces to @xmath81 . and ( a2)-(c2 ) the diffraction phase @xmath80 of the order @xmath82 vs the pulse duration @xmath3 , respectively .
the strength of the pulse is @xmath83 . the blue solid , red dashed and black dash - dotted lines represent results of num , eq .
( [ eq : phase ] ) and rn approximation , respectively .
, width=340 ] speaking of the kd scattering , one has to mention the well - known rn approximation , which neglects the effect of kinetic energy during the pulse and is quite often used in the field of atom optics @xcite .
the transition amplitude of the rn approximation is simply @xmath84 @xcite , with @xmath85 and @xmath86 the bessel function of the first kind .
the scattering phase in rn approximation is thus @xmath87 , where the absolute value of @xmath88 comes from the fact @xmath89 .
this is consistent with the results of eq .
( [ eq : phase ] ) . in fig .
[ fig : phase ] , we show the dependence of the population @xmath90 and the diffraction phase @xmath80 ( l=0,1,2 ) on the pulse duration @xmath57 with @xmath83 , respectively , according to eq .
( [ eq : phase ] ) ( red dashed line ) .
results by exact diagonalization of the hamiltonian ( [ eq : hamiltonian ] ) numerically @xcite ( for convenience we refer to this method as num hereafter ) ( blue solid line ) and rn approximation ( black dash - dotted line ) are shown together for comparison .
as shown by fig .
[ fig : phase](a1)-(c1 ) , the populations given by all methods are consistent with each other when the duration @xmath57 is small .
these three methods deviate as the pulse duration increases .
nevertheless , the trends of the amplitudes from different models are almost the same over a large range of pulse duration . however , the dependences of the diffraction phases on the pulse duration ( fig .
[ fig : phase](a2)-(c2 ) ) disagree qualitatively .
although all methods give the same result of @xmath81 when @xmath91 , the phase @xmath80 given by eq .
( [ eq : phase ] ) and num varies drastically as @xmath57 increases , while that given by rn approximation keeps a constant of @xmath81 .
the difference comes from the inclusion of the kinetic energy in the former two methods .
moreover , the approximate results of both population and phase by eq .
( [ eq : phase ] ) agree excellently with the numerical result when @xmath92s , which shows the effectiveness of our expansion method .
the failure of the approximate results at long duration is because of the finite expansion order and can be corrected by expansion to higher orders .
in sec . [ sec : theory ] , we have shown that the diffraction phases of different orders are not zero . in the one - pulse scenario , it is difficult to measure the effect of the diffraction phase . in this section ,
we show the effect of the diffraction phase using a tl interferometer with a bec .
we first describe our experimental setup and procedure .
the experiment was performed using an elongated @xmath93rb bec created in a magnetic trap as in our previous experiments @xcite .
the condensate , containing @xmath94 atoms in the internal state @xmath95 , has a thomas - fermi radius of @xmath96 m in the radial ( axial ) direction .
the axial and radial frequencies of the magnetic trap are @xmath97 hz and @xmath98 hz , respectively .
the bec in the magnetic trap was illuminated with an optical standing wave created by a retroreflected @xmath99-polarized laser beam along the axial direction ( see fig . [
fig : scheme](a ) ) . losses in the retroreflected beam were negligible .
the laser at the wavelength @xmath100 nm was focused to a waist of @xmath101 m to cover the condensate . after a variable time @xmath49 ,
a second identical pulse was applied ( fig .
[ fig : scheme](b ) . then the light pulse and the magnetic trap
were shut off simultaneously .
the momentum distribution of the condensate was imaged after @xmath102 ms free expansion , long enough for different momentum modes to be resolved ( fig .
[ fig : scheme](c ) ) . the atomic number of each momentum state was determined by integrating the optical density of the absorption image and denoted as @xmath103 .
the relative population of the momentum state @xmath104 can be extracted by @xmath105 . in the experiments ,
the depth of the pulse , which is calibrated by kd scattering , is set to @xmath83 and the durations of both pulses are @xmath106s .
the intensity and duration of each tl pulse are large enough that the diffraction is beyond the rn regime ( the rn regime is usually limited to @xmath107s @xcite ) , as a result of which the amplitude of the second order diffraction is considerable .
the interval @xmath49 between the two pulses is changed every @xmath108s up to @xmath109s . in fig .
[ fig : mainresult](a ) , the relative population @xmath110 as a function of the varied interval @xmath49 is shown by the black circles .
the error bars , corresponding to the standard deviation of three consecutive measurements of the same @xmath49 , are very small and reflect the high repeatability of our experiments .
it is clearly shown that @xmath110 oscillates with @xmath49 periodically , gaining its maximum and minimum near odd times of @xmath111s and multiples of @xmath112s , respectively .
the oscillation amplitude damps as @xmath49 increases .
besides , there is an obvious asymmetry between the rising and falling edges of the oscillation in a period . in the rising edge
there is an obvious turning point , which divides the rising edge into two parts , whose slopes differ apparently .
in contrast , the slope of the falling edge changes little . to see the asymmetry more clearly ,
we show the detail of the first period of the oscillation in fig .
[ fig : mainresult](b ) . of @xmath113 as a function of the interval @xmath4 .
the error bars correspond to the standard deviation of three consecutive measurements of the same @xmath49 .
numerical results by eq .
( [ eq:2pls ] ) ( red solid line ) , gp equation ( blue dashed line ) and rn approximation ( oliver dotted line ) including finite momentum spread are shown for comparison .
( b ) details of ( a ) within the first talbot period .
the two vertical dashed lines indicate the position of maximum of eq .
( [ eq:2pls ] ) and rn approximation , respectively , with distance @xmath114.,title="fig:",width=321 ] + of @xmath113 as a function of the interval @xmath4 .
the error bars correspond to the standard deviation of three consecutive measurements of the same @xmath49 .
numerical results by eq .
( [ eq:2pls ] ) ( red solid line ) , gp equation ( blue dashed line ) and rn approximation ( oliver dotted line ) including finite momentum spread are shown for comparison .
( b ) details of ( a ) within the first talbot period .
the two vertical dashed lines indicate the position of maximum of eq .
( [ eq:2pls ] ) and rn approximation , respectively , with distance @xmath114.,title="fig:",width=321 ] the process of the tl interferometer can be formulated by the evolution operator developed in sec .
[ sec : theory ] . assuming the starting and ending points of the second pulse as @xmath115 and @xmath116 ( see fig .
[ fig : scheme](b ) ) , respectively , the evolution operator of the second pulse is @xmath117 .
the two - pulse scattering process can be then described by @xmath118 since the evolution operator during the interval equals unity in the interaction picture . with the matrix elements @xmath41 of evolution operator @xmath119 ,
we can get the population @xmath120 of the momentum state @xmath55 after the two - pulse sequence : @xmath121 , which can be reduced to @xmath122 with the interference phase @xmath123 .
there are two components in the interference phase .
besides the kinetic phase @xmath124 of the intermediate state @xmath54 during the interval @xmath49 , the diffraction phase @xmath125 enters the interference , with a factor @xmath126 .
this factor @xmath126 comes from the two kd pulses , with each one contributing a phase @xmath127 . as @xmath43 ,
the phases from the two pulses add up rather than cancel each other .
it is this term @xmath128 that we can observe the effect of the diffraction phases . generally speaking
, the phase @xmath125 of different @xmath88 does not equal so that the overall interference phase @xmath129 is shift by a different amount , resulting in a deformation in the dependence of @xmath120 on @xmath49 .
because of its ultra - narrow momentum spread ( the root mean square width of a typical @xmath130rb bec with atom number @xmath131 is @xmath132 ) , the ground state of a bec can be approximated by @xmath133 , which means @xmath62 and @xmath63 .
additionally , as @xmath83 , the highest significantly populated diffraction order @xmath134 @xcite , so @xmath88 can be limited to @xmath135 .
equation ( [ eq:2plsgeneral ] ) can be approximated well as @xmath136 where @xmath137 and @xmath138 .
here @xmath139 is an offset , and @xmath140 is the oscillation amplitude of each frequency component , respectively , with @xmath141 , @xmath142 and @xmath143 . in most experiments realized in the rn regime ,
@xmath144 , hence @xmath145 and @xmath146 are much smaller than @xmath147 , which makes it reasonable to neglect these multi - frequency terms . however , in our case , with the experimental parameters described above , we have @xmath148 with @xmath149 , which can be seen from fig .
[ fig : phase ] .
this means that @xmath150 , @xmath151 and @xmath152 are of the same order , while the former two are only several times larger than the latter one . in this case
, @xmath147 is only about an order of magnitude larger than @xmath145 and @xmath146 , which means that the multi - frequency terms can not be neglected any more . from eqs .
( [ eq:2plsgeneral ] ) and ( [ eq:2pls ] ) we can see that the oscillation of @xmath110 with @xmath49 comes from the cosine functions whose frequencies are multiples of the recoil frequency .
the oscillation period is then the talbot period @xcite @xmath153s .
the periodicity can be understood as follows . during the interval ,
atoms in each momentum state evolve freely according to their kinetic energy , accumulating phases from @xmath154 to @xmath155 .
as the phases change with the interval , the probability amplitudes of atoms in different momentum states interfere constructively and destructively periodically when combined by the second pulse @xcite .
the effect of the phase difference in each cosine function is as follows .
according to eq .
( [ eq:2pls ] ) , there are three frequency components in the oscillation of @xmath110 with @xmath49
. the amplitude ratio of each frequency component determines the number of peaks in one period of the talbot time .
as we have argued , @xmath147 is about an order of magnitude larger than @xmath145 and @xmath146 .
the dependence of @xmath110 on @xmath49 is then determined by the biggest amplitude single - frequency term @xmath156 , giving only one peak over a talbot period .
the phase difference @xmath157 in the single - frequency term shifts the peak by @xmath158 .
the other two multi - frequency terms add small distortion to the peak produced by the single - frequency term .
shoulders appear when the distortion is strong enough to be comparable with the amplitude of the single - frequency term , but not so strong to produce additional peaks .
the shift of the maximum position by the diffraction phase difference @xmath159 results in distortions appearing at different distance to the maximum but with same strength .
the inclusion of the diffraction phase differences @xmath160 and @xmath161 further enhances the asymmetry .
the collaboration of the phase differences results in different distortion strength as what we have observed in the experiments .
numerical result according to eq .
( [ eq:2pls ] ) is shown by the red solid line in fig .
[ fig : mainresult ] .
we see that its agreement with the experimental data is good ( the damping will be discussed later ) . without considering the phase factor @xmath162 in each term , eq .
( [ eq:2pls ] ) is reduced to rn approximation , which is described in details in @xcite . in fig .
[ fig : mainresult ] , the result by rn approximation ( oliver dotted line ) is also shown for comparison .
we can see that the peak of rn approximation locates exactly at @xmath163 .
the shift of the peak by the phase difference @xmath157 is @xmath164 . in our case
, @xmath165 gives @xmath166s , which agrees with the experimental result of @xmath167s ( see the detail in fig . [
fig : mainresult](b ) , the maximum of experimental data lies at @xmath168s while that of rn approximation lies exactly at half the talbot period @xmath111s ) .
besides , the result by rn approximation is symmetric in a period , which further confirms our argument that the asymmetry is induced by the diffraction phase differences .
we notice that in @xcite , bchner _
also study the phase shift of the diffraction process in the contrast interferometer built by gupta _
et al . _
@xcite . however , they ignore the @xmath5 term , which makes the final oscillation monochromatic ( see eq . ( 15 ) in @xcite ) .
although a phase difference between the zeroth and first order diffractions is introduced , the monochromatic oscillation will not give rise to any asymmetry except a trival shift .
as the interval increases , the interaction between atoms in the condensate shows its effect . from fig .
[ fig : mainresult](a ) we can see that , as the oscillation period increases , the theory without considering the interatomic interactions deviates from the experimental data .
the deviation comes from the interatomic interaction , which can be described by the nonlinear term in the one - dimensional gross - pitaevskii ( gp ) equation @xcite @xmath169 here the condensate wavefunction @xmath170 is normalized to the total atom number @xmath171 and @xmath172 is the effective one - dimensional interaction strength @xcite , with @xmath173 the @xmath174-wave scattering length , and @xmath175 the harmonic oscillation frequency in the radial direction . using mature numerical methods @xcite
, we numerically compute the ground state of a bec in a harmonic trap and solve the time - dependent gp equation . substituting the numerically obtained ground state into the time - dependent gp equation , we can get the relative population of the zero momentum state after the two pulses .
with the experimental parameters described previously , the simulation results of different interval @xmath49 are shown by the blue dashed line in fig .
[ fig : mainresult ] .
we can see that its consistency with the experimental data is excellent , even when the interval is as large as @xmath176s .
it is pointed out that , the main effect of the interatomic interaction is to change the momentum distribution of the ground state of the bec in the harmonic trap , which leads to the oscillation damping in fig .
[ fig : mainresult](a ) . in our simulation , when the interatomic interaction is changed , the momentum spread of the ground state of bec changes accordingly , leading to different damping rate .
however , keeping the same ground state , if we change the interatomic interaction strength during the tl interferometer process , the damping rate changes little .
the oscillation damping comes from the momentum distribution of the ground state also can be reflected by the compare with the case without the atomic interaction . in fig .
[ fig : mainresult ] , the initial state @xmath133 is replaced by a gaussian wave packet with a root - mean - square momentum width @xmath177 , then the damping rates from the results of eq .
( [ eq:2pls ] ) and rn approximation are nearly the same as the experimental one .
the interaction absent damping apparently comes from the initial finite momentum spread .
the slightly larger value of the momentum spread may be caused by additional damping mechanics such as the ignored heating effect of the bec .
et al . _
@xcite and campbell _ et al . _
@xcite explained the damping as the finite coherent length of the bec . according to heisenberg s uncertainty principle
, a finite coherent length corresponds to a finite momentum spread , hence that is consistent with our explanation .
in conclusion , we have studied the diffraction of an atom from a standing - wave pulse over a wide regime via the evolution operator method .
the advantage of this method is that it is not limited by the pulse intensity and duration .
approximate analytical expressions for the diffraction amplitudes and diffraction phases of different orders are obtained .
the effect of the diffraction phases is further studied experimentally using a temporal tl interferometer with a bec .
an asymmetry between the rising and falling edges is observed in the population oscillation .
the experimental result agrees well with our analysis including the high - order diffraction phases .
our work implies that the diffraction phase plays an essential role in getting a better quantitative analysis of the matter wave interference experiments , particularly in precision measurements .
we thank peng zhang , biao wu , hui zhai and han pu for helpful discussions .
this work is supported by nkbrsfc ( 2011cb921501),nsfc ( 61027016,10934010,11175246),the rfdp number 20120001110091 and ncet-10 - 0209 . | we study the diffraction phase of different orders via the dyson expansion series , for ultracold atomic gases scattered by a standing - wave pulse .
as these diffraction phases are not observable in a single pulse scattering process , a temporal talbot - lau interferometer consisting of two standing - wave pulses is demonstrated experimentally with a bose - einstein condensate to explore this physical effect .
the role of the diffraction phases is clearly shown by the second standing - wave pulse in the relative population of different momentum states .
our experiments demonstrate obvious effects beyond the raman - nath method , while agree well with our theory by including the diffraction phases .
in particular , the observed asymmetry in the dependence of the relative population on the interval between two standing - wave pulses reflects the diffraction phase differences .
the role of interatomic interaction in the talbot - lau interferometer is also discussed . |
experimental evidence for various types of symmetry breaking in the pseudogap region of the phase diagram of the high-@xmath0 cuprate superconductors has been accumulating in recent years .
neutron scattering experiments discovered a subtle staggered magnetic order in the pseudo - gap region of ybco@xcite and hg - compounds@xcite that could be accounted for by either so - called nematic - spin - nematic order@xcite or circulating current loops@xcite . on the other hand ,
neutron scattering @xcite and nernst effect @xcite measurements on ybco as well as si - stm on bscco @xcite point towards an electronic nematic state .
all these states retain the translational symmetry of the underlying crystal and can thus naturally be described by breaking intra - unit - cell ( iuc ) symmetries .
hence , identifying mechanisms for these symmetry - breaking possibilities and understanding their competition is crucial for understanding the nature of the pseudogap phase .
theoretical investigations of translationally - invariant iuc order have so far been focussing on one particular ordering at a time within simplified models each aimed at the ordering of interest .
nematic and nematic - spin - nematic order have only been studied in one - band models@xcite or in the extreme limit of infinite interactions@xcite .
loop currents , being more dependent on an iuc picture , have been studied in a mean - field picture with additional assumptions@xcite or numerically on small clusters or ladders@xcite . here , we aim at a comprehensive investigation of iuc - ordering possibilities suggested by recent experiments .
a theoretical description of intra - unit - cell order should be based on a microscopic model that allows for structures within the unit - cell .
we thus start from a three - band model for the cuo@xmath1 plane , the so - called emery model@xcite , and consider various on - site and nearest - neighbor ( nn ) interactions ( see fig .
[ fig : unitcell ] ) .
we analyze three distinct iuc orders : nematic , nematic - spin - nematic , and loop currents .
these phases can be distinguished by the respective symmetries they break , both of the point group @xmath2 and time reversal , as is summarized in tab .
[ tab : sym ] . for simplicity , only fourfold rotations , inversion , time reversal and combinations of these are shown . within a self - consistent mean - field scheme we analyze and compare the origins of these phases and compare the influence of the different model parameters on them .
in addition , we show how the o - o on - site and nn interactions result in effective interactions of @xmath3 symmetry in one - band models , thus naturally leading to nematic or nematic - spin - nematic order .
.the broken symmetries distinguishing the different iuc orderings with @xmath4 denoting symmetries broken in the respective phase . for simplicity ,
we restrict the table to the fourfold rotation @xmath5 , the inversion @xmath6 , time - reversal operation @xmath7 as well as combinations thereof . [ cols="^,^,^,^,^,^",options="header " , ]
a comment on the magnitude of the interactions necessary found here is in order : the energy scale of the pseudogap phase in the cuprates is of order 100 kelvin . for any of the above phases to reach to such high temperatures ,
unrealistically large interactions are needed within our mean - field calculation .
looking at the nematic phase for example and taking @xmath8 , @xmath9 needs to be of order @xmath10 as can be deduced from fig .
[ fig : tcvpp](a ) . also , increasing the cu on - site interaction strength @xmath11 to enter a nematic phase leads to unphysically large values .
for the nematic - spin - nematic phase , the respective interactions need to be even larger as can be seen in fig . [
fig : upc ] . finally , for the loop - current phase , we deduce values for the critical interaction strengths from fig .
[ fig : varma ] , that are much higher than realistically expected . however , the aim of our analysis is not to answer whether this phases exist in the cuprates - a mean - field analysis would certainly not be the appropriate approach for such a task - but to analyze this different iuc orderings within the same framework .
we can also draw some conclusions about the competition or coexistence of the iuc - ordered phases from our calculation . even though symmetry allows the two orders discussed in sec .
[ sec : static ] , the nematic and the nematic - spin - nematic , to coexist independently , they are promoted by different interactions , @xmath9 and @xmath12 , each hurting the respective other phase .
in addition , both depend in a mean - field picture on the presence of a van hove singularity .
even if the interactions were tuned in a way as to allow for both instabilities , having one kind of order removes already the high density of states from the fermi level , thus preventing the system from entering the other phase .
in contrast , the loop - current phase is promoted by the same interaction as the nematic phase , @xmath9 , and does not depend on a high density of states at the fermi level .
a deformation of the fermi surface has thus no direct influence on this instability .
being of different symmetry , a nematic and a loop - current phase can therefore coexist in general independently .
the mean - field analysis and our exploration of the rich phase space of the three - band emery model in this paper can serve as a stepping stone towards more sophisticated calculations of iuc orders and their interdependence .
for instance , extension of the calculations in @xcite to the case of three bands might provide further valuable insight .
in particular , investigation of the interplay between these iuc orders and superconductivity in a genuinely strong coupling approach will be of great interest . as superconductivity can not be accessed within mean - field theory with purely repulsive interactions , we left out this important issue altogether .
the possible coexistence of nematic and loop - current phases we find in this work is interesting in light of experimental observations of both iuc nematic order@xcite and iuc staggered magnetism in underdoped ybco@xcite . on the other hand , in hg - compounds
only iuc staggered magnetism has been observed@xcite . in order to test whether coexistence of both orders is a generic feature , we propose a measurement of anisotropy in nernst effect in the presence of a symmetry breaking field on hg - compounds .
for example , an in - plane magnetic field could align possibly existing nematic domains in hg - compounds which is tetragonal otherwise , much in the same manner as in sr@xmath13ru@xmath1o@xmath14@xcite .
in this appendix , the case @xmath15 is investigated for @xmath16 for which we write the hamiltonian as @xmath17 where @xmath18 and @xmath19 with @xmath20 the unperturbed hamiltonian can straight - forwardly be diagonalized yielding the eigenenergies @xmath21 and @xmath22 with the corresponding states @xmath23 given by the eigenvectors @xmath24 in these equations , we introduced @xmath25 , @xmath26 with @xmath27 and @xmath28 for @xmath29 , we can thus express the eigenenergies in powers of @xmath30 in a text - book perturbation - theory expansion , @xmath31 with @xmath32 and @xmath33 the derivatives appearing in the self - consistency equations , and can thus all be expressed analytically through eq .
yielding @xmath34 evaluating the derivatives for the oxygen hole density in eq . , we find using @xmath35 using these derivatives , the self - consistency equation for the oxygen occupation number simplifies to @xmath36
we thank steve kivelson , michael lawler , kyungmin lee , richard scalettar , eduardo fradkin , walter metzner , hiroyuki yamase , thomas maier , shiquan su , carsten honerkamp , and andre - marie tremblay for useful discussions .
we acknowledge support from nsf grant dmr-0520404 to the cornell center for materials research and from nsf grant dmr-0955822 . | motivated by recent experiments on high-@xmath0 cuprate superconductors pointing towards intra - unit - cell order in the pseudogap phase , we investigate three distinct intra - unit - cell - ordering possibilities : nematic , nematic - spin - nematic , and current - loop order .
the first two are fermi - surface instabilities involving a spontaneous charge and magnetization imbalance between the two oxygen sites in the unit cell , respectively , while the third describes circulating currents within the unit cell .
we analyze the three - band emery model of a single cuo@xmath1 layer including various on - site and nearest - neighbor interactions within a self - consistent mean - field approach .
we show how these on - site and further - neighbor repulsions suppress or enhance particular iuc orders .
in particular , we show that the attractive interactions necessary for nematic and nematic - spin - nematic orders in one - band models have their natural microscopic origin in the o - o on - site and nearest - neighbor repulsions in the three - band model .
finally , we find that while the nematic and nematic - spin - nematic orders can not coexist in this framework , the loop - current order can coexist with nematic order . |
the basic model of a pulsar had long been recognized as a perfect conductor ( neutron star ) rotating with an angular velocity @xmath1 in a hiper magnetic field @xmath2 @xcite .
considering the rotational axis and the magnetic axis both aligned in the same sense , charged particles in the neutron star are driven by the magnetic force to the surface according to their signs , with electrons towards the polar regions and ions towrds the equatorial region .
this is the faraday unipolar induction that operates homopolar generator , which extracts the energy of a conducting rigid rotator .
due to its rotation in the presence of a magnetic field , the conducting pulsar surface is not an equipotential surface , contrary to the static case .
if the pulsar were surrounded by a vacuum , the separated charges would accumulate on the pulsar surface by their signs , and a counter electric field would build up inside the neutron star to counteract the magnetic force .
a steady state surface charge distribution or potential distribution would be reached when the two forces inside the pulsar cancel out , and there would be a quadrupole electric field in space surrounding the neutron star . the unipolar induction would cease to operate then .
however , the seminal work of @xcite noted that such a vacuum configuration was unstable , and there had to be a pulsar magnetosphere to channel the surface charges . as a result ,
the counter electric field inside the pulsar would be small , and the unipolar induction would continue to operate .
the steady state potential distribution on the pulsar surface would be uploaded to the equipotential magnetic field lines of the magnetosphere .
quite often , field line breaking near the pulsar is invoked to provide an electron - positron background pair plasma to fill the magnetosphere .
the equation of motion of a charge particle in the magnetosphere is given by @xmath3 with the plasma inertia on the left side and plasma pressure , collision , gravitational force on the right side neglected , comparing to the hiper magnetic and electric fields , the macroscopic structure of the magnetosphere is determined only by the balance between electric and magnetic forces , which is the force - free field model @xcite with @xmath4 although plasma inertia and plasma pressure are neglected in the macroscopic description , these terms in the equation of motion give rise charged particle gyro motion about magnetic field lines and transverse drift motions across field lines in the microscopic single particle description @xcite . as a result of these drifts , macroscopic drift currents and plasma flows , corresponding to the guiding center motions ,
are generated in eq .
[ eqno1 ] , as will be addressed in sec.3 . with the surface charges having ready access to the field lines , plasma density will be high enough in the magnetosphere such that plasma conductivity @xmath5 tends to be infinite ,
the ohm law reads @xmath6 for a quasi - neutral normal magnetohydrodynamic ( mhd ) plasma , the plasma velocity @xmath7 in eq .
[ eqno2 ] amounts to the center of mass velocity . for a charge separated plasma , this equation coincides with eq .
[ eqno1 ] other than an overall fator of the space charge density @xmath8 .
nevertheless , this does not mean that eq .
[ eqno2 ] is redundant , since it describes the magnetosphere in terms of velocities while eq .
[ eqno1 ] does it in terms of fields .
these two descriptions are then coupled through the space charge densities .
probably induced by the lighthouse scenario of pulsar beacons , the pulsar magnetosphere is often interpreted as such that the magnetic field lines are rigidly anchored on the pulsar surface .
the entire magnetosphere thus corotates with the pulsar at its angular velocity , and the charge separated plasma is being dragged along with it .
although for an aligned rotator and under axisymmetry , it is indistinguishable whether or not the magnetosphere is rotating . nevertheless , in an oblique rotator , it could generate the lighthouse beacons and magnetic dipole radiation . in this scenario ,
the rotating magnetic field @xmath2 drags the plasma generating a rigid rotor velocity @xmath7 . applying this rotating plasma velocity @xmath7
back to a stationary frame in space , and considering the magnetic field @xmath2 be stationary , which is indistinguishable from a rotating one , eq .
[ eqno2 ] then gives the magnetospheric electric field @xmath9 in the stationary frame .
we remark that only under the axisymmetric aligned rotator case that the magnetic field can be considered stationary and/or rotating by option .
so the plasma velocity @xmath7 is an input parameter in eq .
[ eqno2 ] , the driving energy is the rotating magnetic field , and the electric field @xmath9 is the response which is called the rotation induced electric field .
some authors introduce a non - rotation induced longitudinal field @xmath10 by taking into consideration of the interstellar plasma potential , or by considering the centrifugal outflow at the light cylinder ( lc ) @xcite .
other authors have considered a very stringent supply of positive charges from the pulsar surface , due to probably the surface physics of neutron star , and proposed that vacuum gaps could be formed separating the charged regions from the pulsar surface and/or separating the charged regions from each other @xcite . in this corotating magnetic field model ,
the unipolar induction in the neutron star is clearly decoupled from the magnetosphere .
the magnetic dipole radiation , the angular momentum outflow of the stellar winds , and the current loop in the presence of @xmath10 are attributed to the braking torque on the pulsar indirectly through the general energy conservation principle .
however , the use of the energy conservation principle implies a direct coupling with the neutron star , which is not reflected in this scenario . as a matter of fact ,
if the magnetosphere field lines are anchored on the pulsar surface , the ( dipole ) magnetic field inside the pulsar will be rotating as well .
consequently , there is no relative rotation between the pulsar and the magnetic field inside .
for this reason , there will be no unipolar induction as well .
as a matter of fact , a steady state magnetosphere decoupled from the pulsar implies that the emission problem of the beacons @xcite can be constructed from the equilibrium magnetosphere .
although the validity of separating equilibrium and emission into two different issues can not be disproved , this approach could have over simplified the pulsar dynamics . recently ,
a time dependent description of magnetosphere where the plasma equations are coupled to the maxwell equations to account for induction effects .
this broader approach has been adopted to study the formation of magnetosphere current sheets @xcite , and the linear instabilities in magnetosphere that could be responsable for emissions @xcite .
to overcome the rotation ambiguities of plasma velocity and magnetic field , we consider the magnetic field @xmath2 be stationary in space , and the electric field @xmath9 is uploaded to the equipotential field lines of the magnetosphere according to the potential distribution on the pulsar surface estblished by the unipolar induction .
this means that the charge separated magnetospheric plasma has a negligible contribution to the electric field there . in this case of stationary magnetic field @xmath2 ,
the driving energy is the uploaded electric field @xmath9 which is the input parameter in eq .
[ eqno2 ] , and the plasma velocity @xmath7 is the response .
the magnetosphere is directly coupled to the unipolar induction of the pulsar .
the charge separation flow inside the pulsar generates a current .
the interaction of this current with the magnetic field produces a direct braking torque in addition to the contributions of dipole radiation and stellar winds which slow down the rotation rate . finally , we remark that magnetic field is a real physical quantity , while magnetic field line is only a physical concept derived from the field line equation to help visualize the action of magnetic field .
it is rather unphysical to consider field lines being rigidly attached to the pulsar surface corotating with it .
the notion of a rotating magnetosphere is probably cultivated in the oblique rotator case , where the magnetosphere is seen to rotate obliquely about the pulsar rotational axis . as obliqueness goes to zero , we then have an aligned rotator . for an oblique rotator
, we choose to phrase the magnetosphere as wobbling ( not rotating ) about the rotational axis .
consequently , as the obliqueness goes to zero , the magnetosphere stops to wobble and becomes stationary in space . here , in this paper , we reconsider the @xcite analysis of charge separated plasmas .
in particular , there are two inconsistencies in their work .
the first inconsistency is the space charge density expression which is incompatible with the curvature of the closed ion magnetic field lines near the pulsar . to resolve this inconsistency
, they resorted to a corotating electron cloud above the closed magnetic field lines .
the second one is the introduction of the far zone quasi - neutral interstellar plasma potential to the near zone pulsar magnetosphere to draw plasma outflows in the open field lines .
the field line that has the potential of the interstellar plasma is the critical field line with no charge flow .
this interstellar emf is inconsistent to the @xmath11 plasma condition in the magnetosphere .
actually , many publications in recent decades have concentrated on the pulsar equation itself without mentioning neither the question of space charge density nor the condition of the critical field line .
these simplifications are equivalent to considering the pulsar magnetosphere as filled with a quasi - neutral normal plasma instead of a charge separated plasma .
the charge separated plasma in this stationary magnetosphere undergoes an @xmath12 drift which happens to have an angular velocity of the pulsar , instead of being dragged along by the corotating magnetosphere .
as for the plasma outflow , we consider it be driven by plasma drifts , not by interstellar emf . finally , with a stationary magnetosphere in space , coupled to the unipolar induction in the pulsar interior , we offer an alternative pulsar beacon mechanism by suggesting a pulsating magnetosphere . to understand this dynamic picture , we assume that the unipolar induction is delivering more power to the magnetosphere than the open field lines can drain .
this leads to an energy built - up in the closed field lines represented by plasma pressure .
although plasma pressure is neglected in eq .
[ eqno1 ] , it appears indirectly through plasma drifts , which could open up the closed field lines near lc .
observationally , radio pulsars are known for their stable repetitive pulse profiles .
however , the pulse profile that characterizes each radio pulsar is obtained as the average over many many shots . from shot to shot , the profiles are far from identical .
therefore , the period of oblique rotator is not well defined from shot to shot , which is not supposed to be in the lighthouse scenario . on the other hand , such shot to shot variations
are expected in a pulsating magnetosphere .
we review the basic concepts of charge separated pulsar magnetosphere and the plasma velocity @xmath7 of eq .
[ eqno2 ] in sec.3 .
the space charge density @xmath8 is derived in the presence of a poloidal and toroidal field , and in the presence of drift velocities in sec.4 . under the concept of guiding center drifts , the idea of a critical field line
is reconsidered in sec.5 .
we then follow the earlier works of @xcite , @xcite , @xcite , @xcite , @xcite , to solve the magnetosphere of an isolated pulsar analytically via the force - free pulsar equation . nevertheless , we take a different approach to split the pulsar equation into two coupled equations through a separating function .
each one of the coupled equations is solved by separation of variables for the global structure of the pulsar magnetosphere in sec.6 .
the standard monopole - like steady state magnetosphere is presented in sec.7 .
a self - consistent description of magnetosphere in terms of fields , velocities , and space charge densities is outlined in sec.8 . under continuous unipolar pumping ,
the standard magnetosphere is generalized to the pulsating magnetosphere in sec.9 , and with conclusions in sec.10 .
it is important to remark that eq .
[ eqno2 ] is written in an inertial frame where the plasma moves with velocity @xmath7 with respect to the stationary magnetic and electric fields . in the plasma moving frame ,
the lorentz transformed electric field is @xmath13 and the ohm law reads @xmath14 there .
this ohm law in the limit of infinite @xmath5 is known as the frozen field condition in mhd plasmas , and happens to be identical to eq .
[ eqno1 ] , and we have @xmath15 and @xmath0 . as a matter of fact , for a charge separated plasma , eq . [ eqno1 ] and eq . [ eqno2 ] amount to the same equation always .
however , eq . [ eqno1 ] describes the magnetosphere in terms of fields , while eq .
[ eqno2 ] considers the magnetosphere in terms of plasma velocities .
the fields and the velocities are then connected through the space charge density . here
we call to the attention that the condition of @xmath11 allows an arbitrary longitudinal velocity @xmath16 along the magnetic field line .
consequently , the plasma velocity can be projected in components parallel and perpendicular to the magnetic field as @xmath17 the frozen field plasma condition shows that @xmath18 is given by the @xmath12 guiding center drift of the gyro motion due to the transverse part of the electric field @xmath9 @xcite , under the constrain of @xmath19 which implies the electric field strength be @xmath20 for a magnetized plasma .
this @xmath18 drift is charge independent , and for a quasi - neutral normal plasma , this drift generates no current and represents only the fluid drift velocity . but for charge separated plasmas , the @xmath12 drift corresponds a current flow .
consequently , we have the picture of a stationary pulsar magnetosphere with plasma drifting with respect to the magnetic field . in cylindrical coordinates @xmath21
, the axisymmetric magnetic field can be represented as @xmath22 where @xmath23 carries the dimension of the poloidal magnetic flux such that @xmath24 is a dimensionless flux function , and @xmath25 represents the relative ratio of the toroidal flux to the poloidal flux .
the function @xmath26 is related to the axial current through @xmath27 from the magnetic field line equation , it is easy to show that the poloidal field lines are given by the contours of @xmath24 @xmath28 the fact that the longitudinal electric field vanishes means that any potential difference along the magnetic field line will be shorted out by plasma mobility . with the longitudinal electric field @xmath11 and the magnetic field @xmath2 represented by eq .
[ eqno3 ] , the transverse electric field in the magnetosphere can be expressed through the equipotential field lines of eq .
[ eqno5 ] as @xmath29 where @xmath30 is the equator - pole voltage drop on the pulsar surface , and @xmath24 is assigned by the corresponding surface potential label . alternatively , we can also project the plasma velocity in poloidal and toroidal components by writing @xmath31 and @xmath32 . from eq .
[ eqno2 ] , we get @xmath33 by axisymmetry , and @xmath34 .
we therefore have respectively @xmath35 loading the potential distribution of the pulsar surface onto the equipotential magnetosphere poloidal field lines , we have @xmath36 where @xmath37 measures the radial position of a point @xmath38 in the magnetosphere . substituting this result to eq .
[ eqno7 ] then gives @xmath39 , and therefore @xmath40 we call to the attention that @xmath41 drifts with an angular velocity of the pulsar .
this equation is often interpreted as the beads on a wire model , where the plasma ( beads ) slides with a velocity @xmath42 along a rotating @xmath43 field line ( wire ) .
we have rederived this equation with a stationary magnetosphere where the rotation comes from the guiding center drift .
we have expressed the plasma velocity in two orthogonal projections .
the first one is with respect to the magnetic field lines , and the second one is with respect to the cylindrical coordinate system .
these two projections are independent to each other .
only when the magnetic field is purely poloidal with @xmath44 that the two expressions are correspondent , and we can identify @xmath45
in the magnetosphere , the electric field @xmath9 is not the only transverse force that causes plasma drift . magnetic field line curvature and field gradient are other two major transverse forces .
therefore , besides the @xmath18 drift , we also have the curvature and gradient b drifts @xcite given by @xmath46 where @xmath47 and @xmath48 are the charge particle energies parallel and perpendicular to the magnetic field .
therefore , these drifts are mass dependent as well as field dependent . in a thermal plasma , these energies
are measued by the plasma pressure .
although plasma inertia and pressure are neglected in eq .
[ eqno1 ] , they appear indirectly through these two drift velocities .
the direction of these two drifts depends on the sign of charge q. even in a quasi - neutral normal plasma , these are current generating drifts , while the electric drift @xmath18 represents the plasma fluid velocity only . in a charge separated plasma ,
all three drifts generate currents .
these plasma drifts are driven by transverse forces , and they are not emf driven velocities . to understand this , we recall that for a uniform magnetic fleld , charge particles follow a circular gyro motion along the magnetic field lines according to the equation of motion due to the plasma inertia . in the presence of a transverse force to the magnetic field such as an electric field , a centrifugal force due to curvature , and an inhomogeneous magnetic field , etc .
, the circular gyro motion will be transformed into a displacing cycloid across the field lines @xcite , such that the gyro guiding center presents a transverse drift .
although eq . [ eqno2 ] gives the guiding center electric drift directly , the gyro motion itself is due to the plasma inertia
. therefore , eq . [ eqno2 ] actually includes the plasma inertia implicitly .
we remark that these drift velocities can be very large and become relativistic
. should the electric field @xmath9 be time varying , there would be a polarization drift as well which we will ignore here . we should point out that in eq .
[ eqno2 ] only the @xmath18 drift is represented on the left side .
the other two current generating drifts , that are contained in @xmath49 on the right side of eq .
[ eqno2 ] , do not appear in eq .
[ eqno2 ] because the plasma conductivity @xmath5 is infinite . for this reason ,
the plasma drift velocity is given by @xmath50 since the plasma drift velocities are obtained from the equation of motion of each charged plasma , the plasma drift velocity @xmath51 is beyond the scope of eq .
[ eqno2 ] , as discussed in sec.1 , in the sense that the mhd expression of @xmath52 differs from @xmath51 because @xmath53 . the space charge density , which connects the fields of eq . [ eqno1 ] and the velocities of eq .
[ eqno2 ] , in the magnetosphere is , by eq .
[ eqno2 ] , @xmath54 writing @xmath55 , considering a general poloidal and toroidal fields , and representing @xmath18 by the projection of eq .
[ eqno8 ] to evaluate the @xmath56 term on the right side , we get @xmath57\ , = \,[1-\epsilon_{0}\mu_{0}(\vec v_{e}-\kappa_{v}\vec b ) \cdot\vec v_{drift}]\rho_{q}\ , = \,-2\epsilon_{0}\vec b\cdot\vec\omega_{0}\,\,\,.\end{aligned}\ ] ] in the particular case of a purely poloidal field , we can identify @xmath58 , and eq .
[ eqno10 ] becomes @xmath59 @xcite had derived a space charge density ( their eq.8 ) with @xmath60 alone that reads @xmath61 in the absence of @xmath62 and @xmath63 , the left side of the above equation looks like the relativistic lorentz factor , which is just coincidental .
this equation predicts @xmath64 for closed ion field lines , which is consistent only to the outer portion of the dipole - like closed ion field lines . for the inner portion with @xmath65
, they postulated a negative charged cloud corotating with the field lines .
we antecipate the result in sec.8 that @xmath66 , such that the @xmath67 term is positive . near the pulsar , @xmath62 and @xmath63 are large , where the plasma pressure is very high , on the left side and we could have @xmath65 . farther out from the pulsar with plasma pressure decreasing
we have @xmath64 .
since @xmath62 and @xmath63 drifts are mass dependent , they would be dominant again as the closed field lines approach lc , due to their relativistic inertia .
we would have @xmath65 on approaching lc .
for this reason , the closed region of a steady state magnetosphere can only extend to some intermediate radius , staying away from lc . although plasma pressure and plasma inertia are not considered in eq .
[ eqno1 ] explicitly , plasma inertia appears implicitly through the guiding center drifts of the gyro motion and plasma pressure through the coefficients of the gradient b and curvature drifts .
we have given a qualitative understanding of these effects in the closed region through the drift velocities , and we will return to this issue in sec.8 .
in terms of cylindrical coordinates , the toroidal components of eq .
[ eqno1 ] , which describes the magnetosphere in terms of fields , gives @xmath68 whereas the poloidal components yield the pulsar equation @xmath69 where @xmath70 is the lc radius . because of the singular coefficient of the highest derivative which vanishes at @xmath71 , eq .
[ eqno12 ] is a singular equation . as indicated by @xcite in sec.iii , the steady state magnetosphere , which is stationary in space ,
is solved within lc , and is then analytically continued beyond lc where the pulsar equation is not valid with the tengential velocity @xmath72 . since plasma parameters like plasma pressure and velocities are not considered in the pulsar equation , analytic continuation of the field structure beyond lc amounts to solving the pulsar equation in that region with solutions on both sides matched across lc , as done numerically by @xcite , @xcite , and others .
analytic solutions of this equation had been extensively studied , yet global solutions were not obtained explicitly @xcite .
so far , solutions are developed by numerical iterrations , under the boundary condition of a dipole field at the center notably by @xcite , @xcite , @xcite , @xcite where the open and closed regions rotate differentially , and @xcite where the braking index is evaluated in terms of the magnetosphere dynamics . in the open field lines , electrons and ions
are assummed to stream out along their respective open field lines with equal flux .
these currents will be connected to form a return loop at a distant load ( interstellar plasma ) .
the electron and ion outflows are thought to be drawn by the action of the far zone quasi - neutral normal interstellar plasma floating potential @xmath73 , where @xmath74 is the potential difference between the far zone interstellar plasma and the near zone electron or ion open field lines , and @xmath75 is a dimensionless scalar potential function . in particular , there will be a field line having the same potential of the interstellar plasma .
this is the critical field line @xmath76 , where there is no charge and no poloidal current flow , that separates the electron and ion field lines .
this means that @xmath77 , and @xmath78 should be stationary at @xmath76 .
this critical field line should pierce lc with @xmath79 .
nevertheless , using the interstellar emf to draw current flows in the open field lines is clearly in confrontation with the @xmath11 condition .
this inconsistency can be removed should we note that the presence of a toroidal and a poloidal magnetic field in the open region can drive a poloidal and a toroidal @xmath18 drift respectively @xmath80 the same can be applied to @xmath62 and @xmath63 drifts .
all these drift velocities will , therefore , produce a component along the magnetic field lines plus a toroidal rotation , as in eq .
[ eqno8 ] , and there is no need to introduce neither the interstellar potential to the near zone nor the concept of a critical field line .
by considering the magnetic field @xmath2 be stationary in space , and the electric field @xmath9 be uploaded to the field lines from the pulsar surface , our model has the plasma velocity @xmath7 as the response in eq .
[ eqno2 ] . with @xmath81
, our model simply implies that the electric potential on the pulsar surface can not be fully uploaded to the magnetosphere to all distances , even beyond lc .
this means that the plasma conductivity @xmath5 is finite in eq .
[ eqno2 ] on approaching lc , and collisional effect has to be taken into account , or equivalently , inertia effect .
likewise is in eq .
[ eqno1 ] .
this should be the subject of further investigations . for the moment
, we persuit eq .
[ eqno12 ] as it is . to construct the standard magnetosphere ,
we denote @xmath82 as the normalized radial coordinate and take a different approach to recast the pulsar equation as @xmath83\ , = \,f(\psi)\,\,\,,\end{aligned}\ ] ] where we have required both sides of the equality be equal to a separating function @xmath84 .
we attempt to solve this equation analytically by choosing @xmath85 which renders @xmath86 where @xmath87 is the normalized axial coordinate .
we write @xmath88 by separation of variables , and eq . [ eqno15a ] gives @xmath89 where @xmath90 is the separation constant .
we have denoted the two constants in eq .
[ eqno14 ] and eq .
[ eqno16b ] by @xmath91 and @xmath90 respectively in quadratic form by choice .
however , @xmath91 and @xmath90 are not necessarily be positive .
as for eq .
[ eqno15b ] , we take a linear @xmath78 and get @xmath92r\,\,\,.\end{aligned}\ ] ] here , @xmath93 has the dimension of an inverse scale length .
we have chosen a minus sign explicitly in eq .
[ eqno17a ] to represent an overall downward axial current in the northern polar region . with the understanding that electrons and ions flow along their field lines by plasma drifts
, @xmath78 does not have to have a stationary point on lc which ables the linear choice in eq .
[ eqno17a ] . such choice of @xmath94 was also used by @xcite and @xcite .
our analytic approach to the pulsar equation has introduced a separating function @xmath95 , a separating constant @xmath90 , and an axial current function @xmath94 .
the @xmath96 function can be solved readily to render @xmath97 with @xmath98 , and @xmath99 with @xmath100 . to solve for @xmath101 , we make use of eq .
[ eqno17b ] to cast eq .
[ eqno16b ] as @xmath102r\ , \\
\nonumber = \,2{\partial^{2}r\over\partial\xi^{2 } } + [ ( a_{z}^{2}+a_{l}^{2}-2m^{2 } ) -a_{l}^{2}\xi^{2}]r\ , \\
\label{eqno20 } = \,2{\partial^{2}r\over\partial\xi^{2 } } + [ a_{\phi}^{2 } -a_{l}^{2}\xi^{2}]r\ , = \,0\,\,\,.\end{aligned}\ ] ] this equation can be solved through a power series .
because of the boundary condition of @xmath103 on the polar axis @xmath104 , we should have @xmath105 as @xmath106 goes to zero .
we therefore have @xmath107 we note that @xmath108 and @xmath109 are the two independent constants of the second order differential equation , eq .
[ eqno20 ] .
other coefficients are generated by the recurssion formula , eq .
[ eqno22 ] .
the coefficients @xmath108 and @xmath109 generate even and odd power terms of the series respectively . thus @xmath108 and @xmath109 are the multipliers of the even and odd series that we can write @xmath110
we note that the radial function @xmath101 is governed by the parameters @xmath111 and the choice of the two independent coefficients @xmath112 . with positive @xmath113 and @xmath114
, the coefficients @xmath115 are alternating in sign , and @xmath101 could have oscillating solutions for small @xmath106 .
this functional form is suitable to construct the open ion field lines and the closed region of the pulsar magnetosphere . as @xmath106 increases
, @xmath101 becomes monotonically increasing , which is suitble for the split monopole structure far from the pulsar . as for the axial function @xmath116 , it is a decreasing function of @xmath117 .
when this is combined with the radial function , together they could generate contours adequate for open field lines . for a given @xmath113
, we could scan @xmath114 over a given range to generate a set of field line morphologies . as a result
, we represent the electron open field lines by taking the poloidal flux function as @xmath118 for simplicity , we consider the even terms of the power series by taking @xmath119 and @xmath120 . we choose to set @xmath121 , say , for the radial solution .
taking @xmath122 , the function @xmath101 is a monotonically increasing function , as is shown in fig.1 which is appropriate for electron open field lines .
we should emphasize that the monotonic functional form of @xmath101 can be obtained under a wide range of parameters , not restricted to the chosen ones .
furthermore , we set @xmath123 , for example , and we have the self - consistent parameters @xmath124 , @xmath125 . with @xmath123 , the axial function @xmath116 is shown in fig.2 , and a set of poloidal field lines is shown in fig.3 . according to eq .
[ eqno3 ] , the poloidal and toroidal magnetic fields are given by @xmath126 where the last equation gives the ratio of the two fields in the magnetosphere . however , this ratio is expressed in terms of their respective fluxes , @xmath23 and @xmath127 , without which a specific ratio can not be determined . for the ion open field lines
, we use the same representation of eq .
[ eqno24 ] , and keep @xmath121 and @xmath123 . however , we change @xmath128 and other parameters become @xmath129 , @xmath130 . by
changing @xmath122 to @xmath128 , the radial function begins to present oscillations for small @xmath106 , which is appropriate for the ion field lines .
the functions @xmath101 and @xmath116 are given in fig.4 and fig.5 respectively .
we note that @xmath101 rises from zero and passes through a maximum and a minimum locally before it takes off monotonically .
the local maximum and minimum can become more pronounced by taking a smaller @xmath114 .
a set of poloidal field lines is shown in fig.6 which reflects the monopole - like solution of a charge separated magnetosphere in the near zone . in the boundary zone
, the plasma tends to be quasi - neutral as it proceeds to meet the interstellar plasma in the far zone .
the @xmath131 part of the open field lines is the solution of the pulsar equation , and the @xmath132 part is the analytic continuation across lc . to construct the closed field lines of the ion plasma , known as the dead zone
, we note that , with ion current outflow and electron return current outflow ( closed at a distant load ) along the open field lines equal in magnitude , @xmath44 in the closed region which requires @xmath133 . to represent the closed field lines , we need a radial function bounded by a root @xmath134 within lc . we also need a bounded axial function .
we therefore write @xmath135 in order to have @xmath101 bounded by a root within lc , we choose @xmath136 and @xmath137 with different signs . under the condition @xmath133 , we take @xmath138 , @xmath139 , and @xmath140 ( @xmath141 ) , we have @xmath142 , and the functions @xmath101 and @xmath143 are shown in fig.7 and fig.8 respectively .
once more , we recall that these parameters are neither unique ones nor restrictive . there is a good range in parameter space that reproduces the closed region .
the boundary condition of @xmath144 on the equatorial plane at @xmath145 is satisfied by @xmath143 .
a corresponding set of poloidal field lines is shown in fig.9 with a different @xmath106 scale of fig.3 and fig.6 for more clarity . in this figure ,
the closed region terminates at an intermediate value of @xmath146 and the reason of which has been discussed in sec.4 .
the same figure is redrawn with the same scale of fig.3 and fig.6 , so that the three figures can be superimposed on each other to get the global features of the magnetosphere , as in fig.10 .
this set of global field lines is a continuous function of the contour value @xmath147 .
more field lines could be added by assigning the intermediate contour values @xmath147 . comparing to the numerical results , such as @xcite , our magnetospheric configuration reproduces well the main global features .
so far , we have introduced plasma drift velocities to remove the inconsistency of the space charge density with the closed field lines , and the incinsistency of the @xmath11 condition with the emf driven plasma outflows . furthermore , we have solved analytically the pulsar equation for the global structure of the magnetosphere . above all
, we have taken the view of a stationary magnetosphere in space under which the plasma drifts with a velocity @xmath7 according to eq .
[ eqno2 ] . as we have already mentioned , the magnetosphere is described in terms of fields by eq .
[ eqno1 ] , and in terms of velocities by eq .
[ eqno2 ] .
we should note that the force - free approximation is only valid for the field - description .
as for the velocity - description , plasma inertia and plasma pressure are the prime factors .
these two descriptions are then connected by eq .
[ eqno10 ] which describes the magnetosphere in terms of space charge densities .
all these three aspects have to be self - consistent among each other . with a general poloidal and toroidal magnetic fields ,
the plasma velocities are given by eq .
[ eqno8 ] , derived from plasma drifts . as for the space charge density , it is reflected by the @xmath148 on the right side of eq .
[ eqno10 ] . before addressing the consistency of the space charge density ,
let us first establish the magnitudes of the @xmath62 and @xmath63 drifts by noting their ratio @xmath149 where @xmath150 is the scale length defined by @xmath151 .
next , to compare with @xmath18 , we rewrite @xmath18 in terms of @xmath24 which reads @xmath152 for a poloidal magnetic field with @xmath153 representing the field line , @xmath154 is directed inward towards the equatorial plane , and @xmath18 corresponds to a toroidal rotation of @xmath58 . for the @xmath63 drift of eq .
[ eqno9b ] , we can evaluate @xmath155 by using the @xmath153 representation to get @xmath156 where @xmath157 has the dimension of an inverse scale length .
equivalently , @xmath158 has the dimension of a reference poloidal mangnetic field . as a result ,
[ eqno9b ] reads @xmath159 we note that this equation is identical in form for a positive charge @xmath160 to eq .
[ eqno28 ] . as for the @xmath62 drift of eq .
[ eqno9a ] , we note that the radius of curvature @xmath161 directs from the center of curvature outward to the field lines .
therefore , @xmath161 and @xmath162 are pointing outward , although they are not parallel .
consequently , both @xmath63 and @xmath62 drifts are in the toroidal direction , as is @xmath18 . as for their amplitudes ,
based on eq .
[ eqno28 ] , eq .
[ eqno30 ] , and eq . [ eqno9a ] , all three velocities have the same @xmath163 scaling , so their strengths can be compared by other factors . for @xmath62 , we note that @xmath164 is small when it is near the pulsar and when it is near the apex on returning back to the conjugate point of the southern equatorial region . in between ,
the field lines are relatively flat , as shown in fig.9 , and @xmath164 is large . as for @xmath63 ,
the ratio of this velocity to @xmath18 is @xmath165 assumming the reference poloidal field @xmath158 comparable to @xmath166 , these two velocities become comparable when the plasma energy ( pressure ) @xmath167 .
the same can also be said for @xmath168 for @xmath62 . this high plasma pressure can be found close to the pulsar surface , and it decreases as @xmath37 increases . for a toroidal magnetic field in @xmath169 direction ,
the @xmath18 drift will be poloidal but in the counter direction of closed field lines . likewise are @xmath63 and @xmath62 . by eq .
[ eqno6 ] , we therefore have @xmath66 , and the factor @xmath67 in eq .
[ eqno10 ] is positive .
should we consider the toroidal magnetic field be in @xmath170 direction , we would get @xmath171 , and the factor @xmath67 would be positive again .
we remark that the space charge density of eq .
[ eqno10 ] is valid for the entire magnetosphere , not just for the closed region . in the presence of poloidal and toroidal fields , the scalar product in the square bracket on the right side of eq .
[ eqno10 ] has a poloidal and a toroidal conribution .
this space charge density has to be calculated together with the fields and plasma drifts in numerical solutions .
we have constructed the standard magnetsphere in sec.7 where the closed region with @xmath44 is described by eq .
[ eqno26 ] . on the other hand , with the self - consistent description of the magnetosphere based on fields , velocities , and space charge densities in sec.8
, there should be a @xmath172 in the closed region . on this question of a toroidal field ,
we note that , in a charge separated magnetosphere , a toroidal magnetic field can arise spontaneously .
if we begin with a pure poloidal magnetic field in the polar region open electron field lines , equilibrium condition requires the electron plasma be at rest .
this is an unstable configuration , since any longitudinal movement of the electron plasma corresponds to a poloidal current that would generate a toroidal magnetic field .
this toroidal magnetic field would enhance the poloidal drift velocity ( current ) itself .
since the field lines are open , an equilibrium configuration of fields and currents could be reached . with @xmath173 for electron field lines , and with small @xmath41 due to small @xmath106 , it is reasonable that the square bracket be positive and thus @xmath65 . as for the ion open field lines ,
the toroidal field here begins to decline due to the ion outflows . with @xmath174 for ion field lines ,
it is likely that we could have @xmath65 in some part of the field line and @xmath64 in other part . as for the ion field lines in the closed region
, there still is a toroidal field due to the imperfect shielding of the electron axial current by the ion outflow in the open region . with @xmath174 , and with the presence of a toroidal field and a high plasma pressure near the pulsar , eq .
[ eqno10 ] could have @xmath65 near the pulsar . farther out with large @xmath106 ,
the toroidal field becomes small , likewise is the plasma pressure , we could expect @xmath64 . with the poloidal field lines described by eq .
[ eqno26 ] , the corresponding toroidal field in the closed region is represented by eq .
[ eqno25b ] .
because of the mirror symmetry between the upper ( north ) and lower ( south ) hemispheres , the toroidal fields have opposite sign in the two parts , which drives an outward radial current sheet on the equatorial plane . on reaching the apex of the closed region
, this current sheet divides into two parts returning from the northern and southern closed poloidal field lines .
we therefore have two current loops , one north and one south , to enhance the corresponding toroidal fields .
this in turn enhances the radial current sheet and the current loops . in a closed finite region , this positive feedback cycle can only be stablized in the presence of dissipations .
naturally , there is the usual toroidal current sheet on the equatorial plane beyond lc for the split monopole poloidal field lines . to find an alternative beacon mechanism
, we first racall that the current model pictures a corotating magnetic field that drags the magnetospheric plasma . in response to this
, a rotation induced electric field is generated in the stationary frame of space . in this picture
, the rotating magnetic field is the prime driver , and the magnetosphere is decoupled from the unipolar induction of the pulsar .
here , we view the unipolar induction inside the pulsar and the magnetosphere outside the pulsar as one coupled system . the unipolar induction delivers charges separated by their signs to the polar and equatorial regions at a given rate .
the current flow inside the neutron star interacts with the magnetic field , generating a braking torque .
the charges are then channelled to the magnetosphere along the field lines , and are ejected to the interstellar space as stellar winds through open field lines , which allows the unipolar induction to operate continuously .
considering the unipolar energy pumping rate exceeds the poynting outflow rate along the open field lines , plasma density and pressure will build up in the magnetosphere , particularly in the closed region . in the outer part of the closed region extending to intermediate @xmath106 with @xmath64 , @xmath63 and @xmath62 are small .
however , as the plasma pressure build - up reaches there , these velocities begin to increase .
eventually , they get to the point of having @xmath65 , flipping open the field lines and releasing the trapped plasma and magnetic energies . furthermore , the field lines can also be flipped open in the outer part of the closed region by the positive feedback on the toroidal field which appears in the @xmath67 factor of eq .
[ eqno10 ] .
once depleted of plasmas , in the sense that plasma density is so low that eq .
[ eqno2 ] is not warranted , the magnetosphere will then be recharged all over to the standard configuration , and the whole cycle starts anew . instead of a stationary magnetosphere
, we therefore have a scenario of a pulsating magnetosphere where the openning of the closed field lines either by unipolar pumping on drift velocities or by positive feedback on the toroidal field works as a magnetic switch , which offers an altenative mechanism for pulsar beacons . in this alternative picture ,
an observer aligned with the magnetic axis sees the beacons separated by the unipolar induction charging period , instead of the pulsar rotation period . here
, we have reached the pulsating picture of the magnetosphere through a steady state approach by viewing the unipolar induction inside the pulsar and the magnetosphere outside as a coupled system . as a matter of fact
, the steady state approach may not be adequate to describe a charge separated magnetosphere .
we call to the attention that a charge separated plasma is a very unstable plasma due to the fact that any plasma flow ( drift ) is a current flow which generates magnetic field that drives drifts . for this reason
, we might have to let go the lighthouse paradigm , and consider the time dependent model with maxwell equations coupled to the plasma equations @xcite .
furthermore , the inherent unstable nature of a charge separated magnetosphere could be the cause of periodic beaming of electromagnetic waves .
we have taken the view of interpreting the magnetic field @xmath2 of the magnetosphere as stationary in space , and the electric field @xmath9 as established by uploading the potential distribution of the pulsar surface to the magnetic field lines .
the plasma velocity @xmath7 is , therefore , the drift response in eq .
[ eqno2 ] which rotates with an angular velocity of the pulsar .
the presence of a magnetosphere allows the pulsar surface charges be channelled to the field lines , and keeps the counter electric field in the pulsar interior low .
this warrants the continuous operation of the unipolar induction that constantly pumps energy to the magnetosphere . in this scenario , the energetics of the magnetosphere
is coupled to the unipolar induction of the neutron star . by introducing the gradient b and curvature plasma drifts and the presence of a toroidal field
, we have removed the inconsistency between the space charge density and the curvature of the dipole - like closed ion field lines . also , the inconsistency between introducing the far zone interstellar plasma floating potential to the near zone open field lines to generate an emf driven plasma flow along open field lines and the @xmath11 plasma condition can be removed as well by drift velocities .
although plasma inertia and pressure are neglected in the force - free pulsar equation , they enter the magnetospheric dynamics implicitly through the @xmath12 , gradient b , and curvature plasma drifts , which are the respective guiding center drifts of the basic gyro motion . since the gradient b and curvature drifts are mass dependent , the closed region should be bounded at a distance within lc .
we have devised a method to solve the pulsar equation analytically for the standard split monopole magnetosphere .
the solutions are given in terms of three parameters , the separation function parameter @xmath95 , the axial current parameter @xmath94 , and the separation constant @xmath90 .
recognizing that the magnetosphere is described by the fields of eq .
[ eqno1 ] under force - free approximation , and by the plasma drift velocities of eq .
[ eqno2 ] where plasma inertia and plasma pressure are prime factors , and these two descriptions are connected by the space charge densities of eq .
[ eqno10 ] , we have presented a self - consistent description of the pulsar magnetosphere . through the coupled system between the unipolar induction inside the neutron star and the magnetosphere outside , and considering the unipolar induction rate exceeds the open field line poynting flux ,
we have constructed a dynamic pulsating magnetosphere . in this pulsating model ,
the accumulated magnetic and plasma energies of the closed field lines can be released periodically by switching open the field lines either as plasma pressure gets high enough or as the toroidal magnetic field gets large enough . as an alternative to the lighthouse paradigm ,
this pulsating magnetosphere offers a different mechanism to generate pulsar beacons . | following the basic principles of a charge separated pulsar magnetosphere @xcite , we consider the magnetosphere be stationary in space , instead of corotating , and the electric field be uploaded from the potential distribution on the pulsar surface , set up by the unipolar induction . consequently , the plasma of the magnetosphere undergoes guiding center drifts of the gyro motion due to the transverse forces to the magnetic field .
these forces are the electric force , magnetic gradient force , and field line curvature force .
since these plasma velocities are of drift nature , there is no need to introduce an emf along the field lines , which would contradict the @xmath0 plasma condition .
furthermore , there is also no need to introduce the critical field line separating the electron and ion open field lines .
we present a self - consistent description where the magnetosphere is described in terms of electric and magnetic fields and also in terms of plasma velocities .
the fields and velocities are then connected through the space charge densities self - consistently .
we solve the pulsar equation analytically for the fields and construct the standard steady state pulsar magnetosphere . by considering the unipolar induction inside the pulsar and the magnetosphere outside the pulsar as one coupled system , and under the condition that the unipolar pumping rate exceeds the poynting flux in the open field lines , plasma pressure can build up in the magnetosphere , in particular in the closed region .
this could cause a periodic openning up of the closed region , leading to a pulsating magnetosphere , which could be an alternative for pulsar beacons .
the closed region can also be openned periodically by the build - up of toroidal magnetic field through a positive feedback cycle . |
the preceding paper @xcite discovered , among other findings , the independence of the sign of the polarization @xmath6 on the boundary conditions ( bcs ) for the one - dimensional ( 1d ) quantum well ( qw ) of the width @xmath7 placed into the uniform electric field @xmath0 that is directed perpendicular to its confining surfaces located at @xmath8 : the polarization @xmath9 of the ground state for any permutation of the dirichlet ( d ) , @xmath10 and neumann ( n ) , @xmath11 edge requirements imposed on the wavefunction @xmath12 is positive for all applied voltages while its excited - state counterparts @xmath13 , @xmath14 , for the small growing fields decrease from zero at @xmath15 to the negative values , pass through the minimum and only after this start to increase crossing zero at the @xmath16- and bc - dependent intensity @xmath17 .
immediately , one wonders : for any kind of the particles , is it possible to observe the total statistically averaged polarization that is negative at the small electric forces ?
analysis below answers this question together with the thermodynamic calculations of the corresponding energy @xmath18 and heat capacity @xmath1 .
following the previous research @xcite , the qw with the particular distribution of the bcs will be denoted by the two characters , where the first ( second ) one corresponds to the edge condition at the left ( right ) interface .
similar to the discussion of the spectrum @xmath19 and polarizations @xmath13 @xcite , all energies will be measured , if not specified otherwise , in units of @xmath20 , which is a ground - state energy of the dd qw , while the unit of the electric field will be @xmath21 , and that of the polarization - @xmath22 , with @xmath23 being the particle mass and @xmath24 denoting the absolute value of the electronic charge .
in addition , heat capacity is expressed below in terms of boltzmann constant @xmath25 .
discussion considers canonical as well as grand canonical ensembles . in this last case ,
the properties are calculated both for fermions and bosons .
also , frequently we draw parallels with the 1d harmonic oscillator ( ho ) with the potential ( in regular units ) @xcite @xmath26 whose energies @xmath27 , upon application of the electric voltage , are @xmath28 for this configuration , the natural units that will be used below are : for the energy , @xmath29 ; for the length , @xmath30^{1/2}$ ] ; for the electric field , @xmath31 ; and for the polarization , @xmath32 .
this type of the statistical ensemble assumes that the system under consideration is in the thermal equilibrium with the much larger bath characterized by the thermodynamic temperature @xmath2 .
the fundamental quantity here is the partition function @xmath33 where the summation runs over all possible quantum states , and the parameter @xmath34 is ( in regular , unnormalized units ) @xmath35 .
the probability @xmath36 of finding particle in the state @xmath16 depends on the temperature and the energy @xmath37 as @xmath38 as a result , the mean value @xmath39 of any physical quantity @xmath40 is calculated as @xmath41 for the @xmath3 particles in the system , this equation has to be multiplied by @xmath3 .
applying these general results to the qw with the different bcs in the electric field @xmath0 , one derives the mean values of the energy @xmath42 and polarization @xmath5 [ canonicalmeanvalue2 ] @xmath43 where in the left - hand side we have explicitly underlined that they are functions of the temperature @xmath2 ( through the parameter @xmath34 ) and electric field [ through the corresponding dependence of @xmath19 and @xmath13 ] .
equivalently , eq .
can be written as : @xmath45 heat capacity at the constant volume @xmath1 is a work that has to be done to change the temperature of the system by one degree and , as a result of this , it is calculated as a derivative of the total energy with respect to the temperature @xmath2 : @xmath46 where regular , unnormalized units have been used . applying this generic definition to the canonical distribution from eq .
, one gets fluctuation - dissipation theorem @xcite @xmath47 where , for convenience of the notation , the subscript @xmath48 has been dropped .
energies @xmath37 and polarizations @xmath6 for the qw were calculated before @xcite while for the ho they are : [ ho_energypolarization ] @xmath49 note that , contrary to the hard - wall qw @xcite , for its ho counterpart the polarization is at any voltage a linear function of the field and is the same for all levels .
accordingly , its mean value for the one particle is equal to @xmath0 too while the energy becomes : @xmath50 as a result , the electric field does _ not _ affect the ho canonical heat capacity , which reads @xcite : @xmath51 one can derive limiting cases of these dependencies : for the small temperatures ( @xmath52 ) : [ ho_canonicallimit1 ] @xmath53 for the large temperatures ( @xmath54 ) : [ ho_canonicallimit2 ] @xmath55 ( a ) heat capacity @xmath56 and ( b ) mean energy @xmath57 as a function of the normalized temperature @xmath58 for the canonical ensemble and pure dirichlet ( dotted line ) , neumann ( solid curve ) and nd ( dashed line ) qw at zero electric field . ] before discussing the electric field influence on the thermodynamic properties of the hard - wall qw , let us address first the voltage - free configuration . plugging in the well known expressions for the zero - field energies @xmath59 into eq .
, one gets after some algebra : [ canonicalmeanenergy5 ] @xmath60 here , @xmath61 , @xmath62 , are theta functions @xcite . for small temperatures , @xmath52
, these equations degenerate to [ canonicalmeanenergy4 ] @xmath63 utilizing transformation properties of the theta functions @xcite [ thetatransformation1 ] @xmath64 one derives the energies in the opposite limit of the high temperatures : [ canonicalmeanenergy6 ] @xmath65 the corresponding heat capacities @xmath1 are calculated by applying the right - most part of eq . to the above dependencies ; in particular , one has for the cold " qw , @xmath52 : [ canonicalheatcapacity1 ] @xmath66 for the hot thermal bath : [ canonicalheatcapacity2 ] @xmath67 mean energy @xmath57 of the canonical ensemble in terms of the electric field @xmath68 and temperature @xmath58 for all permutations of the bcs . in each of the panels , the corresponding type of the edge requirements is denoted by the two characters . ]
statistically averaged energies and corresponding heat capacities are shown in fig . [ electr0 ] .
at the zero temperature , the total internal energy reduces to the ground - state energy and the heat capacity is zero , as it should be . as it follows from eqs . and , their avalanche growth from the @xmath69 values for the purely neumann qw takes place at the smaller temperatures as compared to the mixed bcs , which , in turn , is followed by the quantities for the dirichlet structure .
this is explained by the growing difference between the two lowest energies for just consecutively mentioned bc configurations : the quantity @xmath70 at the zero field is the smallest ( largest ) for the neumann ( dirichlet ) structure : @xmath71 a remarkable feature of the heat capacity dependence is its nonmonotonic behaviour for the neumann qw : at @xmath72 ( @xmath73 ) it reaches a pronounced maximum @xmath74 that is followed by the minimum of @xmath75 located at @xmath76 ( @xmath77 ) . if the maximum is observed quite exactly by keeping only the first term in the parentheses of the right - hand side of eq .
, the emergence and precision of the location and magnitude of the second extremum are described better by keeping more terms in the same expansion .
physically , this nonmonotonicity of the heat capacity is attributed to the structure of the energy spectrum , see eqs . and ;
namely , very small temperature promotes the particle mainly to the first excited level that is only one unit above the ground state , @xmath78 , with the contribution of the other levels being negligibly small due to the almost vanishing exponents in eq . or ,
equivalently , in eqs . and ;
as a result , the heat capacity grows rapidly . for the larger temperatures , the occupations of the higher lying levels become essential
; however , the transitions to them are more difficult since the difference between , e.g. , second and first excited states @xmath79 is three times larger than that between the latter and the ground level .
accordingly , the same speed of the heat capacity change can not be sustained what results in the observed maximum . for the other bcs ,
the ratio @xmath80 is smaller than for the neumann qw , as it follows from eq . :
@xmath81 and @xmath82 ; as a result , for them no extrema are observed on the @xmath83 dependence at @xmath84 .
mathematically , the drop of the nn specific heat is caused by the interplay between the counterbalancing terms @xmath85 and @xmath86 in eq . as the temperature grows . keeping only the first exponent in the parentheses of the right - hand side of this equation produces @xmath87 while the same procedure applied to the other bcs , see eqs . and
, results in @xmath88 and @xmath89 , which are , respectively , three and two times smaller and lie beyond the range of the validity of these expansions . accordingly , for the latter two configurations , it is essential to keep other items in the corresponding series in order for them to be correct at the decreasing @xmath34 , and these extra exponents eliminate the resonance of the first - term approximation while for the neumann qw the ( negative ) second component simply improves the previous result . note that the ho leading term of the capacity expansion from eq . also results in @xmath90 ; however , the subsequent ( all positive ) items in the series wipe out the extremum . very broad and gentle asymmetric maximum
is observed at @xmath91 for the dirichlet qw while for the mixed bc the heat capacity is a monotonically increasing function of the temperature , which , at quite large @xmath2 , rapidly approaches the asymptotic value of one half . on the contrary
, the heat capacity of the symmetric qws reaches the same limit much slower , as eq . asserts and panel ( a ) of fig .
[ electr0 ] exemplifies .
note that the ho internal energy for the high temperatures is twice of that for the hard - wall qw : @xmath92 and @xmath93 in regular units , respectively . from point of view of classical equilibrium statistics that is applicable for @xmath94 ,
this difference is explained by the fact that in the former case the kinetic and potential parts of the motion make equal contributions of @xmath93 to the total energy @xcite while for the latter system it is the kinetic energy only that determines @xmath42 as the qw potential is zero . as a direct consequence of this ,
the qw heat capacities in the same limit are one half of their ho counterpart
. applied electric field modifies the energy spectrum what , in turn , affects the thermodynamic properties of the wells .
it was shown that the voltage increases the difference @xmath95 between the ground and first excited levels for any permutation of the bcs ( the only exception is the nd case at the small fields , see equations ( 50 ) in @xcite ) ; accordingly , the larger temperature is needed to push out the electron from its lowest state .
this is reflected in figs .
[ energycanonical ] and [ heatcapacitycanonical ] where the energy @xmath96 and heat capacity @xmath97 , respectively , are shown .
it is seen that the @xmath98 range where the mean energy does not change appreciably from the ground - state value gets wider for the stronger intensities @xmath0 .
the same is true for the heat capacity where the plateau with its almost zero value grows with the field .
the increasing voltage wipes out the nn minimum of the heat capacity simultaneously moving the maximum to the higher temperatures and increasing its magnitude . for each of the mixed bcs
, it also creates a maximum that was absent at @xmath15 . mentioned above dd extremum of the heat capacity gets narrower and its peak increases with the field growing .
recalling again the language of the classical statistical mechanics @xcite , one qualitatively explains the larger heat capacities at the nonzero fields by the contribution of the electric potential ; namely , the thermally averaged value of the potential energy @xmath99 is : @xmath100 this classical expression is applicable to our quantum system for the large temperatures only : @xmath101 then , the potential contribution to the heat capacity reads : @xmath102 note that , contrary to the ho , the kinetic and potential contributions to the heat capacity in this case , generally , are not equal to each other .
let us also mention once again that the electric field does _ not _ affect at all the ho heat capacity , see eq .
, since it simply shifts all the levels by the same amount , according to eq . . the same as in fig . [ energycanonical ] but for the polarization @xmath103
] fig .
[ polarizationcanonical ] depicts statistically averaged polarizations @xmath104 in terms of @xmath0 and @xmath98 .
growing temperature leads to the decrease of @xmath104 for all electric fields ; however , thermal energy is not strong enough to make the total dipole moment negative : for any bc the polarization stays positive .
to understand the statistical properties better , it is instructive to consider the case of the low temperatures . for the small voltages , as a first approximation , we also accept undisturbed by the field energies from eq . .
then , one has following dependencies : [ polarizationlimit1 ] @xmath105 these equations were derived under the assumption of @xmath106 , but the general property stating that the first - order temperature correction is determined by @xmath107 , holds for any electric intensities . for the small fields , this difference is negative @xcite what naturally explains the decrease of the total polarization with the temperature growing . in the opposite limit of the high voltages ,
the polarizations of the qw with the uniform bcs tend to the same level - independent value of one - half @xcite what requires larger temperatures in order to see the deviation of @xmath5 from its @xmath69 value .
this is exemplified in fig .
[ polarizationcanonical ] where the temperature - independent plateau at @xmath69 widens with the field growing .
for the mixed edge requirements , this limiting quantity is supplemented by the term that is proportional to @xmath108 with its sign being determined by the orientation of the bcs @xcite ; so , for the dn case it is actually possible to observe the increase of the polarization with the temperature growing from zero .
this feature is not shown in the corresponding panel of the figure since it takes place beyond the figure range @xmath109 .
grand canonical distribution is used for the description of the quantum system that , in addition to the thermal balance with the external reservoir , is also in the chemical equilibrium with it .
accordingly , the structure can exchange the energy as well as particles with the heat bath .
so , the number of the quantum corpuscles @xmath3 in it can be changed . the fundamental role in this case
is played by the chemical potential @xmath110 , which is defined from the condition @xmath111 where the upper sign corresponds to the fermi - dirac ( fd ) distribution while the lower one describes bose - einstein ( be ) particles .
physically , the difference between these two statistics is in the fact that each quantum level can not be occupied by more than one fermion ( pauli exclusion principle ) while the arbitrary number of bosons can coexist in the same state .
the distribution function now depends not only on the energies @xmath37 but also on the number of the particles in the system @xmath3 ; namely , for the physical quantity @xmath40 its grand canonical average value @xmath112 is @xmath113 applying the distribution from eq . for the calculation of the heat capacity , eq .
, one finds that its grand canonical value @xmath114 is @xmath115 ^ 2}\,e^{(e_n-\mu)\beta},\ ] ] where the chemical potential , which , in the case of fermions , is also frequently called the fermi level , is calculated , as stated above , from eq . . physically , the value of @xmath110 corresponds to the energy that is needed for changing by one the number of the particles in the system : @xmath116 for calculating its partial derivative with respect to the temperature , one should consider eq . as a condition of zeroing of the implicit function @xmath117 of the chemical potential in terms of the variables @xmath34 and @xmath3 : @xmath118 the rule of differentiating implicit functions states @xcite : @xmath119 as a result , one finds : @xmath120 ^ 2}\,e^{(e_n-\mu)\beta}}{\sum_n\frac{1}{\left[e^{(e_n-\mu)\beta}\pm1\right]^2}\,e^{(e_n-\mu)\beta } } .\ ] ] the boson statistics is used for the particles with the integer spin such as photons or cooper pairs in superconductors while the fd distribution is applied for the system of the constituents with the half integer spins ; for example , the electron with its spin of @xmath121 has , for the same energy , two projections of its spin equal to @xmath122 .
however , in our discussion below we will neglect this fact and will assume that the number of the fermions for each energy @xmath37 is not larger than one .
[ fermigranddirichlet ] depicts the fd chemical potential for the pure dirichlet qw as a function of the electric field and temperature for several numbers @xmath3 .
qualitatively , the same features are characteristic for other bc permutations too .
there are several distinct regions of the fermi energy @xmath123 dependence on the temperature . from its @xmath124 value at @xmath69
it rapidly grows as @xmath125 until it reaches and stays exactly at the value of @xmath126 which is due to the interaction of the two corresponding levels that , at the zero temperature , were the highest occupied and lowest unoccupied states .
the width of this @xmath2-independent plateau is determined by the number of the particles @xmath3 and electric field @xmath0 . for the still higher temperatures , the chemical potential for @xmath127 decreases while for the larger number of particles , @xmath128
, it grows with @xmath98 , reaches maximum and only after that decreases , passes zero at @xmath129 and continues to decline into the negative part of the spectrum . for the nonpositive chemical potentials , @xmath130 , eq .
can be cast into the form @xmath131 for the zero field , @xmath15 , the energy spectrum from eqs . and simplifies this equation as follows : for the ho : [ numbern_3 ] @xmath132 putting here the chemical potential equal to zero , @xmath133 , leads to the calculation of @xmath129 . in
known to us literature @xcite , there are no analytical expressions for these infinite series . however , for the very small @xmath34 , the @xmath134 terms in the above equations make the most significant contributions producing the following dependencies : for the ho : [ chemicalpotentiallimit3 ] @xmath135 superscripts @xmath136 and @xmath137 in eq .
stand for any of the values of @xmath138 and/or @xmath3 .
[ fermigranddirichlet ] manifests that , for the larger @xmath3 , these asymptotics are achieved at the higher temperatures . as a result , the grand canonical mean energy @xmath139 reads in the same limit : [ grandenergylimit1 ] @xmath140 what , by means of eq .
, immediately leads to the associated heat capacities @xmath114 : [ grandcanonicalheatcapacity2 ] @xmath141 a comparison of these remarkable results with eqs . , and confirms the general property , which states that for the large temperatures there is no difference between canonical and grand canonical distributions @xcite .
however , for the small @xmath2 these two statistics produce very different features .
[ heatcapacitygranddirichlet ] shows the fd heat capacity of the pure dirichlet qw in terms of the temperature and electric field for the different @xmath3 corresponding to their counterparts from fig .
[ fermigranddirichlet ] .
it is seen that , for the larger number of the particles , the asymptotics from eq .
is achieved at the higher @xmath2 . at the zero field , a prominent characteristic of the heat capacity dependence for the one particle
( top left panel of fig .
[ heatcapacitygranddirichlet ] ) is a salient maximum @xmath142 observed at @xmath143 , i.e. , at the right edge of the plateau from eq . .
accordingly , we attribute this extremum to the different behaviour of the chemical potential for @xmath127 and @xmath128 ; namely , as it was mentioned during discussion of fig .
[ fermigranddirichlet ] , for one particle the fermi energy decreases after the flat part from eq . while for any other number @xmath3 it grows with @xmath2 . thus , their contributions to the heat capacity from eq . are opposite to each other what results in the resonance that is observed for the one particle only .
even though the shape of this maximum is quite similar to its nn counterpart for the canonical ensemble , see sec .
[ sec_canonical ] , its physical explanation is completely different .
first , we point out that the very similar extrema are calculated also for the nd ( with @xmath144 and @xmath145 ) and pure neumann ( @xmath146 and @xmath147 ) qws too .
the fact that the three @xmath148 are almost the same and the ratios of the three temperatures @xmath149 are practically equal to those of @xmath150 from eq .
, undoubtedly proves that the origin of this effect is the bc independent one and that the interplay between the two lowest states plays a dominant role in it . to understand these resonances ,
let us recall that , for the very small temperatures , the properties of the fd well are determined only by the highest occupied level and its interaction with the nearest ( empty at @xmath69 ) above lying state , what is reflected in the extremely rapid approach by the chemical potential to the energy from eq .
that is located exactly in the middle between them . for @xmath128 , a contribution from the lower lying members in this regime
is negligibly small and can be safely neglected , while for the one - electron qw this addition is absent by definition .
further growth of the temperature increases thermal energy but it is still too `` weak '' to compel the corpuscles , which at @xmath69 lied below the fermi energy , to contribute to the heat capacity . only at the right edge of the plateau ,
the thermodynamic quantum @xmath92 becomes strong enough and forces other particles to donate to @xmath123 and @xmath151 .
therefore , for @xmath128 the heat capacity is a quite smoothly varying function of the temperature .
however , for @xmath127 there are no such additional donors that aid to support the continuous growth of the heat capacity , which can not be sustained by the one particle only . as a result
, the specific heat reaches maximum and drops .
this qualitative physical reasoning can be corroborated by the simple quantitative mathematical analysis .
fermi level from eq
. defines the corresponding mean energy and heat capacity as : [ grandcanonicallimit2 ] @xmath152 on the other hand , for the chemical potential from eq .
these quantities for one fermion , @xmath127 , become : [ grandcanonicallimit3 ] @xmath153,\quad\beta\gtrsim1,\end{aligned}\ ] ] where we take into consideration only the two lowest states . for the pure neumann qw without the field ,
this last expression takes an especially simple form : @xmath154 this function has a pronounced maximum of @xmath146 at @xmath155 . a perfect coincidence with the provided above exact results
justifies a validity of the two - level approximation and proves that the electron transitions between them determine the specific heat resonance .
it is also very instructive to contrast eq . with its canonical counterpart from eq .
in sec . [ sec_canonical ] .
the comparison shows that the magnitude of the grand canonical neumann extremum is almost two times larger and it is achieved at more than two times lower temperature .
applied field @xmath0 smooths out and widens this maximum simultaneously increasing the heat capacity .
similar to the canonical ensemble , this growth is explained by the contribution of the electrostatic potential .
however , the electric influence is drastically decreased by the growing number of the particles in the qw ; for example , right - bottom panels exhibit the almost full independence of the fermi energy and heat capacity on the intensity @xmath0 already for @xmath156 .
this is explained by the properties of the energy spectrum in the electric field when the higher lying states ( which , in the case of the fd distribution , determine the features of the system ) are less affected by the applied voltage @xcite .
[ polarizationgranddirichlett0 ] demonstrates zero - temperature fd polarization for all possible bcs and several numbers @xmath3 .
as the well accommodates more fermions , the total polarization becomes smoother function of the electric field .
figure reveals that , independently of the edge demands , the magnitude of @xmath5 at @xmath157 grows linearly with the voltage and the slope of this almost straight line diminishes with @xmath3 . for any number of fermions ,
the total polarization remains positive at the arbitrary voltage .
nonzero temperature leads to the dependencies that qualitatively are similar to the canonical patterns , fig .
[ polarizationcanonical ] , and , because of this , the corresponding polarizations are not shown here .
total polarization @xmath103 as a function of the electric field @xmath158 at zero temperature , @xmath69 , for different number of fermions @xmath3 that is depicted next to the corresponding curve . ] next , let us discuss bosonic structures .
remarkable experimental observations of the be condensation in the vapors of rubidium @xcite and sodium @xcite spurred an avalanche of the research on the subject predicted almost ninety years ago @xcite , see , e. g. , reviews @xcite .
theoretically , the main effort was devoted to the calculation of the properties of the be systems in the 3d isotropic or anisotropic harmonic traps @xcite and their existence / nonexistence in lower dimensions @xcite .
however , other forms of the confining potentials @xcite , including the 3d box with the periodic @xcite or uniform @xcite bcs , were also discussed with the comparative analysis of their influence of the properties of the trap @xcite . from this point of view , an inclusion of the electric voltage and different bcs presents a generalization of the previous analysis .
moreover , overwhelming majority of the research concentrated on the analysis of the be systems in the thermodynamic limit when the number of the particles and the volume containing them tend to infinity while the the density is kept constant . in this approximation ,
the infinite series above in this section can be safely replaced by the integrals @xcite .
considering @xmath3 changing from one to the large values might help to understand the formation of the be processes with the the number of the particles growing .
first , we state that eqs .
- stay valid for the be statistics too since they were obtained as a result of retaining the first term in the series from eq . .
in the opposite limit of the very small temperatures , it is elementary to derive : [ belimit1 ] @xmath159 ( a ) critical temperatures @xmath160 and ( b ) corresponding to them polarizations @xmath161 from eq . as functions of the applied electric field @xmath158 for different number @xmath162 of bosons that are depicted near the corresponding curves .
solid ( dotted ) lines are for the pure dirichlet ( neumann ) bc while their dash ( dash - dotted ) counterparts denote dn ( nd ) geometry . in panel
( b ) , thin horizontal line is zero polarization . ] an important characteristic of the be system is its critical temperature @xmath4 .
it corresponds to the situation when the chemical potential is equal to the energy of the lowest level , @xmath163 , and the number of the particles in this state @xmath164 is zero what leads to the implicit mathematical equation for finding @xmath165 @xmath166 physically , it is the largest temperature at which the be condensation still can be observed , and at the lower @xmath2 the fraction @xmath167 of the particles in the ground state will increase until at @xmath69 it becomes unity : @xmath168 fig . [
criticaltemppolariz](a ) shows dependencies of the critical parameter @xmath169 on the applied voltage for all possible bcs and several numbers @xmath3 . in accordance with the previous results @xcite , the temperature @xmath4 increases with @xmath3 . in the absence of the fields ,
the lowest ( highest ) temperature is observed for the pure neumann ( dirichlet ) qw what is a reflection of the corresponding spectrum from eq . and
the energy difference between the affiliated states , see eq . .
electric field leads to the modification of the mutual location of the levels on the energy axis ; in particular , at the small voltages , the two lowest states move closer to each other for the nd geometry while the difference @xmath170 grows with the field for all @xmath0 and any other bc configuration @xcite . as a result , the critical temperature for the former edge requirement decreases with the growing from zero field , passes through minimum and then unrestrictedly grows with the electric intensity while for all other bcs it is a continuously increasing function of @xmath0 . at the high voltages ,
the energy spectrum is determined mainly by the condition at the right wall @xcite what leads to almost the same critical temperature for , e. g. , the nn and dn wells .
note that in this regime , contrary to the zero fields , the dirichlet requirement is more favorable to the formation of the be condensate as compared to the neumann interface .
this is explained by the larger level separation for the latter geometry at the high voltages @xcite . as the ground - state polarization is positive for all fields and bcs @xcite
, one can expect that at the onset of the be condensation the total statistically averaged dipole moment will , at the small voltages , be negative .
this is exactly what is observed in panel ( b ) of fig .
[ criticaltemppolariz ] that shows the polarizations @xmath171 corresponding to the critical temperatures @xmath169 from panel ( a ) .
they were calculated from equation @xmath172 and , as stated above , the critical temperature @xmath169 was found from eq . .
the characteristic features of the critical polarizations basically follow the properties of the first excited state : from zero they decrease with the growth of the field , reach minimum after which they increase .
however , for the large voltages many upper lying states are occupied and contribute to the total dipole moment . as a result ,
the high - field @xmath171 for one boson is considerably smaller than @xmath173 .
the absolute value of the negative polarization at the extremum grows with the number of the particles with the largest one , at the fixed @xmath3 , being observed for the pure neumann qw followed by its nd counterpart what is a replica of the similar behaviour for the first excited level @xcite . for the temperatures below @xmath4 ,
the nonzero occupation of the ground state contributes a positive term to the polarization what leads to the gradual disappearance of the negative region of the total dipole moment @xmath5 with the decreasing temperature until at @xmath69 it becomes @xmath174 , which is positive for any bc and arbitrary fields @xcite .
bosonic heat capacity @xmath56 of the pure neumann qw as a function of the applied electric field @xmath68 and temperature @xmath58 for ( a ) @xmath175 , ( b ) @xmath176 and ( c ) @xmath177 .
note that in the last two cases the temperature is scaled in units of the critical temperature @xmath178 while the corresponding @xmath179 axes measure specific heat per particle @xmath180 .
panel ( d ) shows the chemical potential @xmath181 for @xmath177 with the corresponding heat capacity depicted in part ( c ) . ] as a final example , fig .
[ boseneumann ] exhibits evolution of the heat capacity and chemical potential with the varying electric field and temperature for the pure neumann qw .
it is seen that for the small number of bosons , say , @xmath182 in panel ( a ) , the applied voltage leads , at quite warm sample , to the increase of @xmath1 while at the small @xmath2 , the width of the temperature - independent zero - capacity plateau increases with the field .
these features were discussed before for the canonical ensemble . increasing the number of the particles in
the well leads to the suppression of the voltage dependence , as a transition from panel ( a ) to ( b ) with @xmath183 and ( c ) for @xmath184 vividly demonstrates .
no any noticeable field dependence is seen there in the range @xmath109 .
it is well known that for some potentials , such as , e.g. , the 3d isotropic harmonic trap @xcite , the heat capacity has a cusp - like peculiarity as it passes through the critical temperature while for the 1d quadratic potential it is a smooth function of @xmath2 @xcite . fig .
[ boseneumann ] exemplifies that no any peculiarity is observed for the 1d hard - wall potential with neumann surfaces and arbitrary applied electric fields .
our calculations confirm that the same is true for any other bcs .
finally , panel ( d ) shows that the chemical potential @xmath110 is a monotonically decreasing function of both the electric field @xmath0 and temperature @xmath2 .
it is seen that the growing temperature diminishes the voltage influence on the chemical potential .
rigorous mathematical treatment of the qw with miscellaneous bcs under the applied voltage revealed a strong influence of the interplay between them on the thermodynamic properties of the structure .
in particular , without the field the differences of the energy spectrum lead , for the canonical ensemble , to the conspicuous maximum followed by the minimum of the heat capacity @xmath1 on the temperature axis for the nn quantum box while for the other edge requirements no such adjacent extrema are observed .
modification of the specific heat and statistically averaged polarization in the field is qualitatively explained by the influence of the associated electrostatic potential .
numerical calculations , which predicted , for the flat potential with the arbitrary bc , a salient maximum of @xmath1 as a function of @xmath2 for one fermion and its absence for the larger @xmath3 , were corroborated by the two - level model that allows simple analytical treatment with its predictions perfectly coinciding with the exact results . from this , a clear physical explanation of this phenomenon follows that is based on the analysis of the associated fermi energy . it is predicted that the applied field , in general , favors the formation of the be condensate , and the differences and similarities of this process for the different bcs are discussed .
the thermally averaged dipole moment is shown to take the negative values in some ranges of the fields and temperatures .
it is also argued that for the larger number of either fermions or bosons in the qw , the influence of the electric field on the thermodynamic properties diminishes .
dirichlet and neumann conditions are the limiting cases of the so called robin bc @xcite @xmath185 where @xmath186 is an inward unit normal to the surface , and the parameter @xmath187 has a dimension of length and is called the extrapolation distance .
its variation allows a continuous transformation from the dirichlet ( @xmath188 ) to the neumann ( @xmath189 ) situation . without the field ,
especially intriguing are the properties of the qw at the small negative robin lengths , @xmath190 , when , in addition to the positive spectrum , two almost degenerate odd and even levels with the energies @xmath191 are created @xcite .
analysis of the robin qw in the electric field might present an interesting extension of the present research .
this project was supported by deanship of scientific research , college of science research center , king saud university .
100 o. olendski , ann . phys .
( berlin ) * 527 * , 278 ( 2015 ) .
n. dalarsson , m. dalarsson , and l. golubovi , introductory statistical thermodynamics ( academic , amsterdam , 2011 ) .
r. bellman , a brief introduction to theta functions ( holt , rinehart and wilson , new york , 1961 ) . m. abramowitz and i. a. stegun , handbook of mathematical functions ( dover , new york , 1964 ) . i. s. gradshteyn and i. m. ryzhik , table of integrals , series , and products , 8th edn .
( academic , amsterdam , 2014 ) .
a. p. prudnikov , y. a. brychkov , and o. i. marichev , integrals and series vol .
1 ( gordon and breach , new york , 1986 ) .
a. p. prudnikov , y. a. brychkov , and o. i. marichev , integrals and series vol
. 2 ( gordon and breach , new york , 1990 ) . y. a. brychkov , handbook of special functions : derivatives , integrals , series and other formulas ( chapman & hall , boca raton , 2008 ) .
w. rudin , principles of mathematical analysis , 3rd edn .
( mcgraw - hill , new - york , 1976 ) . m. h. anderson , j. r. ensher , m. r. matthews , c. e. wieman , and e. a. cornell , science * 269 * , 198 ( 1995 ) . k. b. davis , m .- o .
mewes , m. r. andrews , n. j. van druten , d. s. durfee , d. m. kurn , and w. ketterle , phys .
lett . * 75 * , 3969 ( 1995 ) .
s. n. bose , z. phys .
* 26 * , 178 ( 1924 )
. f. dalfovo , s. giorgini , l. p. pitaevskii , and s. stringari , rev .
phys . * 71 * , 463 ( 1999 ) .
l. pitaevskii and s. stringari , bose - einstein condensation ( clarendon , oxford , 2003 ) .
a. j. leggett , rev .
* 73 * , 307 ( 2001 ) . c. j. pethick and h. smith , bose - einstein condensation in dilute gases ( cambridge , cambridge , 2004 ) .
w. ketterle and n. j. van druten , phys .
a * 54 * , 656 ( 1996 ) .
n. j. van druten and w. ketterle , phys .
lett . * 79 * , 549 ( 1997 ) .
r. napolitano , j. de luca , v. s. bagnato , and g. c. marques , phys .
a * 55 * , 3954 ( 1997 ) .
w. j. mullin , j. low temp .
106 * , 615 ( 1997 ) .
v. bagnato and d. kleppner , phys .
a * 44 * , 7439 ( 1991 ) .
v. bagnato , d. e. pritchard , and d. kleppner , phys .
a * 35 * , 4354 ( 1987 ) .
e. b. sonin , zh .
fiz . * 56 * , 963 ( 1969 ) [ sov .
- jetp * 29 * , 520 ( 1969 ) ] .
s. greenspoon and r. k. pathria , phys .
rev . a * 9 * , 2103 ( 1974 ) . c. s. zasada and r. k. pathria , phys
. rev . a * 14 * , 1269 ( 1976 ) .
d. f. goble and l. e. h. trainor , can .
* 44 * , 27 ( 1966 ) .
d. f. goble and l. e. h. trainor , phys . rev . * 157 * , 167 ( 1967 ) .
d. f. goble and l. e. h. trainor , can .
* 46 * , 1867 ( 1968 ) .
s. grossmann and m. holthaus , z. phys .
b * 97 * , 319 ( 1995 ) .
r. k. pathria , phys .
rev . a * 5 * , 1451 ( 1972 ) . s. greenspoon and r. k. pathria , phys .
rev . a * 8 * ,
2657 ( 1973 ) . m. n. barber and m. e. fisher , phys .
rev . a * 8 * , 1124 ( 1973 ) . z. r. hasan and d. f. goble , phys . rev . a * 10 * , 618 ( 1974 ) .
h. haugerud , t. haugset , and f. ravndal , phys .
a * 225 * , 18 ( 1997 ) .
t. haugset , h. haugerud , and j. o. andersen , phys .
a * 55 * , 2922 ( 1997 ) .
k. gustafson and t. abe , math .
* 20*(_1 _ ) , 63 ( 1998 ) .
o. olendski and l. mikhailovska , phys .
e * 81 * , 036606 ( 2010 ) . | thermodynamic properties of the one - dimensional ( 1d ) quantum well ( qw ) with miscellaneous permutations of the dirichlet ( d ) and neumann ( n ) boundary conditions ( bcs ) at its edges in the perpendicular to the surfaces electric field @xmath0 are calculated . for the canonical ensemble ,
analytical expressions involving theta functions are found for the mean energy and heat capacity @xmath1 for the box with no applied voltage . pronounced maximum accompanied by the adjacent minimum of the specific heat dependence on the temperature @xmath2 for the pure neumann qw and their absence for other bcs
are predicted and explained by the structure of the corresponding energy spectrum . applied field leads to the increase of the heat capacity and formation of the new or modification of the existing extrema what is qualitatively described by the influence of the associated electric potential .
a remarkable feature of the fermi grand canonical ensemble is , at any bc combination in zero fields , a salient maximum of @xmath1 observed on the @xmath2 axis for one particle and its absence for any other number @xmath3 of corpuscles .
qualitative and quantitative explanation of this phenomenon employs the analysis of the chemical potential and its temperature dependence for different @xmath3 .
it is proved that critical temperature @xmath4 of the bose - einstein ( be ) condensation increases with the applied voltage for any number of particles and for any bc permutation except the nd case at small intensities @xmath0 what is explained again by the modification by the field of the interrelated energies .
it is shown that even for the temperatures smaller than @xmath4 the total dipole moment @xmath5 may become negative for the quite moderate @xmath0 .
for either fermi or be system , the influence of the electric field on the heat capacity is shown to be suppressed with @xmath3 growing .
different asymptotic cases of , e.g. , the small and large temperatures and low and high voltages are derived analytically and explained physically .
parallels are drawn to the similar properties of the 1d harmonic oscillator , and similarities and differences between them are discussed . |
i thank brian smith for discussions .
this research was supported in part by the natural sciences and engineering research council of canada .
99 a. n. kolmogorov , dokl .
nauk sssr 30 , 9 ( 1941 ) .
u. frisch , `` turbulence : the legacy of a.n .
kolmogorov '' , cambridge university press ( 1995 ) .
a. arneodo , et .
al . , europhy .
34 , 411 ( 1996 ) .
j. herweijer and w. van de water , phys .
74 , 4651 ( 1995 ) .
p. tabeling , g. zocci , f. belin , j. maurer , and h. willaime , phys .
e53 , 1613 ( 1995 ) ; f. belin , p. tabeling , h. willaime , physica d93 , 52 ( 1996 ) .
r. benzi , s. ciliberto , c. baudet , g. r. chavarria , physica d80 , 385 ( 1995 ) ; r. benzi , l. biferale , s. ciliberto , m.v .
strugila and t. tripiccione , physica d96 , 162 ( 1996 ) .
b. castaing , y. gagne , and e. hopfinger , physica d46 , 177 ( 1990 ) .
a. n. kolmogorov , dokl .
nauk sssr 32 , 16 ( 1941 ) .
she and e. leveque , phys .
72 , 336 ( 1994 ) .
r. benzi , s. ciliberto , r. tripiccione , c. baudet , e. massaioli and s. succi , phys .
e48 , 29 ( 1993 ) . | from a simple path integral involving a variable volatility in the velocity differences , we obtain velocity probability density functions with exponential tails , resembling those observed in fully developed turbulence .
the model yields realistic scaling exponents and structure functions satisfying extended self - similarity .
but there is an additional small scale dependence for quantities in the inertial range , which is linked to a slow approach to kolmogorov ( 1941 ) scaling occurring in the large distance limit .
.25in.25 in 8.5 in the universal features displayed by fully developed hydrodynamic turbulence are still not fully understood .
kolmogorov ( 1941 ) @xcite showed how a set of statistical quantities known as structure functions are expected depend on the length scale @xmath0 as power laws with predicted exponents .
experimental measurements have indicated that while these predictions are close to the truth , the predicted exponents are not exactly realized . in the face of this experimental input , much effort has been devoted towards understanding the origin of these anomalies in the scaling exponents , while retaining the notion that current experiments are observing an `` inertial range '' where strict power law scaling holds .
in this paper we will investigate what appears to be a loop - hole in this reasoning . in spite of impressive advances made in the experimental studies
, the fact remains that the scaling exponents have been deduced by looking at scaling regions extending over little more than one decade in @xmath0 @xcite .
this leaves open the possibility that a small but significant departure from strict power law scaling is still consistent with the data .
we will argue that this possible departure is sufficient to allow the observed anomalies in the exponents to be nothing more than a transient effect related to intermittency , and that the true power law scaling only occurs on larger scales .
this very large distance scaling could take the form proposed by kolmogorov .
we shall present a model which predicts deviations from power law scaling , and which shows that these deviations may be small enough to have escaped detection thus far .
the model is based on a simple physical picture for the effects of intermittency , which are effects due to the coming and going of coherent structures in the velocity field .
we will model the effects that these structures ( eddies ) have directly on the probability distribution function ( pdf ) for velocity differences .
we are claiming simplicity in our approach , but not uniqueness .
thus the applicability to 3-dimensional hydrodynamical turbulence in particular could be considered speculative , since the model makes no mathematical contact with the navier - stokes equations . on the other hand ,
when the consequences of the model are explicitly worked out they are found to coincide rather well with the data .
only the symmetric part of the observed pdf will be modeled in this paper .
this is denoted by @xmath1 where @xmath2 is the difference in some velocity component at two points separated by a distance @xmath0 .
the model will yield an explicit expression for the pdf which 1 ) for small @xmath0 has the typical sharp peak and broad tails characteristic of intermittent behavior , 2 ) at any finite @xmath0 displays exponential tails at large enough @xmath2 , and 3 ) tends toward a gaussian form in the large @xmath0 limit .
but the model shows that these realistic features may in fact be implying that the values of the scaling exponents are slightly scale dependent , as they evolve from their observed `` anomalous '' values for the values of @xmath0 where they are presently measured , to the pure k41 values in the large @xmath0 limit .
we consider a discrete set of points on the line connecting the two points separated by distance @xmath0 .
we label the points by their distance from one end , @xmath3 where @xmath4 .
we start by assuming that @xmath1 can be approximated by @xmath5{\cal p } ( v_{r}- v_{r_{n}}){\cal p } ( v_{r_{n}}- v_{r_{n-1}})\cdots { \cal p } ( v_{r_{1}}- v_{0 } ) , \label{bb}\ ] ] with @xmath6 .
we assume that the set of points @xmath7 is characterized by a scale @xmath8 such that @xmath9 , where @xmath10 and @xmath11 is a positive constant to be determined below .
the integration over all velocities at the intermediate points may be thought of as a sum over all paths in the space of velocities which start at @xmath12 and end at @xmath13 .
we could thus refer to ( [ bb ] ) as a path integral .
but note that we refrain from taking the continuum limit , @xmath14 with @xmath0 fixed , since the scale @xmath8 has a physical significance . the discretization of the range from 0 to @xmath0 into subregions @xmath15 is associated with our modeling of intermittent behavior .
roughly speaking , we are suggesting that the coming and going of eddies , of sizes comparable to subregion size @xmath16 , produce a _ variable volatility _ in the typical velocity differences @xmath17 .
we make this statement precise by writing @xmath18 as a superposition of gaussians , where we integrate over all values of a volatility parameter @xmath19 which is itself weighted according to a gaussian . to simplify notation
we denote @xmath20 .
@xmath21 the consideration of large and small values of @xmath19 corresponds to the possible presence or absence of eddies in that subregion at various times .
note that the variance of both @xmath22 and @xmath23 is @xmath24 .
( [ bb ] ) is constructed to give the cumulative effect of the variable volatilities occurring in the various subregions .
@xmath8 should be typical of scales on which viscosity influences and damps the formation of eddies .
as such @xmath8 is expected to lie between the dissipative ( kolmogorov ) scale , @xmath25 , and the lower end of the inertial scaling range ( the latter range being characterized by the absence of viscosity effects ) . note that if @xmath26 then the size of the subregion , @xmath16 , and thus the relevant eddy size for that subregion , would increase with @xmath27 .
we proceed by inserting ( [ dd ] ) into ( [ bb ] ) and integrating over the @xmath28 .
we can write the terms in the exponential which depend on @xmath29 in matrix notation @xmath30 where @xmath31 is a matrix and @xmath32 is a vector , with the latter depending on @xmath33 and @xmath34 .
we may complete the square and do the gaussian integrations over @xmath35 , for various values of @xmath36 . from this
we are able to deduce a simple result , where all dependence on the @xmath19 now resides in @xmath37 .
the @xmath38 integrations for fixed @xmath39 then just give the surface area of an @xmath40 dimensional sphere .
@xmath41{\frac{1}{\hat{s}}}{\rm exp}\left({-{\frac{(x_{n+1}-x_{0})^{2}}{2\sigma ^{2}\hat{s}^{2}}}-{\frac{1}{2}}\hat{s}^{2}}\right)\nonumber\\ & = & { \frac{1}{\sigma ( 2\pi ) ^{{\frac{n}{2}}+1}}}\left({{\frac{2\pi ^{{\frac{n+1}{2}}}}{\gamma ( { \frac{n+1}{2}})}}}\right)\times \int_{0}^{\infty } d\hat{s}\hat{s}^{n-1}{\rm exp}\left({-{\frac{(x_{n+1}-x_{0})^{2}}{2\sigma ^{2}\hat{s}^{2}}}-{\frac{1}{2}}\hat{s}^{2}}\right)\nonumber\\ & = & { \frac{\left({(n+1)/2}\right)^{{\frac{n}{2}}}}{\sigma \sqrt{\pi } \gamma ( { \frac{n+1}{2}})}}\int_{0}^{\infty } dss^{n-1}{\rm exp}\left({-{\frac{(x_{n+1}-x_{0})^{2}}{2\sigma ^{2}s^{2}(n+1)}}-{\frac{1}{2}}s^{2}(n+1)}\right)\end{aligned}\ ] ] in the last step we have defined @xmath42 for convenience . finally we will take this result and analytically continue from integer values @xmath40 to positive real values @xmath43 . replacing @xmath40 by @xmath44 gives @xmath45 we note that for large @xmath43 the integral over @xmath46 becomes strongly peaked about @xmath47 , and thus @xmath48 in addition we find , _ for any _ @xmath43 , that the variance in the velocity differences is @xmath49 the variable @xmath43 controls the evolution of the pdf , in the sense that evolution through unit steps in @xmath43 are generated by successive application of the `` evolution operator '' in ( [ dd ] ) . in this sense @xmath43 is an `` evolutionary '' time scale depending on the distance scale @xmath0 .
there is also a dynamical time scale associated with scale @xmath0 , the eddy turnover time , which is simply @xmath50 where @xmath51 is a typical velocity observed on scale @xmath0 . if we associate @xmath51 with @xmath52 from ( [ ee ] ) , and if the evolutionary and dynamical times happen to scale with @xmath0 in the same way , then we obtain @xmath53 .
this gives the standard scaling law @xmath54 .
we now perform the integral over @xmath46 , and rescale the pdf such that the variance is unity for all @xmath43 .
we obtain @xmath55 @xmath56 is the modified bessel function of the second kind . in fig .
( 1 ) we display @xmath57 for various @xmath43 .
$ ] vs. @xmath58 for @xmath59 , top to bottom .
the curves are vertically displaced by @xmath60 respectively . ] when @xmath43 is an even integer , @xmath57 is an exponential times a polynomial in @xmath58 , and in particular for @xmath61 it is purely exponential , @xmath62 . for @xmath63 the pdf is even more strongly peaked and it decreases less quickly than an exponential ( `` stretched exponential '' ) , before becoming purely exponential at large enough @xmath58 .
for @xmath64 we have `` deformed gaussians '' with exponential tails . for any @xmath43 , @xmath65 .
alternatively we may consider moments , @xmath66 , which are all finite and given by @xmath67 the approach to the exponential tails is reflected in @xmath68 .
it is common to describe the evolution of the pdf with @xmath0 in terms of the generalized structure functions , defined by @xmath69 . for @xmath0 in the inertial range
it is usually assumed that @xmath70 .
our model suggests that exactly scale independent exponents do not exist for the typical ranges of @xmath0 considered , and that instead we should consider the local exponents @xmath71 .
we obtain @xmath72 where @xmath73 is the digamma function .
this gives @xmath74 , and in the large @xmath0 limit , @xmath75 .
the latter is k41 scaling when @xmath53 .
although we adopt this value of @xmath11 in the following , we leave open the question of whether @xmath11 is exactly equal to @xmath76 or just close to it .. ] we also stress that we are describing only the symmetric part of the observed pdf , which in the case of the pdf for longitudinal velocity differences also has an asymmetric part .
this asymmetry is reflected by the nonvanishing of the structure functions @xmath77 for odd @xmath78 . in particular
our @xmath79 is not the same as the scaling exponent of @xmath80 , which is constrained to be unity @xcite . in fig .
( 2 ) we display @xmath81 for various @xmath0 , and compare to the large @xmath0 limit values @xmath82 . vs. @xmath78 for @xmath83 , bottom to top .
the dashed line is @xmath82 . ]
the main point is that the scaling exponents approach their asymptotic values very slowly .
it is also of interest that for some range of @xmath84 , our exponents are realistic for a wide range of @xmath78 .
we show this by comparing to the model of she and leveque ( sl)@xcite , which yields @xmath85 , and which is known to fit the current data quite well .
we may phrase the agreement in terms of the relative scaling exponents @xmath86 , which are known to better accuracy @xcite than the individual exponents , and for which the factor @xmath11 in ( [ ii ] ) cancels .
we find , for example , that @xmath87 as a function of @xmath78 is within a few percent of the corresponding sl values up to @xmath88 ! and as @xmath0 varies from 20 to 50 , @xmath89 for example varies from 1.72 to 1.82 , compared to @xmath90 . to see the @xmath0 dependence of the structure functions themselves we display @xmath91 in fig .
( 3 ) .
$ ] vs. @xmath92 with @xmath93 for @xmath94 , bottom to top , each line vertically offset so as to vanish at @xmath95 . ] over some range of @xmath0 the lines appear to be close to straight .
but other than for @xmath96 , which is exactly straight , the lines have a positive curvature which increases for larger @xmath78 .
we will see below how these positive curvatures , or equivalently the scale dependence of the scaling exponents , may be rather subtle to observe experimentally .
but first we must consider the effects of viscosity on the structure functions .
we expect that the effects of viscosity will cause @xmath97 to be replaced by a function @xmath98 which deviates from @xmath99 as @xmath0 approaches the dissipative scale @xmath25 from above .
since a decreasing @xmath43 corresponds to increasing intermittency , and since the cause of intermittency in our picture variable volatility of velocity differences should be damped by viscous effects , we expect that @xmath98 should start to decrease more slowly than @xmath99 .
thus the evolution of the pdf shapes is retarded by viscous effects in the `` intermediate viscous range '' , the range of @xmath0 between @xmath25 and the start of the inertial range .
the implication is that at large scales in the inertial range where @xmath100 it is likely that @xmath101 .
we may also expect that @xmath102 grows with the size of the intermediate viscous range
. for very small @xmath0 where the velocity field is smooth the variance is evolving like @xmath103 , faster than in the inertial range , and thus there must clearly be a decoupling between the evolution of the pdf shapes and the evolution of the pdf variance .
we are thus led to a generalization of our model , where we replace the expression in ( [ aa ] ) by the following . would be determined by unit steps in the function @xmath104 . instead of each point , groups of adjacent points would have a single fluctuating volatility , and the transition from one grouping to the next would determined by unit steps in the function @xmath98 . ]
@xmath105 this new pdf has the property that @xmath106 , and thus we can use @xmath104 to reproduce the observed behavior of the variance even in regions where viscosity or finite size effects are important . the function @xmath98 determines the evolution of the pdf shapes ; that is , the new @xmath107 is obtained from the one in ( [ ff ] ) by replacing @xmath43 with @xmath98 .
we will illustrate the effects of viscosity with specific choices for @xmath98 and @xmath104 .
we consider @xmath108 /\rho}\right)^{2/3}$ ] , which is unity at @xmath109 and approaches @xmath110 at large @xmath0 . to have some dependence on the size of the inertial range , we take @xmath111 for simplicity . ] for @xmath104 in the range @xmath112 we consider @xmath113/c_4}-c_3e^{[1 - 10^{4}{\frac{\eta } { r}}]/c_5}}\right](r/\eta)^{2/3}.\ ] ] we take two examples for @xmath114 which idealize typical data sets , @xmath115 with @xmath116 having a relatively large inertial range ( large reynolds number ) and @xmath117 with @xmath118 having a small inertial range .
our results are not very sensitive to precisely how these functions deviate from @xmath119 behavior outside the inertial range .
note that the intermediate viscous range is larger for @xmath115 .
due to our expectation that @xmath8 should increase with the size of the intermediate viscous range it is natural that @xmath120 .
for illustrative purposes only , we choose @xmath121 and @xmath122 . with this input we can extract all higher order structure functions from ( [ gg ] ) .
the scaling in the inertial range for both cases turns out ( because of our fortuitous choice of @xmath123 and @xmath124 ) to be described by the realistic scaling exponents @xmath125 from ( [ ii ] ) .
we show this in fig .
( 4 ) where we plot @xmath126 .
$ ] vs. @xmath92 with @xmath127 for @xmath128 , bottom to top .
the solid and dashed lines correspond to @xmath115 and @xmath117 respectively . ] of most interest are the @xmath129 curves for the @xmath115 case , where we see ( very slight ! ) evidence of a double hump structure growing more prominent with increasing @xmath78 .
this positive curvature in the middle of the scaling region is arising from the positive curvatures in fig .
( 3 ) , which in turn is a reflection of the gradual approach to k41 scaling .
this is the generic signature of our model , which should be seen when the size of the inertial range and the order @xmath78 are both large enough .
the figure makes it clear , though , that it is a small effect which may be hidden in present data .
our model may also be used to illustrate `` extended self - similarity '' @xcite . in fig .
( 5 ) we plot @xmath91 versus @xmath130 for @xmath131 for the two cases @xmath115 and @xmath117 .
$ ] vs. @xmath132 $ ] for @xmath131 and for @xmath133 , bottom to top .
the lines for the @xmath115 and @xmath117 cases are displaced by 0.5 and @xmath134 respectively .
the dashed straight lines have slopes @xmath87 . ] for comparison we add straight lines with slopes given by the relative exponents @xmath87 .
we see that scaling has been extended to smaller scales than is apparent in fig .
( 4 ) .
similar results are obtained for other choices of the functions @xmath98 and @xmath104 .
it thus appears that such plots are not very sensitive to the deviations from power law scaling we are proposing .
we reiterate that the model yields a universal set of _ local _ scaling exponents , and the scaling exponent from an experiment depends on what distance scale , effectively , the local scaling exponent is being measured at relative to @xmath8 . on the other hand if it is true that @xmath102 increases with the size of the intermediate viscous region , as we are suggesting , then it may be difficult to obtain measurements at distances , in units of @xmath8 , which are very different from each other .
the variability in the scaling exponents is also obscured if different experiments have intermediate viscous regions of similar size , and/or if one is confined to the lower order structure functions , such as @xmath135 and below .
we have seen how viscosity and finite size effects can have the effect of transforming the structure functions in fig .
( 3 ) into the structure functions in fig .
( 4 ) which in turn display extended self - similarity in fig . ( 5 ) .
it is also encouraging to find that realistic values of the scaling exponents emerge when the @xmath8 parameter is within the intermediate viscous range .
but perhaps most important is that the model suggests how evidence for a slow approach to k41 scaling could eventually be uncovered .
we should also differentiate between the evolution of the pdf shapes as a function of @xmath0 ( to which the structure functions are sensitive ) , and the basic set of shapes predicted by the model , given in ( [ ff ] ) and depicted in fig .
( 1 ) .
these probability density functions may be of interest in various other contexts . |
in the last decade , mainly due to the observations by _ beppo_sax and related optical earth based follow up ,
the physics of grbs and their afterglow started to be understood .
however , the intrinsic duration of the observational process , mainly due to the use of device non
optimized for fast response , did not allow up to now to monitor the prompt afterglow , i.e. the emission in the few seconds after the burst ( except in one case , grb990123 ) .
in the next few years , dedicated missions such as heteii or swift and of other space borne high
energy observatories with trigger capabilities ( integral , agile , etc . ) will allow in a few years to accumulate a statistical significant number of grb detections . in this context
, rem will be operational at the european southern observatory of la silla within october 2002 .
rem shares with other telescopes the characteristics of fast pointing and full robotisation but it has the unique feature to be equipped with a high throughput nir camera . any possible optimization of the trigger sources will leave some time in which rem can not observe any grb afterglow due to latitude / longitude constraint .
we estimate such _ idle time _ to be of the order of @xmath2 of rem observing time . during the idle time rem
will be used for multifrequency monitoring of variable objects .
key programmes _ have been identified by the rem science team as particularly suitable for rem : * a * ) monitoring of blazars , * b * ) monitoring of flare and variable stars and * c * ) observation of ir counterpart of galactic black hole ( bh ) candidates .
rem has a classical ritchey
chretien optical scheme mounted in an alt azimuthal configuration with two nasmyth focal stations , a 60 cm primary mirror and a total focal ratio of f/8 .
such optical figures allow a compact structure that provides the needed stiffness for fast motion and windy environmental conditions such as those expected in a fully deployable dome at the la silla observatory .
one of the nasmyth focal station will be equipped with a fully cryogenic nir ( @xmath3 ) camera .
the camera design had been developed with high throughput as major goal .
the camera has a focal reducer scheme and is made of two detached doublets of silica and caf@xmath4 and further silica correction lens .
the pupil of the system is re formed between the doublets allowing to locate a lyot cold stop , imaging filters and grisms for low
less spectroscopy .
the camera has quasi diffraction
limited optical quality .
the dithering , needed for ir image processing is obtained via a wobbling wedge that shifts the image of about 20 pixels in any radial direction from the array center ( conconi et al .
2001 ) . the camera will be equipped with a 512x512 rockwell hawaii lpe hgcdte chip ( 18@xmath5 m pitch ) with a peak efficiency of 64% and values never lower than 56% between 1 and 2.5@xmath5 m .
the camera ( telescope and filter excluded ) is expected to have a transmission of @xmath6% . with a scale of 1.16arcsec / pixel the camera covers a fov of 9.8x9.8arcmin@xmath7 . according to simulations
such an high throughput allows to reach the limit magnitudes reported in table[tab : magn ] .
.limit magnitudes for different s / n and integration time as computed via the rem telescope simulator . [ cols="<,^,^,^,^,^,^,^,^ " , ] [ tab : magn ]
in its essence the rem telescope is a very efficient simultaneous zjhk photometric device . the plate scale chosen makes this telescope of limited use for extended objects or crowded fields but it certainly allow to collect photometry of point like source and is therefore ideal to observe stellar objects in and outside our galaxy . during the observation of a grb rem will acquire continuously frames measuring nir colors of all the stellar objects contained in such a frame . among this object , any variability with time scale smaller or comparable with the time in which rem will insist on the field will be detected .
this will allows to discard variable objects from the list of stars to be used for calibration but it will also allow to discover new variables . moreover also
the colors of the constant stars will be measured with very high precision .
it is however during the idle time when rem will be a unique opportunity for many programs of paramount interest for various branches of stellar astrophysics .
indeed the expected limiting magnitudes of the rem instrument will easily allow to measure a plethora of classes of known variable stars with previously un
conceivable efficiency and precision . in the following section we present a non - exhaustive description of some of the main fields in stellar astrophysics that have given proof in the recent past to gain a lot from good quality observations in the nir .
cepheids and rr lyrae pulsating variables are known to have paramount astrophysical importance related to their use as distance and age indicators at the cosmological level .
such an importance is of course enhanced if we observe at ir wavelengths since the interstellar absorption in the k - band is about ten times smaller than the one in the v - band .
a great gain is also attainable by observing cepheids and rr lyrae at nir wavelength due to the different view of the pulsation given at these and at optical wavelengths .
this because the light variation depends both on the radius variation and the surface brightness ( or effective temperature ) variation but with different weight depending on the wavelength .
another advantage in ir observation of cepheids regards the blanketing . in the 13@xmath5 m region , where the density of metallic absorption lines is low ,
this effect is much reduced the expected performances of the rem telescope allow the observation of the largest number of known cepheids and rrlyrae stars ever observed systematically in z , j , h and k so far .
very bright cepheids will be within the limiting magnitude of the rem instrument even in nearby galaxies . beside gaining a new insight in the pulsation mechanisms ongoing in these stars we
will then have the opportunity to put a further constraint to major distance and age calibrations via the application of independent methods based solely on ir measurements .
t tauri stars are low - mass pre zams stars that were first distinguished due to their photometric variations .
monitoring of the variability of t tauri stars at visible and nir wavelength suggested several causes for such variability derived basically from a picture of the circumstellar environment in which surrounding material in an infalling envelope ( e.g. calvet et al 1994 ) falls onto and dissipates angular momentum in a circumstellar disk .
the presence of magnetic fields produce cool sun like spots on the surface of the star but also channel the disk material on to the stellar surface producing hot isolated spots ( shu et al 1994 ) . in the above framework the variability of t
tauri stars has multiple causes and various characteristics .
the cool starspots modulate periodically the brightness of the star with amplitude slowly changing in time scales of months or years as the spots evolve . typical life
time of these spots is of the order of days or weeks .
on top of these some environmental characteristics such as the changes in extinction as infalling or orbiting material intersects the line of sight , can cause variability .
extinction variations can have amplitudes limited only by the optical depth of the interceding material .
the duration of the event depends on the size of the cloud and its velocity .
such clouds may either originate in the circumstellar environment or represent pristine infalling material .
since the typical amplitude of variation of these stars is between some tenth to some magnitudes rem represents a valid tool to monitor these stars in zjhk with good coverage and s / n ratio : most of the above described time - scales of variation are as well compatible with rem observing strategy .
in addition to cool spots , intense flares are the most remarkable manifestation of stellar magnetic activity .
flare events occur in the atmospheres of several type of stars , from pre ms to post
ms cool stars and involve the whole atmosphere , from the photosphere up to the corona . due to
the rather different physical characteristics of these different layers , with temperatures and densities spanning over several decades , flare flux affects a wide range of wavelengths , from microwaves to x rays , and possibly gamma
stellar flares are quite fast with typical time scales of flux increase to the maximum of the order of 10100s and decreases to the pre flare level on times 10100 longer ( see rodon 1990 ) : a striking case of a multi wavelength phenomenon that , owing to its fast development , requires really simultaneous observations . multi wavelength studies ( see e.g. , rodon et at .
1989 ) have shown that on the occasion of intense flares in the microwave , uv and optical wavebands , the ir flux decreases .
such negative flares are predicted by a non thermal model based on inverse compton effects by fast electron ( gurzadian 1980 ) and a thermal model based on the increased opacity of the @xmath8 ion ( grinin 1976 ) . the missing energy in the k band alone can account for the energy flux increase at all other wavelengths .
multi - wavelength monitoring of stellar flares are therefore relevant for the interpretation of magnetic phenomena
. the possibility of grbs associated with stellar flares has been predicted , among others , by becker and morrison ( 1974 ) , but no observations have been devoted to this aspect . outside of the time devoted to the principal research objective , namely the grb afterglow observations in ir , we intend to use the rem telescope to study negative flares in ir , as well as to ascertain whether any of the grb detected by swift is associated with a stellar flare rather than to an extragalactic phenomenon .
finally , huge flux excess in the far ir have been detected on quiescent dme active stars with iso ( rodon et al .
1999 , leto et al . 2001 ) and it will be interesting to relate those enhanced ir emissions with rem data .
cataclysimic variables ( cvs ) are semi
detached binary systems made up of a white dwarf ( wd ) primary and a main sequence ( ms ) secondary . material donated by the secondary star is funneled through the l1 lagrangian point .
the nir emission in cvs is produced by a variety of components : * 1 * ) the accretion disk in non
magnetic systems or the accretion stream in magnetic systems , * 2 * ) the accretion region near the surface of the wd , * 3 * ) the accretion disk hot spot and * 4 * ) the secondary star . therefore , even photometrically , the nir wavelengths offer a plethora of information about the component stars and the accretion process within cvs . indeed the classical approach to constraint sed ( spectral energy distribution ) models for cvs is to disentangle within time - resolved nir - curves each component of the emission . by browsing the catalogue of cataclysmic variable by ritter and kolb ( 1998 )
we find a consistent number of cvs with @xmath9 magnitude between 10 and 15 .
very little is known about the infrared colours of these stars .
however a number of these objects have been observed in these band providing evidence of @xmath10 band flux larger than @xmath9 band flux .
the typical amplitude of variations in the nir curves is of the order of @xmath11 , well within the expected capabilities of the rem telescope .
low mass x
ray binaries ( lmxrbs ) cointaining either a neutron star or a black hole , accrete matter from the late
type companion by means of an accretion disk .
lmxrbs can be divided into subclasses according to their location in the galaxy : bright galactic bulge sources located near the galactic center and the other lmxrbs in the galactic disk .
the study of the first subclass has been severely hampered by the heavy obscuration , which made even the recognistion of the optical counterpart difficult .
the dichotomy between these two lmxrb subclasses is poorly understood .
the ir band provides us with an ideal window for observing these systems .
rem observation can help studying the variability of these objects in connection with their x ray fluxes .
infrared monitoring of high mass x
ray binaries ( hmxrbs ) is especially important in the case of be companion stars .
these systems are usually transient : large x ray outbursts are associated to shell ejection episodes whereas smaller ones with passeages of the neutron star companion near periastron .
ir monitoring therefore provides a unique opportunity to test the environment in which the neutron star is accreting ( negueruela 1998 ) .
ir monitoring of these sources , which are usually bright and variable ( up to two magnitudes ) , in conjunction with x ray monitoring can shed light on the outburst onset mechanism ( e.g. propeller vs. direct accretion ) as well as lead to the prediction for the occurence of x ray outbursts which could then be monitored much more accurately , with pointing x ray instruments .
brecker k. , morrison p. 1974
, apj 187 , l97 calvet n. , hartmann l. , kenyon s.j . ,
whitney b.a .
1994 , apj 434 , 330 conconi p. , molinari e. , crimi g. 2001 , app .
optics , in press grinin v.p .
1976 , izv .
60 , 179 gurzadian g.a .
1977 , ap .
space sci .
52 , 51 leto g. , pagano i. , rodon m. , marino g. 2001 , csw 12 , asp conf .
( in press ) .
negueruela i. 1998 , a&a 338 , 505 rodon m. 1990 , iau symp .
102 , 371 rodon m. , houdebine e.r . , catalano s. , foing b. , butler c.j . , et al .
1989 , iau coll .
104 , catania ap .
. special publ . , 53 shu f.h . , najta j. , ostriker e. , et al .
1994 , apj 429 , 718 | the main goal of the rem project is the observation of prompt afterglow of gamma ray burst ( grb ) events .
such observations at near infrared ( nir ) wavelengths are even very promising , since they allow to monitor high z ly-@xmath0 absorbed bursts as well as events occurring in dusty star forming regions .
in addition to grb science , a large amount of time ( @xmath1 ) will be available for different scientific targets : among these the study of variability of stellar objects open exciting new perspectives . |
the tradition of using diagrammatic representations of formal structure in logical reasoning stretches back to times when sand or papyrus were the media for visualizations .
the tradition of using interactive software tools in logical reasoning is a recent refinement of an ancient interest .
this stretches back to the age of compact disk distributed software alongside paper printed books , with @xcite as a paradigmatic example , and plenty others , many of which have been presented at this venue @xcite . using didactic software online for logic education and e - learning is an equally respectable tradition going back to the age when java applets roamed on the web , with @xcite as a representative example , and plenty others , many of which have been also presented at this venue @xcite .
latest trends in this evolution are dissemination of logic courseware freely available online , with @xcite as a pertinent example , alongside didactic software tools built on a technology stack up to date with current web standards , with @xcite as a relevant illustration .
this is the general trend for many other fields across the curriculum @xcite .
when compared to what exists in other fields , many reasoning and modeling techniques emerging in current branches of logic remain in many ways still underrepresented , e.g. modal logics , interrogative logics , epistemic logics , etc
. however , consult @xcite as a short list of notable exceptions containing robust software tools for reasoning and modeling techniques emerging in nonclassical logics that have been recently integrated into this tradition , frequently accompanied by web workers or didactically tailored online demos . inside this general landscape
, the paper will revisit classical challenges for teaching logic and reasoning skills and implement possibly new software tools for interactive reasoning tasks driven by automated visualization of logical structure emerging in logics that are often underrepresented in or overlooked by existing approaches .
* overview * the paper is structured as follows : in the first section we introduce the topic and describe the main architecture of the program and some design options .
section [ sec : managing - repositories ] describes the use of the repositories of formal knowledge underlying the program .
section [ sec : interactive - visualizations ] describes the reasoning tasks and interactive visualization tools provided by the software .
sections [ sec : dydactic - touch ] and [ sec : classroom ] explore teaching use case scenarios and provide evidence for didactic relevance .
section [ sec : conclusion ] draws conclusions and hints towards possible directions for further work . * the overall architecture * the rest of this section contains information about implementation and development aspects .
readers only interested in using the program will probably want to skip to section [ sec : managing - repositories ] where the program s features are described . _
the backend repositories of formal knowledge _ the program consists of two layers : remote repositories of formal knowledge are stored and managed on the server side and the client side consists of the user interface to interact with and visualize the repositories formal content .
the backend application layer is a rest - like api built using nodejs and express to provide the usual crud operations on repository items .
the storage layer consists of several document structured databases built using mongodb and mongoose to define repository items @xcite .
the databases contain items from the standard domains in a logic curriculum : propositional logic , modal logic , predicate logic as well as quantified modal logic .
database entries are json documents with each item containing a basic ascii representation of its content , as received from an input device @xcite , together with a set of additional representation formats ( mathjax - latex , d3-svg , unicode , etc . ) used in visualizations , interactive reasoning tasks and natural language information .
the left and middle panels in figure [ figure : main+editor ] illustrate intuitively the list representation and the item features for the formulae library database . section [ sec : managing - repositories ] contains further information about how this data is managed via crud operations on the rest - like api . in total , the databases contain over five hundred items from four logic domains .
moreover , besides the predefined , ready to be used items , and perhaps more importantly , the backend application layer also serves as a tool for students or instructors to add items of their own interest to the database and use the reasoning and visualization tools in an approach to learning or teaching that fits their needs .
the right panel in figure [ figure : main+editor ] shows the formula editor gui , details on how the gui maps to crud actions are in section [ sec : managing - repositories ] .
the server contains mainly databases of formulae libraries and models corpora in the four logic domains mentioned and a repository of interactive reasoning tasks ( dialogic / semantic games , etc . ) using both formulae and models . _ the frontend user interface _ the client side layer of the program uses html5 and css3 for structure respectively presentation and javascript for handling application logic as well as for implementing logic specific functionality .
the framework used to build the frontend application layer is backbone enhanced with marionette modules , together with their standard dependencies jquery and underscore @xcite . for the graphical user interface we use bootstrap and for rendering formulae in a browser mathjax . for displaying logical structure we use svg and for interacting with reasoning tasks visualizations we use d3 @xcite , particularly tree and force graph layouts .
figure [ figure : main+editor ] presents the main user interface components and section [ sec : managing - repositories ] describes standard use scenarios for the gui .
figure [ figure : syntax+tableau+models ] illustrates some visualizations of logical structure and section [ sec : interactive - visualizations ] describes how these can be interactively generated , transformed and explored .
the logic specific application component consists of several modules containing functionality handling formal aspects concerning the syntax and semantics of the logics and dealing with their associated reasoning tasks .
these include a term unification package and an expression parser @xcite , a module for translating syntax into d3 tree layout , a tableau based theorem prover @xcite with a corresponding module to translate the analytic tableau proof trace into a d3 force graph layout with a hierarchical tree structure , and a ( counter)model generator @xcite with a corresponding module for translating the ( closed ) tableau branches into ( counter)example relational structures and further into a d3 force graph layout .
further information about implementation details and the technology stack used are included in the electronic version of this paper and in the documentation available on @xcite .
the program was developed and tested using primarily chromium in an ubuntu desktop environment and will run optimally in webkit - based browsers and using a desktop format factor .
a live demo of the program is openly available online at @xcite .
a first feature of the program consists in providing an interface for accessing and working with the following logic - specific repositories of formal knowledge : 1 . _ formulae libraries
_ provide an interface for accessing databases containing syntactic items from various logical languages .
currently , these include : propositional logic , modal logic , predicate logic , quantified modal logic .
models corpora _ provide an interface for accessing databases containing semantic items associated with various logics .
these are , depending on domain : truth assignments , relational structures , first - order models , qml models .
_ games repositories _ provide an interface for accessing databases containing interactive reasoning tasks .
these are not entries for separate entities but merely documents associating , formulae and/or models with logic modules . to facilitate access to repository items and reasoning tasks , the main interface layout has three component regions containing distinct information panels .
the first panel , in the left in figure [ figure : main+editor ] , is a collection view , its role is to visualize the repository content in an items list format . these can be either a formulae library or a corpus of models from those enumerated above .
the second panel , shown in the middle of figure [ figure : main+editor ] , is an item view , its role is to display item features .
these features are specific to each repository type and logic domain and link to various reasoning tasks , additional natural language information or further symbolic representations for a particular repository item selected from the items list panel .
the third panel , in the right of figure [ figure : main+editor ] , is also an item view .
it will display , depending on the chosen feature , the specific visualization content and/or an interface for interaction .
this can be , fore example , a svg canvas , a editor form , a table , etc .
the specific details of each interactive reasoning task are described in section [ sec : interactive - visualizations ] .
the subviews are interlinked and in a usual workflow will be navigated from left to right by first choosing an item of interest from a collection then a specific feature and finally working with its interactive visualization content in the working details panel .
the main navigation menu also provides links to instances of the three layout regions .
more detailed descriptions of possible workflows applied to concrete examples are presented in section [ sec : dydactic - touch ] . * working with the formula and model editors * the first level of feedback guided interaction with logical structure takes place starting from syntactic aspects and is mediated by the formula and model editors .
the right side panel in figure [ figure : main+editor ] illustrates the editors user interface .
this includes examples of well formed entities for each logic domain and detailed information about the formation rules .
feedback is provided as the user types and parser errors are displayed whenever an incorrect string is entered .
a modular parsing expression grammar is used for input validation .
this extends naturally from logical fragments to more expressive logics , from object language to the specification metalanguage , and from formulae to model syntaxes .
section [ sec : code ] contains an illustration of the the grammar modules .
a concrete editor use case example is given below in section [ sec : dydactic - touch ] .
further details are included in the electronic version of this paper and in the program documentation available on @xcite .
, width=397,height=190 ] * working with repository item features * a second level of feedback guided interaction with logical structure is mediated by reasoning tasks visualizations , accessed through specific item features .
the main feature categories are : 1 .
_ reasoning tasks _ is the main category , it contains various entries such as : visualizing syntactic construction trees , building semantic tableaus , generating countermodels and models , etc .
each task is described in detail in section [ sec : interactive - visualizations ] .
2 . _ item information _ contains various categories of information about the chosen items in natural language , such as name , description , sources , domain , etc .
3 . _ symbolic representation formats _ used for reference , linking , translation , etc . clicking on a formula feature will display its corresponding working details in the right hand side panel of the main layout presented in figure [ figure : main+editor ] .
while the first two panels are used for organization and navigation of formal repositories , the working details panel is used to host and manage most of the interactive aspects by coupling interface actions with a feedback loop allowing the user to generate and/or explore the formal structure underlying and propelling various logical reasoning tasks .
the main interactive feature of the program is visualization of logical structure .
this takes place for several kinds of analysis and reasoning tasks specific to each logic domain . in this section
we describe how to use the main reasoning tasks . ,
width=397,height=135 ] * syntax construction trees * a first analysis task is the interactive generation of syntactic construction trees .
this uses the syntactic formation rules of each logic domain to generate a tree layout of the formula and all its subformulae , all the way down to its atomic components .
the left side of figure [ figure : syntax+tableau+models ] illustrates a syntactic tree layout generated in this fashion for a first order logic formula .
the interface distinguishes between nodes containing fully extended components and the nodes containing further analyzable components .
the fully extended tree branches and the nodes containing atomic syntactic elements are marked with a blue color while the not yet fully analyzed nodes are colored in red , the user can further extend these nodes by clicking on their content . in the illustration
the second branch is not yet fully decomposed , this is marked in red and can be interactively extended further by clicking on the terminal subformula .
concrete teaching context are included in sections [ sec : dydactic - touch ] , [ sec : classroom ] .
* semantic tableaus * a second analysis task is the interactive generation of semantic tableau proofs .
this uses the semantic decomposition rules specific to each logic domain to generate a force tree layout by analyzing the head formula in a branch and so on in all resulting tableau branches , all the way down to exhausted branches .
the middle region of figure [ figure : syntax+tableau+models ] illustrates a semantic tableau generated in this way for an example formula from the modal logic domain .
the interface distinguishes between nodes representing fully expanded branches and the nodes standing for branches that can be further expanded i.e. contain formulae that can be further analyzed .
the layout also distinguishes between closed , i.e. containing contradictory formulae , and open branches , i.e. fully expanded but not containing a formula and its negation .
nodes are labeled by the head position formula in the branch , this is the formula logically analyzed and decomposed in that expansion step . the user can interactively display the full content of a tableau branch by hovering over any node of the tableau .
branches which are not fully expanded yet can be extended by clicking on a terminal node .
this triggers the process of logical decomposition on the head formula in the branch and adds the resulting branch(es ) to the tableau . clicking on an expanded node reverts the tableau expansion step and removes the outgoing branch(es ) . for modal logics in addition to decomposition rules for formulas in the logical object language
the tableau construction process also uses a specification metalanguage and corresponding decomposition rules for hybrid logic with nominals .
these decomposition rules are based on applying metanominals to possible worlds as skolem terms constructed from formulae containing modal operators and using relational properties of accessible successors during tableau expansion .
the theoretical background behind this technique has been described in full detail in @xcite , and the logical grammar used for parsing hybrid logic formulae in the specification metalanguage is also included for reference in section [ sec : code ] .
concrete teaching context are included in sections [ sec : dydactic - touch ] , [ sec : classroom ] . * countermodel generator * a third reasoning task consists in generating countermodels for a given formula .
this starts from the proof trace generated by a semantic tableau and from each open branch in the terminal nodes of the tree layout extracts a corresponding model .
hybrid logic formulae with nominals are used for expressing relational components of a model inside the specification metalanguage .
these are then translated in a force graph layout representing a relational structure which is displayed for visualization and can be further interactively inspected .
the right hand side of figure [ figure : syntax+tableau+models ] presents two graphs generated in this fashion .
the interface distinguishes counterexamples falsifying the formula , which are colored in red , from models satisfying the formula , which are colored in green .
the satisfying models are obtained in a similar way starting from a complementary formula .
each node of the graph is labeled with a relevant extensional component , for modal logic this is the world s propositional valuation .
the user ca obtain further information about the graph by hovering over the nodes .
this will reveal the metanominal corresponding to each possible world .
this nominal is a skolem term generated during tableau construction and contains further data about the stage in the proof trace it was introduced and the formula and other metaterms it was generated from .
concrete teaching contexts are included in sections [ sec : dydactic - touch ] , [ sec : classroom ] .
several additional reasoning tasks are described and/or exemplified in sections [ sec : dydactic - touch ] and [ sec : classroom ] , in the extended version of this paper and the documentation available on @xcite .
these reasoning tasks are either specific to a particular logic domain , such as truth tables for propositional logic , or only relevant for corpora of models , such as model inspection etc .
in this section we will present a series of concrete use case scenarios : in order to illustrate the way in which the program can be useful for didactic purposes .
* use case scenario 1 : exploring formal knowledge repositories * this use case will step through a typical repository navigation .
step ( 1 ) : from the main navigation menu select formulae libraries @xmath0 modal logic , alternatively , you can click on the modal link from the go to a formula library green div in the items list panel in the left of the main layout .
the left side panel will display the content of the modal logic library .
step ( 2 ) : search in the list the formula named distribution axiom ( k ) , alternatively , you can hit ctrl+f and type the formula s ascii content ( # ( p > q)>(#p>#q ) ) or its name , as a shortcut to finding the formula faster ( using native browser search ) .
when you click on the formula , the middle panel will display the formula s features .
step ( 3 ) : inspect the list of available features for the formula listed in the middle panel and find either the name or the domain item .
when you click on the chosen feature the right hand side panel will display additional information about the item of interest . in this case
this will be additional information in natural language about the formula .
item features related with reasoning tasks are explored in the next use case .
* use case scenario 2 : interactively visualizing reasoning tasks * this use case will present step by step a possible scenario of exploring reasoning tasks . repeat steps ( 1 ) to ( 3 ) from the previous use case , but this time use ( # ( p|q)>(#p|#q ) ) to find formula named distribution axiom ( f ) , note the ( f ) at the end which distinguishes it from the previous example .
step ( 4 ) : click on the syntactic tree feature and explore the construction tree by clicking on the nodes of the tree .
note that the syntactic construction of the formula is the same with the one displayed in the previous example ( both formulae generate the same tree structure ) .
step ( 5 ) : click on the semantic tableau feature and fully expand all the tableau branches by clicking on the nodes of the tree layout . by hovering over the nodes in the tableau
you can see the entire structure of the branch containing all the extended formulae up to that point .
continue until all the formulae are logically decomposed and every branch of the tableau is exhausted .
note that the semantic tableau constructed in this case is different from the one for the previous formula .
find the green colored open branch that makes the difference and inspect its content , try to imagine a counterexample starting from it .
step ( 6 ) : click on the model generator feature and scroll the panel of models generated in the working details panel until you find one that is colored in red .
the nodes in the graph represent possible worlds and are labeled by their propositional valuation . hovering over possible worlds in the models
will reveal their construction nominal , compare it with the open branch found in step ( 5 ) .
* use case scenario 3 : update formal knowledge repositories * in the previous use case scenario you have found a counterexample to a formula labeled as an axiom , now you are going to fix this .
if needed , go through steps ( 1 - 6 ) in the previous use case to obtain the open tableau and the countermodel illustrated in figure [ figure : syntax+tableau+models ] ( middle , respectively , right ) .
step ( 7 ) : click on the copy to sandbox library feature .
the sandbox library repository will be displayed in the items list panel .
find the formula and click the edit formula information feature .
the rightmost panel will display a form with editable entries .
step ( 8) : change formula s name from distribution axiom ( f ) to distribution formula ( fails for disjunction ) .
optionally , add a description and sources , then click the submit changes button .
step ( 9 ) : find the modified formula in the sandbox library , click the edit ascii content feature .
the formula editor will open with the formula s ascii content available for editing .
step ( 10 ) : change all disjunctions in the formula to conjunctions , as indicated in the available instructions panels this is done by replacing all occurrences of | with & .
feedback will be provided as you type , when the new formula is parsed correctly , click on the submit formula button .
eventually , repeat step ( 8) : renaming the formula distribution over conjunction and adding the description `` this fails for disjunction '' .
* use case scenario 4 : semantic tableaus expansion heuristics * step ( 1 ) : use the semantic tableau feature to find a counterexample model falsifying the modal formula @xmath1 , theorem ( s5 ) 3 in the modal library , using no more than seven expansion steps .
step ( 2 ) : can you do this with less than seven clicks ?
if yes , how many ? if no , what syntactic property of the formula can be used to prove this for the general case .
several other use cases are possible including features such as : interactive visualization and inspection of models , linking formulae with structures through model checking tasks , playing dialogic or semantic games , etc .
more use cases are described and explored in the electronic version of this paper and in the program documentation available on @xcite .
in this section we describe several possible scenarios of using the program in a classroom setting and the didactic experience that has been accumulated .
the program can be used both during live classroom activities and also remotely by students .
the live teaching classroom settings can integrate the tool alongside traditional teaching methods also during theory exposition but mostly for solving exercises .
the following list describes typical exercise templates ( concrete examples can be randomly selected from repositories ) : exercise model one : : given : a list of several formulae in symbolic notation ( mathjax - latex ) and a matching list of formulae represented as syntax construction trees ( d3-svg ) . required : establish the correct correspondence between the items in the two given lists .
exercise model two : : given : a list of structures in extensional notation ( mathjax - latex ) and a matching list of models represented as labeled graphs ( d3-svg force layout ) . required : establish the correct correspondence between the items in the two given lists .
exercise model three : : given : syntax construction trees of several formulae with matching structure graphs ( pointed in the case of modal logics , i.e. with designated actual worlds ) .
required : decide for each pair if the ( pointed ) model satisfies or not the given formula .
exercise model four : : given : a list of formulae in symbolic notation ( mathjax ) and a matching list of ( pointed ) models in extensional representation ( not as d3-svg graphs ) . required : for each formula in the list select the ( points)/structures that satisfy / falsify it .
exercise model five : : given : a list of ( pointed ) models represented in graphical format ( d3 ) . required : for each structure in the list upload into the formulae library m formulae that are true / false in the model ( at the actual world ) and having ( modal ) depth at least n. exercise model six : : given : a list of ( pointed ) models represented in extensional format .
required : for each structure in the list upload into the formulae library n formulae that are false / true in the model ( at the actual world ) and having ( modal ) depth at least m. exercise model seven : : given : a list of formulae represented as syntactic trees ( d3-svg ) . required : for each formula either upload into the corresponding model corpus a counterexample structure with at least n possible worlds(/objects ) or claim validity otherwise .
exercise model eight : : given : a list of formulae represented symbolically ( mathjax - latex ) . required : for each formula either find and write down in extensional form a counterexample structure with at least m possible worlds(/objects ) or claim validity otherwise .
the previous exercise models can be used in classroom activities , progress evaluations and final exams . to asses the didactic relevance of the program a nonequivalent groups design can be used to measure the average grades in student groups that used the program or not .
another measure can be the response time for accurate answers .
early adopters have ( only recently ) started to use the program in teaching activities and relevant data can be obtained by logging user interaction based on the previous list of exercise templates . as the design of the exercises suggests , the main testing parameter is the usefulness of interactively visualizing logical structure in reasoning tasks .
preliminary results @xcite have been used so far only for internal feedback loop .
these show that the test group significantly outperforms the control group for exercise model one , which is as expected . for exercise model three the grades average is also better in the test group , which is also not surprising .
the really interesting result is that the test group outperforms the control group also for exercise model four ( no visualizations ) .
moreover , when comparing averages for exercise model three with those for exercise models four , inside the same group , the former significantly outperform the later .
this is even more relevant as this outcome holds not only for the test group , but also for the control group .
all these suggest that interactive visualizations of logical structure in reasoning tasks can be didactically relevant . at the moment ,
not enough data is available for the remaining exercise models in order to warrant conclusion validity as a rigorous statistical experiment ( more results will become available online as more data is collected ) .
a final didactic experience that deserves to be mentioned is that the tool facilitates types of exercise that might be impractical ( if not impossible ) to stage using traditional teaching media ( like pen and paper or even chalk and a wide blackboard ) .
consider an example : example 1 : : given : rule of consensus , exception1 from the propositional logic formulae library . required :
find a counterexample structure using the semantic tableau method .
the paper presented a software tool for analysis and interactive engagement in various logical reasoning tasks .
the main feature of the program is to provide the means to intuitively visualize and interactively generate the logical structure that propels customary logical reasoning tasks .
a second feature consists in providing an interface for working with logic - specific repositories of formal knowledge .
starting from this we showed how both aspects have didactic potential and can be integrated in teaching activities to provide an engaging learning experience .
overall , this creates an ecosystem of interactive software tools modeling reasoning tasks driven by automated visualization of logical structure emerging in logics that are often underrepresented in or overlooked by existing approaches .
we also illustrated how this approach can be useful for didactic purposes .
besides presenting a possibly new logical reasoning software suite , the paper also revisited classical challenges for teaching logic and reasoning skills using a completely open source software stack for development , testing , deployment and hosting .
the resulting program is openly available online @xcite , it is implemented completely platform independent and can be run in any modern browser .
topics of further work include adding relational theories to the modal metatheory and customizing the modal tableau prover .
extending the modeled reasoning tasks to currently uncovered logic domains . adding additional layers of interactivity to the interface and a customizable strategy for tableau expansion . adding dynamic logic elements , etc .
.... propsym " proposition " = chars:([p - s0 - 9_\-]+ ) { return chars.join ( " " ) } nominal " nominal " = chars:([i - k0 - 9_\-]+ ) { return chars.join ( " " ) } metanominal " metanominal " = chars:([lt - w0 - 9_\-]+ ) { return chars.join ( " " ) } predicate " predicate " = chars:([p - sa - emn0 - 9_\-]+ ) { return chars.join ( " " ) } function " function " = chars:([f - h0 - 9_\-]+ ) { return chars.join ( " " ) } variable " variable " = chars:([x - zx - z0 - 9_\-]+ ) { return chars.join ( " " ) } constant " constant " = chars:([a - emn0 - 9_\-]+ ) { return chars.join ( " " ) } .... .... start = metaformula / formula formula " formula " = atomic / unary / binary / quanty / metaterm2form metaformula " metaformula " = formula " @ " metaterm quanty " quantified " = " ( " quantifier variable formula ") " quantifier " quantifier " = " $ " / " ! " binary " binary " = " ( " formula " & " formula ") " / " ( " formula " | " formula ") " / " ( " formula " > " formula
") " / " ( " formula " ^ " formula ") " unary " unary " = " ~ " formula / " # " formula / " * " formula atomic " atomic " = nominal / propsym / equality / predicate " ( " term ") " / predicate " ( " term " , " term ") " equality " equality " = " ( " term " = " term ") " term " term " = constant / variable / function " ( " term ") " / function " ( " term " , " term ") " metaterm " metaterm " = metanominal / metafunction " ( " metaterm " , " formula ") " metaterm2form " metaterm2form " = " { " metaterm // reductionaxiom = " < " atomic metafunction " metafunction " = " f * " / " f~ # " .... .... start = model model " model " = left : world " ; " right : model / world world " world " = metanominal " : " extensionnotationslist extensionnotationslist " extensionnotationslist " = left : extensionnotation right : extensionnotationslist / extensionnotation / " " extensionnotation " extensionnotation " = valsymb valuation / relsymb relations / domainsymb domain / monadicsymb monadicext / diadicsymb diadicext extsymb " extsymb " = valsymb / domainsymb / monadicsymb / diadicsymb valsymb " valsymb " = " vl " valuation " valuation " = left : literal " , " right : valuation / literal / " " relsymb " relsymb " = " rl " relations " relations " = left : metanominal " , " right : relations / metanominal / " " literal " literal " = atomic / " ~ " atomic : atomic { return " ~ " + atomic ; } atomic " atomic " = nominal / propsym / equality / predicate " ( " groundterm ") " / predicate " ( " groundterm " , " groundterm ") " domainsymb " domainsymb " = " do " domain " domain " = left : groundterm " , " right : domain / groundterm / " " monadicsymb " monadicsymb " = " mp " monadicext " monadicext " = predicate " { " objectslist " } " monadicextslist " monadicextslist " = left : monadicext " , " right : monadicextslist / monadicext / " " objectslist " objectslist " = left : groundterm " , " right : objectslist / groundterm / " " diadicsymb " diadicsymb " = " dp " diadicext " diadicext " = predicate " { " pairslist " } " diadicextslist " diadicextslist " = left : diadicext " , " right : diadicextslist / diadicext / " " pairslist " pairslist " = left : tuple2 " , " right : pairslist / tuple2 / " " tuple2 " tuple2 " = " ( " groundterm " , " groundterm ") " groundterm " groundterm " = constant / function " ( " groundterm ") " / function " ( " groundterm " , " groundterm ") " .... 50 barwise , jon and etchemendy , john and allwein , gerard and barker - plummer , dave and liu , albert : language , proof and logic , csli publications , stanford , usa , 2000 .
sieg , wilfried : the apros project : strategic thinking and computational logic , logic journal of igpl , vol . 5 , no .
4 , pages : 359 - 368 , 2007 .
van ditmarsch , hans and manzano , mara : proceedings of sicttl06 , tools for teaching logic , salamanca , spain , logic journal of igpl , vol . 5 , no . 4 , 2007 .
blackburn , patrick and van ditmarsch , hans and soler - toscano , fernando and manzano , mara : proceedings of ticttl11 , tools for teaching logic , springer - verlag , 2011 van benthem , johan and van ditmarsch , hans and van eijck , jan and jaspars , jan : logic in action , open logic courseware project , 2009 - 2014 .
minic , tefan and khodadadi , mohammad and schmidt , renate a. and tishkovsky , dmitry : synthesizing and implementing tableau calculi for interrogative epistemic logics , in fontaine , p. , schmidt , r.a . , schulz , s. ( eds . ) , paar , the third workshop on practical aspects of automated reasoning , pp . 109 - 123 , manchester , united kingdom , 2012 .
minic , tefan : dynamic - epistemic logic of questions in inquiry , ph.d .
thesis , institute of logic , language and computation , university of amsterdam , the netherlands , 2011 .
jquery , available online from : http://jquery.com/ , underscore , available online from : http://underscorejs.org/ , backbone , available online from : http://backbonejs.org/ , marionette , available online from : http://marionettejs.com/. mathjax , from : http://www.mathjax.org/ , bootstrap , at : http://getbootstrap.com/ , scalable vector graphics , available online from : http://www.w3.org/graphics/svg/ , data driven documents , available online from : http://d3js.org/ nodejs , available online from : http://nodejs.org/ , expressjs , available online from : http://expressjs.com/ , mongodb , available online from : http://www.mongodb.org/ , benm , available online from : https://github.com/jkat98/benm .
json : javascript object notation : http://www.json.org/ , ascii : american standard code for information interchange : http://www.ascii-code.com/ , crud : persistent storage functions : http://en.wikipedia.org/wiki/create,_read,_update_and_delete .
peg.js , parser generator for javascript , available online from : http://pegjs.org/ , unify.js , package available online from npm : https://www.npmjs.com/package/unify khan academy , available online from : https://www.khanacademy.org/a/0z1s modal logic playground , available online at : http://rkirsling.github.io/modallogic/ mettel2 , metric tesselation logic , a generic tableau prover , demo available online from : http://www.mettel-prover.org./demo.php demo , a demo of epistemic modelling , code and examples available online from : http://homepages.cwi.nl/~jve/demo/ lotrec , logical tableaux research engine companion , demo available online from : http://www.irit.fr/activites/lilac/lotrec/webstart twb , the tableau workbench , a generic tableau framework for building tableau / sequent - based theorem provers for non - classical propositional logics , demo available online from : http://twb.rsise.anu.edu.au/demolist intohylo : inference tools for hybrid logics , source code and demos available online from : http://www.glyc.dc.uba.ar/intohylo/index.php mleantap : a modal theorem prover , at : http://www.leancop.de/mleantap/ , a software tool for interactively visualizing logical structure in reasoning tasks , available online from : http://raeson.mybluemix.net | the paper presents a software tool for analysis and interactive engagement in various logical reasoning tasks . a first feature of the program consists in providing an interface for working with logic - specific repositories of formal knowledge .
a second feature provides the means to intuitively visualize and interactively generate the underlying logical structure that propels customary logical reasoning tasks .
starting from this we argue that both aspects have didactic potential and can be integrated in teaching activities to provide an engaging learning experience . |
the promising optoelectronic properties of spontaneously formed gan nanowires ( nws ) reported in the pioneering works of @xcite and @xcite have triggered world - wide research activities that have led to the demonstration of light - emitting @xcite and light - harvesting devices @xcite based on group - iii - nitride nws . despite this progress ,
several open questions still exist regarding the spontaneous formation of gan nws and their structural and optical properties .
in particular , a prominent band at 3.45 ev has been widely reported in the low - temperature photoluminescence ( pl ) spectra of gan nws grown by plasma - assisted molecular beam epitaxy ( pambe ) on si(111 ) @xcite .
the origin of this band in gan nws has been a subject of a lively debate for almost two decades . in bulk
gan , two different recombination mechanisms are known to manifest themselves by luminescence lines at about 3.45 ev : first , the two - electron satellite ( tes ) of the donor - bound exciton transition [ @xmath0 @xcite and , second , excitons bound to inversion domain boundaries @xcite . the intensity of the tes transitions in bulk gan is about two orders of magnitude lower than that of the related @xmath1line @xcite .
in contrast , the 3.45-ev band in gan nws is often prominent and sometimes even dominates the near band - edge pl spectrum .
nevertheless , this band was ascribed to the tes by @xcite , who proposed that the distortion of the @xmath1wave function near the nw surface would lead to a strong enhancement of this transition .
the same group substantiated this hypothesis by investigating the evolution of the 3.45-ev band with nw diameter and nw density @xcite .
however , investigations of the fermi level pinning in gan nws @xcite as well as polarization - resolved pl and magneto - optical experiments @xcite later refuted the interpretation of the 3.45-ev band as an enhanced tes transition .
inversion domain boundaries ( idbs ) may give rise to intense pl lines in gan films at 3.45 ev @xcite .
an idb denotes a boundary between ga- and n - polar gan for which two different stacking sequences have been proposed by @xcite .
the idb@xmath2notation refers to the specific atomic structure at which each atom remains fourfold coordinated by exclusively forming ga - n bonds across the boundary .
in contrast , the unstarred idb indicates a structure where the formation of ga - ga or n - n bonds would occur .
the idb@xmath2has an exceptionally low formation energy and does not induce electronic states in the band gap , thus facilitating the radiative recombination of excitons bound to these defects @xcite ( again in contrast to the unstarred idb , which entails electronic states in the band gap ) .
@xcite suggested that the 3.45-ev band observed in pl spectra of gan nws is related to the presence of idb@xmath2sin gan nws just as in the bulk .
this suggestion , however , was not supported by investigations of the microstructure of gan nws at that time .
in fact , transmission electron microscopy ( tem ) performed on isolated gan nws invariably demonstrated the absence of extended defects in the nw volume @xcite .
several groups therefore favored point defects as the origin for the 3.45-ev emission .
in early work , in which gan nws were assumed to elongate axially via a ga - induced vapor - liquid - solid mechanism , ga interstitials were proposed as likely candidates for these point defects @xcite .
later , gan nw growth was understood to proceed under n - rich conditions , which were suggested to result in an enhanced formation of ga vacancies near the nw surface @xcite .
@xcite discussed the dependence of the 3.45-ev band on excitation density in terms of both planar defects such as the idb@xmath2and abundant point defects and eventually favored the latter for being most consistent with the whole set of available data .
however , the observation of strong nw - to - nw variations in the intensity of the 3.45-ev emission from single nws @xcite contradicted this conclusion as well .
recently , @xcite presented definitive experimental evidence for the presence of idb@xmath2sin gan nws grown on aln - buffered si(111 ) .
subsequently , the same authors correlated -pl experiments with high - resolution scanning tem performed on single gan nws grown on aln - buffered si @xcite .
they observed a systematic correlation between the presence of idb@xmath2sin the gan nw and transitions at 3.45 ev in its -pl spectrum and thus concluded that these transitions are caused by exciton recombination at idb@xmath2s .
finally , they proposed that the 3.45-ev band is indicative for the presence of idb@xmath2sin gan nws also for other substrates .
it is not uncommon to observe mixed polarities in aln films grown on si(111 ) @xcite , and the coexistence of n- and ga - polar nws on such a film is therefore not an actual surprise . the spontaneous formation of gan nws directly on si(111 ) , however , is largely believed to occur on an amorphous sin@xmath3 interlayer formed during the ( unintentional or intentionally promoted ) nitridation of the si substrate by the n plasma @xcite .
the polarity of gan nws formed directly on such a nitridated si(111 ) surface has been debated for a long time . in earlier work ,
the gan nws were mostly reported to be ga polar @xcite , while recent studies ( which also include ensemble investigations with far better statitics ) indicate that gan nws grow predominantly or even exclusively n polar @xcite . in any case , whether idb@xmath2sform for gan nws on nitridated si(111 ) , and if so , at which density , are open questions . in this paper , we report a comprehensive investigation of the structural and optical properties of gan nws on si(111 ) fabricated with or without intentional substrate nitridation and within a wide range of substrate temperatures .
this investigation focuses on the nature of the 3.45-ev band that is observed for all samples but with varying intensity . in sec .
[ sec2:experimental ] , we briefly describe the samples under investigation and the experimental setups used for our study .
low - temperature cathodoluminescence ( cl ) spectroscopy and tem are employed in sec .
[ sec3:cl ] to investigate the origin of the 3.45-ev band . in accordance with the findings of @xcite
, we attribute this band to the presence of idb@xmath2sin our nws . in sec .
[ sec4:pl ] , we present an in - depth investigation of the 3.45-ev band comprising time - resolved and polarization - resolved pl spectroscopy accompanied by theoretical considerations based on data published by @xcite . the observed doublet structure of the 3.45-ev band is identified to be due to the recombination of localized and delocalized states at the idb@xmath2 . the paper closes with a summary and conclusion in sec .
[ sec5:summary ] .
the gan nw ensembles studied in this work ( see tab .
[ tab : samples ] for an overview ) were grown by pambe on aln - buffered sic(0001 ) ( sample a ) or si(111 ) ( samples b1b7 , c , and d ) .
they were fabricated over the course of six years and in four different pambe systems ( as indicated by the letters in their names ) .
the si(111 ) substrates were intentionally nitridated prior to nw growth for all samples except for sample d. we have intentionally chosen nw ensembles synthesized over a wide range of substrate temperatures @xmath4between 720 and 875 @xmath5c .
all samples were grown under n - rich conditions with the ga and n fluxes adjusted accordingly to account for the increased ga desorption at elevated temperatures @xcite . due to the wide range of conditions ,
the different nw samples exhibited incubation times between a few minutes and hours as well as different nw densities , average nw diameters , and coalescence degrees .
further details on the growth conditions can be found elsewhere ( sample a @xcite , b1b4 @xcite , b5b7 @xcite , c @xcite , and d @xcite ) .
.list of the investigated samples .
the substrate temperature @xmath4during growth , the year of fabrication , and the reference containing growth details are also given . [ cols="^,^,^,^,^ " , ] the morphological and structural properties of the nws were investigated by scanning electron microscopy ( sem ) and tem . for tem ,
cross - sectional specimens were prepared by mechanical grinding , polishing , and subsequent ar - ion polishing .
the tem images were recorded using an acceleration voltage of 300 kv .
the polarity of the nws was determined using convergent beam electron diffraction ( cbed ) complemented by subsequent dark - field tem imaging .
sem and cl spectroscopy were carried out in a field - emission instrument equipped with a cl system and a he - cooling stage .
a photomultiplier tube was used for the acquisition of monochromatic images and a charge - coupled device ( ccd ) camera for recording cl spectra . throughout the experiments ,
the acceleration voltage and the probe current of the electron beam were set to 5 kv and 0.75 na , respectively .
the spectral resolution amounted to 8 mev . due to the strong quenching of the cl intensity of gan nws under the electron beam @xcite , the irradiation time was kept to a minimum . since the @xmath1transition is more prone to quenching than the 3.45-ev band @xcite
, the bichromatic cl images shown in this work consist of two superimposed monochromatic false - color images recorded first at 3.47 and then at 3.45 ev . control experiments performed in reverse order confirmed the validity of our results .
all pl experiments were performed in backscatter geometry with the samples mounted in liquid he - cooled cryostats offering continuous temperature control from 5 to 300 k. all measurements were conducted with nw ensembles except for the polarization - resolved pl experiments , where single nws had been dispersed onto bare si(111 ) wafers prior to the measurement . for continuous - wave pl spectroscopy , the 325-nm line of a hecd laser was used to excite the samples .
the laser was focused to a spot with a diameter of 1 m by a near - ultraviolet microscope objective with a numerical aperture of 0.65 .
the photoluminescence signal was collected by the same objective and dispersed by a spectrometer with an energy resolution of 0.25 to 1 mev .
the dispersed signal was detected by a ccd camera .
polarization - resolved measurements were carried out using a half - wave plate followed by a linear polarizer @xcite .
time - resolved pl measurements were performed by focusing the second harmonic ( 325 nm ) of an optical parametric oscillator pumped by a femtosecond ti : sapphire laser ( pulse width and repetition rate of 200 fs and 76 mhz , respectively ) .
the pl signal was dispersed by a monochromator and detected by a streak camera operating in synchroscan mode .
the energy and time resolutions are 2 mev and 20 ps , respectively .
transition at 3.469 ev , several very narrow lines around 3.45 ev are observed .
the high - resolution tem image in the inset shows the central part of a gan nw of sample a revealing a ga-/n - polar core / shell structure .
the arrows denote the idb@xmath2sbetween the ga - polar core and the n - polar shell . ] prior to a systematic investigation of our gan nw ensembles on si(111 ) , we examine sample a which we already know to contain a very high density of idbs or , most likely , idb@xmath2s .
this sample was part of our study devoted to the role of substrate polarity in the formation of gan nws @xcite .
for this study , we attempted to induce the formation of gan nws on substrates with well - defined polarity , namely , aln / sic@xmath7 and aln / sic@xmath8 . in the present work ,
we discuss further aspects of the sample synthesized at a substrate temperature of 825@xmath5con sic@xmath7 , which is here referred to as sample a. the polar nature of the sic substrate is known to determine the polarity of group - iii - nitride layers deposited on it @xcite , and the aln buffer was consequently found to be al polar . initiating the deposition of gan at conditions typical for the growth of nws resulted in a highly faceted ga - polar gan layer interspersed with sparse vertical nws [
sem images of sample a can be found in figs .
2(c ) and 2(d ) in ref .
we have found the majority of these nws to be n polar due to a si - induced polarity flip at the interface between aln and gan .
consequently , idb@xmath2sform upon coalescence between the ga - polar matrix and the n - polar nws @xcite , and the sample is thus characterized by an exceptionally high fraction of nws containing an idb@xmath2
. transition .
the substrate temperature of each sample is specified in the figure . ]
the pl spectrum of sample a is shown in fig .
[ fig01-pl - tem ] .
the @xmath1transition is observed at 3.469 ev , but the spectrum also exhibits several intense and narrow lines around 3.45 ev .
the slight redshift of the @xmath1transition with respect to the one usually observed for gan nws ( 3.471 ev ) as well as its comparatively large line width suggests that it mainly originates from the ga - polar gan layer .
the strong transitions observed around 3.45 ev are consistent with the high fraction of nws with an idb@xmath2found in our previous analysis [ see fig .
4(b ) in ref . for an example ] .
moreover , further investigations of sample a by tem conducted in the course of the present work revealed in addition the existence of ga-/n - polar core / shell nws analogous to those reported by @xcite ( see the high - resolution tem image displayed in the inset of fig . [ fig01-pl - tem ] ) .
all these findings are in agreement with those reported in ref
. and strongly suggest that the lines observed around 3.45 ev in fig .
[ fig01-pl - tem ] are in fact due to excitons bound to idb@xmath2s .
we will therefore label these transitions in all what follows as @xmath9 .
an interesting finding in this context is the fine structure of the 3.45-ev band visible in fig .
[ fig01-pl - tem ] .
the origin of the distinct narrow lines in the pl spectrum will be addressed in sec .
[ sec4:pl ] .
a prominent band at 3.45 ev is commonly also observed for gan nws grown on si(111 ) @xcite .
figure [ fig02-t - series - norm ] shows the evolution of the low - temperature ( 10 k ) pl spectrum from gan nw ensembles formed on si(111 ) ( samples b1b4 ) under identical conditions except for the substrate temperature .
the linewidth of the @xmath1transition decreases with @xmath4increasing from 720 to 820@xmath5c , indicating a progressive reduction of micro - strain induced by nw coalescence @xcite . in parallel
, the intensity of the 3.45-ev band with respect to that of the @xmath1line also decreases with increasing @xmath4 .
if this band is also related to the presence of idb@xmath2sin gan nws grown directly on si(111 ) , its evolution seems to suggest that nws fabricated at a higher @xmath4exhibit a lower density of idb@xmath2s . in the following ,
we investigate whether this decrease in the relative @xmath9intensity with increasing @xmath4is correlated with changes in the morphology of the nw ensemble .
furthermore , we examine the spatial distribution of the different spectral components for various nw ensembles . figures [ fig03-sem](a ) and [ fig03-sem](b ) display sem bird s eye view images of samples b4 and b1 , which were grown at @xmath10 and 720@xmath5c , respectively .
the morphology of these samples is representative for nw ensembles grown at high and low @xmath4 . for substrate temperatures higher than 750@xmath5c
, the nw ensemble usually exhibits a nw density of about @xmath11 @xmath12 together with a uniform height distribution as shown in fig .
[ fig03-sem](a ) @xcite .
in contrast , for temperatures significantly lower than this value , the ensemble morphology is characterized by a dense ( @xmath13 @xmath12 ) matrix of short and highly coalesced nws interspersed by long nws with a much lower density of @xmath14 @xmath12 [ cf .
[ fig03-sem](b ) ] .
top - view sem images ( not shown here ) indicate that most of the long nws are attached to short nws . to identify a possible correlation between the peculiar morphology of sample b1 and its intense 3.45-ev band , we record the spatial intensity distribution of the two distinct pl bands by cl spectroscopy .
figure [ fig04-cl](a ) shows the superposition of an sem image with a bichromatic cl map recorded at 3.47 and 3.45 ev from sample b1 at 10 k. clearly , the 3.45-ev band ( color coded in green ) originates almost exclusively from long nws , whereas the @xmath1line ( color - coded in magenta ) stems mostly from the top part of short nws .
figure [ fig04-cl](b ) shows the same measurement for sample c , which is another typical representative for a gan nw ensemble grown at comparatively low @xmath4 .
the spatial distribution of the emission at 3.47 and 3.45 ev is the same as for sample b1 , as indeed for all samples exhibiting the morphology characteristic for growth at low substrate temperatures . to clarify the reason for this spatial distribution , we have performed cbed as well as tem imaging on several nw samples grown at low @xmath4 . using cbed , it is straightforward to determine the polarity of the short nws constituting the columnar matrix in these nw ensembles , since their effective diameter is large due to the high degree of coalescence .
the polarity of this matrix was found to be exclusively n polar for both samples b1 and c ( not shown here ) .
for the long nws with diameters below 50 nm , the polarity can not be reliably determined by cbed . however , since we know the polarity of the columnar matrix , we can instead employ dark - field tem and exploit the fact that opposite polarities induce a contrast inversion in images recorded by this technique .
in addition , inverting the diffraction vector @xmath15 should also invert the contrast for both polarities .
representative dark - field micrographs are displayed in fig .
[ fig05-tem ] .
the opposite contrast between the long nw and its short neighbor in fig .
[ fig05-tem](a ) recorded with @xmath16 as well as the contrast inversion in fig .
[ fig05-tem](b ) recorded with @xmath17 demonstrate that the long nw is ga polar .
the situation is more complex in figs .
[ fig05-tem](c ) and [ fig05-tem](d ) . here , the shell of the long nw has the same polarity as the adjacent material , but it clearly has a ga - polar core .
these peculiar polarity core / shell nw structures seem to be identical to those observed in fig .
[ fig01-pl - tem ] and in ref . , suggesting that the mechanism giving rise to the formation of such structures is a general one and does not depend on the substrate .
the long nws are either ga polar [ figs .
[ fig05-tem](a ) and [ fig05-tem](b ) ] or have a core / shell structure with a ga - polar core surrounded by a n - polar shell [ figs .
[ fig05-tem](c ) and [ fig05-tem](d ) ] . in the former case , planar idb@xmath2sare formed at the coalescence boundaries between the long ga - polar nws and the n - polar columnar matrix . in the latter case , the ga-/n - polar core / shell structure results in the formation of an idb@xmath2tube .
idb@xmath2sthus exist either at the junctions between the long nws and the surrounding columnar matrix or directly within the long nws themselves .
it is thus very plausible that the 3.45-ev emission , which arises almost exclusively from the long nws [ see fig .
[ fig04-cl ] ] , originates from the @xmath9complex .
ga polar or [ ( c ) and ( d ) ] to exhibit an ga-/n - polar core / shell structure .
an inverted diffraction vector @xmath15 also inverts the contrast between the ga- and n - polar material . ]
we have shown so far that nw ensembles grown at a lower @xmath4exhibit both short and long nws and that the optical transition at 3.45 ev is due to exciton recombination at idb@xmath2sin long thin nws .
however , with increasing @xmath4 , the nw ensembles are getting more homogeneous in diameter and length [ fig .
[ fig03-sem](a ) ] , and the intensity of the transition at 3.45 ev strongly decreases in comparison to that of the @xmath1as depicted in fig .
[ fig02-t - series - norm ] .
figure [ fig06_pl - cl](a ) shows for sample b5 grown at @xmath18@xmath5cthat the intensity of the line at 3.45 ev for nw ensembles grown at high substrate temperatures can be two orders of magnitude smaller than that of the @xmath1line . in the bulk
, the tes of the @xmath1is also centered at 3.45 ev and its intensity is about two orders of magnitude smaller than that of the @xmath1transition @xcite .
consequently , the question arises whether the weak 3.45-ev band observed in fig .
[ fig06_pl - cl](a ) is related to idb@xmath2sat all .
figure [ fig06_pl - cl](b ) shows a bichromatic cl map of sample b5 and fig .
[ fig06_pl - cl](c ) depicts cl spectra taken at 10 k on individual nws . clearly , the @xmath1and the 3.45-ev emission lines do not coincide spatially , ruling out the standard two - electron satellites as a possible origin of the 3.45-ev band and suggesting instead that the 3.45-ev band also arises from the presence of idb@xmath2sin these high-@xmath4nw ensembles .
remarkably , the density of nws with dominant @xmath9transitions is comparable to that observed for low-@xmath4ensembles ( cf .
[ fig04-cl ] ) , demonstrating that the density of idb@xmath2sdoes not change significantly with substrate temperature .
this finding seems to contradict the evolution of the relative intensity of the @xmath9band with @xmath4as depicted in fig .
[ fig02-t - series - norm ] .
however , plotting the same data on an absolute intensity scale as done in fig .
[ fig07-t - series - abs ] reveals that the intensity of the @xmath9band is actually not significantly reduced with increasing @xmath4 .
the decrease of the relative intensity of the @xmath9band with increasing @xmath4is instead caused by the drastic increase in the @xmath1emission intensity , reflecting the reduced concentration of nonradiative point defects at high @xmath4@xcite .
we have established in the previous section that the 3.45-ev transition in gan nws on si(111 ) is related to idb@xmath2sregardless of the substrate temperature . in this section ,
we investigate the optical properties of excitons bound to idb@xmath2sin more detail .
in particular , we critically examine the consistence of our experimental results with the properties expected theoretically for this particular bound exciton state .
we focus on samples which exhibit a detailed fine structure in their pl spectra .
one basic property of the @xmath9transition can be deduced directly from the fact that the idb@xmath2is a planar defect , which laterally extends over several tens of nm and vertically spans the entire nw length . in analogy to the result in ref . for basal - plane stacking faults
, we hence expect the density of states of the @xmath9to be two - dimensional . as a result of the large number of states available , the @xmath9transition should be difficult to saturate even for high excitation conditions . indeed , the @xmath9transition has been observed to scale linearly with excitation density even after the higher - energy @xmath1transition has started to saturate @xcite .
using density - functional theory , @xcite calculated the electronic potential in the vicinity of an idb@xmath2based on the stacking sequence proposed by @xcite .
this potential , depicted in fig .
[ fig08_eff - mass ] , acts as a barrier for electrons and as a quantum well for holes : an idb@xmath2can therefore be seen as a type - ii quantum well . to get quantitative information on the properties of the electronic state associated with the idb@xmath2
, we calculated the wavefunction and the energy of an exciton in the potential profile obtained by @xcite using the variational approach described in ref . .
the result of our calculations is shown in fig .
[ fig08_eff - mass ] .
the calculations yield an electron - hole overlap , defined as the absolute square of the overlap integral between the electron ( @xmath20 ) and hole ( @xmath21 ) wavefunctions , of @xmath22 . therefore , despite the type - ii band alignment across the idb@xmath2plane , the @xmath9transition possesses a large oscillator strength in agreement with the high intensity observed experimentally [ cf . figs . [ fig04-cl ] and [ fig06_pl - cl](b ) ] . for isotropic electron and hole masses of @xmath23 and @xmath24 , with @xmath25 denoting the mass of the free electron , the energy of this transition @xmath26 is found to be 3.445 ev , i.e. , close to the experimental value .
@xcite also predicted a significant mixing between the @xmath27 and @xmath28 valence bands due to the idb@xmath2 . to verify this prediction ,
we have performed polarization - resolved pl experiments at 40 k on single nws from sample b7 dispersed on a si substrate . a typical polarization - resolved pl map taken on a single nw
is shown in fig .
[ fig09_polarization ] . in agreement with previous reports @xcite , the free a exciton ( @xmath29 ) and the @xmath1transitions are polarized perpendicular to the nw axis ( @xmath30 ) , whereas the @xmath9band is polarized parallel to the nw axis ( @xmath31 ) .
the three lowest optical transitions in strain - free bulk gan obey selection rules such that the @xmath32 transition is allowed only with in - plane ( @xmath30 ) polarization , the @xmath33 transition with both in- and out - of - plane ( @xmath31 ) polarization , and the @xmath34 transition mostly with out - of - plane polarization .
the out - of - plane polarization of the @xmath9band evident in fig . [ fig09_polarization ] thus seems to suggest that it originates from a pure @xmath34 transition , i.e. , the c exciton @xcite . for a nw with sub - wavelength diameter
, however , we have to bear in mind that the dielectric contrast strongly suppresses in - plane polarized emission and thereby artificially enhances the out - of - plane component @xcite .
the strong polarization of the @xmath9band along the nw axis ( @xmath31 ) thus provides clear evidence for the mixing of the @xmath27 and @xmath28 valence bands in the idb@xmath2 , in agreement with the theoretical result of fiorentini @xcite .
note also the pure in - plane polarization for the @xmath35transition , demonstrating that the opposite behavior observed for the @xmath9transition is a consequence of the peculiar potential induced by the idb@xmath2and not a characteristic of excitons bound to planar defects in general . as derived from effective mass calculations applied to a planar idb@xmath2structure using the potentials given in ref . .
for the electron , the dotted line reflects the self - consistent effective potential .
the inset shows an enlarged section of the conduction band edge around the idb@xmath2 , clearly showing the reduction of the conduction band edge in its immediate vicinity due to the accumulation of holes .
the values for the bandgap @xmath36 and the transition energy @xmath26 are also given . ]
axis of gan , is oriented along 90@xmath5as indicated by the dashed line . ]
c , respectively .
the spectral windows containing the @xmath9and the @xmath1have been normalized independently to the strongest transition . ]
the mixing between @xmath37 and @xmath28 states proposed in ref . and observed in fig .
[ fig09_polarization ] is also consistent with the magneto - optical behavior of the @xmath9transition reported in ref . .
the land factor @xmath38 was found to be close to zero for the @xmath9 , while for the @xmath1transition in gan nws it was observed to be 1.75 , a value similar to that reported for the bulk @xcite . as both
the electron and hole land factors are extremely sensitive to valence band mixing @xcite , it is in fact not surprising to measure very different values of @xmath38 for the @xmath1and the @xmath9lines .
all properties of the @xmath9transition discussed so far are consistently described by the electronic potential computed by @xcite . however
, this model does not provide any explanation for the fact that the @xmath9band often exhibits a doublet structure . as shown first by @xcite ,
the band at 3.45 ev actually consists of two lines centered at about 3.449 and 3.455 ev @xcite .
this finding is confirmed by the pl spectra of samples b6 , b7 , and d at 40 k as shown in fig .
[ fig10-doublets ] .
for these three samples , the lower and higher energy component of the doublet is centered at about 3.452 and 3.458 ev , respectively .
note that the splitting between these two lines is almost identical to that between the @xmath1and the @xmath29transitions .
figure [ fig11_t - p - dependencies ] shows the evolution of the @xmath9doublet with increasing excitation density and temperature for sample b6 and b7 , respectively . at low temperatures and excitation densities ,
the @xmath9doublet is dominated by the lower energy transition [ @xmath39 at 3.452 ev .
however , with increasing excitation density [ fig . [
fig11_t - p - dependencies](a ) ] or temperature [ fig .
[ fig11_t - p - dependencies](b ) ] , the higher energy transition [ @xmath40 takes over and eventually dominates the pl spectrum . at around 40 k ,
carriers start to escape from the idb@xmath2 , which manifests itself in a quenching of the @xmath9doublet @xcite . as noted in ref .
, the excitation power and temperature dependences of the @xmath9doublet are similar to the ones observed for the @xmath1and @xmath29transitions . in view of the fact that idb@xmath2sact as quantum wells
, we thus attribute the high - energy line of the @xmath9doublet to the recombination of excitons free to move along the idb@xmath2plane and the low - energy one to excitons localized within this plane .
localization within the plane of a quantum well usually occurs at well width fluctuations or due to alloy disorder @xcite , which clearly can not be the origin of the intra - idb@xmath2localization observed here .
following the results reported for the localization of excitons in i@xmath41 basal - plane stacking faults @xcite , we propose that the short - range potential of shallow donors such as si and o distributed in the vicinity of the idb@xmath2induce the localization of excitons within the idb@xmath2plane .
to confirm this idea , we compare the excitation dependence of the intensity ratio at 10 k between the @xmath42and @xmath43lines with that between the @xmath29and the @xmath1 in fig .
[ fig11_t - p - dependencies](c ) .
both ratios remain nearly constant for low excitation powers , but increase together for powers higher than 18 w. this finding indicates that the density of localized states within the idb@xmath2sis comparable to the equivalent density of donors in nws and thus confirms that intra - idb@xmath2exciton localization occurs due to donors . note that we computed the characteristic extent of the exciton wavefunction perpendicular to the idb@xmath2plane to be 5 nm . assuming that the diffusion length of excitons in an idb@xmath2is 100 nm @xcite , a donor density of 10@xmath44 @xmath45 is found to be sufficient to localize excitons within the idb@xmath2plane , a value close to those reported for unintentionally @xmath46-doped gan nws @xcite . in ensemble measurements
, the @xmath43band typically has a line width of several mev , whereas single nws exhibit numerous sharp lines in this region @xcite .
the small differences in transition energies originate from the varying distances between the involved donor and the idb@xmath2plane @xcite .
this effect can also be seen very clearly for sample a in fig .
[ fig01-pl - tem ] . due to the low nw density of @xmath47 @xmath12 of this sample , only a small number of nws
is probed simultaneously , and the individual narrow lines can be resolved even in an ensemble measurement . the line width of the sharp @xmath43lines in fig .
[ fig01-pl - tem ] and also in fig . 3 of ref .
is resolution limited , demonstrating that these lines stem from the radiative decay of bound excitons .
in contrast , ensemble spectra such as the ones shown in figs . [ fig10-doublets ] and [ fig11_t - p - dependencies ] contain contributions from about @xmath48 nws , and the individual transitions can no longer be resolved , but blend together to an @xmath43band with a line width of several mev @xcite . note that @xcite observed a fine structure of the @xmath9band in bulk gan already in 2001 .
the @xmath42band exhibits a linewidth of a few mev as expected for delocalized states . finally , we have performed time - resolved pl experiments to obtain quantitative information on the capture efficiency of excitons by idb@xmath2sand on the intra - idb@xmath2localization process .
figure [ fig12_trpl ] shows pl transients of sample b6 measured at 10 k for the @xmath29and @xmath1transitions as well as for the @xmath42and @xmath43lines .
while the initial increase of the @xmath29pl intensity is almost instantaneous , it takes 90 ps for the @xmath1pl intensity to reach a maximum .
the latter time corresponds roughly to the characteristic time for the capture of excitons by donors with a density of about @xmath49 @xmath45 @xcite . in agreement with the results reported in ref . , the @xmath29and @xmath1pl decay in parallel at longer times as a result of the quasi - thermalization between these exciton states . using an exponential fit , the effective decay time for the @xmath29and @xmath1is found to be 195 ps , very similar to values obtained in previous reports @xcite .
this decay is much faster than expected for the radiative decay of the @xmath1complex and is due to the nonradiative decay of the @xmath29at point defects @xcite .
analogously to the @xmath1and @xmath29transitions , the initial increase of the @xmath43intensity is delayed compared to that of the @xmath42 .
interestingly , the rise time of the @xmath42intensity is only 47 ps .
therefore , the capture of excitons by idb@xmath2sis more efficient than their trapping by neutral donors due to the idb@xmath2s large capture cross - section .
this situation differs from observations for the rise time of the @xmath35transition at 10 k @xcite , which is limited by the inefficient transport of excitons from one donor to the next @xcite . for time delays longer than 600 ps , the @xmath43and @xmath42decay in parallel , again demonstrating quasi - thermalization between these states .
finally , for time delays longer than 1 ns , the pl decay of the coupled @xmath29and @xmath1states slows down and asymptotically approaches that of the @xmath9doublet ( not shown ) , indicating full thermalization between the @xmath29 , @xmath1and @xmath9states .
this process is the origin of the biexponential pl decay reported for the @xmath1 in refs . . and @xmath29transitions as well as of the @xmath43and @xmath42lines of sample b6
the arrows denote the respective rise times , and the solid lines are exponential fits . ]
the 3.45-ev band observed in low - temperature pl and cl spectra of spontaneously formed gan nws on si(111 ) arises from planar or tubular idb@xmath2s . while the former are due to the coalescence of adjacent ga- and n - polar nws , the latter form at the interface of ga-/n - polar core / shell nws .
the intensity ratio between the @xmath9and the @xmath1transitions decreases with increasing @xmath4 .
this decrease is a consequence of the reduction in the density of nonradiative point defects with increasing @xmath4 , leading to an increase in the absolute intensity of the @xmath1line .
in contrast , the absolute intensity of the @xmath9band is neither directly governed by @xmath4nor by the presence or absence of an intentional nitridation step .
the same applies to the abundance of idb@xmath2sobserved in spatially resolved , bichromatic cl maps .
these results confirm the idea of @xcite that the 3.45-ev luminescence band in gan nws signifies the presence of idb@xmath2sregardless of the substrate .
in fact , we now know with certainty that idb@xmath2soccur in gan nw ensembles synthesized by pambe on aln - buffered si(111 ) @xcite , on nitridated si(111 ) , and on sic(0001 ) .
the exceptionally low formation energy of idb@xmath2sis obviously an important factor promoting their frequent occurrence .
however , a prerequisite for the formation of an idb@xmath2is the simultaneous presence of both ga- and n - polar material . at present , it remains entirely unclear why an apparently constant fraction of the gan nuclei on all of these different substrates are ga polar . we also do not understand how ga - polar nws can evolve from these nuclei despite our inability to synthesize them intentionally on cation - polar substrates @xcite and at substrate temperatures at which gan(0001 ) usually decomposes . finally , it is unclear how the peculiar ga-/n - polar core / shell nws form , which have been observed by different groups and on different substrates .
the lack of knowledge regarding these apparently universal and basic phenomena demonstrates that the nucleation and formation of gan nws in pambe are still far from being completely understood .
concerning the electronic and optical properties of idb@xmath2s , it is helpful to imagine them as a thin type - ii quantum well that binds holes . the coulomb attraction exerted by these holes is strong enough to bind electrons , and the resulting @xmath9state decays radiatively and thus gives rise to intense light emission . the change in the symmetry of the fundamental hole state in
idb@xmath2sstrongly modifies the polarization and the behavior of the @xmath9when subjected to magnetic fields .
donor atoms distributed in the vicinity of the idb@xmath2plane localize the exciton within the idb@xmath2plane . as a result of this localization
, the @xmath9band resolves into a doublet with the lines at low and high energy being associated with localized and free @xmath9states , respectively .
their excitonic nature is manifested by the observation of a fine structure of the low energy line consisting of sharp , resolution - limited peaks . finally , while the capture of excitons by idb@xmath2stakes less than 50 ps , quasi - thermalization between the near - band edge excitons and the @xmath9takes much longer , resulting in the nonexponential decay usually observed for the @xmath1transition in gan nws at low temperatures .
analogously to stacking faults @xcite , idb@xmath2sdo not suffer from fluctuations in layer thickness or composition as conventional quantum wells and thus offer unique possibilities for the study of low - dimensional excitons @xcite .
particularly interesting in this context are the tubular idb@xmath2sformed in ga-/n - polar core / shell nws and intersections of these tubular idb@xmath2swith stacking faults forming perfect crystal - phase quantum rings . since the electronic states associated with all of these quantum structures are shallow @xcite , they should be of little practical relevance for conventional devices
. however , we envisage that their exceptionally well - defined properties make them ideal model systems for an understanding of quantum effects important for a future generation of optoelectronic devices .
the authors thank pierre lefebvre for fruitful discussions , vincent consonni and caroline chze for providing additional samples , and uwe jahn for a critical reading of the manuscript .
c. acknowledges funding from the fonds national suisse de la recherche scientifique through project 161032 .
this work was partly supported by the german bmbf joint research project monalisa ( contract no .
01bl0810 ) , by the deutsche forschungsgemeinschaft within sfb 951 , and by marie curie rtn parsem ( grant no .
mrtn - ct-2004 - 005583 ) .
60ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1143/jjap.36.l459 [ * * , ( ) ] link:\doibase 10.1016/s0022 - 0248(97)00386 - 2 [ * * , ( ) ] link:\doibase 10.1143/jjap.43.l1524 [ * * , ( ) ] link:\doibase 10.1021/ja404043k [ * * , ( ) ] link:\doibase 10.7567/apex.8.042302 [ * * , ( ) ] link:\doibase 10.1103/physrevb.62.16826 [ * * , ( ) ] http://link.aip.org/link/?japiau/101/113506/1 [ * * , ( ) ] link:\doibase 10.1063/1.2980341 [ * * , ( ) ] link:\doibase 10.1063/1.3062742 [ * * , ( ) ] link:\doibase 10.1103/physrevb.81.045302 [ * * , ( ) ] link:\doibase 10.1063/1.4775492 [ * * , ( ) ] link:\doibase 10.1002/pssc.200674901 [ * * , ( ) ] link:\doibase 10.1063/1.1609632 [ * * , ( ) ] link:\doibase 10.1063/1.3556643 [ * * , ( ) ] link:\doibase 10.1103/physrevb.82.045320 [ * * , ( ) ] link:\doibase 10.1063/1.1390486 [ * * , ( ) ] link:\doibase 10.1063/1.3656987 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.77.103 [ * * , ( ) ] in http://scholar.google.com/scholar?hl=en{&}btng=search{&}q=intitle:tem+study+of+(+ga+,+al+)+n+nanocolumns+and+embedded+gan+nanodiscs{#}0[__ ] , , ( , ) p. link:\doibase 10.1007/s11664 - 006 - 0102 - 4 [ * * , ( ) ] link:\doibase 10.1063/1.2204836 [ * * , ( ) ] link:\doibase 10.1016/j.jcrysgro.2006.11.036 [ * * , ( ) ] link:\doibase 10.1007/s12274 - 010 - 0061 - 1 [ * * , ( ) ] link:\doibase 10.1063/1.4923024 [ * * , ( ) ] link:\doibase 10.1063/1.4927826 [ * * , ( ) ] link:\doibase 10.1063/1.3633522 [ * * , ( ) ] link:\doibase 10.1103/physrevb.83.035310 [ * * , ( ) ] link:\doibase 10.1063/1.2899944 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.2953087 [ * * , ( ) ] link:\doibase 10.1088/0957 - 4484/21/19/195202 [ * * , ( ) ] link:\doibase 10.1109/jstqe.2010.2098396 [ * * , ( ) ] http://prb.aps.org/abstract/prb/v84/i24/e245302 [ * * , ( ) ] link:\doibase 10.1021/nl203872q [ * * , ( ) ] link:\doibase 10.1021/nl302664q [ * * , ( ) ] link:\doibase 10.1021/nl400200 g [ * * , ( ) ] link:\doibase 10.1063/1.4905651 [ * * , ( ) ] link:\doibase 10.1063/1.1554776 [ * * , ( ) ] link:\doibase 10.1063/1.3267151 [ * * , ( ) ] link:\doibase 10.1016/j.jcrysgro.2010.12.081 [ * * , ( ) ] link:\doibase 10.1021/acs.cgd.5b00690 [ * * , ( ) ] link:\doibase 10.1063/1.3464956 [ * * , ( ) ] http://www.mrs.org/s{_}mrs/doc.asp?cid=26415{&}did=321215 [ * * , ( ) ] link:\doibase 10.1088/1367 - 2630/17/3/033040 [ * * , ( ) ] link:\doibase 10.1088/0957 - 4484/22/29/295714 [ * * , ( ) ] link:\doibase 10.1088/0957 - 4484/25/45/455702 [ * * , ( ) ] link:\doibase 10.1063/1.4822110 [ * * , ( ) ] link:\doibase 10.1063/1.4935522 [ * * , ( ) ] link:\doibase 10.1103/physrevb.93.115305 [ * * , ( ) ] link:\doibase 10.1063/1.4749789 [ * * , ( ) ] link:\doibase 10.1063/1.2794402 [ * * , ( ) ] link:\doibase 10.1103/physrev.114.90 [ * * , ( ) ] `` , '' in link:\doibase 10.1002/9781118984321.ch3 [ _ _ ] , ( , ) pp .
link:\doibase 10.1103/physrevb.90.195309 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/0038-1098(81)90401-4 [ * * , ( ) ] link:\doibase 10.1103/physrevb.80.153309 [ * * , ( ) ] link:\doibase 10.1103/physrevb.83.245326 [ * * , ( ) ] link:\doibase 10.1063/1.4868131 [ * * , ( ) ] link:\doibase 10.1103/physrevb.88.075312 [ * * , ( ) ] link:\doibase 10.1063/1.3142396 [ * * , ( ) ] link:\doibase 10.1103/physrevb.90.165304 [ * * , ( ) ] | we investigate the 3.45-ev luminescence band of spontaneously formed gan nanowires on si(111 ) by photoluminescence and cathodoluminescence spectroscopy .
this band is found to be particularly prominent for samples synthesized at comparatively low temperatures . at the same time , these samples exhibit a peculiar morphology , namely , isolated long nanowires intersecting a dense matrix of short ones .
cathodoluminescence intensity maps reveal the 3.45-ev band to originate primarily from the long nanowires .
transmission electron microscopy shows that these long nanowires are either ga polar and are joined by an inversion domain boundary with their short n - polar neighbors , or exhibit a ga - polar core surrounded by a n - polar shell with a tubular inversion domain boundary at the core / shell interface . for samples grown at high temperatures , which exhibit a uniform nanowire morphology ,
the 3.45-ev band is also found to originate from particular nanowires in the ensemble and thus presumably from inversion domain boundaries stemming from the coexistence of n- and ga - polar nanowires . for several of the investigated samples , the 3.45-ev band splits into a doublet .
we demonstrate that the higher - energy component of this doublet arises from the recombination of two - dimensional excitons free to move in the plane of the inversion domain boundary .
in contrast , the lower - energy component of the doublet originates from excitons localized in the plane of the inversion domain boundary . we propose that this in - plane localization is due to shallow donors in the vicinity of the inversion domain boundaries . |
the ultra - relativistic outflows that power gamma - ray bursts ( grbs ) are thought to be collimated into narrow jets ( for reviews see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the evidence for this is rather indirect , however , since their images are usually unresolved , and in the best case ( grb 030329 ) the late time radio afterglow image was only marginally resolved @xcite .
the different lines of evidence for jets in grbs include analogy to other astrophysical relativistic outflow sources such as active galactic nuclei or micro - quasars ( e.g. * ? ? ?
* ) , the difficulty in transferring enough energy to ultra - relativistic ejecta in a spherical explosion of a massive star ( for long duration grbs ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , extremely large isotropic equivalent energies in some grbs ( with @xmath3 in grb080916c * ? ? ?
* ) , and an achromatic steepening of the afterglow lightcurves of some grbs that is attributed to a jet ( known as a jet break ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
therefore , there is very little direct observational information about the jet angular structure and dynamics , which make it difficult to interpret grb afterglow observations and infer from them important physical parameters such as the jet energy and opening angle , the external density profile , and the microphysical parameters of the relativistic collisionless shock powering the afterglow emission .
most studies of grb jet dynamics during the afterglow phase have focused on a roughly uniform jet with well defined , sharp edges .
we shall also focus on such a uniform jet , and only briefly remark on the expected relation to jets with a smoother angular structure ( also known as `` structured jets '' ) .
the jet dynamics have been studied both analytically @xcite and numerically , using two dimensional special relativistic numerical simulations @xcite , as well as an intermediate approach where the dynamical equations are integrated over the radial profile of the thin shocked region , thus reducing the set of partial differential equations to one dimension @xcite .
let us consider a uniform double - sided jet of total energy @xmath4 , initial half - opening angle @xmath0 , and initial lorentz factor @xmath5 .
grb observations suggest that typically @xmath6 . at early times , as long as @xmath7 , the bulk of the jet is causally disconnected from its edge and thus evolves as if it where part of a spherical flow with an energy @xmath8 , following the spherical @xcite self - similar solution .
this early phase corresponds to radii @xmath9 , where the jet radius @xmath10 is defined as the radius where @xmath1 for a spherical flow with @xmath11 . at @xmath12
the bulk of the jet is in causal contact with its edge and the jet can in principal rapidly expand sideways .
however , the degree of lateral spreading at this stage , which strongly affects the dynamics , is not well known . therefore , the jet dynamics at @xmath12 are still controversial .
in particular , the radius @xmath13 at which the flow ( or jet ) becomes non - relativistic , still remains uncertain .
the sedov length for a spherical flow with the true jet energy , @xmath14 ( i.e. the radius where it sweeps up a rest mass energy equal to its own energy and becomes non - relativistic ) , @xmath15 , is very close to @xmath10 .
therefore , in order for the jet to be already close to spherical when it becomes non - relativistic ( i.e. at @xmath16 ) , it must expand sideways very quickly and become close to spherical already near @xmath10 [ i.e. @xmath16 can not be @xmath17 .
this is indeed roughly what happens in simple analytic models , where the jet half - opening angle , @xmath18 , starts growing exponentially with radius near @xmath10 , and the jet quickly becomes close to spherical and non - relativistic at a radius @xmath19 , which is larger than @xmath10 only by a logarithmic factor and @xmath20 likely arises from the fact that in simple analytic models the swept - up mass at the radius where the jet becomes spherical is smaller than the external rest mass within a sphere of the same radius .
] , while at @xmath21 the flow quickly approaches the newtonian , spherical , self - similar sedov - taylor solution .
numerical simulations , however , suggest that most of the energy remains within the initial jet half - opening angle @xmath0 until the flow becomes mildly relativistic , and only then does the flow start to gradually approach spherical symmetry @xcite . under the crude approximation that the jet does not expand sideways and
keeps evolving as a conical section of a spherical flow up until the radius where it becomes non - relativistic , the latter is given by @xmath22 .
in this case the flow is still highly non - spherical at @xmath16 , and only very gradually approaches spherical symmetry @xcite .
this clearly shows that without lateral expansion @xmath16 is significantly larger , by a factor of @xmath23 ( which is @xmath24 for @xmath25 ) , than if there is fast lateral expansion at @xmath12 .
thus , the dynamics of the flow at small radii ( @xmath26 ) and at large radii ( @xmath27 ) are reasonably well known , while at intermediate radii ( @xmath28 ) they are still controversial . for typical values of @xmath29
this range of radii may appear rather small , @xmath30 for exponential lateral expansion and @xmath31 for no lateral expansion up to @xmath32 with @xmath33 .
however , it corresponds to a large range in observed times ( over which the corresponding afterglow emission reaches us ) , of @xmath34 and @xmath35 , respectively , since the observed time scales as @xmath36 , and @xmath37 decreases by a large factor ( of @xmath38 ) within this range of radii .
most simulations so far were for @xmath39 , or even wider initial jet half - opening angles .
recently , however , @xcite and later @xcite have performed simulations also for narrower initial jets , @xmath40 .
@xcite have found that significant lateral spreading starts when @xmath37 drops below @xmath38 , as predicted by analytic models , and tried to reconcile the apparent discrepancy with analytic models by attributing it to their small range of validity after significant lateral spreading starts ( @xmath41 ) for the typical modest values of @xmath0 used in the simulations .
@xcite disagree with this conclusion , and we address this dispute in
[ sec : dis ] .
more recently , @xcite has argued that significant lateral spreading is expected only at a later stage , when @xmath37 drops below @xmath42 ( rather than @xmath38 as obtained in simple analytic models ) , based on an analytic consideration ( which we address in [ sec : recipe ] and appendix [ sec : comp - prev ] , and find to be in error ) .
thus , there appears to be an ongoing debate on these important issues .
here we try to reconcile the apparent differences between the analytic and numerical results , in light of this recent debate .
the different relevant critical radii are discussed in
[ sec : r ] . in
[ sec : recipe ] we discuss the recipe for lateral expansion used by analytic models , and derive a new recipe that takes into account the non - spherical nature of the shock driven by the jet into the external medium . in
[ sec : model ] we construct an analytic relativistic model , which includes both the traditional recipe and our new recipe for the jet lateral expansion .
it is also shown that while the region of interest and validity of the analytic model ( corresponding to @xmath41 ) increases as @xmath0 decreases , @xmath18 reaches lower values , resulting in a narrower jet at the time when the analytic solution becomes invalid . because this relativistic model breaks down in a region of interest ( both for typical grb parameters and for comparison with simulations ) , in
[ sec : gen ] we generalize it so that it would be valid also at low @xmath37 and high @xmath18 , using two different assumptions on the accumulation of the swept - up external mass ( in [ sec : trumpet ] and [ sec : conical ] ) .
according to the results of these models ( [ sec : model - results ] ) , a phase of rapid exponential lateral expansion exists only for sufficiently narrow initial jet half - opening angles , of approximately @xmath43 for @xmath44 , respectively . in [ sec : comp - sim ] we compare our analytic models to numerical simulations ( with a modest @xmath39 ) and find reasonably good agreement ( with no exponential lateral expansion in both cases ) , where the differences between the two recipes for the lateral spreading have a smaller effect on the agreement with numerical simulations compared to the generalization of the model to small @xmath37 and large @xmath18 .
the implications of our results are discussed in [
sec : dis ] .
using the approximate equation for energy conservation ( for @xmath45 ) , @xmath46 , where @xmath47 is the swept - up rest mass at radius @xmath48 for a spherical flow in an external density @xmath49 ( with @xmath50 ) , and the definition of the jet radius @xmath10 as the radius where @xmath51 for a spherical flow of energy @xmath52 , we obtain @xmath53}^{1/(3-k ) } = 2^{1/(3-k)}r_{\rm s}(e_{\rm jet})\ .\ ] ] similarly , the sedov radius of a spherical flow with @xmath54 is @xmath55}^{1/(3-k ) } = \theta_0^{-2/(3-k)}r_j = { \left(\frac{\theta_0 ^ 2}{2}\right)}^{-1/(3-k)}r_{\rm s}(e_{\rm jet})\ .\ ] ] two extreme assumptions on the degree of lateral spreading , which likely bracket the true jet dynamics are : 1 .
mildly relativistic lateral expansion in the jet co - moving frame , and 2 .
no lateral spreading until the jet becomes non - relativistic .
assumption 1 , which is made in most semi - analytic models , results in exponential growth of @xmath56 , until the jet becomes quasi - spherical and non - relativistic at @xmath57 assumption 2 was so far studied mainly by @xcite , and leads to @xmath58 in this case the jet is still very far from spherical symmetry at @xmath16 , and thus approaches spherical symmetry only after the radius grows by a factor @xmath59 , of a few or several . moreover , since the radius of the sedov - taylor solution scales as @xmath60}^{2/(5-k ) } \sim { \left(\frac{et^2}{a}\right)}^{1/(5-k)}\ , \ ] ] and at the non - relativistic transition time , @xmath61 , the sedov - taylor radius of a spherical flow with the true jet energy is much smaller than the jet radius at that time , @xmath62 , the flow approaches spherical symmetry only at the time @xmath63 when @xmath64 , which corresponds to @xmath65 ( see eq
. 6 of * ? ? ?
note that this is much smaller than the factor ( @xmath59 ) by which the radius grows over the same time . for assumption 1
similar arguments imply @xmath66 , where @xmath67 can be expected .
( i.e. its extent along its symmetry axis ) and lateral size @xmath68 as a function of the lab frame time @xmath69 for two extreme assumptions on its degree of lateral spreading : ( 1 ) mildly relativistic lateral expansion in the jet co - moving frame , and ( 2 ) no lateral spreading until the jet becomes non - relativistic .
the jet becomes spherical when @xmath70 and @xmath68 become equal , which occurs well after the jet becomes non - relativistic , and then joins the sedov - taylor solution .
[ fig : r1 ] ] fig .
[ fig : r1 ] shows the jet radius @xmath71 ( i.e. its extent along its symmetry axis ) and lateral size @xmath68 as a function of the lab frame time @xmath69 for these two assumptions .
the region where the dynamics for these two extreme assumptions differ is basically where the dynamics are most uncertain , and corresponds to the range of radii @xmath72 ( i.e. a factor of @xmath73 in radius ) , and ( lab frame ) times @xmath74 ( or a factor of @xmath75 in time ) . altogether ,
the ordering of the different radii is @xmath76 or @xmath77
the `` traditional '' basic underlying model assumptions used for the analytic modeling of relativistic jet dynamics during the afterglow phase ( e.g. , * ? ? ? * ; * ? ? ?
* ) are ( i ) a uniform jet within a finite half - opening angle @xmath18 with an initial value @xmath0 that has sharp edges , ( ii ) the shock front is part of a sphere at any given lab frame time @xmath69 , ( iii ) the outer edge of the jet is expanding sideways mildly relativistically , with @xmath78 in the local rest frame of the jet ( where quantities are denoted with a prime ) , ( iv ) the jet velocity is always in the radial direction and @xmath79 . under these assumptions ,
the jet dynamics are obtained by solving the 1d ordinary differential equations for the conservation of energy and particle number . )
highly relativistic ( @xmath80 ) jet , the equation for the conservation of linear momentum in the direction of the jet symmetry axis is almost identical to the energy conservation equation .
when the jet becomes sub - relativistic the conservation of energy and linear momentum force it to approach spherical symmetry , and once it becomes quasi - spherical then again the momentum conservation equation becomes irrelevant . ] the lateral expansion speed in the lab frame ( i.e. the rest frame of the central source and the external medium ) is @xmath81 , where @xmath82 is its lateral component of the 4-velocity ( which is lorentz invariant , so that @xmath83 ) , while @xmath84 is its radial component .
primed quantities are measured in a frame moving at @xmath85 in the radial direction , so that @xmath86 and @xmath87^{1/2 } = \beta'_\theta$ ] .
the usual assumption @xcite is that @xmath78 , which corresponds to @xmath88 as is shown in the next section , @xmath89 directly determines the jet lateral expansion rate in the lab frame .
here we derive a new physically motivated recipe .
it relies on the fact that for any shock front , with an arbitrary shape , the local velocity vector of the material just behind the shock front as measured in the rest frame of the upstream fluid ahead of the shock ( i.e. the lab frame in our case ) , @xmath90 , is normal to the shock front ( i.e. in the direction of the shock normal , @xmath91 , at that location ; * ? ? ?
* ) , namely @xmath92 a simple way to understand this result is that as each fluid element passes through the shock it samples only the local conditions ( and is not aware of the large scale or global shock front geometry ) and locally the shock normal is the only preferred direction in the upstream rest frame ( e.g. , the pressure gradients that accelerate the fluid element are in the @xmath93 direction and thus accelerate it in the @xmath91 direction ) . for an axisymmetric shock ( with no dependence on the azimuthal angle @xmath94 ) , eq .
( [ n_beta_hat ] ) immediately implies that the angle @xmath95 between the shock normal , @xmath91 , and the radial direction , @xmath96 , which is defined by @xmath97 , satisfies @xmath98 where @xmath99 is the polar angle measured from the jet symmetry axis . since @xmath100 , we have @xmath101 , where @xmath102 is the angular scale over which @xmath103 varies significantly , and we have assumed that @xmath103 decreases with @xmath99 , as is usually expected . since for @xmath80 and @xmath104
we also have @xmath105 , eq .
( [ eq : alpha ] ) implies that @xmath106 . for a roughly uniform jet of half - opening angle @xmath18
we have @xmath107 , and therefore @xmath108 which is our new recipe for lateral expansion .
( [ eq : new - recipe ] ) was first derived in the context of grbs by @xcite .
recently it was rederived by @xcite , based on an earlier work by @xcite .
@xcite has argued that eq .
( [ eq : new - recipe ] ) implies a negligible lateral expansion as long as @xmath109 suggesting that with this model one obtains a slow sideways expansion , as seen in the numerical simulations .
however , as we show later , this formula results in a slower lateral expansion ( compared to the usual recipe , i.e. eq . [ [ eq : old - recipe ] ] ) only as long @xmath110 ( the standard condition for the onset of significant lateral expansion ) , but once @xmath111 this formula leads to a faster sideways expansion .
we also show later that other factors , namely the break down of the ultra - relativistic and small angle approximations , are the main cause for the discrepancy between the existing simple analytic models and the numerical simulations . for completeness
we discuss the details of lyutikov s and shapiro s work in appendix a.
we turn now to compare the traditional recipe for the lateral expansion speed , @xmath112 ( eq . [ [ eq : old - recipe ] ] ) , with our own new simple recipe , @xmath113 ( eq . [ [ eq : new - recipe ] ] ) , which was derived in the previous section . these recipes are implemented here within the semi - analytic model for the jet dynamics of @xcite .
the main results are provided here and we refer the reader to that work for more details on that model . broadly similar semi - analytic models , with some variations , were used earlier by other authors ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the lateral size of the jet , @xmath68 , and its radius , @xmath114 , are related by @xmath115 .
the evolution of @xmath68 is governed by @xmath116 and therefore @xmath117 where we have conveniently introduced the parameter @xmath118 that enables us to analyze these two different recipes together .
the external density is assumed to be a power law in radius for which the shock lorentz factor decreases with radius for a spherical adiabatic blast wave during the self - similar stage of its evolution @xcite .
] , @xmath119 .
the total swept - up ( rest ) mass , @xmath47 , is accumulated as @xmath120 where the factor of @xmath121 is since a double sided jet is assumed .
as long as the jet is relativistic , energy conservation takes the form @xmath122 , which implies that @xmath123 , and @xmath124 one can numerically integrate equations ( [ dtheta_dr ] ) , ( [ dm_dr ] ) , and ( [ dgamma_dr ] ) thus obtaining @xmath56 , @xmath47 , and @xmath125 .
alternatively , one can use the relation @xmath122 ( energy conservation ) which reduces the number of free variable to two , and solve equations ( [ dtheta_dr ] ) and ( [ dgamma_dr ] ) . changing to normalized dimensionless variables @xmath126 ,
@xmath127 and @xmath128^{1/(3-k)}r / r_j$ ] , gives @xmath129 where the initial conditions at some small radius @xmath130 ( just after the deceleration radius ) are @xmath131 note that by definition , @xmath132 .
( [ dtheta_drt ] ) and ( [ dgamma_drt ] ) imply @xmath133 for @xmath134 the second term on the r.h.s of eq .
( [ dy_j_drt ] ) dominates , implying @xmath135 , which is consistent with eq .
( [ ic ] ) .
this suggests that the two terms become comparable at @xmath136 that is given by @xmath137}\ .\ ] ] while @xmath138 for @xmath139 , it can reach very low values ( @xmath140 ) as @xmath141 approaches @xmath142 .
we are interested here mainly in @xmath143 , for which @xmath144 still approximately holds .
we do note , however , that the lower values of @xmath145 for higher values of @xmath141 result in an earlier onset of significant lateral expansion for such higher @xmath141-values .
now let us examine what happens at @xmath146 .
if we assume that the first term becomes dominant then eq .
( [ dy_j_drt ] ) would imply @xmath147^{1/(1+a)}$ ] , which in turn implies that the second term would be dominant ( since @xmath148^{(3+a)/(1+a)}r^{3-k}\gg 1 $ ] ) , rendering the original assumption inconsistent .
the same applies if the opposite assumption is made , that the second term is dominant ( in this case @xmath149 which implies that the first term would be dominant , @xmath150 ) .
this implies that the two terms must remain comparable , implying @xmath151 .
a similar conclusion can be reach by taking the ratio of equations ( [ dtheta_drt ] ) and ( [ dgamma_drt ] ) which implies that @xmath152 a more careful examination shows that they must cancel each other to leading order , and the first two leading terms for @xmath153 are given by @xmath154 substituting equation ( [ y_j ] ) into equations ( [ dtheta_drt ] ) and ( [ dgamma_drt ] ) yields @xmath155 and @xmath156\ , \quad \\ \label{gamma_scaling } \gamma & \approx & \frac{1}{b}\,r^{-\frac{2(3-k)}{(3+a)^2 } } \exp\left[-\frac{(3+a)}{(3-k)(1+a)}\,r^\frac{(3-k)(1+a)}{(3+a)}\right]\ , \quad\quad\end{aligned}\ ] ] where the normalization coefficient @xmath157 is determined numerically . for @xmath134 we have @xmath158 for details ) , for either a uniform ( @xmath159 ) or a wind - like , stratified ( @xmath160 ) external density profile , and for either the old ( @xmath161 ) or our new ( @xmath162 ) recipe for the jet lateral expansion speed .
the dynamical range in this figure is unrealistically large , and it is shown mainly in order to demonstrate the properties of this solution , and show how well our analytic approximation for @xmath163 works ( the _ dashed green lines _ in the middle and bottom panels , which are practically on top of the numerical results ) .
[ fig : jet_dyn_1d ] ] fig .
[ fig : jet_dyn_1d ] shows the results of our model in terms of the normalized jet half - opening angle @xmath164 and lorentz factor @xmath165 ( as well as their product , @xmath166 ) as a function of the normalized radius @xmath167^{1/(3-k)}r / r_j$ ] .
the results are shown both for a uniform external medium ( @xmath159 ) , which is the main focus of this work , as well as for a stellar wind ( @xmath160 ; this is included mainly for completeness and is only briefly discussed in [ sec : dis ] ) .
the dynamical range in this figure is unrealistically large , and it is shown mainly in order to demonstrate the properties of this solution , and show how well our analytic approximation for @xmath163 works ( the dashed green lines in the middle and bottom panels , which are practically on top of the numerical results ) . the excellent agreement between our semi - analytic results ( the numerical solution of eqs .
[ [ dtheta_drt ] ] and [ [ dgamma_drt ] ] ) and analytic formulas ( eqs . [
[ theta_scaling ] ] and [ [ gamma_scaling ] ] ) shows that our analytic results ( including eqs . [ [ y_j ] ] and [ [ y_j1 ] ] ) can be safely used in order to analyze the result of this model .
this good agreement was also used in order to find the exact values of the numerical coefficient @xmath157 that determines the normalization for @xmath99 and @xmath168 .
, which were found to be @xmath169 , @xmath170 and @xmath171 . our new recipe for the lateral expansion speed ( eq . [ [ eq : new - recipe ] ] ) results in a slower initial lateral expansion compared to the old recipe at @xmath172 , where @xmath173 .
however , at larger radii , @xmath174 , where @xmath175 it results in a faster lateral expansion . but only for a uniform external density ( @xmath159 ) and for three different values of the initial jet half - opening angle : @xmath176 ( _ green _ ) , @xmath177 ( _ red _ ) and @xmath39 ( _ blue _ ) . the old ( @xmath161 ) and new ( @xmath162 ) recipes for the jet lateral expansion are sown by _ dashed _ and _ solid _ lines , respectively . in the top two panels the lines for different @xmath0 and the same @xmath118 values coincide ( see text for details ) .
the values of @xmath178 are indicated in the bottom panel for reference .
[ fig : jet_dyn_1d_2 ] ] figure [ fig : jet_dyn_1d_2 ] shows similar results for a uniform external medium ( @xmath159 ) and for three different values of the initial jet half - opening angle , @xmath40 . since the dynamical equations ( eqs .
[ [ dtheta_drt ] ] and [ [ dgamma_drt ] ] ) involve only the normalized variables @xmath99 , @xmath168 and @xmath179 , and the initial conditions ( eq . [
[ ic ] ] ) for @xmath99 and @xmath168 depend only on the initial normalized radius @xmath180 , the lines for these normalized variables in the top two panels for the different @xmath0 values exactly coincide . was used , but in the limit @xmath181 the dependence of the solution on @xmath180 goes away at @xmath182 . ]
the two bottom panels show the un - normalized quantities @xmath18 and @xmath37 for our three values of @xmath0 . in the bottom panel
we have added for comparison the sedov radius , @xmath178 , for a spherical flow with the same isotropic equivalent energy the jet started with .
we define @xmath16 for our model as the radius where formally @xmath183 ( at which point this model clearly breaks down ) .
figure [ fig : jet_dyn_1d_3 ] is similar to fig .
[ fig : jet_dyn_1d_2 ] but the jet radius @xmath48 is normalized by the radius @xmath184 instead of @xmath185^{-1/(3-k)}r_j = [ ( 3-k)/4]^{-1/(3-k)}r_{\rm s}(e_{\rm jet})$ ] . but
shown as a function of the jet radius @xmath48 normalized by @xmath186 instead of of @xmath185^{-1/(3-k)}r_j = [ ( 3-k)/4]^{-1/(3-k)}r_{\rm s}(e_{\rm jet})$ ] , where @xmath187 is the isotropic equivalent energy in the jet , while @xmath188 is its true energy .
[ fig : jet_dyn_1d_3 ] ] , defined by @xmath189 , normalized by @xmath10 , as a function of @xmath0 . for comparison
, @xmath178 is also shown ; the two radii , @xmath16 and @xmath178 , become similar at @xmath190 but are very different for @xmath191 . *
lower panel * : the value of the jet half - opening angle , @xmath18 , at @xmath16 , where our simple analytic relativistic model breaks down .
[ fig : theta_j_r_nr ] ] fig .
[ fig : theta_j_r_nr ] shows @xmath192 and @xmath193 as a function of @xmath0 .
it can be seen that @xmath16 depends on @xmath0 only logarithmically ( as can also be seen from eq .
[ [ gamma_scaling ] ] ) , while @xmath194 is simply a power of @xmath0 .
it is also evident that @xmath195 for @xmath25 , and its value increases with @xmath0 ( while @xmath196 decreases with @xmath0 ) .
this can also be seen from eq .
( [ y_j ] ) , using the leading order term in @xmath179 and the definition @xmath197 , which imply that @xmath198 , while @xmath199 ( or @xmath16 ) decreases ( logarithmically ) with @xmath0 . for @xmath200
the jet becomes non - relativistic and the model breaks down at smaller values of @xmath167^{1/(3-k)}r / r_j$ ] compared to @xmath201 , which is consistent with the fact that the jet also starts to spread sideways significantly at smaller values of @xmath179 , of the order of @xmath202}$ ] .
however , we are primarily interested here in @xmath201 .
the model breaks down when @xmath37 drops to 1 ( or even slightly earlier ) . as can be seen from figs .
[ fig : jet_dyn_1d][fig : jet_dyn_1d_3 ] , it breaks down earlier for larger @xmath0 values , and its region of validity ( especially at @xmath203 ) decreases as @xmath0 increases .
in particular , for the value of @xmath39 , which was most widely used so far in numerical simulations ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* while an even larger value of @xmath204rad was used in some works @xcite ; @xcite ) , this dynamical range is very narrow , and the asymptotic exponential growth of @xmath18 with @xmath48 is not reached before the model breaks down ( at @xmath205 or @xmath206 ) . even for @xmath176 , which was used in the most recent simulations @xcite and is at the low end of the values inferred from afterglow observations , the asymptotic exponential regime is only barely reached before the model breaks down ( in agreement with the conclusions of @xcite ) . note that in this ( limited ) region of validity of this semi - analytic model our new recipe might still result in smaller or comparable values of @xmath207 ( i.e. of @xmath18 for a fixed @xmath0 , at a given radius for a fixed @xmath188 ) compared to the old recipe .
the discussion about when this model breaks down is expanded in [ sec : comp - sim ] , where we compare the analytic models to numerical simulations .
because of this important limitation of our relativistic analytic model , in the next section we generalize it so that it would not break down when the jet becomes sub - relativistic or wide .
in order to avoid the breakdown of the model at small lorentz factors @xmath37 or large jet half - opening angles @xmath18 , we construct here simple generalizations of the analytic model studied in the previous section , which do not require the jet to be very narrow ( @xmath208 ) or highly relativistic ( @xmath209 ) .
two variants are introduced , named the trumpet model ( in [ sec : trumpet ] ) and the conical model ( in [ sec : conical ] ) , according to the shape of the region from which the external medium is assumed to have been swept up by the jet ( before it becomes spherical ) .
the rate at which the jet half - opening angle , @xmath18 , increases depends on the lateral velocity at the edge of the jet , @xmath210 , as @xmath211 , or @xmath212 a crude approximation for for the comoving 4-velocity of the lateral expansion ( @xmath213 ) , which would roughly correspond to the sound speed both in the relativistic and in the newtonian regimes , is @xmath214 .
this would modify the traditional recipe to @xmath215 or @xmath216 . in our recipe since the 4-velocity @xmath103 , unlike @xmath217 or @xmath37 , generally varies significantly with @xmath99 both in the relativistic and in the newtonian regimes , so that @xmath218 in both regimes , while @xmath219 only in the relativistic regime and @xmath220 only in the newtonian regime . ]
@xmath221 $ ] .
therefore , just as before , we still have @xmath222 in this model we follow the usual assumption that the external rest mass is swept - up by a working area consisting of the part of an expanding sphere of radius @xmath48 within a half - opening angle @xmath56 .
thus , the total swept - up ( rest ) mass , @xmath47 , for a double - sided jet is accumulated as @xmath2234\pi r^2\rho_{\rm ext}(r ) = [ 1-\cos\theta_j(r)]4\pi ar^{2-k}\ .\ ] ] energy conservation takes the approximate form @xmath224 , implying @xmath225 , and @xmath226u^3(r)\ .\ ] ] thus , in terms of @xmath167^{1/(3-k)}r / r_j$ ] we have @xmath227\ , \ ] ] where the initial conditions at some small radius @xmath228 ( just after the deceleration radius ) , corresponding to @xmath180 , are given by @xmath229 here we note that the usual assumption that leads to eq .
( [ dm_dr_app ] ) neglects the external matter at the sides of the jet . because of this , when eventually @xmath18 reaches @xmath230 at @xmath231 and is thus assumed to be fully spherical , the amount of swept - up external rest mass at @xmath231 calculated according to eq .
( [ dm_dr_app ] ) will be significantly smaller than that originally within a sphere of the same radius .
therefore , here in the conical model we adopt an alternative approach of using for the rest mass of the swept - up matter , that originally within a cone of half - opening angle @xmath18 , @xmath232\frac{4\pi}{(3-k)}a r^{3-k}\ .\ ] ] this still has the drawback of assigning the same lorentz factor to all of the swept - up external matter , even though that at the sides of the jet should have a significantly smaller 4-velocity than that near the head of the jet . using a slightly different normalized radius , @xmath233^{1/(3-k)}r$ ] , energy conservation ( @xmath234 ) and eq .
( [ dtheta_dr_app ] ) imply @xmath235^{(1+a)/2}\theta_j^a}\ , \ ] ] where the initial conditions at some small radius @xmath236 , corresponding to @xmath237 , are given by @xmath238 ) , and for a uniform external medium ( @xmath159 ) .
results are shown for three different values of the jet initial half - opening angle : @xmath176 ( in _ green _ ) , @xmath239 ( in _ red _ ) , and @xmath39 ( in _ blue _ ) .
for reference we also indicate the values of @xmath240 in the top panel , some relevant values of @xmath18 in the two middle panels , as well as the values of @xmath178 and @xmath241 in the bottom panel .
[ fig : jet_dyn_1d_2b ] ] but shown ( 1 ) only for our relativistic model ( _ green _ , _ red _ , and _ blue _ lines for @xmath176 , 0.1 , and 0.2 , respectively ) and trumpet model ( _ black _ , _ magenta _ , and _ cyan _ lines for @xmath242 , 0.1 , and 0.2 , respectively ) , ( 2 ) for both the old recipe ( @xmath243 ; _ dashed lines _ ) and our new recipe ( @xmath244 ; _ solid lines _ ) for the jet lateral expansion , and ( 3 ) as a function of the jet radius @xmath48 normalized by @xmath186 instead of of @xmath185^{-1/(3-k)}r_j = [ ( 3-k)/4]^{-1/(3-k)}r_{\rm s}(e_{\rm jet})$ ] .
[ fig : jet_dyn_1d_3a ] ] with the normalized radius @xmath179 , for @xmath44 ( _ top to bottom panels _ ) , where all models use our new recipe for the lateral spreading of the jet ( @xmath244 ) .
results are shown for @xmath245 ( _ using different colors _ ) while the values of @xmath246 and the critical radius @xmath145 ( given by eq .
[ [ eq : r_c ] ] , where the lateral spreading is expected to become significant ) are shown for reference .
[ fig : jet_dyn_1d_new2c_k012 ] ] and @xmath159 .
we show the jet 4-velocity , @xmath103 ( _ upper panel _ ) , half - opening angle , @xmath18 ( _ middle panel _ ) , as well as its normalized radius @xmath247 and lateral size @xmath248 ( _ bottom panel _ ) , as a function of the normalized lab frame time , @xmath249 , for our relativistic ( _ green lines _ ; until it breaks down at @xmath250 ) , trumpet ( _ red lines _ ) and conical ( _ blue lines _ ) models .
the _ solid _ and _ dashed _ lines are , respectively , for our new recipe ( @xmath244 ; eq . [
[ eq : new - recipe ] ] ) and the old recipe ( @xmath161 ; eq .
[ [ eq : old - recipe ] ] ) for the jet lateral expansion .
[ fig : r_t_lab4 ] ] but as a function of the observed time , @xmath251 , at which photons from the front of the jet reach an observer located along its symmetry axis , normalized by its value at the jet break time @xmath252 . [
fig : r_t_los4 ] ] figures [ fig : jet_dyn_1d_2b ] , [ fig : jet_dyn_1d_3a ] and [ fig : jet_dyn_1d_new2c_k012 ] depict a comparison of these two models with the relativistic model .
all three models agree at early times , while the jet is still highly relativistic , narrow and hardly expanded sideways .
the approximations of our relativistic model hold well at this stage and the difference in the swept - up mass between the trumpet and conical models is still very small . at later times , however , the three models show a different behavior .
the main effect of the relaxation of the small @xmath99 and ultra - relativistic approximations is that for typical values of @xmath253 the region of exponential growth of @xmath18 with @xmath48 largely disappears , and is replaced by a much slower , quasi - logarithmic growth .
this can most clearly be seen by comparing the results of the relativistic model ( from [ sec : model ] ; _ solid lines _ in figs .
[ fig : jet_dyn_1d_2b ] and [ fig : jet_dyn_1d_new2c_k012 ] , and green , red or blue lines in fig .
[ fig : jet_dyn_1d_3a ] ) and the trumpet model ( from [ sec : trumpet ] ; _ dot - dashed lines _ in figs . [
fig : jet_dyn_1d_2b ] and [ fig : jet_dyn_1d_new2c_k012 ] , and black , magenta or cyan lines in fig .
[ fig : jet_dyn_1d_3a ] ) .
these two models share the same assumption on the accumulation of the swept - up external medium , and differ only by relaxing in the trumpet model the requirements of @xmath80 and @xmath208 .
the results of these two models are very close at early times while @xmath80 , but diverge as @xmath37 becomes more modest and the simple relativistic model reaches the exponential regime .
this can also be seen in fig .
[ fig : jet_dyn_1d_new2c_k012 ] through the fact that @xmath254 for the two models start diverging when @xmath18 becomes modest and the small angle approximation breaks down .
the main difference between the trumpet and conical models is that for the conical model the swept up mass at a given radius @xmath48 is larger than for the trumpet model , resulting in a smaller @xmath37 and therefore also a larger @xmath18 , i.e. a faster evolution of @xmath18 and @xmath37 with @xmath48 .
since the larger swept - up mass comes from the sides of the jet , it becomes important only once the jet starts expanding sideways significantly , which occurs at @xmath145 ( see eq .
[ [ eq : r_c ] ] ) .
this can be clearly seen in fig .
[ fig : jet_dyn_1d_new2c_k012 ] , where the _ dot - dashed _ ( or _ solid _ , which practically coincide at early times ) and _ dashed _ lines , for the trumpet ( or relativistic ) and conical models , respectively , start diverging near @xmath145 . note that this remains valid for all @xmath141-values , while @xmath145 decreases with @xmath141 .
[ fig : jet_dyn_1d_new2c_k012 ] also shows that for sufficiently small values of @xmath0 , roughly @xmath255 for @xmath201 and even somewhat smaller @xmath0 values for larger @xmath141 values , there is still a phase of quasi - exponential lateral expansion for @xmath256 or @xmath257 . for such extremely small values of @xmath0
the difference between the conical and trumpet models becomes large during the exponential sideways expansion phase , where the lateral expansion is faster in the conical model .
we note , however , that such extremely narrow initial jet half - opening angles are below the smallest values that have so far been reliably inferred from grb afterglow modeling , so that they might not be very relevant in practice .
[ fig : r_t_lab4 ] and [ fig : r_t_los4 ] show the jet dynamics according to our different analytic models , for @xmath239 .
it can be seen the the differences between the various models are rather small until the point where our relativistic model breaks down .
the behaviour of the jet radius ( @xmath71 ) and lateral size ( @xmath68 ) as a function of the lab frame time ( @xmath69 ) shows a lot of similarities to the analytic expectations ( compare the _ bottom panel _ of fig .
[ fig : r_t_lab4 ] to fig .
[ fig : r1 ] ) .
this , again , demonstrates that our new recipe for the lateral spreading of the jet results in slower lateral expansion compared to the old recipe ( and is closer to assumption 2 of no lateral spreading _ dashed red lines _ in fig .
[ fig : r1 ] ) at early times when @xmath258 but faster lateral expansion at late times when @xmath259 ( i.e. closer to assumption 1 of fast lateral spreading _ solid blue lines _ in fig .
[ fig : r1 ] ) .
we turn now to a comparison of our analytic models with the results of full 2d special relativistic hydrodynamic simulations .
to do so one needs first to define which quantities should be compared .
this , however , is not unique and can be done in different ways . for the 4-velocity , @xmath103 , and as one ( out of a few ) reference value for the jet half - opening angle , @xmath18 ,
we use the weighted mean over the energy @xmath260 in the lab frame ( excluding rest energy ) of @xmath103 and @xmath99 , respectively , @xmath261 for the jet radius ( or parallel size , @xmath262 ) and lateral size ( @xmath68 ) we use : @xmath263 these averages reduce to @xmath264 ( or @xmath265 ) for a spherical flow . in order to perform a proper comparison to our analytic models ,
we need to calculate similar averages for our jet , which at any given time is the part of a thin spherical shell within a cone of half - opening angle @xmath18 .
thus , the radial integration drops out and we are left only with an integral over @xmath266 between @xmath267 and 1 , @xmath268 we can see that @xmath264 for @xmath269 , as it should . similarly , one can calculate @xmath270 as a proxy for @xmath18 in our models , @xmath271 this shows that @xmath272 for @xmath208 , while @xmath273 for @xmath269 ( which is the value for any spherical flow , also one with a radial profile ) and @xmath274 for @xmath275 .
one can also calculate the angle out to which a fraction @xmath276 of the energy is contained ( or the energy @xmath277 percentile ) , @xmath278\ , \ ] ] and compare it to the corresponding value from the numerical simulations . and
@xmath159 , between our analytic models ( _ thin lines _ ) and the results of 2d special relativistic hydrodynamic simulations ( from * ? ? ?
* ; * ? ? ?
* ) of a jet with initial conditions of a conical wedge of half - opening angle @xmath0 taken out of the @xcite self - similar solution ( _ thick dot - dashed black line _ ) , in terms of the jet 4-velocity ( @xmath103 ) , half - opening angle ( @xmath18 ) as well as normalized parallel ( @xmath279 ) and perpendicular ( @xmath280 ) sizes .
the _ green _ , _ red _ and _ blue _ lines are for our relativistic , trumpet , and conical models , respectively .
thin solid lines are for our new recipe for lateral expansion ( @xmath244 ) while thin dashed lines are for the old recipe ( @xmath161 ) .
[ fig : comp_sim2_k0 ] ] but for three different ways of quantifying the jet half - opening angle , @xmath18 .
the _ top panel _ and _ middle panel _ show two different energy percentiles , @xmath281 and @xmath282 , respectively , i.e. the values of @xmath99 up to which 95% and 75% of the energy is contained .
the _ bottom panel _ shows the weighted mean over the energy , @xmath270 .
[ fig : comp_sim1a_k0 ] ] figures [ fig : comp_sim2_k0 ] and [ fig : comp_sim1a_k0 ] show a comparison ( for @xmath201 and @xmath39 ) between the results of our analytic models and of 2d special relativistic hydrodynamic simulations ( from * ? ? ?
* ; * ? ? ?
* ) , when quantifying all of them as discussed above .
as can be seen from fig .
[ fig : comp_sim2_k0 ] , our models provide a reasonable overall description of the full hydrodynamic simulations , and thus appear to catch the basic underlying physics , despite their obvious simplicity .
[ fig : comp_sim1a_k0 ] shows three different ways of quantifying the jet half - opening angle , namely the weighted mean over the energy , @xmath270 ( _ bottom panel _ ) , and two different energy percentiles , @xmath282 ( _ middle panel _ ) and @xmath281 ( _ top panel _ ) , i.e. the values of @xmath99 up to which 75% and 95% of the energy , respectively , is contained .
it can be seen that @xmath281 provides the best match between our analytic model and the numerical simulations . for @xmath282 or @xmath270
the match is not as good ( though even then the difference is not very large ) .
this might be attributed to the fact that our analytic models assume a uniform energy per solid angle , @xmath283 , within the jet opening angle ( @xmath284 ) , while in practice ( or in the numerical simulations ) it drops towards the outer edge of the jet .
the drop in @xmath285 from the jet axis towards its edge causes both smaller values of @xmath270 and smaller values of @xmath286 for the lower energy percentiles ( or @xmath276-values ) relative to a uniform jet with the same @xmath286 for a large energy percentile ( or @xmath276-value ; e.g. , @xmath287 in our case ) .
the results for our new recipe for the jet sideways expansion are somewhat closer to the numerical simulations compared to the usual recipe for @xmath270 and @xmath282 , while the usual recipe is perhaps slightly closer for @xmath281 .
both the analytic models and the numerical simulations show that the flow becomes spherical more than a decade in time after it becomes sub - relativistic ( which may be quantified as the time when @xmath288 ) .
this can be attributed to the fact that once the flow becomes sub - relativistic its sound speed quickly drops , and so does the rate of lateral expansion . moreover , as the flow gradually becomes more spherical the lateral gradients become smaller , which makes the flow approach spherical symmetry more slowly .
the numerical simulations show that @xmath286 corresponding to lower energy percentiles ( or @xmath276-values ) approach their asymptotic values for a spherical flow at later times .
this shows that the transfer of energy to larger @xmath99-values is the slowest near the center of the jet and larger near its edges , which may in turn be attributed to the lateral gradient ( say , of @xmath285 ) in the jet , which are smallest near its center and largest near its edge .
in this work we have introduced a new , physically motivated recipe for the lateral expansion of the jet ( in [ sec : recipe ] ) .
it is based on the jump conditions for oblique shocks of arbitrary 4-velocity , which imply that the velocity of fluid just behind the shock front ( in the downstream region ) is in the direction of the local shock normal ( i.e. perpendicular to the shock front at that location ; @xmath289 , eq . [ [ n_beta_hat ] ] ) in the upstream rest frame ( which in our case is identified with the rest frame of the external medium and the central source ) .
our new recipe for the lateral expansion rate of the jet ( @xmath113 , eq .
[ [ eq : new - recipe ] ] ) has an extra factor of @xmath290 in the denominator relative to the usual recipe that has been used so far ( @xmath112 , eq . [ [ eq : old - recipe ] ] ) . this results in slower lateral expansion relative to the usual ( or old ) recipe at early times when @xmath291 , but faster lateral expansion at later times when @xmath259 , i.e. once the lateral expansion becomes significant .
next ( in [ sec : model ] ) , we have implemented our new recipe as well as the old recipe in a simple analytic model for the jet dynamics , which is valid only for high lorentz factors ( @xmath80 ) and narrow jet half - opening angles ( @xmath208 ) .
this model shows an exponential lateral expansion for @xmath292 , like previous analytic models of this type .
however , we demonstrate that for typical values of the initial jet half - opening angle ( @xmath293 ) this model is valid only over a very limited dynamical range for @xmath294 , so that the asymptotic exponential lateral expansion regime is hardly reached before the model breaks down .
this leads to a reasonable agreement with numerical simulations over this limited range ( as shown by @xcite and in
[ sec : comp - sim ] ) .
this motivated us ( in [ sec : gen ] ) to generalize our relativistic model so that it would be valid for any values of @xmath37 and @xmath18 .
this was done by switching to the 4-velocity @xmath103 ( instead of @xmath37 ) as the dynamical variable that we evolve ( so that it would vary significantly in both the relativistic and the newtonian regimes ) , and systematically not relying on any relativistic or small angle approximations .
moreover , we have implemented two different assumptions for the accumulation of the swept - up external rest mass , corresponding to a different variant of the model .
the trumpet model makes the usual assumption that the working surface is the part of a sphere of radius @xmath48 within a cone of half - opening angle @xmath56 .
the conical model assumes that all the mass within a cone of half - opening angle @xmath56 was swept - up , so that once the flow becomes spherical the swept - up mass is equal to that originally within a sphere of the same radius .
our relativistic , trumpet and conical models all agree at early times when the jet is still highly relativistic , narrow and hardly expanded sideways ( @xmath295 ) . at this stage
the approximations of our relativistic model hold well and there are only very small differences in the swept - up mass between the trumpet and conical models .
however , at later times when @xmath296 the relativistic model enters a phase of rapid , exponential sideways expansion and it quickly breaks down , before becoming spherical .
we note , however , that for a stratified or stellar wind like external medium ( @xmath160 ) the jet is closer to being spherical than for a uniform or ism like external medium ( @xmath159 ; see _ bottom panel _ of fig .
[ fig : theta_j_r_nr ] ) when the relativistic model breaks down . for the trumpet and conical models , which are valid for any @xmath37 or @xmath18
, the phase of rapid , exponential sideways expansion largely disappears for typical values of @xmath253 .
this occurs because the jet is no longer ultra - relativistic soon after @xmath37 drops below @xmath38 , and once it becomes mildly or sub - relativistic its sound speed and therefore its rate of lateral expansion decrease compared to the ultra - relativistic regime .
the conical model evolves somewhat faster than the trumpet model , since it accumulates external rest mass also from the sides of the jet , and thus it slows down faster . the smaller @xmath37 results in turn in even faster lateral expansion rate and a larger @xmath18 ( at a given radius @xmath48 or lab frame time @xmath69 ) .
we compared ( in [ sec : comp - sim ] ) our analytic models to the results of 2d special relativistic hydrodynamic simulations ( from * ? ? ?
* ; * ? ? ?
* ) , finding that they provide a reasonable description of the numerical results at all times .
therefore , they can be used for analytic calculations of the afterglow emission , and would provide more realistic results compared to previous analytic models .
the main factor that significantly improves the agreement with simulations compared to previous analytic models is the fact that we have generalized the model to be valid also at modest lorentz factors @xmath37 and large jet half - opening angles @xmath18 .
both our analytic generalized ( trumpet and conical ) models and the numerical simulations show that the jet first becomes sub - relativistic and only then gradually approaches spherical symmetry over a long time . for typical initial half - opening angles ( @xmath297 )
the phase of rapid exponential lateral spreading is largely eliminated and it is replaced by a quasi - logarithmic increase in @xmath18 with radius @xmath48 or lab frame time @xmath69 . @xcite have stressed that while noticeable sideways expansion starts for @xmath298 , this initially involves only a small fraction of the total jet energy in its outer parts , and the central parts of the jet that carry most of its energy take longer to start spreading their energy to wider angles . while it is true that the jet does not remain uniform ,
the differences in the early growth of the angles @xmath286 containing different fractions @xmath276 of the jet energy , normalized by their initial value , is not very large
less than a factor of 2 in lab frame time or radius between @xmath287 and @xmath299 , and tend to become smaller for narrower @xmath0 .
this can be seen from figure 4 of @xcite , which also shows that as @xmath0 is gradually decreased down to @xmath300 , its initial growth becomes steeper and it looks as if an early phase of exponential growth starts to develop , contrary to what is claimed in @xcite .
therefore , we conclude that ( i ) although the uniform jet approximation used in our analytic models is obvious rather crude , it nonetheless provides a reasonable description of the energetically dominant part of the jet , and ( ii ) the prediction of our analytic models that an early exponential sideways expansion phase should exist for sufficiently small @xmath0 is not only consistent with the existing simulations , but these simulations even show a hint for the development of such a phase .
this should obviously be tested more thoroughly by simulations that reach even lower values of @xmath0 .
a phase of exponential lateral spreading was first found by @xcite and @xcite using a simple analytic model .
later , @xcite found a self - similar solution with a similar scaling .
our main conclusion ( which is in agreement with @xcite ) is that such a phase will occur in practice only for jets that are initially extremely narrow ( with @xmath255 or so ) , while for more modest values of @xmath297 that are more typically inferred in grb jets , such a phase effectively does not exist .
this basically reconciles the long lasting apparent discrepancy between analytic models and numerical simulations .
we thank fabio de colle for sharing the results of his numerical simulations .
this research was supported by the erc advanced research grant `` grbs '' .
abdo , a. a. , et al .
2009a , science , 323 , 1688 blandford , r. d. , & mckee , c. f. 1976 , phys .
fluids , 19 , 1130 cannizzo , j. k. , gehrels , n. , & vishniac , e. t. 2004 , apj , 601 , 380 de colle , f. , granot , j. , ramirez - ruiz , e. , & lopez - camara , d. 2011a , apj in press ( arxiv:1111.6890 ) de colle , f. , ramirez - ruiz , e. , granot , j. , & lopez - camara , d. 2011b , submitted to apj ( arxiv:1111.6667 ) frail , d. a. , et al .
1997 , nature , 389 , 261 fruchter , a. s. , et al . 1999 ,
apj , 519 , l13 granot , j. 2007 , rev .
a&a , 27 , 140 granot , j. , miller , m. , piran , t. , suen , w. m. , & hughes , p. a. 2001 , in `` grbs in the afterglow era '' , ed .
e. costa , f. frontera , & j. hjorth ( berlin : springer ) , 312 granot , j. , & ramirez - ruiz , e. 2011 in `` gamma ray bursts '' , eds .
kouveliotou , c. , woosley , s. e. & wijers , r. a. m. j. , cambridge university press , chapter 11 ( arxiv:1012.5101 ) granot , j. , ramirez - ruiz , e. , & loeb , a. 2005 , apj , 618 , 413 gruzinov , a. 2007 , preprint ( arxiv:0704.3081 ) halpern , j. p. , et al .
2000 , apj , 543 , 697 harrison , f. a. , et al . 1999 ,
apj , 523 , l121 kompaneets , a. s. 1960 , soviet phys .
doklady , 5 , 46 kulkarni , s. r. , et al .
1999 , nature , 398 , 389 kumar , p. , & granot , j. 2003 , apj , 591 , 1075 kumar , p. , & panaitescu , a. 2000 , apj , 541 , l9 laumbach , d. d. , & probstein , p. f. 1969 , j. fluid mech . , 35
, 53 lyutikov , m. 2011 , preprint ( arxiv:1106.0025 ) meliani , z. , & keppens , r. 2010 , a&a , 520 , l3 moderski , r. , sikora , m. , & bulik , t. 2000 , apj , 529 , 151 oren , y. , nakar , e. , & piran , t. 2004 , mnras , 353 , l35 panaitescu , a. , & mszros , p. 1999 , apj , 526 , 707 perna , r. , & vietri , m. 2002 , apj , 569 , l47 pihlstrm , y.m . ,
taylor , g.b . ,
granot , j. , & doeleman , s. 2007 , apj , 664 , 411 piran , t. 2000 , physics reports , 333 , 529 piran , t. 2005 , rev .
phys . , 76 , 1143 price , p. a. , et al .
2001 , apj , 549 , l7 rhoads , j. e. 1997 , apj , 487 , l1 rhoads , j. e. 1999 , apj , 525 , 737 sari , r. , piran , t. , & halpern , j. 1999 , apj , 519 , l17 shapiro , p. r. 1979 , apj , 233 , 831 tan , j. c. , matzner , c. d. , & mckee , c. f. 2001 , apj , 551 , 946 taylor , g. b. , frail , d. a. , beasley , a. j. , & kulkarni , s. r. 1997 , nature , 389 , 263 taylor , g. b. , frail , d. a. , berger , e. , & kulkarni , s. r. 2004 , apj , 609 , l1 van eerten , h. j. , zhang , w. , & macfadyen , a. 2010 , apj , 722 , 235 van eerten , h. j. , meliani , z. , wijers , r. a. m. j. , & keppens , r. 2011 , mnras , 410 , 2016 van eerten , h. j. , & macfadyen , a. i. 2011 , preprint ( arxiv:1105.2485 ) van eerten , h. j. , van der horst , a. , & macfadyen , a. i. 2011 , preprint ( arxiv:1110.5089 ) wygoda , n. , waxman , e. , & frail , d. a. 2011 , apj , 738 , l23 zhang , w. , & macfadyen , a. i. 2009 , apj , 698 , 1261
we compare here our formulation for the jet lateral expansion rate , eq .
( [ eq : new - recipe ] ) , with earlier work .
this formula was first derived within the context of grbs by @xcite , who provided two different derivations .
the first follows our line of argument and is based on the orthogonality of the shock front and the velocity of the fluid just behind it in the rest frame of the fluid ahead of the shock ( eq . [ [ n_beta_hat ] ] ) .
the second derivation involves an analysis of the dynamical equations integrated over the radial profile . a result similar to eqs .
( [ eq : alpha ] ) and ( [ eq : new - recipe ] ) was also recently derived by @xcite , based on an earlier work by @xcite .
@xcite has argued that it implies a negligible lateral expansion as long as @xmath80 unless @xmath301 , suggesting that with this model one obtains a slow sideways expansion , as seen in the numerical simulations .
however , we note that the condition @xmath302 corresponds to @xmath303 .
this requirement is too extreme since @xmath303 would result in a quasi - spherical flow within a single dynamical time ( since in that case @xmath304 ) . as is well known ( see also
[ sec : model ] ) , the traditional recipe for lateral expansion ( eq . [ [ eq : old - recipe ] ] ) , namely @xmath305 , already gives an asymptotic exponential growth of @xmath18 with @xmath48 ( i.e. very rapid lateral expansion ) .
the earlier work by @xcite discusses two possible approximations for the dynamics of a non - spherical relativistic blast wave , both based on a thin shell approximation for the layer of shocked external medium that carries most of the energy , but with different additional assumptions : ( i ) the quasi - radial approximation ( used in the newtonian regime by * ? ? ?
* ) in which each part of the shock is assumed to move in a radial trajectory as if it were part of a spherical flow with the same local conditions ( and in particular the same energy per solid angle , excluding rest energy , @xmath306 ) , and ( ii ) the @xcite approximation , that the pressure behind the shock is uniform , i.e. the same at all locations behind the shock and is proportional to the average energy density in the region bounded by the shock front .
the first approximation assumes that the energy per solid angle in the flow ( excluding rest energy ) does not change and remains equal to its initial value , @xmath307 . in this sense
, it basically assumes no lateral expansion ( as the jet retains its initial angular structure in @xmath308 indefinitely ) , so that this is a model assumption in this case rather than a result . the second approximation , which was originally used by @xcite in the newtonian regime , does not appear to be very appropriate for the relativistic regime where the angular size of causally connected regions is @xmath309 , so that that the local dynamics of a small portion of the flow should not be affected by the average energy per unit volume in the whole flow , which may be dominated by regions that are not in causal contact with it .
a simple example of how the @xcite approximation violates causality in the relativistic regime is that for a uniform external medium it implies that the velocity of the shock front is uniform ( i.e. depends only on the lab frame time , but not on the location along the shock front ; * ? ? ?
* ) , which necessarily implies that the flow must approach spherical symmetry within a few dynamical times
. becomes essentially independent of @xmath99 , since its local value is dominated by the recently shocked material , where the shock lorentz factor is uniform .
] this obviously violates causality , since as we discussed in the introduction , a roughly uniform jet with reasonable sharp edges can not expand sideways significantly as long as @xmath310 , from causal considerations ( since its bulk is not in causal contact with its edges , and it does not `` know '' that it is not part of a spherical flow and should thus start expanding sideways ) . @xcite reaches the conclusion that the two approximations give the same result in the extreme relativistic limit only because he explicitly assumed that in the quasi - radial approximation the energy per solid angle , @xmath283 , is not only independent of time , but also independent of the location along the shock front ( this can be seen from the fact that his energy integral is independent of @xmath99 ) .
this assumption quickly leads to a quasi - spherical flow for a spherical external density profile , and the non - spherical solutions obtained by @xcite arise since he considered an exponential atmosphere , which is a highly non - spherical external density profile .
the problem of interest for us , namely the dynamics of grb jets during the afterglow phase , involves a non - uniform initial distribution of the energy per solid angle , @xmath311 , and in such a case the two approximations are not equivalent in the extreme relativistic limit .
therefore , we conclude that neither of these two approximations appears to be appropriate for studying the dynamics or degree of lateral spreading of grb jets during the afterglow phase . | the dynamics of grb jets during the afterglow phase have an important effect on the interpretation of their observations and for inferring key physical parameters such as their true energy and event rate .
semi - analytic models generally predict a fast lateral expansion , where the jet opening angle asymptotically grows exponentially with its radius . numerical simulations , however , show a much more modest lateral expansion , where the jet retains memory of its initial opening angle for a very long time , and the flow remains non - spherical until it becomes sub - relativistic , and only then gradually approaches spherical symmetry . here
we suggest a new analytic model based on a new physically derived recipe for the lateral expansion .
we also generalize the model by relaxing the common approximations of ultra - relativistic motion and a narrow jet opening angle .
we find that the new analytic model fits much better the results of numerical simulations , mainly because it remains valid also in the mildly relativistic , quasi spherical regime .
this model shows that for modest initial jet half - opening angles , @xmath0 , the outflow is not _
sufficiently _ ultra - relativistic when its lorentz factor reaches @xmath1 and therefore the sideways expansion is rather slow , showing no rapid , exponential phase . on the other hand , we find that jets with an extremely narrow initial half - opening angle , of about @xmath2 or so , which are still sufficiently ultra - relativistic at @xmath1 , do show a phase of rapid , exponential lateral expansion .
however , even such jets that expand sideways exponentially are still not spherical when they become sub - relativistic . |
in the context of inflationary cosmology , initial seeds of today s structures are generated in the de - sitter phase of an inflationary universe , as a consequence of quantum vacuum fluctuations ( see e.g. @xcite for a comprehensive overview of the theory of cosmological fluctuations and @xcite for a recent introductory overview ) . the standard way of analyzing these fluctuations is through linear perturbation theory .
however , the linear analyses is limited by the non - linear nature of the einstein equations , and in particular , the presence of perturbations is likely to affect the evolution of the background cosmology at the non - linear level .
the back reaction of short wavelength gravity waves on an expanding friedmann - robertson - walker ( frw ) cosmology is a well - understood problem @xcite .
however , in the context of inflationary and post - inflationary cosmology , the scalar metric fluctuations ( fluctuations coupled to energy density and pressure perturbations ) are believed to dominate over the effects of gravity waves , and thus , it is of great interest to understand the possible back reaction of these scalar perturbations on the cosmological background .
furthermore , during inflation , the phase space of infrared modes ( defined as modes with wavelength greater than the hubble radius ) grows exponentially , whereas the phase space of the ultraviolet modes does not grow and since the amplitude of the associated metric fluctuations of these infrared modes does _ not _ decrease in time , the back reaction of these infrared modes may grow to be significant .
there have been different approaches to address this question . in @xcite ,
the effective energy - momentum tensor formalism of @xcite was generalized to study the back reaction of infrared modes on the spatially homogeneous component of the metric .
the result was that , in a slow - roll inflationary background , the back reaction of infrared modes takes the form of a negative cosmological constant .
this was later confirmed in @xcite using very different techniques .
the objections ( see @xcite ) to these analyses are mainly concerned with the fact that @xcite do not consider a local observable quantity , and even if the effect is not simply a gauge artifact , it is still not clear how a local observer could distinguish the back reaction of long - wavelength modes from an initial condition ambiguity in a homogenous universe . more importantly , calculating an `` observable '' from the spatially averaged metric
will not in general give the same result as calculating the spatially averaged value of the observable . a later work by abramo and woodard in @xcite tried to resolve this problem by identifying a local physical variable which describes expansion rate of the universe , then calculating the back reaction of cosmological perturbations on this quantity .
a different approach was introduced in @xcite , where a simple variable describing the local expansion rate was considered and calculated up to second order in the metric fluctuations in a model with a single inflation field . in this approach ,
the leading contribution of back reaction of infrared modes to this variable was evaluated . when evaluated at a fixed value of the matter field @xmath0 , the only physical clock available in this simple system
, the dependence of the expansion rate on the clock time takes on exactly the same form as in an unperturbed background .
thus , the leading infrared back reaction terms had no locally measurable effect in this system .
similar conclusions were reached in @xcite . however , the fact that during inflation the phase space of infrared modes is growing in time , i.e. modes are exiting the hubble radius , leads us to the conjecture that there might be a locally measurable effect just due to these crossings , and if we include the corrections due this effect , the results for single scalar field models may change .
since these modes are generated in the quantum mechanical ultraviolet limit , handling them in general is very difficult .
attempts , so far , in order to model the evolution through writing a classical langevin - like equation for the coarse - grained scaler field , the so - called _ stochastic approach _
@xcite , fail to include the inflaton and metric fluctuations self - consistently . in this work ,
for the first time , we devise a self - consistent stochastic framework to study the evolution of the coarse - grained inflaton .
while the approach uses the results of linear ( metric+inflaton ) perturbation theory for the ultraviolet and transitional modes , it is rigorously extended to the _ non - linear _ infrared regime , which allows a consistent study of the evolution of fluctuations through their entire history .
we then apply this framework to the problem of back reaction in inflationary cosmology , and investigate how / when the non - linear effects may change properties / predictions of the standard slow - roll inflation .
the structure of the paper is as follows : in section ii , we explain how the linear theory of perturbations can be self - consistently generalized to the non - linear infrared regime , including both inflaton and metric fluctuations .
we also introduce @xmath1 , the local expansion rate , and how it is related to metric perturbations . in sec .
iii , we show that the quantum generation of fluctuations can be modelled as a stochastic term in the equation of motion for @xmath1 , coarse - grained over local hubble patches . in sec .
iv , we apply this equation to a chaotic inflationary scenario with a quadratic potential @xmath2 and evaluate the local expansion rate with back reaction corrections due to infrared modes .
v contains the generalization of the result from sec .
iv to any power - law potential . in sec .
vi , we investigate if back reaction can ever significantly change the inflationary predictions for our observable universe .
finally , sec .
vii summarizes our results and concludes the paper . throughout this paper
, we use the natural units , and set the planck mass equal to unity , i.e. @xmath3 .
our analysis is based on the consistent quantization of the linearized metric and matter perturbations in a classical expanding background space - time . in this framework
, it can be shown that , in the linear regime , the sasaki - mukhanov parameter @xmath4 @xcite , defined as : , the hubble expansion rate by @xmath5 , and the scalar matter field by @xmath0 . ] @xmath6 \ , = \ , \frac{a\dot{\varphi}}{h}\zeta,\ ] ] obeys the equation of motion of a free scalar field , on scales much smaller than the hubble radius . here
, @xmath7 is the bardeen parameter @xcite : @xmath8 where @xmath9 is the scalar metric perturbations in the longitudinal gauge , in a spatially flat universe : @xmath10 we use the following ansatz for our non - linear metric : @xmath11 which is a generalization of a metric with linear scalar perturbations .
this ansatz , though not a general solution , represents the most important sector of the metric , as any such metric can be continuously connected to a metric with linear scalar perturbations , which is the dominant sector if perturbations are generated during slow - roll inflation @xcite .
we consider the simplest model of matter with a single scalar field @xmath0 , the inflaton .
we also use a perfect fluid model to describe the expansion of the universe .
the velocity four - vector field @xmath12 can be considered to be tangential to a family of world lines of a matter fluid in a general space - time .
this four - vector is normalized such that @xmath13 where @xmath14 runs over the space - time indices . in terms of this four - vector , the local expansion rate @xmath1
is defined by @xmath15 where @xmath16 is the local hubble constant .
it represents the local expansion rate of the tangential surfaces orthogonal to the fluid flow .
it is easy to verify that , for our specific choice of gauge , @xmath1 follows the following relation , @xmath17 where a prime denotes the derivative with respect to @xmath18 , the proper time , and is related to the coordinate time via @xmath19 .
the lagrangian and the energy - momentum tensor for the inflaton are given by : @xmath20 it can be shown that for slow - roll inflation , in the long - wavelength limit , @xmath1 satisfies the following condition @xcite , @xmath21 in @xcite it is demonstrated that , in these limits , the generalized bardeen parameter : @xmath22 remains constant in the classical non - linear regime @xcite . here
, @xmath23 is a function of the inflaton field , obtained by integrating the @xmath24 einstein constraint equations , and satisfies @xmath25 it is easy to see that for linear perturbations @xmath26 reduces to the usual bardeen parameter @xmath7 , which is known to remain constant for adiabatic super - hubble modes .
is forced to be equal to @xmath27 by the off - diagonal @xmath28 equations for all theories in which the off - diagonal components of @xmath29 vanish to the linear order .
this is in particular the case for scalar field matter . ] combining the above equation , with eqs .
( [ thetapsi ] ) and ( [ defzeta ] ) , we see that @xmath30 in the classical long - wavelength limit . in the next section ,
we show how quantum fluctuation affect eq .
( [ zetaprime ] ) .
in this section , we use coarse - graining on the scale of the hubble radius to provide a framework to analyze back reaction , and obtain a new derivation of the equation of motion for stochastic inflation @xcite , which takes into account the metric fluctuations . let us define the coarse - grained generalized bardeen parameter , @xmath31 , in the fourier space , as : @xmath32 where @xmath33 is the step function ( not to be mistaken with @xmath1 , the local expansion rate ) which is zero for @xmath34 , and is one for @xmath35 . and tends to zero as @xmath36 goes to infinity fast enough to eliminate the ultraviolet divergence . ]
taking the time derivative of the above equation ; the time derivative of @xmath37 vanishes for infrared modes and the time derivative of @xmath33 is a delta function at @xmath38 , which yields @xmath39 this simply implies that @xmath40 in real space is related to @xmath41 in the following way @xmath42 two - point function of @xmath41 can be described in terms of its power spectrum , @xmath43 , defined by : @xmath44 assuming that hubble - scale perturbations are linear , @xmath43 can be derived in the linear theory of quantum generation of perturbations @xcite , @xmath45 using this knowledge for @xmath41 and assuming @xmath46 ( the slow - roll condition ) in the above equation , we can calculate the two - point function for @xmath47 .
note that @xmath5 in eqs .
( [ cg]-[pwrspc ] ) is the local hubble constant , and can be replaced by @xmath48 . combining eqs .
( [ xk]-[pwrspc ] ) , after performing few simple integrations , we arrive at : @xmath49 where @xmath50 and @xmath51 .
let us express @xmath52 as @xmath53 where the variable @xmath54 is a random gaussian field with : @xmath55 now , plugging the coarse grained value of @xmath56 from eq .
( [ defxi ] ) into the left hand side of eq .
( [ zetaprime ] ) , we may obtain the coarse - grained field equation : @xmath57 \ , .\ ] ]
in this section , we apply the coarse - graining method , developed in the previous section , to a chaotic inflationary scenario with a quadratic potential , and investigate the evolution of the background during inflation under the effect of second - order back reaction . here , we assume that the potential for the inflaton field , @xmath0 , has a quadratic form : @xmath58 plugging this into eq .
( [ thetav ] ) , and then ( [ maineqn ] ) , the equation of motion for @xmath1 reduces to @xmath59 this equation can be solved perturbatively , taking @xmath54 as the perturbation variable .
@xmath1 can be approximated by @xmath60 where @xmath61 , @xmath62 and @xmath63 are zeroth , first and second order in @xmath54 , respectively .
as the first step in solving eq .
( [ eqnquad ] ) , we note that @xmath61 is simply the expansion rate in the absence of perturbations , @xmath64 where our convention for @xmath18 is so that it has a negative initial value @xmath65 and goes to zero towards the end of inflation , so that @xmath1 decreases as inflation proceeds .
now we substitute @xmath61 back into eq .
( [ eqnquad ] ) to find @xmath66 , @xmath67 this step is easily followed by calculation of @xmath63 .
@xmath68 since @xmath54 is a random gaussian variable , we are not interested in the spontaneous value of @xmath1 , and instead investigate the statistical behavior of @xmath69 , the expectation value of the expansion rate over time as more infrared modes cross the hubble radius and may build up a back reaction effect .
as @xmath61 does not have a @xmath54 term , and the expectation value of @xmath62 vanishes over time , we have : @xmath70 where @xmath71 can be derived from eqs .
( [ theta2 ] ) and ( [ xi2 ] ) , which finally enables us to calculate expansion rate with corrections due to stochastic back reaction : @xmath72 contrary to our expectation from @xcite , we can see that the back reaction term has a positive value and it is increasing with time , as if it is speeding up inflation .
however , as we see in the next section , this is merely a result of our choice to average over hyper - surfaces of constant _ proper _ time , while other choices of time hyper - surfaces may lead to different forms , or even a different sign , for the back reaction term . nevertheless , as both @xmath1 and @xmath18 are observable
, we may conclude that the back reaction _ does not vanish _ during inflation and
since its magnitude is increasing with time , it may become important if inflation lasts long enough .
notice that this correction is , in fact , due to wavelengths larger than the hubble radius as , using eq .
( [ theta2 ] ) , @xmath71 could be expressed as an integral over @xmath36 , the wave number instead of @xmath18 , which would yield @xmath73 and the above integral is only over modes which have already crossed the hubble radius . in this sense , this is an infrared back reaction effect .
in this section , we generalize the calculations of the last section to power law inflaton potentials .
also , rather than proper time , we use the inflationary e - folding number as our time variable , as it has more direct and intuitive observational relevance . the number of e - foldings since the beginning of inflation is @xmath74 using eq .
( [ thetapsi ] ) , we can see that @xmath75 may be used as a measure of e - folding : @xmath76 and thus , in order to find how @xmath1 changes as a function of @xmath77 , it is sufficient to express it in terms of @xmath75 . combining eq .
( [ twopointzeta ] ) with @xmath78 , we can re - write eq .
( [ maineqn ] ) as @xmath79,\ ] ] where @xmath54 in the above equation like is , again , a random gaussian field , defined by @xmath80 in the case of single scalar field inflationary models with a power law potential @xmath81 using the procedure described in the previous section , eq .
( [ maineqn2 ] ) can be solved to obtain @xmath1 in terms of @xmath75 : @xmath82 note that we assumed @xmath83 towards the end of inflation .
the result may be simplified for the case of a quadratic potential : @xmath84 we see that , if averaged over the hyper - surfaces of constant e - folding rather than constant proper time , the second order back reaction has a negative sign , i.e. it slows down inflation .
we again believe that this is a real effect , as both @xmath1 and @xmath77 are observable parameters .
let us calculate the ratio of the second order term in eq .
( [ backp ] ) to the zeroth order term @xmath85 if @xmath86 , close to the end of inflation . currently , the most stringent constraints on the inflationary potential is due to the observations of the cosmic microwave background anisotropies by the wmap satellite @xcite .
the observed amplitude of anisotropies at the observable horizon scale require : @xmath87 with @xmath88 being the derivative of @xmath89 with respect to @xmath0 .
then we can use the simple slow - roll prediction ( first term in eq .
[ thet ] , along with eq .
[ thetav ] ) to relate @xmath0 to @xmath75 , and express @xmath90 in terms of @xmath75 : @xmath91 assuming that @xmath92 ( eq . [ net ] ) , and that inflation ends when @xmath93 ( in planck units ) , eq .
( [ ratio ] ) can be expressed in terms of @xmath94 : @xmath95 although , this result is specific to the power law inflaton potentials , we expect a similar order of magnitude for any smooth potential that is consistent with observational bounds
. does this mean that back reaction will always be negligible during inflation ?
of course , the number of e - foldings , @xmath77 , could be arbitrarily large .
indeed , after enough e - foldings , we end up with the self - similar chaotic inflationary picture which results from the stochastic formulation @xcite .
however , only @xmath96 of these e - foldings could fit in our local hubble patch today , and any effect on the power spectrum due to the back reaction of the super - hubble modes is inaccessible to observations .
another possibility is that the inflaton potential may not be smooth .
the observational bounds on @xmath90 hold from the observable horizon ( @xmath97 ) down to the scale of the lyman-@xmath14 forest ( @xmath98 ) @xcite , constraining only about 10 e - foldings in the inflaton potential .
an unusually flat portion in the inflaton potential ( see fig . [ signal ] ) may yield a significantly larger amplitude of fluctuations , e.g. @xmath99 , and be inaccessible to cosmological observations if it happens at small enough scales .
however , such a large amplitude may result in the production of primordial black holes ( pbhs ) , as the modes enter the hubble radius , which results in further constraints on @xmath94 . to study the production of pbhs @xcite ,
let us look at the comoving scale of @xmath100 , with the primordial amplitude of fluctuations @xmath94 .
this scale enters the hubble radius when @xmath101 , which happens at redshift @xmath102 , roughly given by @xmath103 where @xmath104 and @xmath105 are today s hubble constant and density parameter respectively , while @xmath106 is the redshift of radiation - matter equality . as the modes of scale @xmath107 enter the hubble radius , a fraction , @xmath108 , of the hubble patches have a fluctuation amplitude larger than @xmath109 , and
thus may form pbhs , where @xmath108 is given by @xmath110 \frac{d\zeta}{\sqrt{2\pi \delta^2_{\cal r } } } \sim \delta_{\cal r } \exp[-\delta^{-2}_{\cal r}/2].\ ] ] the current density in pbhs in units of the critical density , @xmath111 , is given by @xmath112 where @xmath113 , is the fraction of the mass in a hubble patch that goes into the pbh @xcite . requiring that @xmath114 yields @xmath115 for @xmath116 , eq . ( [ ff ] ) gives @xmath117 , and thus eq .
( [ omega ] ) , in combination with large scale structure observations , require @xmath118 other astrophysical observations may put tighter constraints on a population of pbhs in the universe @xcite . instead of going through these constraints ,
let us give an example .
assume at @xmath119 , @xmath116 and @xmath120 .
( [ om ] ) , @xmath121 , while the pbh mass , @xmath122 is @xmath123 , with @xmath124 , being the hubble mass at the hubble entry : @xmath125 therefore , with our choice of parameters , about @xmath126 of the mass of the universe will end up in @xmath127 black holes .
this is consistent with current astrophysical constraints on such objects @xcite . to summarize this section
, we showed that if we assume a smooth inflaton potential , in single matter field models infrared back reaction ( during the last 70 e - foldings of inflation ) should be negligible , at least to the 2nd order ( one loop ) in the amplitude of perturbations .
however , if we relax this assumption , it is possible to construct an inflaton potential that allows a significant back reaction effect during the slow - roll inflationary period , without breaking any observational constraints .
we use a stochastic coarse - graining method to approach the problem of infrared back reaction in a single scalar field inflationary cosmology . using the coarse - grained generalized bardeen parameter ,
we self - consistently derive the stochastic langevin equation that governs the evolution of the local expansion rate . solving this equation to second order ( one loop ) for chaotic inflation with a power law potential
, we see that there is a non - vanishing average back reaction term due to the infrared modes that have left the hubble radius .
although , the size of this effect will be negligible ( within the observable universe ) for smooth inflaton potentials , we show that it is possible to have non - smooth potentials which allow significant back reaction happening on sub - horizon scales , without breaking any observational constraints .
a possible implication of such non - smooth potentials is the generation of a population of massive primordial black holes , which could be further used to constrain or verify these models .
we would like to thank robert brandenberger for many useful insights during the course of this research .
g.g . would like to thank brown university for a dissertation fellowship which this research was in part supported with .
10 v. mukhanov , h. feldman , and r. brandenberger , _ phys .
rep . _ * 215 * , 203 ( 1992 ) .
r. h. brandenberger , arxiv : hep - th/0306071 .
d. brill and j. hartle , _ phys .
rev . _ * 135 * , b271 ( 1964 ) ; + r. isaacson , _ phys .
rev . _ * 166 * , 1272 ( 1968 ) .
v. f. mukhanov , l. r. abramo and r. h. brandenberger , _ phys .
lett . _ * 78 * , 1624 ( 1997 ) [ arxiv : gr - qc/9609026 ] .
l. r. abramo , r. h. brandenberger and v. f. mukhanov , _ phys .
d _ * 56 * , 3248 ( 1997 ) [ arxiv : gr - qc/9704037 ] .
l. r. abramo and r. p. woodard , _ phys .
d _ * 60 * , 044010 ( 1999 ) [ arxiv : astro - ph/9811430 ] .
w. unruh , arxiv : astro - ph/9802323 .
l. r. abramo and r. p. woodard , _ phys .
d _ * 65 * , 043507 ( 2002 ) [ arxiv : astro - ph/0109271 ] .
g. geshnizjani and r. brandenberger , _ phys .
d _ * 66 * , 123507 ( 2002 ) [ arxiv : gr - qc/0204074 ] .
s. rasanen , jcap * 0402 * , 003 ( 2004 ) [ arxiv : astro - ph/0311257 ] .
a. liddle and d. lyth , _ phys .
* 231 * , 1 ( 1993 ) .
n. afshordi and r. h. brandenberger , phys .
d * 63 * , 123505 ( 2001 ) [ arxiv : gr - qc/0011075 ] .
a similar conserved quantity is derived in d. s. salopek , and j. r. bond , _ phys .
d _ * 42 * , 3936 ( 1990 ) .
a. starobinski , in _ current topics in field theory , quantum gravity and strings _ , eds .
de vega and n. sanchez , lecture notes in physics , vol . *
206 * , p. 107
( springer , heidelberg , 1986 ) ; d. s. salopek , and j. r. bond , _ phys . rev .
d _ * 43 * , 1005 ( 1991 ) .
croft , et al .
, _ astrophys
. j. _ * 581 * , 20 ( 2002 ) [ arxiv : astro - ph/0012324 ] . for a modern account of pbh formation calculations , see a.m. green , a.r .
liddle , k.a .
malik , and m. sasaki,[arxiv : hep - th/0310265 ] i. hawke , and j.m .
stewart , _ class .
* 19 * , 3687 ( 2002 ) .
j. yoo , j. chaname , and a. gould , _ astrophys .
j. _ * 601 * , 311 ( 2004 ) [ arxiv : astro - ph/0307437 ] ; n. afshordi , p , mcdonald , and d.n .
spergel , _ astrophys .
j. lett . _
* 574 * , 71 ( 2003 ) [ arxiv : astro - ph/0302035 ] . | we introduce a self - consistent stochastic coarse - graining method , which includes both metric and scalar field fluctuations , to investigate the back reaction of long wavelength perturbations in single - scalar driven inflation , up to the second ( one loop ) order .
we demonstrate that , although back reaction can not be significant during the last 70 e - foldings of inflation with a smooth potential , there exist non - smooth inflaton potentials which allow significant back reaction , and are also consistent with cosmological observations .
such non - smooth potentials may lead to the generation of massive primordial black holes , which could be further used to constrain / verify these models . |
polymers combining chemically different monomeric units on their structure found very much theoretical and experimental interest , as , for instance , diblock copolymers , where in the case of a linear chain one block of the chain consists of one type of monomers , while the other part of another type of monomers .
these macromolecules have already found particular interest as they form different structures depending on their structural parameters , thermodynamics conditions , etc . for
a standard theoretical review one could see ref .
@xcite and references therein .
we underline that the motivation for studying such systems stems from the fact that these materials are used in industrial scale @xcite and also in very advanced technological applications ( as an example one may look a very recent review ref .
@xcite ) . as an extension ,
regarding possible complexity in the structure or other aspects , of diblock copolymers is regarded multiblock copolymers , where in their simplest case more than two blocks comprise a linear polymeric chain and the blocks are composed either of type a or type b monomers .
the phase diagram of multiblock copolymer melts @xcite resembles the phase diagram of diblock copolymer chains @xcite , where the scaling parameter @xmath2 controls the phase behavior of the system , as it has also been discussed in simulation studies @xcite . for the most symmetrical case
the lamellar structure is predicted , an approach that is particular valid in the strong segregation limit .
the case of multiblock copolymer chains is particularly interesting even when only two type of blocks ( a , b ) alternating along a linear chain in an infinite dilute solution is considered @xcite . in this case
, it is sufficient to study the behavior of an isolated chain taking into account only interactions ( often of short range ) within the chain .
a model describing a multiblock copolymer chain could also exhibit some interest due to its close relation to various toy - models @xcite , which try to mimic the behavior of various biomacromolecules in several processes @xcite .
such biomacromolecules are rather formed by periodically repeated structural units ( `` monomers '' ) along their chain .
simple ( coarse - grained ) models @xcite incorporating the key aspects for a specific biomacromolecule are able to capture efficiently their relevant properties for which the model is designed . the phase behavior of a single multiblock copolymer chain has been extensively studied by means of molecular dynamics ( md ) simulations by using a standard bead spring - model and considering interactions between beads according to a standard model ( lennard - jones ( lj ) fluid ) @xcite .
of course the use of md simulations imposes certain limits in the range that the parameters may vary , for example , the range of the temperature and the chain length .
however , it has been shown that the different regimes of interest for the phase behavior can all be accessed by simulations @xcite .
multiblock copolymers composed of two different types of blocks ( a and b ) alternating along the linear chain and length of a block a being equal to the length of block b that adopt coil - like structures in a good solvent or at temperatures close to the @xmath3 @xcite have been studied .
the chemical difference of monomers should actually be kept responsible for an expansion in the chain dimensions with respect to the equivalent homopolymer chains ( same total number of monomers ) under the same thermodynamic conditions @xcite . below a certain temperature @xcite a multiblock copolymer chain collapses due to the incompatibility between monomers and the solvent obtaining the so - called globule formation and different scenarios of phase behavior
have been discussed @xcite .
simulations have confirmed the validity of the parameter @xmath2 in defining the phase behavior of a di / multi - block copolymer system , but it has also been discussed that the geometry of the microphase separated regions is controlled by the total number of blocks @xmath0 , as well as other parameters , i.e. the relative size and arrangement of the blocks @xcite .
the same has been shown to be true for a single multiblock copolymer chain under poor solvent conditions @xcite .
the phase behavior of such macromolecules has been discussed within the frame of the analysis of the formed clusters ( which contain two or more blocks of the same type ) and of the probability that a cluster of one type of blocks contains all a blocks while another cluster contains all b blocks @xcite , i.e. a single lamellar structure with an a - b interface between blocks of type a and blocks of type b. thus , when this probability is unity we fall into the above single lamellar case .
then , for the range of temperatures ( @xmath4 , @xmath5 ) and of chain lengths ( @xmath6 , where @xmath7 is the block length ; blocks a and b have the same length ) accessible to simulations this type of phase behavior has been seen for small number of blocks @xmath0 and for values of block length @xmath8 .
another scenario suggests that the above type of phase behavior takes place with a certain probability ( @xmath9 , @xmath10 being the number of clusters ; such probability can be very high or very small depending on the values of @xmath7 , @xmath0 , and @xmath11 ) .
the third scenario corresponds to the case that the probability @xmath12 ( the formation of only two clusters with monomers of different type never occurs ) and a symmetric variation in the number of clusters @xmath10 around an average value ( @xmath13 , @xmath0 is the total number of blocks of the chain ) is seen .
of course , the analysis of the probability is performed separately for clusters of type a monomers and for clusters of type b monomers .
otherwise the a and b beads that belong to the same chain would always belong to a single cluster containing all the beads of the chain .
similar conclusions have been also drawn by a similar model @xcite , where also the geometry of these cluster phases has been analyzed in detail .
guided by the theory @xcite , one would rather expect in the long chain limit that a ground - state - type structure would be a single lamellar domain , where an interface between all a- and b - type blocks is formed , similarly to what is known for multiblock copolymer melts for symmetric choice of parameters .
such a structure would have much less ( unfavorable ) a - b contacts compared to a multidomain structure of a and b clusters , which is kinetically favored in simulations .
recently , the static properties of the formed clusters for such systems have been discussed @xcite , where one is also able to detect in addition to the phase boundaries the globule - coil crossover and the proximity of the @xmath3 temperature through such an analysis . in this manuscript
, we will show further , that properties such as the bond correlation and asphericity can provide a rather accurate evidence for detection of full phase separation ( single lamellar structure ) in multiblock copolymer chains .
additionally , we will argue that these properties can provide information on the length @xmath7 and the number of blocks @xmath0 that this phase behavior is taking place .
furthermore , we provide a complete understanding and a broad discussion of the static properties of such macromolecules under poor solvent conditions for the blocks and the multiblock chain as a whole , at least for the most symmetric case that is considered here , and for chain lengths and thermodynamic conditions accessible to our simulations . as we have chosen in our study the same potential parameters as for the standard lj fluid @xcite , our results could also serve as a benchmark for similar polymeric and biological systems .
we discuss our results only in the poor solvent regime , which is the most interesting case due to the different phase behavior that has been observed when the chain is in the collapsed state ( globule state ) @xcite .
the rest of the paper is organized as follows . in sec .
[ sec:2 ] we describe our simulation model and the methods to analyze our results . in sec .
[ sec:3 ] we present our results for the whole multiblock chain and the individual blocks and we discuss particular details of the studied properties pertinent to the phase behavior of the multiblock chains . finally , in sec .
[ sec:4 ] we summarize our conclusions in connection with recent results on the phase behavior of symmetric linear multiblock copolymers .
as a detailed study of the systems under consideration is computationally demanding for reasons that will be discussed below , we have focused only on symmetric linear multiblock copolymers , where blocks of type a beads and blocks of type b beads alternate along the chain .
then the length of all blocks , irrespective of whether being of type a or b , is @xmath7 , the number of a - type blocks @xmath14 is equal to the number of b - type blocks @xmath15 , with @xmath16 an even number denoting the total number of blocks .
a multiblock copolymer chain is described by the parameters shown schematically in fig .
[ fig1 ] , where the different colors correspond to the different types of blocks , which alternate along the chain . to model our chains we use the standard bead - spring model @xcite to describe a multiblock copolymer chain , where all beads interact with a truncated and shifted lj potential @xmath17 and nearest neighbors bonded together along the chain
also experience the finitely extensible nonlinear elastic potential @xmath18 , @xmath19 being the distance between the beads .
@xmath20 + c , \quad r \leq r_c \quad,\ ] ] where @xmath21 , @xmath22 denote the different type of monomers , and the constant @xmath23 is defined such that the potential is continuous at the cut - off ( @xmath24 ) .
henceforth units are chosen such that @xmath25 , @xmath26 , the boltzmann constant @xmath27 , and in addition the mass @xmath28 of beads is chosen to be unity . for simplicity , @xmath29 , but @xmath30 , in order to create an unmixing tendency between monomers a and b belonging to different blocks as done in previous studies @xcite , and as is used for a standard system ( lj fluid ) @xcite .
therefore , @xmath31 was kept the same throughout our simulations and @xmath32 ( @xmath33 ) was mainly varied by tuning the temperature @xmath11 .
the potential of eq .
( [ eq:1 ] ) acts between any pair of beads , irrespective of whether they are bonded or not . for bonded beads additionally the potential @xmath18 acts , @xmath34 , \quad 0 < r \leq r_0.\ ] ] where @xmath35 , and
hence @xmath36 is the maximal distance between bonded beads .
we use the standard choice @xmath37 and @xmath38 .
we recall that for linear chains the @xmath3 temperature for the present model has been roughly estimated as @xmath39 ( a more accurate value @xcite @xmath40 could only be established for chain length @xmath41 , @xmath42 ) .
thus , we have considered in our study the range of temperatures @xmath43 . moreover , in previous discussion of this system @xcite we showed that the chain enters in the globule regime at temperature @xmath44 .
therefore , phase separation between a and b monomers takes place below this temperature . for homopolymer chains
this boundary is sharper compared to the case of multiblock copolymers where also appears smoother as the number of blocks @xmath0 increases for constant chain length @xmath6 as the a - b junctions between blocks increase .
of course , the temperature that we enter the globule regime ( upon lowering the temperature ) is the same for homopolymer and copolymer chains , because in our model the location of this boundary depends only on the solvent ( controlled by temperature in our simulations ) , which is the same for a homopolymer chain , as well as for type a and type b beads of a copolymer chain , but in the case of copolymer chains it only appears smoother .
further on our simulation method , the positions @xmath45 of the effective monomers with label @xmath46 evolve in time @xmath47 according to newton s equation of motion , amended by the langevin thermostat @xmath48 where @xmath49 is the total potential acting on the @xmath46th bead due to its interactions with the other beads at sites @xmath50 , @xmath51 is the friction coefficient , and @xmath52 the associated random force .
the latter is related to @xmath51 by the standard fluctuation - dissipation relation @xmath53 following previous work @xcite , we choose @xmath54 , the md time unit @xmath55 being also unity for our choice of units .
equation ( [ eq:3 ] ) is integrated using the leap - frog algorithm @xcite , with a time step @xmath56 and utilizing the gromacs package @xcite .
simulation of polymer chains under poor solvent conditions is notoriously difficult as relaxation times become exceedingly high .
therefore , we describe here briefly our simulation procedure .
after equilibration at a `` high '' temperature ( in the coil regime ) , we collect a number of independent samples ( typically @xmath57 ) , which we use as initial configurations for slow cooling runs , as is done in previous studies @xcite . for longer chains , temperatures higher than @xmath58
were used in order to facilitate the procedure of obtaining initial independent samples .
we note here that the solvent is taken into account in our model only implicitly by tuning the temperature @xcite .
then , decrease of the temperature corresponds to higher incompatibility of the implicit solvent with the monomers . for each cooling history , we lower the temperature in steps of @xmath59 and run the simulations for @xmath60 md steps using the final configuration at each ( higher ) temperature as starting configuration for the next ( lower ) temperature . at low enough temperatures @xcite ( typically @xmath61 , at a rather high distance from the @xmath3 temperature ) , where dense `` clusters '' of a few neighboring blocks are formed ,
it is not possible to run simple md simulations long enough to sample the phase space adequately .
therefore , using this procedure of independent cooling histories is indispensable for obtaining meaningful statistics .
this large statistical effort prevents the study of exceedingly long chains for this simulation model . at temperatures close to @xmath3 , coil - like
structures are formed and the differences between multiblock copolymer chains with different @xmath0 , @xmath7 are rather small , expecting that at @xmath62 these differences will completely disappear . at temperatures below
@xmath44 the chains collapse and blocks of the same type of monomers form clusters with other monomers belonging to different blocks , due to the unfavorable interactions between a and b beads @xcite .
the static properties of these clusters have been described in detail in previous communication @xcite . here , we focus on the static properties of the whole chain ( of length @xmath6 ) as well as the static properties of individual blocks of @xmath7 effective monomers . due to the symmetric choice of our parameters ,
the averages of the static properties for a and b blocks should not differ .
therefore , the calculation for averages of the properties of the blocks should be taken over all blocks @xmath0 , irrespective of their type , and their properties will be denoted with the subscript `` b '' , for example the average value of the gyration radius of individual blocks would read @xmath63 while those for the whole chains without this index , namely as @xmath64 .
we discuss below the properties that might need some more details .
+ we follow the definition of standard textbooks @xcite to discuss aspects related to the persistence length of the polymer chains , where the persistence length is defined from the decay of bond orientational correlations along the chain
. however , one should be careful when defines the persistence length with this correlation function as it has been discussed elsewhere @xcite .
nevertheless , we show in this work that this property is extremely accurate in indicating the appearance of full phase separation of the blocks in symmetric multiblock copolymers .
we give here a brief definition of this property . defining the bond vectors @xmath65 in terms of the monomer positions @xmath66 as @xmath67 , @xmath68 , the bond orientational correlation is defined as @xmath69 note @xmath70 , with @xmath71 being the average bond length .
for this model @xmath72 and hence @xmath73 , of course .
considering the limit @xmath74 and assuming gaussian chain statistics , one obtains an exponential decay , since then @xmath75 $ ] , and thus @xmath76 , l_{p}^{-1}=-\ln\langle cos\theta(1 ) \rangle.\ ] ] for chains at the @xmath3 point @xcite or in melts @xcite @xmath77 below @xmath3 temperature this correlation function falls - off rapidly at small @xmath78 depending strongly on temperature @xmath11 . in our study
we also consider properties related to the shape of multiblock copolymers .
we follow the description of theodorou and suter @xcite to define the apshericity and acylindricity of a mutliblock copolymer chain .
first , we define the gyration tensor as @xmath79 , % \]\ ] ] where @xmath80 is the position vector of each bead , which is considered with respect to the center of mass of the beads @xmath81 , and the overbars denote an average over all beads @xmath82 . when the gyration tensor of the whole chain is considered , then @xmath83 . for the blocks @xmath84 , the gyration tensor and the properties
are calculated for each block separately and then an average over all blocks irrespective of their type is taken .
as mentioned earlier , this is perfectly valid due to the symmetry of our model .
the gyration tensor is symmetric with real eigenvalues and a cartesian system that this tensor is diagonal can always be found , i.e. , @xmath85 where the axes are also chosen in such way that the diagonal elements ( eigenvalues of @xmath86 ) @xmath87 , @xmath88 , and @xmath89 are in descending order ( @xmath90 ) .
these eigenvalues are called the principal moments of the gyration tensor . from the values of the principal moments
, one defines quantities such as the asphericity @xmath91 , @xmath92 when the particle distribution is spherically symmetric or has a tetrahedral or higher symmetry , then @xmath93 .
the acylindricity @xmath94 @xmath95 is zero when the particle distribution is in sync with a cylindrical symmetry .
therefore , the acylindricity and asphericity are the relevant quantities that would describe geometries that could be relevant in the case of a homopolymer or multiblock chain .
these quantities are taken with respect to @xmath96 , to the sum of the eigenvalues , i.e. the square gyration radius of the chain , which we also have calculated independently on our original cartesian coordinates in order to check our results .
assuming an ellipsoidal shape and based on the calculation of the above eigenvalues of the gyration tensor , one can define an effective volume expressed by the following relation @xcite @xmath97{3 } \prod_{i=1}^{3 } \sqrt[]{\lambda_{i}},\ ] ] where @xmath98 , @xmath99 , and @xmath100 , are the eigenvalues of the radius of gyration tensor .
then , the effective radius of a sphere with the same volume as this ellipsoid ( @xmath101 ) is given by the geometrical mean of individual radii @xmath102{3 } \prod_{i=1}^{3 } \sqrt[6]{\lambda_{i}}$ ] .
this can be compared with the volume of an effective sphere defined by the gyration radius @xmath103{\sum^{i=1}_{3 } \lambda_{i}}$ ] .
such an analysis is not in the scope of the present study .
another property that provides information on the microscopic properties of multiblock chains is the number of contacts between beads of the same or different type .
smaller number of contacts a - b suggests a tendency towards phase separation .
the number of contacts can be mathematically described by the following formula @xcite @xmath104 where @xmath105 is the absolute value of the distance between two sites of monomers @xmath106 in the multiblock copolymer chain , and @xmath107 the corresponding radial distribution function .
equation ( [ eq:13 ] ) means that a pair of monomers [ @xmath21 , @xmath108 , @xmath109 is defined to have a pairwise contact if their distance is less than @xmath110 . for the distance @xmath110
we have used the standard stillinger @xcite neighborhood criterion for monomers .
we followed the standard choice @xmath111 and checked that qualitatively very similar results were obtained if one chooses @xmath112 a bit smaller than this choice ( larger values of @xmath112 are physically hardly significant , since then the particles are too weakly bound , due to the rapid fall - off of the lj potential ) .
therefore , a pair of monomers being an absolute distance less than @xmath113 apart define a `` contact '' .
then , the numbers presented in this manuscript will denote the average number of neighbors per monomer .
+ + + +
firstly , we discuss properties related to the acylindricity and asphericity of the chains . the acylindricity defines the deviation of the shape of the chain from a cylindrical geometry ; correspondingly the asphericity expresses the deviation from a spherical shape or tetrahedral or higher symmetry with respect to the total dimensions of the chains .
then fig .
[ fig2]a presents the dependence of the acylindricity of individual blocks on average for multiblock copolymer chains of the same total length @xmath114 on the block length @xmath7 .
thus , the number of blocks for each chain is @xmath115 .
this choice for the total length of the multiblock chain allows for the observation of all interesting regimes related to the phase behavior of such macromolecules @xcite . in fig .
[ fig2]a one can observe very clearly a regime for intermediate values of @xmath116 where the curves show small deviations from one another .
the cases of @xmath117 and @xmath118 are particular . for @xmath118 (
not shown ) the acylindricity drops to zero . at lower temperatures
larger deviations for higher values of @xmath7 are seen .
nevertheless from such a plot we are not able to find any hint that relates to the well - known behavior of these systems @xcite . looking at the dependence of the above systems with different @xmath7 on the temperature @xmath11 ( fig .
[ fig2]b ) it is clearer that small blocks show very small dependence on the temperature due to the very small size of the blocks ( @xmath119 ) , as it also happens in the plot of fig .
the graph of fig .
[ fig2]b indicates that for small values of @xmath7 , i.e. @xmath120 , the blocks rather do not show any dependence on the temperature , while for @xmath121 ( @xmath122 ) , stronger dependencies on the the temperature @xmath11 are seen .
this suggests that stronger deviations from a cylidrical symmetry persist for multiblock chains of higher @xmath7 , as @xmath6 remains constant .
this behavior is also related to the behavior of the size of the cluster formations in these systems , which has been discussed elsewhere @xcite . from the results of fig .
[ fig2 ] we do not see any direct correlation to the phase behavior of the multiblock chains by examining the shape of the individual blocks . on the other hand , the asphericity of the whole multiblock chain of total length @xmath114 shows interesting behavior when a plot with the temperature is attempted .
we remind the reader that the total chain length @xmath114 is high enough to consider all the different cases of phase behavior of symmetric multiblock copolymers , which have already been discussed in detail @xcite .
figure [ fig3 ] then illustrates two different regimes . at rather high temperatures ( @xmath123 ) the different multibock chains ( of the same total length @xmath6 )
exhibit small differences as we are close to @xmath3 conditions .
such small differences are related to the number of proxima a - b interactions at the junctions that connect the consecutive blocks a - b blocks .
the values on the @xmath124-axis of this plot indicate strong deviation of the chain s shape from a spherical one , due to the excluded volume interactions in combinations with the incompatibility between monomers of type a and b at the junctions of a and b blocks . at high values of @xmath7 ( small @xmath0 except for the case @xmath125 which will be separately discussed below ) we observe strong differences at low temperatures .
the case @xmath126 reaches at temperature close to @xmath127 a plateau - like regime .
we can even observe a slight increase as temperature decreases and the incompatibility of a and b monomers increases .
as @xmath0 increases ( @xmath7 decreases ) this plateau - like dependence appears correspondingly at lower temperature @xmath11 .
knowing already the phase behavior of these systems @xcite , these temperatures correspond to those that the multiblock copolymer chain separates in two clusters with a single a - b interface forming between them , where each cluster contains only beads of one type with a permanent interface a - b separating the two clusters .
the transition of multiple a - b interfaces to a single a - b interface becomes also noticeable in the behavior of asphericity , for high @xmath0 , i.e. @xmath120 , where this plateau now has disappeared .
therefore , the phase behavior of the multiblock chains is nicely reflected to a quantity that describes the overall formation of the chain , without taking into account the measurement of any microscopic quantities which relate to the incompatibility between a and b beads .
for the case @xmath125 this transition to the phase separated clusters can not be concluded from the results of fig .
[ fig3 ] , because in this case the chain always has only these two clusters of a and b beads either in a coil like formation at higher temperatures or a globule - like formation at lower temperatures . in this case of
@xmath125 we rather see a small decrease in the quantity @xmath128 , as the chain obtains a globular formation
. also , the values of @xmath128 for @xmath125 are higher than the other cases indicating the stronger presence of a - b interface . in this case
a smaller number of unfavorable contacts a - b exist ( fig.4b of ref .
@xcite ) . for all @xmath0 shown in fig .
[ fig3 ] a more spherical formation is obtained at lower temperatures . in particular this
spherical symmetry is strongly present for the multiblock chains of small @xmath7 , where also we do not have the appearance of this large a - b interface and therefore the curves vary smoothly with the temperature @xmath11 .
the above description is valid for all cases where full phase separation occurs without restrictions on the length @xmath7 or number of blocks @xmath0 and reflects totally the full phase separation behavior of multiblock copolymers .
+ + now we turn our focus on acylindricity and asphericity for the blocks and the whole multiblock chains on their dependence with the block length @xmath7 and the number of blocks @xmath0 , so without the fact that results correspond to chains of the same total length @xmath6 .
figure [ fig4]a shows such results for the average acylindricity of individual blocks and its dependence on the block length @xmath7 for a multiblock chain containing @xmath129 blocks .
it has been shown that for this number of blocks phase separation in two clusters a and b is taking place for high enough block lengths ( @xmath8 ) following our simulation protocol .
however , the plot of this quantity for individual blocks does not reflect the different phase separation behavior for such systems . at temperatures close to the @xmath3 ,
i.e. @xmath58 , an almost straight line is seen suggesting that chain blocks appear with the same overall formation irrespective of their block length @xmath7 . at lower temperatures down to @xmath130 deviations
are seen from the above behavior and the chains with higher @xmath7 show a higher dependence on the temperature @xmath11 .
larger blocks tend to obtain a more cylindrical - like shape as the temperature decreases at @xmath130 , something that is seen for all cases .
however , it becomes clear that even for small @xmath7 the increase of the number of blocks @xmath0 favors a more cylindrical shape of individual blocks on average . in the case that the chain length @xmath7 is high enough that full phase separation can take place ( @xmath131 , fig .
[ fig4]b ) , i.e. , one cluster contains the beads of all blocks of type a and another cluster beads of all blocks of type b , variations due to temperature can become now more apparent . for @xmath131
we observe the same tendency that is seen for @xmath132 , but the chains are long enough to observe the differences attributed to different temperatures .
this can also be concluded by examining the @xmath0 dependence of multiblock chains with various block lengths @xmath7 .
the graphs of fig .
[ fig4 ] show that the shape of individual blocks is affected by the number and block of lengths , but phenomena of phase separation can not be identified on an individual block - based analysis as a function of block length @xmath7 or number of blocks @xmath0 .
+ + + plots showing the overall asphericity of multiblock chains as a function of @xmath7 or @xmath0 ( fig .
[ fig5 ] ) neither do reveal any entry to the phase separation regime of multiblock chains and the behavior is similar to what is seen for the average properties of the individual blocks .
asphericity shows an independence on @xmath0 and @xmath7 for temperatures close to @xmath3 ( i.e. @xmath58 ) , while @xmath128 decays with @xmath7 or @xmath0 obtaining a more globular formation with decreasing temperature and increasing @xmath0 or @xmath7 , indicating the entry to the regime where the multiblock chain collapses in a poor solvent .
showing the corresponding plots of fig .
[ fig4 ] for the asphericity of the blocks ( fig .
[ fig6 ] ) , we draw the same conclusions
. however , the asphericity of the blocks shows a more pronounced relation to the entry of the globule regime .
higher block lengths exhibit a higher dependence on the change of temperature and decreasing the temperature from @xmath133 to @xmath130 ( fig .
[ fig6]a ) the slope of the curve does not change , implying that regarding asphericity of individual blocks there is a stronger dependence on @xmath7 , rather than @xmath0 ( figs .
[ fig6]b ) . as in the case of fig .
[ fig3 ] where we have illustrated the asphericity of a multiblock chain as a function of temperature , the phase behavior of multiblock copolymer chains can be perfectly reflected to other properties , as for example the orientational correlation function of eq .
( [ eq:5 ] ) .
then fig . [ fig7 ] shows examples of this quantity for different cases . for a system not exhibiting full phase separation this orientational correlation ( not shown here ) for high values of distance @xmath78 ( defined in number of beads along the chain )
is usually neglected , since in this range of @xmath78 we rather observe random scattered points indicating a loss of correlation . however , as we will see this range of @xmath78 could provide very important information in the case of multiblock copolymer chains , when a single interface between blocks a and b forms for various @xmath0 and @xmath7 at low temperatures @xmath11 . for @xmath78 close to zero ( fig .
[ fig7 ] ) one is not able to fit any exponents , rather there occurs a very fast decay , which is more pronounced at lower temperatures .
it is remarkable that this initial decay is independent of the varied parameters ; an obvious interpretation is that in the scale of a few subsequent bonds the chain maintains a high local flexibility .
we focus now on our aim which is to relate the phase behavior to the bond orientational correlations , which is very well shown in figs .
[ fig7]a and b. figure [ fig7]a shows results for a multiblock copolymer chain with @xmath134 . at temperatures
@xmath135 and @xmath136 we observe peaks which are repeated periodically , as @xmath78 increases .
we can clearly observe four different peaks , which correspond to the @xmath137 consecutive a - b blocks respectively .
the centers of these peaks correspond exactly to an integer multiple of the total length of an a and b block , which are joint together one after the other .
these peaks appear for the range of temperatures that a complete phase separation of a and b blocks takes place ( a single interface between a and b blocks is formed ) .
then the values at small @xmath78 ( @xmath138 ) correspond to correlations of bonds within the length of a block a and b. again , the orientational correlation function is able to provide very good information for intermediate and high @xmath78 for the phase behavior of multiblock copolymer chains under poor solvent conditions . figure [ fig7]b shows results for the case @xmath139 at @xmath140 for chains of different block length @xmath7 . at this temperature , systems that exhibit a full phase separation between blocks
a and b show symmetrical peaks at positions @xmath141 , where @xmath142 takes values @xmath143 .
the number of contacts of beads does not also reveal directly any correspondence to the phase behavior of multiblock copolymers ( fig .
[ fig8 ] ) , although these numbers are a direct result of the microscopic interactions between monomers .
the number of contacts a - a ( which is equal to the number of contacts b - b due to the symmetry of our model ) decreases monotonically with increasing temperature as beads come apart , but with a lower rate for temperatures @xmath123 , as we can see from fig . [ fig8]a .
we have found @xcite that this is the temperature that the chain enters the collapsed state , i.e. effects due to the poor solvent are becoming very important until the chain obtains a globular formation at even lower temperatures . for different number of blocks
@xmath0 , this change in the slope sets in at different temperatures . of course , for symmetric multiblock copolymer chains of the same total length @xmath114 and increasing @xmath0 it is expected that the number of contacts a - a will become lower .
increasing the number of blocks and joining them one after the other , the number of contacts a - b increases .
this is better shown in fig .
[ fig8]b . counting the number of contacts between beads irrespective of their type ( fig .
[ fig8]c ) , we can see that the increasing number of blocks @xmath0 for chains of the same total length @xmath6 results in smaller amount of contacts between beads .
this is obviously attributed to unfavorable contacts a - b ; beads a and b prefer to be apart , reducing in this way the number of contacts @xmath144 for chains composed of the same amount of beads a and b. again , such plots do not provide a clear information on the regime that this phase separation between blocks in a multiblock copolymer chain takes place .
we rather see a change in the slope of the curves upon entering the collapsed state , and the results of fig .
[ fig8 ] suggest that any transitions take place gradually .
as the bond orientational correlation has shown a direct correlation on the phase behavior of a multiblock chain , we have also looked at other properties as a function of @xmath0 , @xmath7 , and @xmath11 , as it is for example the dihedral angle @xmath145 which is formed by four consecutive monomers and taking the average of this quantity along the chain for all @xmath146 tetrads disregarding in this way effects attributed to different sequences along the chain . as the orientational parameters for small correlation distances @xmath78 in terms of monomer count ( e.g. @xmath147 ) has shown , the chain keeps its local flexibility irrespective of the interactions a , b , or even the chain architecture @xcite .
therefore , small differences in @xmath145 are expected with variation of the parameters @xmath0 , @xmath7 , and @xmath11 in the case of multiblock copolymers , as it also appears in fig . [ fig9 ]
. then , fig .
[ fig9]a illustrates the behavior of @xmath145 with temperature @xmath11 .
increase of the temperature induces a slight increase in @xmath145 for all chains with different number of blocks , that is a tiny local stretching of the chains .
the increase in the number of blocks @xmath0 shows that @xmath145 also increases , but this change is rather small due to reasons that have already been discussed . in this case
all chains have the same total length @xmath114 . if we choose a rather intermediate block length ( as for example in fig .
[ fig9]b , @xmath148 ) , we can see the dependence on @xmath0 .
however , in this case and for @xmath120 , @xmath145 has reached a plateau . for small number of @xmath0
there is a sharper increase of @xmath145 with @xmath0 , which is also more pronounced at lower temperatures and in the range @xmath149 .
but , this is not the whole story , @xmath145 depends also on the block length @xmath7 ( see fig . [
fig9]c ) . here
we show an example for @xmath150 , which corresponds to a point belonging to the plateau regime of part ( c ) of fig .
@xmath145 provides interesting information on the local behavior of multiblock copolymer chains , with small differences arising by variation of the parameters and no effects related to the phase behavior of the multiblock copolymers can be found , as expected , since @xmath145 is a particularly local property of the chain .
figure [ fig10]a shows the dependence on @xmath7 for a choice of intermediate number of blocks @xmath150 for different temperatures . differences due to temperature are expected to increase as the block length @xmath7 increases .
we see that small differences result from the variation of @xmath7 for different temperatures .
similar behavior we see for other values of @xmath0 .
interestingly the size of individual blocks also depends on the number of blocks @xmath0 ( fig .
[ fig10]b ) at lower temperatures , but at higher temperatures the gyration radius of the blocks does not change with @xmath0 .
again , these are results for an intermediate value of block length @xmath148 , where a full phase separation between a and b blocks can take place at low temperatures .
one could argue that the number of blocks does not play any role when the chain has a coil formation , while in the collapsed state a @xmath0-dependence occurs at low number of blocks @xmath0 .
however , a connection of phase separation or a partial phase separation between blocks is not seen for this property of individual blocks by examining all relevant cases and possible variations of parameters .
nevertheless , it shows us that the dependence on @xmath0 of the individual size of the blocks seen under poor solvent conditions , is smeared out at high number of blocks @xmath0 .
similar effects are seen for other choices of @xmath7 . unfortunately , increasing the number of blocks @xmath0 for the interesting range of @xmath7 and apply further exploration of this phenomena is rather limited in the present simulations . assuming a rather small number of blocks @xmath0 , and knowing that full phase separation is favored by a small number of blocks , we plot the squared average gyration radius of the whole multiblock chain ( fig . [ fig11]a ) . at temperatures close to @xmath3 one
can see a straight line ( mind the double logarithmic scale ) , as expected .
deviations from this behavior are seen for lower temperatures .
the dependence on the number of blocks @xmath0 is shown in fig .
[ fig11]b . in both figures
we can rather see the same behavior for variation of @xmath0 and @xmath7 , showing that the overall dimensions of the multiblock copolymer chain rather do not depend very much on the way that the blocks are distributed along the chains .
of course , increasing the number of blocks favors a more elongated chain at lower temperatures , due to the higher number of unfavorable interactions between a and b monomers .
similar conclusions are drawn by examining all combinations of @xmath0 , @xmath7 , and @xmath11 .
in this manuscript we have presented results for a single multiblock chain and its individual blocks varying parameters such as the block length @xmath7 , the number of blocks @xmath0 , and the temperature @xmath11 .
the phase behavior of multiblock copolymer chains in a dilute solution has been discussed in detail elsewhere @xcite , as well as the static properties of the clusters composed of blocks a or b @xcite , which are formed due to the microphase separation of monomers a and b. in this manuscript we have tried to make a connection between the phase behavior of these macromolecules and the static properties of the chain and of the individual blocks on average .
we have shown that properties related to the chain as a whole can reveal significant information on the phase behavior of multiblock copolymers , while study of the individual blocks does not provide such information .
the presence of a single a - b interface between blocks of different type play the major role in the behavior of multiblock chains .
the effect of the presence of this interface is reflected to properties such as the bond orientation correlation function and the asphericity of the whole multiblock chains . in particular , we have shown that correlations for intermediate and high distances @xmath78 ( measured in number of monomers ) along the chains for the orientation correlation function which , is generally not interesting for a homopolymer chain , can show very interesting effects due to the presence of a well defined a - b interface in the case of multiblock copolymers .
furthermore , it is remarkable that the study of properties of the individual blocks can reveal dependence on parameters such as the number of blocks @xmath0 as the chain gets into the collapsed state .
this dependence is lost for higher number of blocks .
phase separation that is due to the microscopic interactions between monomers and the incompatibility of a and b monomers can be mainly measured with properties related to the chain as a whole , or on the basis of a cluster analysis for multiblock copolymer chains , as was done previously @xcite .
the results presented in this manuscript , in combination with recent results in the phase behavior of multiblock copolymer chains , give a detailed description for these systems , at least for the most symmetrical case , where an even number of blocks of a and b type monomers alternate along a linear multiblock chain with @xmath151 being the lengths of blocks a and b respectively .
it is interesting to see that the change of composition of such symmetrical chains can lead to different scenarios of phase separation . introducing any kind of asymmetry in our systems or the consideration of an additional type of monomers in the structure of multiblock chains could result in different and interesting behavior for these systems .
the use of a standard generic coarse - grained model to describe our chains could act as a reference system for the study of these complicated systems .
moreover , the behavior of multiblock copolymers can be parallelized to that of various biological macromolecules which are formed by periodically repeated chemical units ( `` monomers '' ) along their chain .
a detailed discussion on the properties of multiblock copolymer chains using a simple model for a variety of parameters @xmath0 , @xmath7 , and @xmath11 is useful to understand the behavior of these chains , before any other effects , due to , e.g. charges , more complicated interactions , architecture , etc .
, come into play .
we envisage that the present contribution could also stimulate further work on these topics .
p.e.t . is grateful for financial support by the austrian science foundation within the sfb vicom ( grant f41 ) .
he also acknowledges financial support through a max planck fellowship awarded by the max planck institute for polymer research at an initial stage of this work . | we use a standard bead - spring model and molecular dynamics simulations to study the static properties of symmetric linear multiblock copolymer chains and their blocks under poor solvent conditions in a dilute solution from the regime close to theta conditions , where the chains adopt a coil - like formation , to the poorer solvent regime where the chains collapse obtaining a globular formation and phase separation between the blocks occurs .
we choose interaction parameters as is done for a standard model , i.e. , the lennard - jones fluid and we consider symmetric chains , i.e. , the multiblock copolymer consists of an even number @xmath0 of alternating chemically different a and b blocks of the same length @xmath1 . we show how usual static properties of the individual blocks and the whole multiblock chain can reflect the phase behavior of such macromolecules .
also , how parameters , such as the number of blocks @xmath0 can affect properties of the individual blocks , when chains are in a poor solvent for a certain range of @xmath0 .
a detailed discussion of the static properties of these symmetric multiblock copolymers is also given .
our results in combination with recent simulation results on the behavior of multiblock copolymer chains provide a complete picture for the behavior of these macromolecules under poor solvent conditions , at least for this most symmetrical case .
due to the standard choice of our parameters , our system can be used as a benchmark for related models , which aim at capturing the basic aspects of the behavior of various biological systems . |
in the seminal paper @xcite , roger davies and collaborators showed that early - type galaxies ( etgs ) have divers kinematics , ranging from fainter and rotational supported to brighter and dispersion dominated systems .
this picture is effectively still with us ( e.g. * ? ? ?
* ; * ? ? ?
* ) , although the focus has somewhat changed . in particular , the usage of integral field spectrographs and the analysis of kinematic maps enable a more detailed analysis of the internal structure of galaxies.=-2 the atlas@xmath0 project @xcite is a survey of morphologically selected etgs observable with the sauron integral - field spectrograph @xcite mounted on the william herschel telescope . the galaxies selected were brighter than -21.5 magnitude in the 2mass k@xmath1 band and within 42 mpc in order to properly exploit sauron s short wavelength range . out of about 900 such galaxies ,
260 etgs were selected , based purely on their morphology .
the sample galaxies ( or their subsets ) were observed with sauron , iram 30 m single - dish telescope , carma array , westerbork radio synthesis array , the isaac newton telescope ( int ) and the megacam at the canada - france - hawaii telescope . in this contribution , i will review the analysis of the sauron kinematics maps and of ground and space based imaging , looking for evidence that can be used to constrain the assembly processes of atlas@xmath0 galaxies .
etgs can be divided into fast and slow rotators @xcite , a division which to the first order follows the brightness ( or mass ) @xcite .
the distinction between fast and slow rotators is empirically defined in the space of the specific angular momentum calculated within one effective radius , @xmath2 , and the observed ellipticity ( also estimated within one effective radius ) , such that @xmath3 .
this empirical division follows another empirical finding that the velocity maps of etgs either exhibit regular or non - regular velocity features @xcite .
the regularity of a velocity map is measured using kinemetry @xcite , a generalisation of the method of isophotometry ( e.g. * ? ? ?
* ) to higher order moments of the line - of - sight velocity distribution ( losvd ) .
for the velocity maps , the best fitting ellipse is parameterised by @xmath4 , where @xmath5 is the amplitude of rotation and @xmath6 is the eccentric anomaly .
deviations from the cosine law , parameterised with the higher order fourier harmonics , can be quantitatively measured @xcite and used to separate regular velocity maps , such as the ` spider ' diagrams of the discs galaxies , from non - regular , complex looking velocity maps , often containing kinematically distinct components , strong velocity twists or even no mean rotation .
@xcite showed that more than 80% of atlas@xmath0 galaxies have regular velocity maps , which , using the criteria for distinction between fast and slow rotators , translates into 86% of fast rotators.=-2 the velocity maps can provide one further evidence of the remarkable regular structure found in the majority of etgs .
following @xcite , we can define the _ kinematic misalignment angle _
, @xmath7 , as the angle between the photometric major axis and the orientation of the velocity map .
the kinematic misalignment angle is a powerful tool to constrain the shape of galaxies , as when @xmath7 is different from zero , the galaxy can not have an axisymmetric shape . in @xcite
, we measured the photometric position angle at large radii ( @xmath8 effective radii ) and the kinematic position angle within the sauron velocity maps ( typically within one effective radius ) and found that galaxies with regular velocity maps ( fast rotators ) are typically nearly axisymmetric systems ( their @xmath9 ) . on the other hand ,
galaxies with non - regular velocity maps ( slow rotators ) show strong kinematic misalignments ( up to @xmath10 ) , suggesting many of them are triaxial systems .
are the regular velocity maps and the axisymmetric structure of the vast majority of etgs a consequence of hidden discs ?
what fraction of the etg population actually can be considered disc dominated galaxies ?
the first step in answering these questions can be achieved by analysis of images . in @xcite
, we decomposed the surface brightness profiles , extracted from the sdss @xcite and int @xcite images , by fitting two component models consisting of a @xcite fitting function representing the bulge , and an exponential function , representing a disc . from this decomposition we excluded galaxies with ( obvious ) bars , as they would require a three component model and bars live in discs .
the results show a general correspondence between fast / slow division and the existence of discs : discs are typically found and , actually , dominate fast rotators .
light profiles of some slow rotators , however , suggest embedded and small discs could be found in this population as well .
a remarkable result is that the distribution of the disc - to - total light ratios ( d / t ) in fast rotators covers essentially the full range , from small disc contributions to fast rotators which are essentially lacking a bulge - like component).].=-2 due to the deprojection effects , finding discs in the photometry is essentially degenerate @xcite .
it should be , however , more likely to determine the internal structure from kinematics . to demonstrate this , i will here make use of the third gauss - hermite coefficient , @xmath11 , in the parametric expansion of the losvd ( e.g. * ? ? ?
this parameters measure the asymmetric deviations of the losvd from a gaussian and it was shown to tightly anti - correlate with the mean velocity of the same losvd ( e.g. * ? ? ?
this anti - correlation is a direct consequence of the existence of fast moving stars , as one would expect in an embedded disc . in galaxies which have a significant disc component or have a single srsic profile of a low ( @xmath12 ) index , the anti - correlation produces a specific shape in the @xmath13 , as shown on fig .
[ f : fig ] ( top left ) .
the same general shape is found in fast rotators , but a very different shape ( no anti - correlation ) is seen for slow rotators ( fig .
[ f : fig ] ; bottom left ) .
the implication is simple : fast rotators are indeed disc dominated systems .
discs , whether large or small and embedded , are present in the majority of etgs , and support a new classification scheme @xcite , which should substitute the classical hubble diagram .
= -2 relation for all atlas@xmath0 galaxy spectra with @xmath14 km / s and an error on @xmath15 .
top panel shows the contours for galaxies with significant disc contribution or galaxies with single srsic profiles of a low srsic index ( dashed ) .
bottom panel shows the distribution for fast and slow rotators ( dashed ) .
_ right : _ angular momentum vs ellipticity for the atlas@xmath0 sample .
the green solid line separates fast ( above ) and slow ( below ) rotators .
various symbols show the properties of nuclear light profiles , as stated on the legend .
, title="fig:",scaledwidth=38.0% ] relation for all atlas@xmath0 galaxy spectra with @xmath14 km / s and an error on @xmath15 .
top panel shows the contours for galaxies with significant disc contribution or galaxies with single srsic profiles of a low srsic index ( dashed ) .
bottom panel shows the distribution for fast and slow rotators ( dashed ) .
_ right : _ angular momentum vs ellipticity for the atlas@xmath0 sample .
the green solid line separates fast ( above ) and slow ( below ) rotators .
various symbols show the properties of nuclear light profiles , as stated on the legend .
, title="fig:",scaledwidth=52.0% ]
nuclear surface brightness profiles of massive galaxies are known to show a bend downwards or a deficit of light with respect to a srsic fit at large radii ( e.g. * ? ? ?
* ; * ? ? ?
galaxies with such profiles are often said to have _ cores _ , while galaxies whose profiles stay consistent with the outer srsic fit or even show a nuclear excess of light , we will call here _ core - less_. core galaxies are typically more massive @xcite , as much as are slow rotators and a natural question is whether there is a clear correspondence between slow rotators and cores . in @xcite , we investigated this issue by matching the observations in the hst archives with atlas@xmath16 galaxies .
to a first approximation , fast rotators are core - less galaxies , while slow rotators are cores , but there are some exceptions ( fig [ f : fig ] , right ) .
notably there is a population of fast rotators with cores .
they are kinematically consistent with being fast rotators with large discs ( e.g. regular kinematics , @xmath17 , signifiant @xmath13 anti - correlation ) , but have @xmath18 and they are not far from the line separating fast and slow rotators .
there are also some core - less slow rotators , while for a number of slow rotators their nuclear profiles are not know .
even if one removes some special cases among slow rotators , it is obvious there is no 1:1 correspondence between cores and slow rotators and core - less and fast rotators.=-2
what are core - less slow rotators and how are they made ?
how is it possible to make , or preserve , a core in a disk dominated fast rotator ?
these types of galaxies clearly require special formation paths .
if the cores are created by the merging black holes of similar masses ( brought together by a major merger event ) , while the core - less ( light excess ) profiles are made in nuclear starburst events , a fine tuning between the duration of the starburst and the binary evolution is necessary for the creation of both core fast rotators or core - less slow rotators . creating the cores in fast rotators will require a longer evolution of the bh binary than the duration of the starburst , while the cores in the slow rotators need to be refilled by new material after the black holes are merged .
this means that slow rotators can be produced in dissipative , while fast rotators in dissipation - less merges .
there are indication that this is indeed the case @xcite , but simulating the observed galaxies still requires some work.=-2
i thank the organisers for a very interesting meeting and the members of atlas@xmath0 for making this project a success . | in this contribution i show that by combining imaging and integral - field spectroscopy it is possible to unravel the internal structure of galaxies .
in particular , i will present the photometric and kinematic evidence for discs linking them with stellar angular momentum content of early - type galaxies .
furthermore , i show that the existence of both fast rotators with core and slow rotators with core - less nuclear light profiles challenges the standard formation scenarios for fast and slow rotators and suggests new pathways of mass assembly . |
one key requirement for a starburst is the presence of a reservoir of dense , cold gas that can be efficiently converted to stars . for galaxies entering rich clusters
this is especially important , because they are expected to be affected by mechanisms that can remove cold gas from the haloes and discs of infalling galaxies ( e.g.ram-pressure stripping , gunn & gott 1972 ) or prevent further cooling of gas within galaxies dark matter halos ( starvation or strangulation , e.g. larson , tinsley & caldwell 1980 ; bekki , couch & shioya 2002 ) .
this environmental dependence has a profound influence on cluster galaxies evolutionary histories , the net effect of which is the eventual termination of star formation ( there is virtually no residual star formation in the cores of local clusters ) . the observational evidence for the gradual truncation of star formation in clusters is the conspicuous disappearance of star - forming disc galaxies in the cores of rich clusters since @xmath42 . it has been proposed that the spiral galaxies in distant clusters must be transforming into passive lenticular ( s0 ) galaxies , since the fraction of s0s in rich clusters is believed to increase at the same time that the spirals begin to vanish ( dressler et al.1997 ) .
however , the detailed nature of the physics controlling this evolution is still poorly understood .
for example , what process is responsible for the transformation of the bulge - to - disc ratio of the spiral population ? ( kodama & smail 2001 ) .
a growing body of observational evidence suggests that distant clusters contain significant populations of dust obscured starburst galaxies in their peripheral regions ( e.g. duc et al .
2000 ; biviano et al . 2004
; coia et al . 2004 ; geach et al . 2006 ; marcillac et al.2007 ; bai et al .
2007 ; dressler et al . 2008 ; koyama et al . 2008 ; fadda et al .
geach et al .
( 2009 ) proposed that the population of luminous infrared galaxies ( lirgs ) residing in the ` infall ' population of @xmath43 clusters could be examples of spiral galaxies undergoing an episode of bulge growth via circumnuclear starburst , and therefore excellent candidates for the progenitors of local massive s0s .
geach et al .
( 2009 ) estimate that , given the cluster lirgs current sfrs , 10@xmath44 of stars could be built up within a few 100myr .
although this simple model suggests that these galaxies could evolve onto the bright end of the local cluster s0 luminosity function by @xmath45 , we still lack crucial observational constraints , such as the likely duration of the starburst , where it occurs in the cluster , and limits on the final stellar mass of the bulge . in this letter
we refine our model by asking a simple question : what is the mass of the cold molecular gas reservoir in these starbursts ? using the iram plateau de bure interferometer
we have conducted a pilot study of the cold gas properties of two dusty starburst galaxies in the distant cluster cl0024 + 16 .
the initial results are presented here . throughout
we assume @xmath46 in units of 100kms@xmath7mpc@xmath7 , @xmath47 and @xmath48 .
the projected scale at @xmath4 in this model is 5.3kpc/@xmath25 , and the luminosity distance is @xmath49mpc .
the two galaxies chosen for this pilot study were selected as two of the brightest in our _ spitzer space telescope _ infrared spectrograph ( irs ) survey of 24@xmath1m - selected members of cl0024 + 16 ( @xmath4 ) .
full details of the mid - infrared observations can be found in geach et al .
( 2006 & 2009 ) .
both galaxies have @xmath50mjy , and implied far - infrared luminosities @xmath51 ; they are in the lirg class .
the far - infrared luminosity is estimated from the luminosity of the 7.7@xmath1 m polycyclic aromatic hydrocarbon ( pah ) emission line , extrapolating from the tight correlation between pah strength and total infrared luminosity observed in local star forming galaxies ( smith et al.2007 ; geach et al .
both galaxies are in the outskirts , or infall region , of the cluster : they large clustocentric radius , @xmath52 and @xmath53 , where @xmath54 is the virial radius , @xmath2@xmath55mpc ( treu et al .
further details of the two galaxies is provided in table 1 .
the observations were conducted using the iram plateau de bure interferometer on 31 july 2008 ( mipsj002621.7 ) and 25 august 2008 ( mipsj002721.0 ) as part of program s035 .
the exposure times were 9.5hr and 8.2hr for each source respectively using 5 antennae .
the observing conditions were excellent in terms of atmospheric phase stability , however any anomalous and high phase - noise visibilities were flagged .
data were calibrated , mapped and analysed using the iram gildas software ( guilloteau & lucas 2000 ) .
phase and flux calibration was performed using 3c454.3 and 0007 + 106 .
secondary flux calibrators also included the sources mwc349 , 3c454.3 , 0119 + 115 , 3c84 , 2145 + 067 .
we targeted the co(10 ) 115.27ghz rotational transition , which at @xmath56 is redshifted into the 3 mm band , with @xmath57ghz .
the central frequency of the 3 mm receiver was set to the frequency of the redshifted co(10 ) line at the systemic redshift of each galaxy derived from optical spectroscopy ( table1 ) .
the correlator was set - up with 2.5mhz spacing ( 2@xmath5864 channels , 320mhz bandwidth ) to comfortably cover the co line should it be offset from the systemic redshift , and to take into account a broad velocity profile .
the noise per 18kms@xmath7 channel for the two observations is 1.4mjy beam@xmath7 and 1.3mjy beam@xmath7 .
we detect co(10 ) emission in both galaxies with significances of @xmath60 and @xmath61 respectively .
note that these observations represent the highest redshift co detection in un - lensed lirgs yet achieved ( see melchior & combes 2008 ) .
in figure1 we present velocity integrated emission maps , and the spectra extracted from the peak pixel of these are shown in figure2 .
we estimate the line luminosities by fitting the spectra with gaussian profiles , although in the case of mipsj002721.0 , the broad line is better fit with a double gaussian ( there is a marginal improvement on the @xmath62 fit ) .
we discuss the physical interpretation of the line profiles in 3.2 .
we calculate @xmath21 ( in the usual units of kkms@xmath7pc@xmath28 ) from the integrated line emission according to e.g. solomon & vanden bout ( 2005 ) : @xmath63 .
the line luminosities are given in table1 with 1-@xmath64 errors estimated by repeatedly re - evaluating the line fits after adding noise to the spectra .
the ` noise ' is randomly drawn from a gaussian distribution with a width equivalent to the variance in the data sampled in the extreme wings ( @xmath65kms@xmath7 ) of the line .
we take the standard deviation in the resulting @xmath21 values from 1000 of these ` bootstrap ' realisations as the 1@xmath64 uncertainty in the measured value .
it is assumed that the co luminosity linearly traces the molecular gas mass : @xmath66 ( e.g. young & scoville 1991 ; solomon & vanden bout 2005 ) with the conversion factor @xmath67 typically taken as either @xmath68 for ` normal ' spirals ( the galactic conversion factor ) , or @xmath69 for ultraluminous infrared galaxies
( ulirgs , @xmath70 ) . the cluster starbursts are not quite in the ulirg class , but are at the extreme end of what would be considered a ` normal ' ir - bright galaxy ( see gao & solomon2004 ) .
we therefore adopt the more conservative ulirg conversion when estimating the h@xmath59 gas mass ( table 1 ) , and we discuss the implications of this in 3.4 . in this sense ,
the estimated h@xmath59 masses could be considered lower limits .
deep , high - resolution ( [email protected]@xmath25 ) optical imaging of these galaxies ( fig . 1 ,
inset ) allows us to constrain their orientation .
mipsj002621.7 appears to be nearly face - on to the line of sight , with a bright core and distinctive arms .
mipsj002721.0 is slightly more complicated ; an un - sharp masked _
r_-band image reveals a ring of bright knots and a bright core or bar that appears to be slightly extended to the north - east .
it could also be orientated close to face - on .
the co(10 ) spectrum for mipsj002721.0 exhibits a double peaked line profile ( fig .
2 ) , and we show in figure 3 that the peak of the co emission traced by this velocity shear describes a locus that runs @xmath210kpc across the direction of the extended optical emission .
this could be interpreted as a rotating disc , with an angular separation between the velocity peaks of @xmath71 or @xmath72kpc .
note that the peak of the co emission at the [ o ii ] redshift is centred on the bright central core .
an alternative situation is that this galaxy is in the process of a merger along the line - of - sight , and the co(10 ) emission profile originates from emission from two gas - rich clumps .
@xmath733727 emission line is indicated by the vertical dashed lines , and expected to be accurate to @xmath230kms@xmath7 .
the systemic co redshifts agree with the optical redshifts ( czoske et al .
2001 ; moran et al .
2006 ) , although in the case of mipsj002721.0 the broad blue - shifted profile is best fit by a double gaussian with a peak separation of @xmath74kms@xmath7 .
this could be evidence of a rotating disc , or could also be interpreted as a merger .
note that redshifts have been corrected for heliocentric velocity ( see table 1).,scaledwidth=49.0% ] @xmath733727 emission line is indicated by the vertical dashed lines , and expected to be accurate to @xmath230kms@xmath7 .
the systemic co redshifts agree with the optical redshifts ( czoske et al .
2001 ; moran et al .
2006 ) , although in the case of mipsj002721.0 the broad blue - shifted profile is best fit by a double gaussian with a peak separation of @xmath74kms@xmath7 .
this could be evidence of a rotating disc , or could also be interpreted as a merger .
note that redshifts have been corrected for heliocentric velocity ( see table 1).,scaledwidth=49.0% ] finally , we can use the co line widths to estimate the galaxies dynamical masses . for a rotating disc of radius @xmath75 ,
the dynamical mass enclosed within @xmath75 is @xmath76 ( neri et al .
2003 ) . one way of estimating the likely size of the molecular disc is to force these galaxies to fall on the global schmidt - kennicutt relation , connecting star formation surface density to the gas surface density : @xmath77 ( schmidt 1959 ; kennicutt 1998 ) .
assuming the gas masses and sfrs given in table1 , this implies discs of radius @xmath20.4kpc and @xmath21kpc for mipsj002621.7 and mipsj002721.0 respectively .
these values are not unreasonable , given that observations of local lirgs indicate large concentrations of gas in their inner few kpc ( sanders & mirabel 1996 ) . however
, clearly the latter value is at odds with the @xmath210kpc separation of the velocity peaks in mipsj002721.0 , and of course , these radii would further increase if we had adopted a higher co - to - h@xmath59 mass conversion .
assuming @xmath78 for mipsj002621.7 , we find @xmath79 .
this ignores the systematic uncertainty on the inclination angle , which would clearly affect these mass estimates .
the orientation of mipsj002721.0 is similarly uncertain it could be a face - on merger , or an inclined disc . assuming it is a disc of radius 5kpc , with @xmath80 ( appropriate for a random orientation ) , this yields @xmath81 . in a merger model
this estimate would increase by a factor @xmath22 ( neri et al .
2003 ; genzel et al .
2003 ) .
if we assume that the far - infrared luminosity is dominated by dusty star - formation , then the star formation efficiency can be represented by @xmath82 . for these galaxies , we have @xmath83 and @xmath84 .
this is in the range observed in local ( @xmath85 ) spirals , which have @xmath86 ( solomon & vanden bout2005 ) .
in contrast , much larger values of @xmath87 are observed in ulirgs and smgs over a wide range of redshift ( @xmath88 ) implying higher sfes , perhaps driven by more violent mergers and interactions ( downes , solomon , & radford 1993 ; tacconi et al .
2008 ; coppin et al . 2008 ) .
recently , daddi et al . ( 2008 ) have presented evidence for similar ` low - efficiency ' star formation in two _ bzk_-selected galaxies at @xmath89 .
those authors argue that while the most luminous galaxies such as ulirgs and smgs are indeed special cases of intense , high - efficiency star - formation , the _ general _ case could be a less efficient ( yet still vigorous ) mode of star - formation fuelled by a large reservoir of gas .
unfortunately a lack of co observations of lirgs at intermediate redshift makes such comparisons challenging .
more extensive surveys in the alma era will no - doubt dramatically improve our understanding of this issue .
given our measured gas mass , we can estimate the gas depletion timescale from the galaxies current sfrs ( table 1 ) . assuming the activity continues at the observed rate , and that all the gas is converted to stars , the lifetime of the starbursts are @xmath2100myr and @xmath2160myr for mipsj002621.7 and mipsj002721.0 respectively .
note that this is much shorter than the time it takes for a galaxy on a radial orbit to traverse the distance from the outskirts of the cluster to the core , which is of order 25gyr for cl0024 + 16 ( treu et al .
the implication of these short time - scales is that self - exhaustion of the gas - reservoirs dominates these galaxies star formation histories .
moreover it is only the environment within a few 100kpc of the cluster core that can ram - pressure strip the disc gas ( and thus truncate star formation ) .
the major episode of star - formation for these very active galaxies will be long - since over by the time the galaxies reach the core .
note that this may not be true of the ` general ' star forming population , where lower sfrs mean that gas is not exhausted in such a dramatic way , and allows star - formation in less active systems to endure further into the cluster environment ( e.g. moran et al.2007 ) .
we have treated these galaxies as ulirgs in terms of h@xmath59 mass estimation , but how would our conclusions change if we applied the galactic conversion of @xmath90 ?
the obvious implication is that our derived gas masses would increase by a factor 6@xmath58 , and so the galaxies descendants could have stellar masses of order @xmath210@xmath91 .
if the majority of this additional mass is built in the bulge region , then this further supports a model where spiral galaxies in distant clusters can evolve into some of the most massive s0 galaxies in the cores of local clusters .
the longer gas - depletion timescales ( @xmath2500myr and @xmath2900myr ) are still shorter than the time it takes the galaxies to reach the virial radius in cl0024 + 16 ; our conclusion that the majority of the bulge growth occurs in the outskirts of the cluster still holds .
these galaxies are clearly different in nature to the milky way they are forming stars at @xmath215@xmath58 the galactic rate for example .
although these galaxies are not as extreme as ulirgs , it is likely that their mode of star formation is more similar to ulirgs than the ` quiescent ' mode of star formation that occurs in the discs of local spirals .
it could be driven by a circumnuclear mode where cold gas is being funnelled into the central regions for example .
taking this into account , we assume that the ulirg conversion factor is most appropriate here . _
r_-band images of mipsj002721.0 .
the main morphological details are a ring of knots and a central bright core which is extended to the north - east .
the four images are overlaid with the co(10 ) emission slices taken at 70kms@xmath7 intervals .
the change in peak position represents the velocity shear across the galaxy .
the contours are taken from the cleaned map , and are at integer multiples of the noise , @xmath20.65mjybeam@xmath7 .
the separation of the velocity peaks correspond to @xmath72kpc.,scaledwidth=49.0% ] this pilot study is not extensive enough to make strong conclusions about the potential role of the cluster environment on the gas properties of these galaxies ; a more complete survey is needed .
however we can make some comments about the possible mechanisms that might be operating , the likely fate of these galaxies , and the implications this has for models of galaxy evolution in rich clusters . at their large clustocentric radii ,
these galaxies are not affected by ram - pressure stripping ( although weak pressure of the icm on the discs of the galaxies could promote star formation in the disc ; see bekki & couch 2003 and the discussion in geach et al .
loosely bound gas in the halos of the galaxies could be gradually removed ( moran et al .
2007 ) , but the additional mass such a reservoir could supply in the short - term is probably insignificant compared to the large masses already in the discs and cores .
so , provided the molecular reservoirs are securely bound to the galaxy , the cold gas mass we measure now is likely to contribute to their final stellar masses .
their current stellar masses ( estimated from their @xmath92-band luminosities , see geach et al .
2009 ) are likely to be in the range @xmath93@xmath94 , implying that the current episode of star formation could double their stellar mass ( if the bursts are half - way through their lifetime ) . assuming most of the current star formation occurs in the central regions , then relative fading of the disc component compared to the maturing bulge will contribute to the morphological transformation necessary for them to evolve into a massive s0 ( kodama & smail 2001 ) .
these galaxies appear to be undergoing significant evolution in the outskirts of the cluster , with their activity potentially driven by mergers and interactions .
so how important is the central cluster environment in transforming spirals to s0s , and maintaining the evolution of the morphology - density relationship ?
our observations suggest that the cluster environment is not required to terminate star formation in the progenitors of s0s ; the key environmental effect might be more subtle .
one can speculate on the likely processes that the galaxies will experience as they are virialised ( long after they have used up their gas ) .
first , the lack of accretion of additional gas from a cooling halo will prevent additional star formation .
it is unlikely that the galaxies will experience further mergers , and so they can not acquire cold gas that way ; all that remains is morphological transformation .
if the galaxies are on radial orbits , then high speed passes in the cluster potential and close encounters could promote tidal disruptions or even removal of any remaining spiral disc ( moore , lake & katz 1998 ; gnedin 2003 ) , completing the transformation to s0 .
we have presented new iram co detections of two lirgs in the outskirts of the rich cluster cl0024 + 16 at @xmath4 . from the co(10 )
line luminosities , we measure h@xmath59 gas masses of @xmath95@xmath6 and @xmath96@xmath6 in the two galaxies .
this assumes a conversion between co luminosity and h@xmath59 mass that is applied in ulirgs ; if we adopt the galactic conversion factor , these estimates would increase by a factor @xmath26@xmath58 .
the infrared - derived sfrs are @xmath260@xmath6yr@xmath7 , and so they will exhaust their reservoirs in as little as @xmath2150myr .
our observations hint that dusty starbursts in the outskirts of cl0024 + 16 can assemble @xmath97 of stellar mass long before they reach the cluster core , implying that the cluster environment is unlikely to have a significant influence on the burst .
however , it is interesting to note that one of the galaxies shows evidence of dynamic disturbance , potentially linked to a merger . if the galaxies are destined to evolve into s0s , then the main transformation required after the gas reservoirs have been exhausted is fading of the disc relative to the bulge .
potentially this could be accompanied by some morphological disturbance when the galaxies reach the high density core . on a final note ,
we highlight the fact that if the critical epoch of star formation occurs @xmath212gyr before the galaxies reach the virial radius , and these starbursts are feeding the local s0 population , then there should be a sizable population of post - starburst ( k+a ) galaxies at @xmath2@xmath98 in rich clusters at @xmath990.3 ( couch & sharples 1987 ) .
detection of such an ` intermediate ' population would be a key piece of evidence in the model where starbursts triggered early during cluster infall at high - redshift are rapidly exhausting their gas reservoirs and sinking to the bottom of the cluster potential well where they are destined to complete their evolution into local massive s0s .
we warmly thank roberto neri and philippe salom at iram for their assistance in obtaining and reducing this data .
we also thank mark swinbank for helpful comments .
this work is based on observations carried out with the iram pdbi .
iram is supported by insu / cnrs ( france ) , mpg ( germany ) and ign ( spain ) .
jeg is supported by the u.k .
science and technology facilities council ( stfc ) .
kekc is a stfc fellow at durham university .
irs and rse acknowledge the royal society and stfc . | we present iram plateau de bure interferometer 3 mm observations of co _
j_(1@xmath00 ) emission in two 24@xmath1m - selected starburst galaxies in the outskirts ( @xmath223@xmath3 ) of the rich cluster cl0024 + 16 ( @xmath4 ) .
the galaxies inferred far - infrared luminosities place them in the luminous infrared galaxy class ( lirgs , @xmath5 ) , with star formation rates of @xmath260@xmath6yr@xmath7 .
strong co _ j_(1@xmath00 ) emission is detected in both galaxies , and we use the co line luminosity to estimate the mass of cold molecular gas , @xmath8 . assuming @xmath9 , we estimate @xmath10@xmath11 for the two galaxies .
we estimate the galaxies dynamical masses from their co line - widths , @xmath12@xmath13 , implying large cold gas fractions in the galaxies central regions . at their current rates
they will complete the assembly of @xmath14 and double their stellar mass within as little as @xmath2150myr . if these galaxies are destined to evolve into s0s , then the short time - scale for stellar mass assembly implies that their major episode of bulge growth occurs while they are still in the cluster outskirts , long before they reach the core regions .
subsequent fading of the disc component relative to the stellar bulge after the gas reservoirs have been exhausted could complete the transformation of spiral - to - s0 .
[ firstpage ] clusters : galaxies , clusters : individual : cl0024 + 16 , galaxies : starburst , evolution @lcccccccccc target & @xmath15 & @xmath16 & @xmath17 & @xmath18 & sfr & @xmath19 & @xmath20 & @xmath21 & @xmath22 & ( h m s ) & ( @xmath23 @xmath24 @xmath25 ) & & ( @xmath26 ) & ( @xmath6yr@xmath7 ) & ( kms@xmath7 ) & ( kms@xmath7 ) & ( 10@xmath27 k km s@xmath7
pc@xmath28 ) & ( @xmath29 ) mipsj002621.7 & 00 26 21.7 & + 17 19 26.4 & 0.3803 & @xmath30 & @xmath31 & 144@xmath3213 & @xmath33 & @xmath34 & @xmath35 mipsj002721.0 & 00 27 21.1 & + 16 59 49.9 & 0.3964 & @xmath36 & @xmath37 & 158@xmath3234 & @xmath38 , @xmath39 & @xmath40 & @xmath41 |
trojans of jupiter , which reside in the neighbourhood of @xmath2 and @xmath3 lagrangian points , serve as an important test of the planetary migration theory ( morbidelli et al .
their inclination distribution , namely the large spread of @xmath4 , can be explained as a result of chaotic capture during a brief period when jupiter and saturn encountered a 1:2 mean - motion resonance .
moreover , the late heavy bombardment provides the timing of this resonant encounter @xmath5gyr ago ( gomes et al .
it is thus important to understand the population of trojans accurately .
there are several unresolved problems regarding trojans , however , for example the number of families , which is a stringent constraint for collisional models .
roig et al .
( 2008 ) studied as many as ten suggested families , using relatively sparse sloan data and spectra .
they noted most families seem to be heterogeneous from the spectroscopic point of view , with one exception the c - type eurybates family . as we argue in this paper , the number of families ( with parent - body size @xmath6 ) is indeed as low as one .
another strange fact is the ratio of @xmath2 and @xmath3 trojans .
szab et al .
( 2007 ) used sloan colour data to reach fainter than orbital catalogues and estimated the ratio to @xmath7 .
there is no clear explanation for this , since the chaotic capture as a gravitational interaction should be independent of size or @xmath8 membership .
any hypothesis involving collisions would require a relatively recent disruption of a huge parent body , which is highly unlikely ( obrien and morbidelli 2008 , d. obrien , personal communication ) .
this is again related to the actual observed number of trojan families .
bro and vokrouhlick ( 2008 ) studied another resonant population , the so called hilda group in the 3/2 mean - motion resonance with jupiter , and reported only two families : hilda and schubart with approximately 200 and 100 km parent bodies .
this number might be in accord with low collisional probabilities , assuming the hilda family is very old and experienced the late heavy bombardment ( bro et al .
2011 ) .
levison et al . ( 2009 ) compared the observed distribution of d - type asteroids and the model of their delivery from transneptunian region .
they found a good match assuming the d - types ( presumably of cometary origin ) are easy - to - disrupt objects ( with the strength more than 5 times lower than that of solid ice ) .
note that trojan asteroids are a mixture of c- and d - type objects and we have to discriminate between them with respect to collisional behaviour .
all of the works mentioned above are a good motivation for us to focus on asteroid families in the trojan population .
the paper is organised as follows .
first , we describe our data sources and methods in section [ sec : methods ]
. a detailed study of orbital and physical properties of families ( and other ` false ' groupings ) is a matter of section [ sec : families ] .
section [ sec : evolution ] is devoted to the modelling of long - term dynamical evolution .
finally , there are concluding remarks in section [ sec : conclusions ] .
we use the symplectic swift integrator ( levison & duncan 1994 ) for orbital calculations .
our modifications include a second order scheme of laskar & robutel ( 2001 ) and on - line digital filters , which enable us to compute suitable resonant proper elements : libration amplitude @xmath9 of the @xmath10 oscillations , where @xmath11 is the osculating semimajor axis of an asteroid and @xmath12 that of jupiter , eccentricity @xmath13 and inclination @xmath14 .
( in figures , we usually plot a mean value @xmath15 of semimajor axis plus the libration amplitude @xmath9 . )
we employ their definition from milani ( 1993 ) .
the source of initial osculating elements is the astorb catalogue , version @xmath16 ( oct 31st 2010 ) .
there are actually two independent filters running in parallel : in the first one , we sample osculating elements every 1yr , compute mean elements using the filter sequence b , b with decimation factors 3 , 3 ( refer to quinn , tremaine & duncan 1991 ) a store this data in a buffer spanning 1kyr .
we then estimate the libration frequency @xmath17 by a linear fit of @xmath18 , where @xmath19 , @xmath20 are the mean longitudes of an asteroid and jupiter and @xmath21 for @xmath2 or @xmath3 respectively .
the revolution of angle @xmath22 must not be confined to the interval @xmath23 , of course .
the amplitude of @xmath9 is computed for the already known @xmath17 by a discrete fourier transform .
finally , an off - line running - average filter with a window 1myr is used to smooth the data . of mean longitudes
anyway , there is a linear relation between @xmath9 and @xmath25 . ] in the second filter , we compute proper eccentricity @xmath13 and proper inclination @xmath14 by sampling osculating elements ( 1yr step ) , computing mean elements using a filter sequence a , a , b and decimation factors 10 , 10 , 3 , and then we apply a frequency modified fourier transform ( nesvorn & idlichovsk 1997 ) , which gives us the relevant proper amplitudes .
the values of the resonant elements agree very well with those listed in the astdys catalogue by kneevi & milani ( see figure [ trojans - hcm_20100915_follow.proper.out_dfd_0myr_astdys4_e ] ) .
there are only few outliers , probably due to a different time span of integration .
we computed proper elements for 2647 @xmath2 and 1496 @xmath3 trojan asteroids .
this sample is roughly twice larger than previously analysed .
the ratio of populations valid for @xmath26 asteroids is thus @xmath27 . the overall distribution of trojans in the @xmath28 space is shown in figure [ trojans_l4_arer_sizes ]
note there is only one cluster visible immediately in the bottom - left panel around ( 3548 ) eurybates .
the reason is its tight confinement in inclinations ( @xmath29 ) . to aid a comparison.,width=226 ] [ cols="^,^ " , ] [ 9799_tab ] let us compare trojan clusters to the well known asteroid families in the outer main belt ( figure [ main_belt_families_nv_all ] ) .
most families ( e.g. , themis , koronis , eos ) exhibit a steady increase of @xmath1 until they merge with another families or the entire outer main belt .
eurybates , aneas and 1988 rg@xmath30 are the only trojan clusters which behave in a similar fashion . the veritas family ( dynamically young , nesvorn et al
. 2003 ) exhibits a different behaviour for a large interval of @xmath31 the number of members @xmath1 remains almost the same , which indicates a clear separation from the background population . with respect to the @xmath32 dependence ,
the ennomos cluster is similar to veritas .
dependence for seven outer main - belt families .
if we would consider only a subset of asteroids brighter than @xmath33 , which is an approximate observational limit for trojans , the @xmath32 dependencies would be qualitatively the same , only slightly shifted to larger cut off velocities.,width=302 ] at first , let us assume a single value of albedo for all family members .
it is a reasonable assumption if the family is of collisional origin .
we can then calculate sizes from absolute magnitudes and construct size - frequency distributions .
figure [ l4_size_distribution ] shows a comparison of sfd s for the clusters detected by the hcm is flat .
size - frequency distribution is not very sensitive to this selection anyway . ] and for the whole population of @xmath2 and @xmath3 trojans .
a slope @xmath34 of the cumulative distribution @xmath35 is an indicative parameter . for @xmath2 and @xmath3
trojans it equals to @xmath36 and @xmath37 in the intermediate size range 15 to 60 km .
( these numbers are compatible with the study of yoshida & nakamura 2008 . )
the slope is steeper at large sizes .
the uncertainties are mainly due to a freedom in selection of the size range and the difference between @xmath2 and @xmath3 sfd s does not seem significant .
the clusters have typically similar slope as background ( within 0.1 uncertainty ) , thought sometimes the results are inconclusive due to small number of members .
the eurybates family with @xmath38 slope is on the other hand significantly steeper than the mean slope of the whole trojan population .
] there are two more groups which exhibit a relatively steep slope , namely laertes in @xmath2 ( @xmath39 ) and 1988 rg@xmath30 in @xmath3 ( @xmath40 )
. we should be aware , however , that even the background exhibits a trend with respect to inclinations ( see figure [ trojans - hcm_20090623_trojans_sini.lsm_all ] ) .
slope @xmath34 typically decreases with inclination @xmath14 , which is especially prominent in case of the @xmath2 cloud .
we have to admit if we compare the eurybates family to its surroundings only ( @xmath41 ) , the difference in slopes is not so prominent
. an interesting feature of the @xmath3 cloud is a dip in the interval @xmath42 .
this corresponds to the approximate location of the 1988 rg@xmath30 group .
the @xmath43 dependence among trojans is not unique .
e.g. low - inclination bodies in the j3/2 resonance also have the sfd steeper than background ( @xmath44 versus @xmath45 ) , without any clear family and a few big interlopers
. may be , this feature reflects different _ source reservoirs _ of low- and high - inclination bodies among trojans and j3/2 ?
it may be also in concert with a colour inclination dependence reported by szab et al .
( 2007 ) .
we also test albedo distributions dependent on size , since the measurements by fernndez et al .
( 2009 ) suggested small trojans are significantly brighter and thus smaller .
large asteroids have @xmath46 while small @xmath47 .
this is a significant change of the sfd , which occurs around the size @xmath48 .
the sfd thus becomes shallower below this size , e.g. for eurybates we would have @xmath49 and for l4 trojans @xmath50 , so the sfd s become comparable with respect to the slope . thought , as we stated above , for a real collisional family we expect the albedo distribution to be rather homogeneous and independent of size . of the size - frequency distributions @xmath51 for @xmath2 and @xmath3 trojans and their dependence on the inclination @xmath14 .
the range of diameters for which the sfd s were fitted is @xmath52 , @xmath53 .
thin lines were calculated for different ranges , which were varied as @xmath54 , @xmath55 .
their spread indicates the uncertainty of @xmath34 in a given interval of @xmath14 .
the populations are observationally complete down to @xmath56 , because the characteristic change of slope due to incompleteness occurs at smaller sizes ( see also yoshida and nakamura 2008).,title="fig:",width=302 ] of the size - frequency distributions @xmath51 for @xmath2 and @xmath3 trojans and their dependence on the inclination @xmath14 .
the range of diameters for which the sfd s were fitted is @xmath52 , @xmath53 .
thin lines were calculated for different ranges , which were varied as @xmath54 , @xmath55 .
their spread indicates the uncertainty of @xmath34 in a given interval of @xmath14 .
the populations are observationally complete down to @xmath56 , because the characteristic change of slope due to incompleteness occurs at smaller sizes ( see also yoshida and nakamura 2008).,title="fig:",width=302 ] we used the sloan digital sky survey moving object catalogue version 4 ( sdss - moc4 ) to check the families are spectrally homogeneous , as we expect . due to a larger uncertainty in the @xmath57 colour in sdss - moc4
, we used the color indices @xmath58 and @xmath59 , where @xmath60 ( defined by parker et al .
2008 ) .
the result is shown in figure [ pc1_vs_pc2_err_eurybates ] .
it is clearly visible that the distribution of the eurybates family in the space of @xmath61 colours is different from the trojan background . on contrary ,
the 1988 rg@xmath30 group covers essentially the same area as the background .
the aneas is only slightly shifted towards larger @xmath58 and @xmath59 with respect to the background .
there is a lack of data for the ennomos group three bodies are not sufficient to compare the colour distributions .
alternatively , we may use principal component analysis of the sdss colour indices .
we use only data with uncertainties smaller than 0.2mag , which resulted in 70887 records .
we calculated eigenvalues ( @xmath62 , 0.0532 , 0.0249 , 0.0095 ) , corresponding eigenvectors and constructed the following three principal components ( trojanov 2010 ) : @xmath63 which have a clear physical interpretation : pc@xmath64 corresponds to an overall slope , pc@xmath65 is a variability in the @xmath57 band , and pc@xmath66 a depth of the @xmath67 absorption band .
the eurybates family is different from trojans in all three principal components ( mean pc@xmath64 of the eurybates members is smaller , pc@xmath65 and pc@xmath66 larger ) .
the aneas group has the same distribution of pc@xmath65 and pc@xmath66 as trojans and the 1988 rg@xmath30 group is similar to trojans even in all three components .
hence , we confirm the eurybates family seems distinct in color even in the fourth version of the sdss - moc .
this fact is consistent with previous work of roig et al .
( 2008 ) , who used third version of the same catalogue and classified eurybates family members as c - type asteroids .
finally , note that de luise et al .
( 2010 ) pointed out an absence of ice spectral features at 1.5 and @xmath68 on several eurybates members and yang and jewitt ( 2007 ) concluded the same for ( 4709 ) ennomos .
this puzzling fact may indicate that pure ice covers at most 10% of the ennomos surface .
colours for the l@xmath69 trojans ( gray dots ) and the eurybates family ( black dots with error bars ) .
the distributions differ significantly in this case .
right panel : a similar comparison for the l@xmath70 trojans and the 1988 rg@xmath30 group , which seem to be indistinguishable.,title="fig:",width=151 ] colours for the l@xmath69 trojans ( gray dots ) and the eurybates family ( black dots with error bars ) . the distributions differ significantly in this case .
right panel : a similar comparison for the l@xmath70 trojans and the 1988 rg@xmath30 group , which seem to be indistinguishable.,title="fig:",width=151 ] we use a simple model of an isotropic disruption from farinella et al .
the distribution of velocities `` at infinity '' follows the function @xmath71 with the exponent @xmath72 being a free parameter , @xmath73 a normalisation constant and @xmath74 the escape velocity from the parent body , which is determined by its size @xmath75 and mean density @xmath76 .
the distribution is cut at a selected maximum allowed velocity @xmath77 to prevent outliers .
we typically use @xmath78 .
the orientations of velocity vectors in space are assigned randomly .
we assume the velocity of fragments is independent on their size .
there are several more free parameters , which determine the initial shape of the family in the space of proper elements : initial osculating eccentricity @xmath79 of the parent body , initial inclination @xmath80 , as well as true anomaly @xmath81 and argument of perihelion @xmath82 at the time of impact disruption .
an example of a synthetic family just after disruption and its comparison to the observed eurybates family is shown in figure [ eurybates - impact4_f0_ae ] .
usually , there is a significant disagreement between this simple model of impact disruption and the observations .
synthetic families usually look like thin ` filaments ' in the @xmath28 space , which are curved due to the mapping from osculating elements to resonant ones . on the other hand ,
observed groups among trojans are much more spread .
however , this only indicates an importance of further long - term evolution by chaotic diffusion and possibly by planetary migration . in case of the ennomos group members ,
they are distributed mostly on larger semimajor axes than ( 4709 ) ennomos , thought isotropic impact disruptions produce fragments distributed evenly on larger and smaller @xmath11 .
may be , it is an indication of an anisotropic velocity field ? or a different parent body of this cluster ?
if asteroid families are very old , planetary migration might influence their current shape . in order to study of late stages of planetary migration , which is caused by interactions with a planetesimal disk ,
we construct the following model .
we treat the migration analytically within a modified version of the numerical symplectic swift - rmvs3 integrator ( levison & duncan 1994 ) , which accounts for gravitational perturbations of the sun and four giant planets and includes also an energy - dissipation term , as described in bro et al . (
the speed of migration is characterised by the exponential time scale @xmath83 and the required total change of semimajor axis @xmath84 .
we use an eccentricity damping formula too , which simulates the effects of dynamical friction and prevent an unrealistic increase of eccentricities ( morbidelli et al .
the amount of damping is determined by the parameter @xmath85 .
we try to adjust initial orbital parameters of planets and the parameters of migration in such a way to end up at currently observed orbits .
the integration time step is @xmath86 days and the time span is usually equal to @xmath87 , when planetary orbits almost stop to migrate . on long time scales ,
the yarkovsky thermal force might cause significant perturbations of orbits .
we use an implementation of the yarkovsky thermal effect in the swift n - body integrator ( bro 2006 ) .
it includes both the diurnal and seasonal variants .
the yorp effect ( thermal torques affecting spin states ; vokrouhlick et al .
2006 ) was not taken into account in our simulations .
the reason is that the respective time scale @xmath88 is of the order 100my to 1gyr so as a ` zero ' approximation , we neglect the yorp effect on these `` short '' time scales and keep the orientations of the spin axes fixed . for trojan asteroids captured in a zero - order mean motion resonance ,
the yarkovsky perturbation only affects the position of libration centre ( moldovan et al .
note that the perturbation acts ` instantly ' there is no systematic secular drift in eccentricity nor in other proper elements which is an important difference from first - order resonances , where a @xmath13-drift is expected ( bro & vokrouhlick 2008 , appendix a ) .
this is another reason we do not need a detailed yorp model here .
the thermal parameter we use are reasonable estimates for c / x - type bodies : @xmath89 for the surface and bulk densities , @xmath90 for the surface thermal conductivity , @xmath91 for the heat capacity , @xmath92 for the bond albedo and @xmath93 for the thermal emissivity .
in this section , we briefly discuss properties of selected clusters : eurybates , ennomos , aneas and 1988 rg@xmath30 .
we focus on these four clusters , since they seem most prominent according to our previous analysis .
the eurybates family can be detected by the hierarchical clustering method for cut off velocities @xmath94 , when it merges with menelaus ( see figure [ l4_nv_random ] ) . yet
, we do not rely just on the hcm !
another selection criterion we use is a ` meaningful ' shape of the family and its changes with respect to @xmath31 .
a very important characteristic of the eurybates family at low values of @xmath31 is a tight confinement of inclinations ( @xmath14 within 0.01 ) .
it breaks down at @xmath95 , so we consider this value as an upper limit . the eurybates family is also confined in semimajor axis , being approximately twice smaller than other groups .
the diameter of the parent body is @xmath96 for albedo @xmath97 if we sum the volumes of the known bodies . of course ,
in reality it is slightly larger due to observational incompleteness .
if we prolong the slope of the sfd @xmath98 down to zero we obtain @xmath99 .
the geometric method of tanga et al .
( 1999 ) gives an upper limit @xmath100 .
spectral slopes of family members are rather homogeneous and correspond to c / p - types ( roig et al .
2008 ) .
the cluster around ( 4709 ) ennomos can be recognised for a wide interval of cut off velocities @xmath101 when it stays compact and confined in inclinations @xmath102 . very probably , there are several interlopers , because we can count 4 to 10 asteroids in the surroundings , i.e. , in the same volume of the @xmath28 space ( see figure [ trojans_l5_arer_sizes_ennomos ] ) . since small bodies dominate the ennomos group we suspect large bodies might be actually interlopers .
a very intriguing feature is a high albedo of ( 4709 ) ennomos @xmath0 measured by fernndez et al .
apart from other explanations , the authors speculated it may result from a recent impact which covered the surface with pristine ice .
if true the relation between the fresh surface and the collisional family might be a unique opportunity to study cratering events .
we can not exclude a possibility that ( 4709 ) ennomos is actually an interloper and the family is not related to it at all .
nevertheless , our hypothesis is testable : family members should exhibit a similarity in spectra and albedos .
the only information we have to date are sdss colours for three members : 98362 , 2005 yg@xmath103 are probably c - types and 2005 ar@xmath104 is a d - type . in case
new data become available we can remove interlopers from our sample and improve our analysis .
the size distribution of the ennomos group is barely constrained , since small bodies are at the observational limit .
moreover , removal of interlopers can change the sfd slope completely ( from @xmath105 to @xmath106 or so ) .
the minimum parent body size is about @xmath107 if all members have high albedo @xmath108 .
the aneas group looks like a middle portion of the @xmath3 cloud with approximately background density .
it spans whole range of semimajor axes , as background asteroids do .
the minimum size of a hypothetical parent body is @xmath109 ( for albedo @xmath110 ) .
this size is very large and an impact disruption of such body is less probable ( see section [ sec : collisions ] ) .
the size - frequency distribution is shallow , with approximately the same slope as background . according to roig et al .
( 2008 ) the colours are rather homogeneous and correspond to d - types , with @xmath111 of probable interlopers .
the group around asteroid ( 11487 ) 1988 rg@xmath30 again looks like a lower portion of the @xmath3 cloud at low inclinations , with @xmath112 .
the sfd is steeper ( @xmath113 ) than surroundings in @xmath3 and the resulting parent body size @xmath114 is relatively small .
the colours seems heterogeneous ( roig et al .
2008 ) and we can confirm this statement based on the new sdss - moc version 4 data .
the remaining clusters ( hektor , teucer , sinon , 1986 wd , laertes , asios , polydoros , panthoos , etc . )
may be characterised as follows : ( i ) they have a density in @xmath28 space comparable to that of background ( surroundings ) ; ( ii ) when identified by the hcm their semimajor axes span the whole range of trojan region ; ( iii ) the slopes of their sfd s are comparable to the background ; ( iv ) they are often inhomogeneous with respect to colours ( according to roig et al .
these reasons lead us to a conclusion that these clusters are not necessarily real collisional families .
we try to model long - term evolution of the eurybates family . at first
, we generate a synthetic family ( consisting of 42 bodies ) by an impact disruption of the parent body with required size .
then we integrate the synthetic family and compare it at particular time to the observed eurybates family .
the time span of the integration is 4gyr .
the main driving mechanism is slow _
chaotic diffusion _ ( the yarkovsky effect is present but inefficient in the trojan region ) .
initially , the spread of inclinations of the synthetic family is consistent with the observed one . on the other hand ,
the shape in @xmath115 elements is clearly inconsistent .
since the inclinations evolve only barely we focus on the evolution of in the @xmath115 plane ( see figure [ eurybates - impye2_ae_500myr ] ) .
the point is the synthetic family , namely the ` filament ' structure , has to disperse sufficiently .
after 500myr it is still recognisable but after 1gyr of evolution it is not .
so we may constrain the age of the eurybates family from 1 to 4gyr .
distributions : initially the ks distance is @xmath116 and the probability @xmath117 , which means the distribution are incompatible . at @xmath118 ,
the values are @xmath119 and @xmath120 , which indicates a reasonable match . ]
a similar analysis for the ennomos group indicates that chaotic diffusion is faster in this region ( given the large inclination ) and the most probable age thus seems to be from 1 to 2gyr . beyond 2gyr
the inclinations of the synthetic family become too large compared to the observed ennomos group , while the eccentricites are still compatible .
we try to model aneas and 1988 rg@xmath30 groups too ( see figure [ aneas - impye2_ae_4gyr ] ) . in these two cases
, there is a strong disagreement between our model and observations .
the observed groups are much larger and chaotic diffusion in respective regions is very slow .
even after 4gyr of orbital evolution , the synthetic family remains too small .
the only free parameter which may substantially change our results is the initial velocity distribution . theoretically , the distribution might have been strongly anisotropic
. however , we can not choose initial velocities entirely freely , since their magnitude should be comparable to the escape velocity from the parent body , which is fixed by the size @xmath121 and only weakly dependent on a - priori unknown density @xmath76 .
another solution of this problem is possible if we assume families are very old and they experienced perturbations due to planetary migration . the major perturbation acting on trojans are _ secondary resonances _ between the libration period @xmath122 of the asteroid in the j1/1 mean - motion resonance with jupiter and the period @xmath123 of the critical argument of jupiter
saturn 1:2 resonance ( morbidelli et al .
2005 ) @xmath124 where @xmath125 is a small integer number .
typical libration periods are @xmath126 and @xmath123 changes as planets migrate ( it decreases because jupiter and saturn recede from their mutual 1:2 resonance ) .
( the so called great inequality period ) thought it is weaker than @xmath123 .
we find no asteroids perturbed by secondary resonances connected with @xmath127 or @xmath128 which are present ` en route ' .
neither uranus nor neptune play an important role . ]
all synthetic families are strongly unstable when @xmath129 and even during later stages of migration with @xmath130 the eccentricities of family members are perturbed too much to match the observed families like eurybates or ennomos ( see figure [ eurybates - mig3_dampe_ae_at_00000300 ] ) .
there are practically no plausible migration scenarios regardless of time scale @xmath83 which would produce a sufficiently compact group , unless jupiter and saturn are almost on their current orbits .
we tested @xmath131 and even for @xmath132 as small as @xmath133 and @xmath134 the perturbation was too strong .
the reason is that one has to avoid @xmath135 secondary resonance to preserve a low spread of a synthetic family .
let us conclude if any of trojan families was created during planetary migration and if the migration was smooth ( exponential ) the family can not be visible today .
however , we can not exclude a possibility that final stages of migration were entirely different , e.g. , similar to the ` jumping - jupiter ' scenario ( morbidelli et al .
2010 ) . in this case ) .
top panel : the state at 0myr , middle : 100myr , bottom : the respective orbital evolution of jupiter and saturn .
the family is almost destroyed and it is definitely incompatible with the observed eurybates family.,width=313 ] we studied a possibility that some families can not be identified because the breakup occurred at the outskirts of the stable libration zone and some fragments were ejected outside the j1/1 resonance .
we thus chose 30 largest asteroids near the edge of the @xmath2 libration zone and we simulated breakups of these asteroids which create families with 30 fragments each .
we assumed the diameter of all parent bodies equal to @xmath136 and their density @xmath137 .
the breakups always occurred at the same geometry @xmath138 , @xmath139 . after the breakup , we calculated proper elements of the family members and plotted their distribution ( see figure [ ballistic_transport_ei ] ) .
we can see all 30 synthetic families can be easily identified . in most cases ,
more than 95% of family members remained within the stable libration zone .
we can thus conclude that the ejection of fragments outside the libration zone does not affect the number of observed families among trojans .
we can estimate collisional activity by means of a simple stationary model .
trojan collisions play a major role here , because trojans are detached from the main belt . in case of eurybates , the target ( parent body ) diameter @xmath140 , the mean impact velocity @xmath141 ( delloro et al . 1998 )
, the strength @xmath142 ( benz & asphaug 1999 ) and thus the necessary impactor size ( bottke et al . 2005 ) @xmath143 number of @xmath144 projectiles among @xmath2 trojans is @xmath145 and we have @xmath146 available targets .
an intrinsic collision probability for trojan trojan collisions @xmath147 ( delloro et al .
1998 ) and corresponding frequency of disruptions is @xmath148 over the age of the solar system @xmath149 ( after the lhb ) , we have a very low number of such events @xmath150 .
this number seems to be in concert with only one @xmath151 family currently observed among trojans . while the number of observed families with @xmath152 is about 20 ( durda et al .
these two numbers are comparable at least to order - of - magnitude . ] in a less - likely case , the material of the eurybates parent body was very weak and its strength may be at most one order of magnitude lower , @xmath153 ( see leinhardt & stewart 2009 , bottke et al .
we then obtain @xmath154 and @xmath155 , so the conclusion about the low number of expected trojan families remains essentially the same .
the parent body of aneas group is 1.5 larger and consequently the resulting number of events is more than one order of magnitude lower . on the other hand , clusters with smaller parent bodies ( @xmath156 ) or significantly weaker ( @xmath157 )
might be more frequent . during the late heavy bombardment epoch
we may assume a substantial increase of collisional activity ( levison et al .
hypothetical old families were however probably ` erased ' due to the late phases of planetary migration ( see section [ sec : migration ] ) unless the migration time scale for jupiter and saturn was significantly shorter than the time scale of the impactor flux from transneptunian region which is mainly controlled by the migration of uranus and neptune .
increasing number of trojan asteroids with available proper elements enables us to get new insights into this important population . essentially , new faint / small asteroids filled the ` gaps ' in the proper - element space between previously known clusters and today it seems most clusters are rather comparable to background .
one should be aware that the number of families among trojans may be low and one should not take the number of @xmath158 families as a rule . only the c - type eurybates family fulfils all criteria to be considered a collisional family .
this is probably also true for the newly discovered ennomos group .
moreover , there might be a potentially interesting relation between the high - albedo surface of ( 4709 ) ennomos and the collisional family thought we do not have enough data yet to prove it independently ( by colours , spectra or albedos ) .
note there may exist clusters among trojans which are not of collisional origin .
they may be caused by : ( i ) differences in chaotic diffusion rates ; ( ii ) @xmath159-dependent efficiency of original capture mechanism ; or ( iii ) it may somehow reflect orbital distribution in source regions .
we can not exclude a hypothetical existence of old families which were totally dispersed by dynamical processes , e.g. , by perturbations due to planetary migration which is especially efficient in the trojan region .
finally , note there seem to be no d - type families anywhere in the solar system neither in the trojan region , nor in the j3/2 ( bro et al .
2011 ) and cybele regions ( vokrouhlick et al .
2010 ) . may be , the d - type parent bodies are too weak and the target is completely pulverized during a collision ?
this might have important implications for collisional models of icy bodies .
we thank david vokrouhlick , david nesvorn , alessandro morbidelli and william f. bottke for valuable discussions on the subject .
we also thank lenka trojanov for the pricipal component analysis of the sdss - moc4 data and gonzalo carlos de ela for his review which helped to improve the final version of the paper . the work has been supported by the grant agency of the czech republic ( grant 205/08/p196 ) and the research program msm0021620860 of the czech ministry of education .
we acknowledge the usage of computers of the observatory and planetarium in hradec krlov and observatory and planetarium of prague .
99 beaug c. , roig f. , 2001 , icarus , 153 , 391 benz w. , asphaug e. , 1999 , icarus , 142 , 5 bottke w.f .
, durda d.d .
, nesvorn d. , jedicke r. , morbidelli a. , vokrouhlick d. , levison h.f .
, 2005 , icarus , 175 , 111 bottke , w.f .
, nesvorn d. , vokrouhlick d. , morbidelli , a. , 2010 , aj , 139 , 994 bro m. , vokrouhlick d. , 2008 , mnras , 390 , 715 bro m. , vokrouhlick d. , morbidelli a. , bottke w.f . , nesvorn d. , 2011 , mnras , submitted delloro a. , marzari f. , paolicchi p. , dotto e. , vanzani v. , 1998 , a&a , 339 , 272 durda d.d . , bottke w.f . , nesvorn d. , enke b.l .
, merline w.j .
, aspaugh e. , richardson d.c . , 2007 ,
icarus , 186 , 498 farinella p. , froeschl c. , gonczi r. , 1994 , in milani a. , di martino m. , cellino a. , eds . , asteroids , comets , meteors 1993 .
kluwer academic publishers , dordrecht , p. 205
fernndez y.r . , jewitt d.c . , ziffer j.e . , 2009 ,
aj , 138 , 240 fernndez y.r . ,
sheppard s.s . , jewitt d.c . , 2003 , aj , 126 , 1536 gomes r. , levison h.f . , tsiganis k. , morbidelli a. , 2005 , nature , 435 , 466 kneevi z. , milani a. , 2003 , a&a , 403 , 1165 laskar j. , robutel p. , 2001
, celest .
astron . , 80 , 39 leinhardt z.m . , stewart , s.t .
, 2009 , icarus , 199 , 542 levison h.f . , duncan m. , 1994 , icarus , 108 , 18 levison h.f .
, bottke w.f . , gounelle m. , morbidelli a. , nesvorn d. , tsiganis k. , 2009 , nature , 460 , 364 de luise f. , dotto e. , fornasier s. , barucci m.a .
, pinilla - alonso n. , perna d. , marzari f. , 2010 , icarus , 209 , 586 milani a. , 1993 , cel .
astr . , 57 , 59 moldovan r. , matthews j.m . , gladman b. , bottke w.f . , vokrouhlick d. , 2010 , apj , 716 , 315 morbidelli a. , levison h.f . , tsiganis k. , gomes r. , 2005 , nature , 435 , 462 morbidelli a. , brasser r. , gomes r. , levison h.f . ,
tsiganis k. , 2010 , aj , accepted nesvorn d. , bottke w.f . , dones l. , levison h.f . , 2002 , nature , 417 , 720 nesvorn d. , bottke w.f .
, levison h.f . ,
dones l. , 2003 , apj , 591 , 486 nesvorn d. , vokrouhlick d. , bottke w.f . , 2006 , science , 312 , 1490 obrien d. , morbidelli a. , 2008 , lpi cont .
, 1405 , 8367o parker a. , ivezi . , juri m , lupton r. , sekora m.d . , kowalski a. , 2008 , icarus , 198 , 138 quinn t.r . , tremaine s. , duncan m. , 1991 , aj , 101 , 2287 roig f. , ribeiro a.o . , gil - hutton r. , 2008 , a&a , 483 , 911 szab gy.m .
, ivezi . ,
juri m , lupton r. , 2007 , mnras , 377 , 1393 idlichovsk m. , nesvorn d. , 1997 , celest .
dynam . astron . ,
65 , 137 tanga p. , cellino a. , michel p. , zappal v. , paolicchi p. , delloro a. , 1998 , icarus , 141 , 65 trojanov l. , 2010 , bc thesis , univ .
hradec krlov vokrouhlick d. , bro m. , bottke w.f .
, nesvorn d. , morbidelli a. , 2006 , icarus , 182 , 118 vokrouhlick d. , nesvorn d. , bottke w.f .
, morbidelli a. , 2010 , aj , 139 , 2148 yang b. , jewitt d. , 2007 , aj , 134 , 223 yoshida f. , nakamura t. , 2008 , pasj , 60 , 297 zappal v. , cellino a. , farinella p. , milani a. , 1994 , aj , 107 , 772 | we study orbital and physical properties of trojan asteroids of jupiter .
we try to discern all families previously discussed in literature , but we conclude there is only one significant family among trojans , namely the cluster around asteroid ( 3548 ) eurybates .
it is the only cluster , which has all of the following characteristics : ( i ) it is clearly concentrated in the proper - element space ; ( ii ) size - frequency distribution is different from background asteroids ; ( iii ) we have a reasonable collisional / dynamical model of the family .
henceforth , we can consider it as a real collisional family .
we also report a discovery of a possible family around the asteroid ( 4709 ) ennomos , composed mostly of small asteroids .
the asteroid ( 4709 ) ennomos is known to have a very high albedo @xmath0 , which may be related to a hypothetical cratering event which exposed ice ( fernndez et al .
2003 ) .
the relation between the collisional family and the exposed surface of the parent body is a unique opportunity to study the physics of cratering events . however , more data are needed to confirm the existence of this family and its relationship with ennomos .
[ firstpage ] celestial mechanics minor planets , asteroids methods : @xmath1-body simulations . |
for many years people were puzzling over the question if quantum mechanics as the underlying theory to describe non - classical phenomena is complete .
after the einstein , podolsky , and rosen ( epr ) paradox@xcite , published in 1935 , it was john steward bell who presented in 1964 a seminal theorem to tackle this question.@xcite he accepted the epr conclusion as a working principle and developed what is called a local hidden variable model ( lhvm ) in which measurement outcomes are completely predetermined and only affected locally .
the assumptions of this model impose linear constraints on experimental input - output correlations in a bipartite system which have become known as bell inequalities .
an important feature in the context of an experimental test of bell s inequality is entanglement@xcite , a quantum mechanical phenomenon which occurs if the state of a particle pair can not be written as independent single particle states .
furthermore , due to entanglement , bell showed that it is impossible to reproduce quantum correlations in the framework of a lhvm which implies a non - local behavior of nature .
but since he assumes in his work perfect correlations , the practical realization of a corresponding experiment is challenging .
for this reason clauser , horn , shimony , and holt ( chsh ) published in 1969 a bell type inequality@xcite which allows to experimentally test a lhvm . only a few years later , in 1972
, the first experiment@xcite in the framework of non - locality tests was performed using entangled photons .
due to the strong resistance against decoherence photon - based systems are ideal for transmitting quantum states over large distances and provide entanglement in different degrees of freedom like their polarization@xcite , transverse or orbital angular momentum ( see refs .
@xcite ) , or energy - time.@xcite towards the end of the century many experiments are engaged in testing the chsh inequality.@xcite in 2002 , the chsh inquality was generalized to what is known today as the collins - gisin - linden - masser - popescu ( cglmp ) inequality@xcite allowing for the investigation of quantum states with dimension @xmath0 .
we present in this work the manifestation of quantum entanglement in bipartite 2-level as well as 3-level systems , also referred to as qubits and qutrits , by using photons entangled in the energy - time domain .
experimentally , we modulate in full coherent control@xcite the photon spectra via a 1-dimensional spatial light modulator ( slm ) allowing us to measure a cglmp bell parameter above the local variable limit for maximally and partially entangled 2-qubit and 2-qutrit states.@xcite the corresponding measurement settings are optimized using the most general su(3 ) unitary transformation .
a cglmp experiment involves two parties renowned as alice ( @xmath1 ) and bob ( @xmath2 ) .
each party performs two different measurements @xmath3 and @xmath4 on a shared 2-qudit state yielding measurement outcomes @xmath5 for alice and @xmath6 for bob .
the total number of possible outcomes thereby depends on the dimension of the distributed quantum state , i.e. either @xmath7 for qubits or @xmath8 for qutrits . in this way
, a set of joint probabilities @xmath9 can be measured which determines the corresponding bell parameter @xmath10 . if @xmath11 the correlations hidden in the set of joint probabilities can be explained by a lhvm .
whereas for a maximally entangled 2-qubit state the inequality in eq .
is maximally violated by the cirelson bound@xcite , i.e. @xmath12 , the bell parameter @xmath13 is maximized to @xmath14 by a _ non_-maximally entangled 2-qutrit state.@xcite it is therefore of interest to measure the bell parameters @xmath15 and @xmath13 in dependence of a variable degree of entanglement . in the following we thus focus on a 2-qubit state @xmath16 and a 2-qutrit state @xmath17 with a parameter @xmath18 $ ] quantifying the degree of entanglement .
the joint probabilities in eq . are obtained via projective measurements @xmath19 with @xmath20 and complex coefficients @xmath21 . in order to maximize the bell parameter for each value of @xmath22 , optimal measurement settings @xmath23 and @xmath24 , with @xmath25 ,
were determined by numerical optimization , i.e. @xmath26 in ref .
@xcite , a numerical search and an analytic calculation revealed a maximal violation @xmath27 of the cglmp inequality under the measurement settings first given in ref .
@xcite and a non - maximally entangled 2-qutrit state with @xmath28 .
the eigenvectors of the corresponding measurement operators refer to a unitary phase operator followed by a symmetric multiport beamsplitter for each party .
we , however , numerically optimize the measurement settings for each @xmath22 incorporating the most general su(3 ) unitary transformation as a measurement bases . for @xmath29 the optimality of the settings can be justified by comparing the value of the optimized bell parameter with the corresponding outcome of the horodecki theorem.@xcite the latter provides the maximal value of the chsh bell parameter for a given biparite qubit state with respect to its bell operator . in the case of @xmath30 no such theorem exists .
the experimental setup can be subdivided into three parts and is schematically depicted in fig .
[ fig : lxil ] . by pumping a periodically poled @xmath31 ( ppktp ) crystal with a quasi - monochromatic @xmath32 ( coherent verdi v5 ) laser centered at 532 nm , collinearly phase - matched type-0 spontaneous parametric down - conversion ( spdc ) is exploited . in this way , energy - time entangled photons centered at 1064 nm are prepared with a power of 1.5 @xmath33w and a spectral width of approximately 105 nm .
the corresponding biphoton state can be described by first order perturbation theory @xmath34 where we omit the leading order vacuum state . in general @xmath35
is a non - separable function , denoted as the joint spectral amplitude , and @xmath36 are operators which create the idler ( @xmath37 ) and signal ( @xmath38 ) photon with relative frequency @xmath39 by acting on a combined vacuum state .
note , that in ref .
@xcite the amount of entanglement of a similar state with a pulsed pump field was quantified in the frequency domain by means of an experimentally determined fedorov parameter . in the subsequent manipulation
part , group - velocity dispersion control is established by means of a four - prism compressor , composed of n - sf11 equilateral prisms , which at the same time serves to spatially align the spectrum of the entangled pairs at the symmetry axis of the experimental setup .
a symmetric 2-lens arrangement is used to image the entangled pairs from the center of the spdc crystal to the center of a sum - frequency generation ( sfg ) detection crystal .
the arrangement is such that a magnification factor of 6 is achieved at the slm plane .
the slm ( jenoptik , slm - s640d ) is used to discretize the continuous two - photon state of eq . into a discrete @xmath40-dimensional subspace spanned by the orthonormal product states
@xmath41 with @xmath42.@xcite thereby , we generally obtain a 2-qudit state @xmath43 with @xmath44 .
experimentally , a discretization scheme is applied on the slm via a complex transfer function @xmath45 where each spectral component of the idler and signal photon can be independently manipulated in amplitude @xmath46 and phase @xmath47 . to obtain a frequency - bin discretization the basis function @xmath48
is chosen according to @xmath49 if we impose for all @xmath50 that @xmath51 , i.e. adjacent bins do not overlap , and further take into account the continuous wave pump field in the form of a dirac delta function , eq .
is restricted to its diagonal form @xmath52 coincidences are detected through sfg which provides a time window with femtosecond temporal resolution.@xcite the up - converted sfg photons are finally detected by a single photon counting module ( spcm , i d quantique , id100 - 50-uln ) . the measured signal is then described by @xmath53 and is equal to the signal of a projective measurement @xmath54 for a direct product state @xmath55 and @xmath56 of eq . . by identifying the idler and signal photon with alice ( @xmath57 ) and bob ( @xmath58 ) , respectively , the probabilities in eq . are measured by projective measurements according to eq . with @xmath59 of eq . .
however , since a sfg process demands a local interaction between the involved photons , the following results can not be considered as a test of the non - local structure of nature .
though , it was shown that the bell parameter related to @xmath60 can be used as an entanglement witness.@xcite as a consequence , the violation of eq
. signifies entanglement in the 2-qudit state under consideration .
due to the shape of the spdc spectrum ( fig .
[ fig : lxil ] ) the state of eq . entails unequally distributed probability amplitudes and therefore belongs to the class of non - maximally entangled qudit states . in order to experimentally obtain states according to eq . and
eq . we first maximize the entanglement by means of the procrustean method of entanglement concentration.@xcite to equate the amplitudes to @xmath61 we discard photon combinations with high probabilities by adjusting the amplitudes @xmath46 of the corresponding frequency bins in the transfer function of eq . .
after having prepared a maximally entangled state , the reduced entanglement ( @xmath62 ) is then obtained by decreasing the amplitudes @xmath63 of the bins associated with @xmath64 .
the states are first characterized in terms of their density matrices by performing quantum state tomography for various values of @xmath22 .
further , by applying the optimized measurement settings mentioned in section [ sec : cglmp ] , the bell parameters @xmath65 are measured at different degrees of entanglement .
quantum state tomography ( qst ) fully reconstructs a quantum state in terms of its density matrix . to demonstrate the ability of our experimental setup to perform qst we focus on various 2-qutrit states according to eq . .
the corresponding density matrix can be expressed as a linear combination of generalized gell - mann matrices @xmath66 and the @xmath67-dimensional identity matrix @xmath68.@xcite to determine the coefficients @xmath69 we chose @xmath70 in eq . according to a tomographically complete set of projective measurements provided in ref .
@xcite . in order to retrieve a positive semidefinite @xmath71 out of the tomographic measurements
, we employ a maximum likelihood estimation method for high dimensional states established in the aforementioned reference . the resulting 2-qutrit density matrices for @xmath72
$ ] are depicted in fig .
[ fig : qstresults ] .
the fidelity@xcite @xmath73 ^ 2 $ ] is calculated with respect to the target density matrix @xmath74 given by the pure state in eq . .
the 2@xmath75 uncertainties on @xmath76 are calculated through a monte carlo method by randomly adding normally distributed noise to each measurement outcome and recomputing the fidelity . in order to experimentally determine @xmath77 of eq . for a specific value of @xmath22 ,
a number of 16 and 36 different projective measurements are required for qubit- and qutrit - based systems , respectively . for various values of @xmath22 , fig .
[ fig : results ] presents experimental data , i.e. @xmath78 , together with theoretical predictions for the bell parameters @xmath79 and @xmath80 using optimized measurements settings .
the theoretical curves are thereby scaled to the experimental data using a symmetric noise model @xmath81 where the mixing parameter @xmath82 and the @xmath40-dimensional identity operator @xmath83 are used .
therefore , deviations from a pure state due to white noise and experimental imperfections can be quantified by @xmath84 $ ] .
the value of the bell parameter for @xmath85 then scales as @xmath86 .
we observe a smaller sensitivity to @xmath22 of the bell parameter for qubits compared to qutrits which is in accordance with theoretical predictions . as discussd , in both systems ,
higher bell parameter values are obtained after optimizing the corresponding experimental settings . to motivate the use of the optimized su(3 ) operators ,
we additionally depict in fig .
[ fig : results ] the bell parameter @xmath80 now optimized with respect to a symmetric multiport beamsplitter combined with a unitary phase operator as discussed in ref . @xcite ( red dot - dashed line ) . in this case
the measurement settings @xmath87 and @xmath88 to be optimized solely consist of phase angles whereas the full su(3 ) involves also optimized amplitudes in the projection states .
the red dot - dashed line shows the necessity of the su(3 ) optimization procedure since smaller values of the bell parameter are obtained , especially for small @xmath22 , compared to the su(3 ) case ( solid red line ) .
in summary , we have demonstrated the quantum state reconstruction of maximally and non - maximally entangled qutrits encoded in frequency bins by a slm . the latter provides a versatile tool to manipulate a quantum state which allowed to study the cglmp bell inequality for a 2-qubit and a 2-qutrit state with a variable degree of entanglement .
the bell test measurements presented here can be readily extended to a more general class of qutrits and projection measurements , and , thus , can help to tackle the experiments imposed by the numerical bell test studies in ref .
apart from the cglmp inequality , there exists a more general class of bell inequalities which demands @xmath67-input settings per party instead of only two.@xcite the experimental examination of such a multisetting bell inequality for qutrits is currently under study .
this research was supported by the grant pp00p2_133596 funded by the swiss national science foundation . | we experimentally study the violation of the cglmp inequality for entangled 2-qubit and 2-qutrit states with different degrees of entanglement using numerically optimized measurement settings . the qudits are encoded and manipulated in the frequency spectrum of broadband energy - time entangled photons by taking into account a spatial light modulator .
the latter allows to discretize the spectrum into bins . by controlling each frequency bin individually ,
the generation of maximally and non - maximally entangled qutrits is verified through quantum state tomography . |
in a scenario dubbed `` chain inflation '' @xcite it was suggested that the dynamics of inflation may be dominated by a series ( chain ) of phase transitions , each induced by the quantum nucleation of bubbles ( tunneling events ) .
provided that each phase transition occurs much less than one hubble time after the previous one , this scenario is free from the problem of `` old inflation '' @xcite that the space still in the previous phase continues to inflate and prevents the completion of a phase transition .
the need for several phase transitions per hubble time , and the requirement that inflation persist for at least @xmath2 hubble times to solve the flatness and horizon problems , means that the scenario requires a potential with a large number of minima , tunneled through in sequence .
this is a challenge for model building but not an insurmountable one .
the density perturbations arising from chain inflation have been investigated in @xcite . a simplified model of chain inflation
was proposed in @xcite . however , in these works either the density fluctuation has to be determined numerically , or features of standard slow roll inflation ( with vacuum fluctuations instead of fluctuations from tunneling ) are assumed in the calculation .
thus an analytical calculation for the perturbation theory is still absent for chain inflation . in this paper
we demonstrate three properties of chain inflation which seem not to have been addressed , and which are each troubling from the point of view of building realistic models .
first , when the chain of tunneling events takes place in a single field direction , the collision of two bubble walls will trigger a nucleation of the next phase .
second , the rate at which the field , responsible for the chain of phase transitions , moves along its potential is controlled by the rate of bubble nucleations .
but since bubble nucleations are stochastic in nature , the fluctuations in the field value , which become curvature perturbations , are controlled by the poisson statistics of nucleation events .
to meet current observational limits on curvature perturbations requires of order @xmath1 phase transitions per hubble time during the observable part of inflation , requiring a potential with of order @xmath3 or more local minima to tunnel through successively .
the third problem is that it is challenging to provide a potential where the nucleation rate between minima is large enough to support chain inflation , in a field theory that remains valid to scales that are much higher than the mass of the field .
our paper is organized as follows : in section [ sec : setup - backgr - dynam ] , we set up our model and discuss the dynamics of bubble collisions which are essential to understanding the overall evolution of the inflaton . in section [ sec : evolution ]
we compute the time dependence of the mean field which constitutes the homogeneous and isotropic background . in section [ sec : power - spectrum ] , we consider the density perturbations and we calculate the power spectrum analytically , showing that it takes a simple form .
section [ sec : models ] investigates the simplest explicit realization of a chain inflation model and identifies a narrow region of parameter space in which it satisfies observational and internal consistency requirements . in section [ sec : non - gaussianities ] , we consider three- and higher - point correlations .
we consider generalizations in section [ sec : generalizations ] and conclude in section [ sec : conclusion ] .
we consider a simple model of chain inflation , with a single scalar field whose potential is a combination of an oscillation and a slope .
specifically , we consider @xmath4 where @xmath5 is a slowly varying potential and @xmath6 , @xmath7 are constants . a potential of this kind is illustrated in figure [ fig : potential ] . in practice
the pertinent features of the potential are that each minimum is lower than the last , and that the nucleation rate between each pair of minima is approximately the same .
we will also assume that the mean spacetime separation between nucleation events is large compared to the size of a bubble at nucleation .
otherwise , a description in terms of bubble nucleations does not make sense .
since nucleation events must in turn be common on the hubble scale , this ensures that the instanton method @xcite applies .
we will also assume that most of the energy from the potential drop between minima goes into accelerating the bubble wall .
unlike the case in @xcite , there is a chain of metastable vacua . as we now argue , in this case collisions of bubble walls trigger the subsequent phase nucleation .
( this phenomenon has been previously discussed in ref .
@xcite . ) specifically , numbering the potential minima from high to low as @xmath8 , when an interface between minima 1 and 2 meets another interface between 1 and 2 , they pass through each other and become interfaces between minima 2 and 3 . to see why , note first that the walls become very relativistic and lorentz contract to much narrower than the `` natural '' width set by the mass ( second derivative of the potential ) .
the process of walls crossing therefore occurs on length and time scales shorter than the natural scale in the potential .
therefore it is controlled by the kinetic term .
scalar field dynamics governed by only the kinetic term is free field theory , which is solved by the linear wave equation . according to the linear wave equation , when two interfaces meet , in which the field changes from value @xmath9 to value @xmath10 , they pass through each other , turning into interfaces where @xmath11 changes from @xmath10 to @xmath12 .
this is precisely the same behavior as colliding kinks in the sine - gordon model .
these remarks apply both in our case , and to the case of a potential with only two minima .
when there are only two minima , the region between the bubble walls is displaced from the potential minimum and the field there oscillates , with a time scale set by the parameters of the potential .
these oscillations exhaust the energy of the bubble walls , which slow and stop , completing the transition .
but in our case , the new value of the field after the walls have crossed is again a minimum of the potential , and the field stays there . after passing each other ,
the walls are again interfaces connecting a higher to a lower minimum , and they continue to accelerate .
we have verified these statements by simulating wave interactions in 1 + 1 dimensional classical field theory using the potential from eq .
( [ eq : sample - potential ] ) .
some snapshots of the field configuration as two walls meet are provided in figure [ fig : collide ] as an illustration .
collision of bubble walls .
short - dashed lines show @xmath11 at two moments before the walls meet ; solid lines show @xmath11 at a series of times as the walls meet ; long dashes show two times after the walls have crossed .
when the walls cross , they switch from connecting minima 1 and 2 , to connecting minima 2 and 3 . , scaledwidth=60.0% ] since the bubble walls do not stop or dissipate their energy when they collide , no thermal radiation is created by bubble collisions .
thus we do not need to worry about the perturbations and back - reaction from the radiation components created by the wall collisions . since energy in bubble walls redshifts like radiation , at rest , @xmath13 $ ] and then boosting by a large factor for a bubble moving near the speed of light .
the diagonal part after boosting is @xmath14 $ ] from which it follows that the pressure is @xmath15 . ]
they also do not come to dominate the energy density of the system .
however , there is another complication of nucleation in a potential with multiple minima such as that given in eq .
( [ eq : sample - potential ] ) .
when the `` slope '' is relatively large compared to the `` oscillation , '' quantified by the parameter @xmath16 , not all of the potential energy liberated by a bubble wall goes into accelerating the wall .
some goes into creating ripples in the @xmath11 field , as illustrated in figure [ ripples ] , which shows moving bubble wall profiles for @xmath17 . solving the field equations for a single accelerating bubble wall using the potential of eq .
( [ eq : sample - potential ] ) , we find that for @xmath18 these ripples grow large enough that they overshoot the next barrier and induce the next phase transition .
this sets off a cascade in which @xmath11 `` runs away '' down the potential .
so @xmath19 is required to ensure that bubble nucleations only induce a single phase transition and do not set off a catastrophic run down the potential .
this puts a lower limit on the potential barrier relative to the potential drop between minima .
( note that at @xmath20 the minima disappear ; so the limiting value of @xmath21 is quite close to where the potential becomes monotonic . )
profile of a bubble wall , moving towards the right , for several values of @xmath21 the ratio of `` slope '' to `` oscillation '' heights in the potential .
the larger the slope , the larger the `` ripples '' which appear behind the wall . for a critical value of @xmath21 just larger than the
largest shown , the ripples nucleate the next transition.,scaledwidth=60.0% ] this requirement on the potential has consequences for the bubble nucleation action . applying the procedure of coleman _
@xcite to find the nucleation action @xmath22 for the potential in eq .
( [ eq : sample - potential ] ) ( without making any thin - wall approximation ) , we find the result depends only on @xmath21 and @xmath23 : @xmath24 , where the function @xmath25 is fit by @xmath26 on the interval @xmath27 .
in particular , at the critical value @xmath28 , we find that @xmath29 .
( for smaller values of @xmath21 , @xmath25 is even larger , so this represents the best case for getting a large tunneling rate . )
the nucleation action can only be made small by choosing a large value for @xmath30 .
but this quantity is also the maximal value of the fourth derivative of the potential .
provided that the potential can be regarded as the effective potential of a canonically normalized scalar field in a four - dimensional quantum field theory , the potential is only stable against large loop corrections if @xmath31 is small enough , , defined at some physical scale @xmath32 , then the potential as measured at a more infrared scale @xmath33 will have a smaller value @xmath34 .
the condition @xmath35 is therefore unavoidable if we demand that the theory _ exist _ at an energy scale a few times higher than the mass scale of the potential @xmath36 .
] say , @xmath37 . imposing this condition
, we find @xmath38 .
this leads to a bubble nucleation rate @xmath39 which may be too small to allow the phase transitions to complete in a hubble time .
we will see that this requirement makes it difficult , though not impossible , to find consistent models of chain inflation .
we now proceed to study how the mean field @xmath40 evolves in chain inflation , which is quite different from the slow - roll dynamics of conventional inflation models .
we work in comoving coordinate @xmath41 and conformal time @xmath42 . in these coordinates ,
light cones are straight lines with slope @xmath43 .
the tunneling rate can be written as a function of time as @xmath44 where @xmath45 is the tunneling rate in comoving coordinates and conformal time , and @xmath46 is the tunneling rate in physical coordinates .
we assume @xmath46 is a constant , or varying slowly along the chain of meta - stable vacua
. varies significantly in one hubble time , the perturbation spectrum will have a large tilt , in contradiction to observations . ] to model the phase transition dynamics , a random variable @xmath47 is placed on every spacetime point , denoting the probability density for a spontaneous ( in other words , not from bubble wall collision ) tunneling event to take place . as we discussed , the collision of bubble walls does not stop the propagation of a bubble wall ; so any spontaneous nucleation event along the past light - cone of some spacetime point @xmath48 will cause a bubble wall to induce a phase transition at @xmath48 , and any spontaneous nucleation event _ inside _ the past light cone of @xmath48 has caused such a transition along the past world - line of an observer at @xmath48 .
therefore the scalar field value at a point @xmath48 is determined by a sum over all spontaneous nucleation events in the past light cone .
we find @xmath49\ , x(x',\tau')~,\end{aligned}\ ] ] where @xmath50 is the distance between neighboring minima in field space , and the integral with the heaviside @xmath51 function is over the interior of the past light cone of @xmath48 .
the random variables @xmath52 obey the poisson distribution @xmath53 note that the sum of poisson random variables is also a poisson random variable , with @xmath54 thus the summation over the past light cone ( analog of @xmath55 in equation ( [ eq : sum - pois ] ) ) takes the form @xmath56 where the constant @xmath57 is the field value at an initial time @xmath58 , and @xmath59 denotes proper times of a comoving observer such that @xmath60 .
here we have used the approximation @xmath61 . because of the exponential expansion of space , this approximation should hold for any reasonably early @xmath58 .
equation ( [ eq : lambda ] ) means that @xmath62 obeys the distribution @xmath63 from properties of the poisson distribution ( @xmath64 for @xmath65 ) , the expectation value of @xmath62 is @xmath66
in this section , we first calculate the position space two point function and then fourier transform to get the power spectrum . before doing so , it is helpful to first have an intuitive understanding of the fluctuations .
as illustrated in figure [ fig : prop ] , bubble walls move nearly at the speed of light , and the collision of bubble walls does not affect their dynamics . when there is a fluctuation at an initial slice ( denoted by @xmath67 and @xmath68 ) , it propagates at the speed of light to @xmath69 and @xmath70 at a later slice . since fluctuations can not propagate very far in the comoving diagram , they are conserved on super - hubble scales . the propagation of fluctuations for single field chain inflation .
the lower heavy ( red ) line denotes an initial slice .
the fluctuation on this initial slice is propagated to later slices , for example the one denoted by upper heavy ( green ) line , and becomes conserved on super - hubble scales.,scaledwidth=80.0% ] to calculate the position - space two - point correlation @xmath71 , we note that @xmath62 and @xmath72 can be expressed in terms of equation ( [ eq : phi_rand ] ) . in the overlapping region between the past light cone of @xmath73 and @xmath74 ,
the random variables in the summation for @xmath62 and @xmath72 are identical . on the other hand , outside the overlapping region , the random variables in the summation for @xmath62 and @xmath72 are independent .
the latter results in zero expectation value for correlations , and the former leads to the connected part of the two - point function @xmath75 where @xmath76 is the intersection volume of the two past light cones at time @xmath77 . , @xmath78 of the distribution pois@xmath79 to derive this result . ] here we have put the start of chain inflation at @xmath80 . in other words we assume chain inflation begins several e - folds before the largest currently observable scales exit the horizon .
otherwise there can be modifications to the lowest observable multipoles , which depend on the details of what happens before chain inflation , an issue we do not want to address here .
the integration in equation ( [ eq:2pt_pos ] ) has an ir divergence , proportional to @xmath81 .
the ir divergence is a characteristic of random walk behavior in position space , which is also present in standard slow roll inflation .
we do not need to worry about this divergence because it is absent after the fourier transformation .
the fourier space two - point correlation function is given by @xmath82 inserting equation ( [ eq:2pt_pos ] ) into ( [ eq:2pt_mom ] ) , interchanging the order of integration , and then taking the limit @xmath83 to consider the super - hubble modes , we obtain @xmath84 using the relation @xmath85 where @xmath86 is the curvature perturbation in the uniform energy density slice , and the time derivative of equation ( [ eq : phi ] ) , we have @xmath87 comparing to the relation @xmath88 gives the normalization of the power spectrum ,
@xmath89 inserting the observed value @xmath90 , we find one of our main results , that @xmath91 .
the number of tunnelings per unit hubble time and volume has to be surprisingly large .
our result disagrees with previous computations of the normalization of the power spectrum @xcite , whose validity was assumed by subsequent papers such as ref.@xcite .
our different finding is in part due to the fact that previous references did not consider bubble nucleations triggered by bubble wall collisions .
our methodologies have further differences ; ref .
@xcite calculated the spectrum numerically and obtained @xmath92 in contrast to our result @xmath93 .
@xcite obtained the conventional result @xmath94 where @xmath95 is the hubble slow roll parameter .
this arises from assuming vacuum initial conditions to solve the equation of motion for the perturbations , which is not valid for chain inflation .
the spectral index is @xmath96 when @xmath97 is negligible , the spectrum is necessarily red since @xmath98 .
however , @xmath99 depends on how @xmath100 vary with @xmath11 , with indeterminate sign and magnitude .
therefore chain inflation does not make a definite prediction for the spectral tilt ; it can generate a large tilt of either sign .
in particular , since we saw that @xmath46 is extremely exponentially sensitive to @xmath21 the ratio of slope to oscillation height in the potential , the slightest evolution of @xmath21 along the potential will significantly change @xmath46 .
it appears to require a carefully tuned potential to ensure that the spectral tilt is _ not _ large .
in this section we investigate the extent to which one can find simple working models of chain inflation , taking into account all the phenomenological constraints and requirements of consistency .
we take the simplest possible potential of the form ( [ eq : sample - potential ] ) , with @xmath101 where @xmath5 is constant and @xmath102 .
this form is assumed to hold during inflation ; for @xmath11 exceeding some maximum value @xmath103 , @xmath5 must become field - dependent and reach a minimum in order for inflation to end .
motivated by the discussion in section [ sec : setup - backgr - dynam ] , we will take the optimal value for @xmath21 , @xmath104
so as to alleviate as much as possible the tension between having a large enough tunneling rate and a small enough value of the effective self - coupling of the field , @xmath105 we further assume that @xmath106 so that the hubble parameter is approximately constant during inflation , @xmath107 .
to estimate the tunneling rate , we use @xcite @xmath108 where @xmath22 is the action of the bounce solution , and @xmath109 is the mass scale from the fluctuation determinant , which we estimate using the curvature of the potential , @xmath110 .
the bounce action was already computed in section [ sec : setup - backgr - dynam ] , @xmath111 . using ( [ gammaeq ] ) and @xmath112 we can express the requirement @xmath113 , coming from the normalization of the power spectrum , in terms of the model parameters : @xmath114 using the observed value of the spectral index , @xmath115 @xcite , we obtain a constraint from eq .
( [ nseq ] ) .
although @xmath46 is strictly constant in the present model , one could imagine that the parameters of the model vary slowly with @xmath11 and thus provide a source of deviation from the harrison - zeldovich spectrum @xmath116 through @xmath117 .
it is however unlikely that the two terms @xmath118 and @xmath99 should be finely tuned against each other to give a small result .
we therefore demand that @xmath119 ( the 68% c.l .
this leads to @xmath120 . using ( [ bchoice ] ) and ( [ normeq ] ) to eliminate @xmath121 and @xmath5 ,
this leads to an upper bound on @xmath122 , depending upon the effective self - coupling @xmath123 , @xmath124 unshaded regions in the @xmath125-@xmath126 plane are those for which the spectral index is not too red and the reheat temperature is above 10 mev .
the wedge - like region labeled `` allowed '' additionally fulfills the consistency requirement @xmath127.,scaledwidth=80.0% ] for our description in terms of nucleation of individual bubbles to be consistent , the bubbles must not be strongly overlapping . for the thick - walled bubbles we are considering ,
the initial radius of the bubble is of order the inverse mass scale , @xmath128 .
since @xmath129 , there must be room for @xmath1 such bubbles in a 4-dimensional hubble volume : @xmath130 . using @xmath131 and ( [ normeq ] ) to eliminate @xmath5 ,
this becomes an upper limit on @xmath125 , @xmath132 which for the case @xmath133 is solved by @xmath134 .
this of course is far weaker than the field theoretic consistency bound @xmath127 discussed in section [ sec : setup - backgr - dynam ] .
one would like to ensure that the change in @xmath11 remains less than @xmath135 over the course of the field s evolution .
assuming there were @xmath136 e - foldings of inflation , the total change in @xmath11 implied by eq .
( [ eq : phi ] ) is @xmath137 where @xmath122 is the change in @xmath11 between successive minima of the potential .
we demand that @xmath138 for @xmath139 .
this gives the upper bound @xmath140 , which is weaker than ( [ spectlim ] ) and thus does not play an important role . on the other hand
, the requirement of a large enough reheating temperature @xmath141 does give an interesting constraint on @xmath126 that is complementary to ( [ spectlim ] ) .
the reason is that @xmath142 and from ( [ normeq ] ) @xmath143 , so low values of @xmath126 as needed to satisfy ( [ spectlim ] ) may also lead to excessively low reheat temperatures .
if we generously assume such efficient reheating that @xmath144 and that @xmath145 mev so that big bang nucleosynthesis can proceed normally , we obtain @xmath146 we have summarized the most important constraints in figure [ fig : const ] , which plots them in the @xmath125-@xmath126 plane . if @xmath147 mev , there is an interval of @xmath148 $ ] where it is possible to find consistent models ; at @xmath149 gev , more preferable for being able to generate a baryon asymmetry , the range shrinks to @xmath150 $ ] if one is willing to relax the field - theoretic consistency condition @xmath151 for some reason , it becomes much easier to find viable models .
unlike the case of perturbative quantum fluctuations from the vacuum , in chain inflation the perturbations come from light - cone shaped bubbles . thus one might worry about the gaussianity of the scenario .
however , as there are a great number of bubbles per hubble time and hubble volume , the poisson - distributed field fluctuation approaches a gaussian distribution .
further , as we shall show , all the non - gaussianity from @xmath152-point correlations turns out to be small .
the three - point correlation function can be calculated similarly to the case in section [ sec : power - spectrum ] .
note that for @xmath153 , @xmath154 . we have @xmath155 where @xmath156 is the volume of the intersection region of three balls with radius @xmath157 .
writing in terms of @xmath86 , and comparing to the definition of @xmath0 @xcite , @xmath158 we find @xmath159 .
higher point correlations can be estimated in the same way . for @xmath153 , @xmath160 , @xmath161 , @xmath162 .
for the four - point function , the @xmath163 term must be subtracted to remove the disconnected part .
thus the connected part has @xmath164 . in the language of non - gaussian estimators
, this corresponds to @xmath165 .
similarly , the five- and six - point estimators @xmath166 and @xmath167 defined in @xcite are also of order one .
this statement generalizes to arbitrary @xmath152-point correlation functions .
thus non - gaussianity for single - field chain inflation is small at any order .
in this section we briefly discuss three generalizations of single - field chain inflation , namely the curvaton scenario , the multi - field case , and a noncanonical kinetic term .
we start with the curvaton scenario , where there is a second field whose rolling ( or other dynamics ) dominates the change in the total energy density . in this case
, the fluctuation in the chain tunneling direction ( still denoted by @xmath11 here ) is an isocurvature perturbation .
this isocurvature perturbation converts to a curvature perturbation later by oscillation of @xmath11 around its minimum , or if the chain of tunnelings ends after the reheating associated with the inflaton .
suppose there is no direct interaction ( except via gravity ) between the inflaton sector and the @xmath11 field .
then the component curvature perturbation @xmath168 is conserved , which is related to the total curvature perturbation @xmath86 as @xmath169 where @xmath170 , @xmath171 , @xmath172 , @xmath173 are the total energy density and pressure and the energy density and pressure for the @xmath11 field respectively .
@xmath168 is related to the field fluctuation as @xmath174 .
after the curvaton @xmath11 decays , @xmath175 becomes constant , thus @xmath86 is subsequently conserved .
the power spectrum can be calculated similarly to the last section as @xmath176 since it is suppressed by @xmath177 , the original fluctuation in the @xmath11 field may be large .
this reduces the number of minima needed in the @xmath11 potential , but at the same time it increases the non - gaussianity @xmath178 , which should be the best estimator for non - gaussianity , and is potentially observable .
it is also interesting to generalize the simplest scenario to a multi - field case , where a chain of tunnelings can take place in more than one field direction .
whenever successive tunnelings occur in orthogonal field - space directions , the collision of bubble walls will not trigger a new phase nucleation . in this case , for a given nucleation rate @xmath46
, an individual observer will undergo fewer phase transitions . specifically , the mean inter - bubble spacing is @xmath179 , the mean time between phase changes for a comoving observer is @xmath180 , and so the mean number of phase changes per hubble time is @xmath181 , not @xmath182 as we found for the single - field case .
perturbations in this scenario remain correlated on scales shorter than the hubble scale .
the reason is that , while individual bubble walls no longer propagate hubble scale distances , _ information _ about bubble nucleations still can . as an illustration ,
consider figure [ fig : multi ] . in the figure , one region of space ( near @xmath121 ) reaches @xmath183 before another region ( near @xmath21 ) .
but the probability that @xmath21 will encounter another phase change and get to @xmath184 is determined by the likelihood of a nucleation occurring in the shaded area in its past light cone .
this region is larger , and hence the chance of a nucleation is larger , than the shaded area in @xmath121 s past light cone .
therefore , unevennesses in the phase boundary allows regions which are `` behind '' to `` catch up , '' at least on sub - hubble scales .
this means that perturbations from the stochastic nature of nucleations get averaged over roughly hubble - scale volumes . however , on super - hubble scales no such communication is possible .
therefore the progress of the field @xmath11 on super - hubble scales is still determined by the independent , poisson - distributed number of nucleations in the past light cone .
hence we estimate that the power spectrum is again @xmath185 , similar to the power spectrum for single - field chain inflation .
we have not been able to evaluate the order-1 coefficient analytically in this case .
but this is enough to show that the tunneling rate must be comparable to that in the single - field case , though the number of minima in the potential can be substantially smaller ( @xmath186 rather than @xmath187 ) .
the probability for @xmath21 in the next step of the chain is larger than that of @xmath121 , even if bubble wall collision no longer triggers a new bubble nucleation .
, scaledwidth=80.0% ] a third possible generalization of chain inflation is to allow for a noncanonical kinetic term .
there is some motivation to do so since ref .
@xcite pointed out that the tunneling rate in a generic scalar field theory with false minima can be increased using the string - motivated dbi action with lagrangian @xmath188 if @xmath189 is a constant , the tunneling action can be reduced by at most a factor of order unity ; to achieve a greater reduction , one must choose @xmath189 to be proportional to the term @xmath190 that gives the barrier , and tune @xmath191 .
@xmath22 is then reduced by the factor @xmath192 .
however taking @xmath193 seems quite contrived , so we consider to what extent the modest reduction in @xmath22 afforded by constant @xmath189 can help to alleviate the tension between various constraints discussed in section [ sec : models ] .
it is amusing to notice that the tunneling action @xmath22 can still be computed exactly within the thin - wall approximation ( even though we did not use this approximation in our previous analysis ) for the model of section [ sec : models ] and the dbi action with constant @xmath189 . in the thin - wall approximation , @xmath22
is reduced relative to its usual value by the factor @xmath194 where @xmath195 .
when @xmath196 we recover the usual result , and when @xmath197 we obtain the maximum possible reduction in @xmath22 for constant @xmath189 , a factor of @xmath198 .
let us take this as indicative of how well one might hope to do in the case of interest where the thin - wall approximation is not valid .
( we are not enthusiastic enough about this generalization to do a full numerical solution of the bubbles . ) the most important constraints , those on the spectral index and the reheat temperature , are both relaxed since the spectral index bound goes like @xmath199 while the reheating bound has the form @xmath200 .
we find that the minimum allowed value of @xmath125 decreases to @xmath201 for @xmath202 mev , and to @xmath203 for @xmath204 gev .
the model remains rather strongly coupled , but less so than for a standard kinetic term .
to conclude , we reconsidered models of chain inflation focusing on the single - field case , and giving what we consider to be the first correct derivation of the spectrum of fluctuations , taking into account their poisson nature as a result of the dynamics of bubble collisions .
we showed that the power spectrum can be calculated analytically and that consistency with the size of density perturbations measured today requires @xmath1 tunnelings per hubble volume per e - fold .
therefore single - field models of chain inflation need a potential with at least @xmath205 successive minima , a significant challenge for model building .
the spectral tilt depends on details of the potential ; it can be either red or blue and need not be small .
non - gaussianities in the model are small , of order unity .
thermal contributions to the fluctuations are negligible in this scenario because the energy stored in bubble walls remains in them after they collide .
a challenge for chain inflation is posed by a standard requirement of consistency for any field theory . under renormalization ,
couplings run , and large couplings run very fast . in order to have any nonnegligible range of scales over which the theory remains valid , the effective coupling given by the fourth derivative of the potential must not be much larger than @xmath206 .
we have shown that this requirement , together with constraints on the spectral index and reheating temperature , leaves only a small interval of parameter space that gives a large enough nucleation rate .
on the other hand , the method used to calculate the power spectrum and non - gaussianity in this model may find application for cases where other discrete processes instead of a continuous process take over the role of quantum fluctuations during inflation for example , if the scale of inflation is very high and physics becomes discrete near the planck scale .
we thank bret underwood for discussion . yw
is supported by grants from mcgill university , fonds qubcois de la recherche sur la nature et les technologies ( fqrnt ) , the institute of particle physics ( canada ) and the foundational questions institute .
this work was supported in part by the natural sciences and engineering research council of canada .
k. freese , d. spolyar , `` chain inflation : bubble bubble toil and trouble , '' jcap * 0507 * , 007 ( 2005 ) .
[ hep - ph/0412145 ] .
k. freese , j. t. liu , d. spolyar , `` inflating with the qcd axion , '' phys .
rev . * d72 * , 123521 ( 2005 ) .
[ hep - ph/0502177 ] .
r. easther , j. t. giblin , l. hui and e. a. lim , `` a new mechanism for bubble nucleation : classical transitions , '' phys .
d * 80 * , 123519 ( 2009 ) [ arxiv:0907.3234 [ hep - th ] ] ; j. t. giblin , l. hui , e. a. lim and i. s. yang , `` how to run through walls : dynamics of bubble and soliton collisions , '' phys .
d * 82 * , 045019 ( 2010 ) [ arxiv:1005.3493 [ hep - th ] ] .
a. ashoorioon , k. freese and j. t. liu , `` slow nucleation rates in chain inflation with qcd axions or monodromy , '' phys .
d * 79 * , 067302 ( 2009 ) [ arxiv:0810.0228 [ hep - ph ] ] ; a. ashoorioon and k. freese , `` gravity waves from chain inflation , '' arxiv:0811.2401 [ hep - th ] .
e. komatsu _ et al .
_ [ wmap collaboration ] , `` seven - year wilkinson microwave anisotropy probe ( wmap ) observations : cosmological interpretation , '' astrophys .
j. suppl .
* 192 * , 18 ( 2011 ) .
[ arxiv:1001.4538 [ astro-ph.co ] ] . | we investigate density perturbations in the chain inflation scenario , where the inflaton undergoes successive tunneling transitions along one field direction .
first we show that when the bubble walls associated with such a phase transition meet , they induce the next phase transition and continue to propagate as phase interfaces
. then we present an analytical calculation of the density fluctuations and an estimate of non - gaussianities such as @xmath0 , which we find to be small ( of order 1 ) . to get the right amplitude for the power spectrum
, there have to be @xmath1 phase transitions per hubble time , a significant model building challenge .
we find that working models of chain inflation must be rather strongly coupled , and thus have a very limited range of validity as effective field theories .
we discuss generalizations to the multiple field case , the curvaton scenario , and noncanonical kinetic terms . |
we consider a region , @xmath0 , in @xmath1 with a smooth boundary @xmath2 .
we assume that there is a riemannian metric , @xmath3 , on @xmath0 that is conformal to the euclidean metric with conformal factor @xmath6 where @xmath7 is strictly positive .
this means that @xmath8 , where @xmath9 is the euclidean metric , or , in cartesian coordinates @xmath10 , @xmath11 .
we analyze the inverse problem of reconstructing @xmath3 based on measurements of the curvature of wavefronts produced by point diffractors located inside @xmath0 and propagated according to the wave operator on @xmath12 .
indeed , geodesics for the metric @xmath3 are rays following the propagation of singularities by a parametrix corresponding to this wave operator . in the seismic context
@xmath13 is the wave speed .
as originally formulated by dix @xcite , this type of data may in some cases be reconstructed from reflection data by variation of source and receiver locations at the surface of the earth .
in particular , it is possible to recover from reflection data the shape operator for the wave front produced by a given point diffractor where the rays beginning at the diffractor intersect @xmath2 orthogonally @xcite .
dix developed a procedure , with a formula , for reconstructing one - dimensional wave speed profiles in a half space from reflection data . since dix various adaptations have been considered to admit more general wave speed functions in a half space .
we mention the work of shah @xcite , hubral & krey @xcite , dubose , jr .
@xcite , and mann @xcite .
we consider here the case of higher dimensional regions with riemannian metrics conformal to the euclidean metric .
our method is different in the cases of @xmath5 and @xmath14 and in fact we expect better results in the case @xmath14 .
the problem is closely related to the problem where broken geodesics are observed on the boundary @xcite or when the cauchy data of the solution of the wave equations are observed on the boundary , see e.g. @xcite and references therein . assuming that we know @xmath13 and all of its derivatives on @xmath2 ( c.f .
@xcite ) , we may extend @xmath13 to a function , which we will also denote by @xmath13 , on a complete manifold @xmath15 compactly containing @xmath0 ( in fact we can take @xmath16 ) .
the corresponding extended metric is also denoted simply be @xmath3 . as described in detail below , we measure the curvature of generalized spheres for @xmath3 centered at `` diffraction '' points intersected with an open subset of @xmath17 . from these data , and assuming we know @xmath13 in @xmath18 , we show an explicit method to determine the function @xmath13 in the cartesian coordinates @xmath10 along geodesics of @xmath3 which connect the diffractions points to the measurement region .
this method can be viewed as a generalization of the work of iversen and tygel @xcite .
we now proceed to introduce the concepts and notations necessary for the statement of our main results . for any @xmath19 ( @xmath20 indicates the unit sphere bundle with respect to @xmath3 ) we will write @xmath21 for the geodesic with initial data @xmath22 , @xmath23 . for @xmath24 in the domain of @xmath21
we let @xmath25 denote the set of times @xmath26 such that @xmath26 is conjugate to @xmath24 along @xmath21 ( by this we mean that @xmath27 is a critical point for @xmath28 ) . for the moment we fix @xmath29 and use the notation @xmath30 for @xmath31 .
we will refer to the image of the set @xmath32 under the exponential map as the _ generalized sphere of radius @xmath33 centered at @xmath34_. when @xmath35 is in the domain of @xmath36 and @xmath37 there is a small portion of the generalized sphere of radius @xmath38 centered at @xmath39 containing @xmath40 that is an embedded submanifold @xmath41 of @xmath15 .
indeed , in this case we can define a vector field @xmath42 in a neighborhood of @xmath40 by writing for @xmath43 in a small neighborhood of @xmath44 : @xmath45 geometrically , @xmath42 gives the outward pointing normal vector fields to a part of the generalized sphere centered at @xmath46 near @xmath40 .
the shape operator @xmath47 of @xmath41 at @xmath40 is then given by @xmath48 for all @xmath49 , where @xmath50 is the levi - civita connection for @xmath3 .
for the reconstruction of @xmath13 we assume that @xmath51 is known for all @xmath52 such that @xmath53 . in reflection
seismology one refers to @xmath41 as the ( partial ) front of a point diffractor located at @xmath54 .
[ notfig ] we now introduce a mapping which defines local coordinates in which we will perform our initial calculations .
we begin with picking a large @xmath55 in the domain of @xmath36 such that @xmath56 .
next , let us take local coordinates @xmath57 on @xmath58 such that @xmath59 defines @xmath60 and suppose @xmath61 is the coordinate map .
we will assume that the image of these maps , @xmath62 , is the same for all @xmath63 .
for @xmath64 , let @xmath65 be the geodesic with a special choice of initial data : @xmath66 .
then we define coordinates , @xmath67 , on some set by the inverse of the map @xmath68 we define @xmath69 and so that the map @xmath70 defines a local parametrization on @xmath71 .
it may not be a global parametrization because it may not be injective .
the @xmath67 local coordinates on @xmath71 , basically , are riemannian normal coordinates centered at @xmath72 , but parametrized in a particular way : @xmath73 can be thought of as a parametrization of part of the sphere of radius @xmath63 in @xmath74 , and then @xmath24 corresponds to the radial variable in @xmath74 .
figure [ wfig ] shows a possible depiction of @xmath71 and the coordinates defined there .
we note that the domain @xmath71 includes @xmath75 .
we also define for some @xmath76 @xmath77 in our reconstruction we cover @xmath0 by sets like @xmath71 , and then recover @xmath13 on each of these regions separately . in the case that there are conjugate points to @xmath63 along @xmath36 we will also have to cover some regions with more than one such set . note
that along @xmath36 the coordinate vectors @xmath78 are jacobi fields , and are defined even at the conjugate points .
finally , we introduce frames @xmath79 defined by parallel translation along @xmath80 such that for @xmath81 @xmath82 where we are using the notation @xmath83 and so @xmath84 points in the opposite direction of @xmath85 . to simplify the presentation
we adopt this notation , @xmath83 , throughout the paper .
also we write @xmath86 for the corresponding dual frame ; that is @xmath87 where @xmath88 denotes the usual pairing of @xmath89 and @xmath90 .
[ fig : coord ] in the sequel will also consider the shape operators @xmath91 when @xmath60 is replaced in the above construction by another point in @xmath58 represented in the coordinates by @xmath73 .
we thus have for each @xmath73 and @xmath92 such that @xmath24 and @xmath26 are not conjugate along @xmath80 a tensor @xmath93 .
we represent @xmath94 using the frames constructed above as @xmath95 for fixed @xmath63 and @xmath73 we will also use the notation @xmath96 .
note that immediately from the definition we have @xmath97 for all @xmath98 and because of this in what follows when we write @xmath99 ( respectively @xmath100 ) without indices we will actually be referring to the @xmath101 matrix @xmath102 ( respectively @xmath103 ) with @xmath98 , @xmath104 .
the data for our recovery are the matrix elements @xmath105 , and their first three derivatives with respect to @xmath26 , for @xmath106 and @xmath26 not conjugate to @xmath107 along @xmath80 . in @xcite
, we obtain the following result : it is possible to uniquely determine the riemannian metric @xmath3 in a neighborhood of @xmath108 in riemannian normal coordinates having origin at the point @xmath72 ( this can be done for general metrics , not just ones which are conformally euclidean ) . here
, we cast this result into an algorithm , and construct a conversion from the mentioned coordinates to cartesian coordinates , which is the main contribution of this paper .
essentially , we generalize the time - to - depth conversion in dix original method to multi - dimensional manifolds with riemannian metrics conformal to the euclidean metric and , roughly speaking , show that if we measure near the point @xmath60 the shape operators of the wave fronts of waves diffracted from the points @xmath109 , we can then determine the wave speed @xmath13 near the geodesic @xmath36 .
the theoretical contributions of this work are thus contained in the following two theorems .
[ thm1 ] suppose that @xmath110 and @xmath111 is open and nonempty , @xmath112 , @xmath113 is a set of the form constructed above ( see ( [ wdef ] ) ) using the metric @xmath114 , and @xmath115 for all @xmath116 .
we also assume that @xmath117 is complete .
if @xmath118 is known then from the data @xmath119 ( see ( [ shape_coord ] ) ) for all @xmath63 in an open subinterval of @xmath120 , @xmath64 , and @xmath121 it is possible to recover @xmath13 in a neighborhood of @xmath122 in cartesian coordinates . in dimension
strictly larger than @xmath123 the reconstruction only involves solving ordinary differential equations along the rays of @xmath3 .
note that the sets @xmath25 are always discrete , and so the hypotheses of the theorem assume we have data except for at a discrete set of times .
we also stress here that the reconstruction is local along geodesics .
that is , to reconstruct @xmath13 in a neighborhood of @xmath36 we only require measurements of the shape operator near the point @xmath60 for generalized spheres @xmath58 centered at points @xmath109 with radii @xmath124 .
we recall @xmath125 is representation of the shape operator of these generalized spheres in the local coordinates and that we have @xmath126 . in the case
that the dimension @xmath4 the conversion from riemannian normal to cartesian coordinates involves solving a system of @xmath127 nonlinear ordinary differential equations . in the two dimensional case
the construction requires the solution of a cauchy problem for an elliptic operator ( i.e. we must solve the scalar curvature equation ( [ scal ] ) with known boundary data ) .
the discretization of the system is directly related to the available `` density '' of scatterers . in order to state our second result
, we must introduce generalized distance functions . as @xmath128 is complete ,
the map @xmath129 is surjective , and by sard s theorem , the set @xmath130 of critical values of @xmath131 has measure zero .
suppose @xmath132 and let @xmath133 be such that @xmath134 .
then @xmath135 has a neighborhood @xmath136 such that @xmath137 is a diffeomorphism ; one says that @xmath138 is a local inverse of @xmath131 corresponding to @xmath135 . in @xmath139 , we define the function , @xmath140 , by @xmath141 and call this function the generalized distance or travel time function from the point @xmath34 associated to direction @xmath135 ; the wave fronts observed from a point source at @xmath34 give us its level sets .
we may now state our second theorem .
[ thm2 ] let @xmath0 and @xmath142 be as in theorem [ thm1 ] and suppose that @xmath2 is smooth and @xmath143 is an open set such that for all @xmath144 there is a non - normalized geodesic @xmath145 for @xmath3 such that @xmath146 ) \subset w'$ ] , @xmath147 , @xmath148 and @xmath149 .
we use the notation @xmath150 , and also assume that @xmath151 is known .
further suppose that @xmath152 is an indexing set such that there is a bijective map from @xmath152 to @xmath153 and label @xmath154 according to this map .
then from knowledge of @xmath155 for @xmath156 we can recover the wave speed @xmath13 in @xmath157 .
this generalizes an earlier result which says that the boundary distance functions of a compact riemmanian manifold given on the whole boundary determine the manifold uniquely , see @xcite and ( * ? ? ?
* section 3.8 ) .
the structure of the rest of the paper is as follows . in section [ sec : prelim ]
we go over some background from riemannian geometry that is necessary .
section [ step1 ] reviews the first step of the recovery procedure in which we reconstruct the metric @xmath3 in the coordinates @xmath67 given by the map @xmath70 .
then sections [ step2 ] and [ 2d ] describe the second step of the recovery procedure in which the geodesics and wave speed are reconstructed in cartesian coordinates respectively in the case of three and higher dimensions and the case of two dimensions . finally , the proofs of theorems [ thm1 ] and [ thm2 ] are given in section [ sec : proofs ] , and section [ sec : conc ] contains some concluding remarks .
we also provide detail on the conversion from travel time measurements to measurements of the shape operator of generalized spheres for the case when @xmath2 is flat and @xmath13 is constant in a neighborhood of @xmath2 in [ appenda ] .
we summarize the basic differential equations from riemannian geometry that we will use .
we mention some general references to riemannian geometry @xcite . in this work
we will use the conventions from @xcite for the curvature tensor and related quantities in local coordinates .
we evaluate the geodesics by solving @xmath158 where @xmath159 are the christoffel symbols .
it is also possible to find the solutions of ( [ gamma ] ) using the hamiltonian flow for the hamiltonian @xmath160 .
although we will not use the hamiltonian formulation here , we note that it gives the system @xmath161 for the geodesics . from this , we see that , in terms of seismic ray tracing , the geodesics may be identified with generalized image rays . in our case , assuming isotropy ( i.e. that the metric is conformally euclidean ) , we have @xmath162 it is more convenient to work with @xmath163 than @xmath13 and we will generally do so throughout the remainder of the paper although it is clear that recovery of @xmath163 is equivalent to recovery of @xmath13 . to make the notation more concise below we introduce the shorthand @xmath164 as mentioned above , @xmath165 denotes the parallel translation of @xmath166 along @xmath167 from @xmath107 to @xmath24 .
thus , for every @xmath168 and @xmath169 , @xmath170 forms a basis for @xmath171 .
the invariant formula for parallel translation is @xmath172 if we introduce matrices which give the frames @xmath173 in cartesian coordinates @xmath174 then the invariant formula implies that these satisfy @xmath175 or @xmath176 and since the coordinates @xmath73 on @xmath58 are known , we also know the initial conditions , @xmath177
. the riemannian curvature tensor in any coordinate system is given as a @xmath178 tensor field by @xmath179 or as a @xmath180 tensor field as @xmath181 the ricci curvature tensor is given by the trace @xmath182 and the scalar curvature is @xmath183 . in the cartesian coordinates the riemmanian and ricci curvature tensors are given respectively by the following formulae in terms of @xmath163 : @xmath184 where @xmath185 is the kulkarni - nomizu product ( in coordinates , see @xcite for the invariant formula ) , @xmath186 is the hessian matrix of second derivatives , @xmath187 is the ( euclidean ) gradient , and @xmath188 is the euclidean norm of @xmath187 ; @xmath189 also , when the dimension is @xmath5 the scalar curvature satisfies the so - called scalar curvature equation @xmath190 where @xmath191 is the laplace - beltrami operator corresponding to @xmath3 given in any coordinate system @xmath192 by @xmath193 here @xmath194_{j , k=1}^2 ) $ ] and @xmath195_{j , k=1}^2 $ ] is the inverse matrix of @xmath196_{j , k=1}^2 $ ] . from these formulae we see a fundamental difference between the two dimensional case and the case of three or more dimensions . in two dimensions
the ricci curvature and scalar curvature give only the laplacian of @xmath163 , and so to find @xmath163 from these curvature tensors would require the solution of an elliptic equation .
on the other hand , in three or more dimensions the ricci curvature tensor depends on all the second partial derivatives of @xmath163 , and in general we can recover a formula for @xmath186 in terms of the ricci curvature and @xmath187 .
indeed , if we define @xmath197 then from ( [ ric0 ] ) we may calculate @xmath198 this is possible in three dimensions , but not in two dimensions , and is the reason we must consider the two cases separately .
we will also write the riemannian curvature on the geodesic @xmath167 in the frame obtained by parallel transport as @xmath199 or @xmath200 recalling that @xmath201 , we also write @xmath202 and for fixed @xmath63 @xmath203 for the directional curvature operator which we reconstruct in the first step of our procedure .
note that as with @xmath204 , for any @xmath98 @xmath205 , and so when we write @xmath206 without indices we will actually be referring to the corresponding @xmath207 matrix .
we continue in the next sections to describe the actual reconstruction algorithm .
the reconstruction procedure consists of two steps . in the first step
we consider only the single geodesic @xmath36 and reconstruct @xmath208 , and then the metric @xmath3 as a function of @xmath24 for @xmath209 . in the second step ,
we determine @xmath163 , and therefore also @xmath13 , in a neighborhood of @xmath36 by also varying @xmath73 and @xmath63 . following the geometric analysis of
@xcite we now describe the first step of the procedure which is itself broken up into a number of substeps below . 1 .
let @xmath210 , @xmath211 represent @xmath207 matrices .
we solve the autonomous system of ordinary differential equations for @xmath212 , @xmath213 in which @xmath214 for @xmath215 .
this system is supplemented with initial data , @xmath216 applying picard s theorem in a standard way , we see that the equations ( 12)-(13 ) have unique solution on some interval @xmath217 $ ] ( for details on this see @xcite ) . in practice , we may use a runge - kutta method to solve the system numerically for @xmath218 .
the system will not generally have a solution all the way up to @xmath63 ; in this case , we must divide the interval @xmath219 $ ] into several subintervals , and reconstruct on each of these in turn as described below in substep 3
we extract the directional curvature operator , @xmath220 note that this matrix @xmath221 is @xmath207 , but recall that from this we can recover the full directional curvature operator @xmath222 since the @xmath223th row and column of @xmath222 are both equal to zero .
[ step1fig ] on line 1 .
then substep 1 recovers @xmath224 in region 1 by solving ( [ eq a ] ) .
next substep 2 gives @xmath221 for @xmath225 by ( [ eq ainit ] ) , and then substep 3 gives @xmath204 on line 2 .
returning to substep 1 again gives @xmath224 in region 2 .
then as before substep 2 gives @xmath221 for @xmath226 , and substep 3 gives @xmath204 on line 3 .
continuing in this way we reconstruct @xmath206 as far along @xmath36 as we like.,title="fig : " ] + in general the first two steps only reconstruct @xmath227 for @xmath225 where @xmath228 since we may not able to solve ( [ eq a ] ) all the way up to @xmath63 . here
, we describe how to find @xmath227 for @xmath24 in the entire interval @xmath219 $ ] .
the idea is to replace @xmath107 by @xmath229 , and then use knowledge of @xmath230 for @xmath231 .
indeed , let us fix @xmath232 .
now we find the coordinates of the jacobi fields corresponding to the coordinate vectors fields of @xmath67 along @xmath36 with respect to the parallel frame .
these can be found by solving the system @xmath233 with initial conditions @xmath234 we can then recover @xmath235 by the equation @xmath236 now we may return to substep 1 and solve ( [ eq a ] ) with ( [ final init .
data ] ) replaced by @xmath237 this then allows us to use ( [ eq ainit ] ) to recover @xmath221 for @xmath24 up another time @xmath238 say with @xmath239 . if @xmath238 is less than @xmath63 we repeat this same procedure again with @xmath229 replaced by @xmath238 , and so on .
by results in @xcite there is a lower bound on the size of each step we take ( i.e. a lower bound on @xmath240 ) , and so by induction we eventually recover @xmath227 for @xmath24 on the entire interval @xmath219 $ ] . for a visual depiction of how the first three substeps proceed see figure [ step1fig ] .
we now obtain the metric in the @xmath67 coordinates given by the coordinate map @xmath70 along @xmath36 , which we write as @xmath241 , by the formula @xmath242 where @xmath243 we assume in the hypotheses of theorem [ thm1 ] that @xmath244 is known , and @xmath245 ( resp .
@xmath246 ) are the components of the coordinate vector @xmath247 ( resp .
@xmath248 ) with respect to cartesian coordinates . thus @xmath249 is known . by adjusting the choice of @xmath60 we also find the metric with respect to the @xmath67 coordinates where they are defined ( i.e. on all of @xmath71 ) .
by the procedure described in the previous section we can reconstruct the metric @xmath250 in the coordinates @xmath67 everywhere in the domain @xmath71 where those coordinates are defined . thus we can also reconstruct the ricci curvature tensor in these coordinates and also the scalar curvature , which we will use below . in this section ,
we show how to determine the velocity function @xmath251 and the geodesics @xmath252 in the cartesian coordinates from this information .
the reconstruction is done initially up to the first conjugate point to @xmath63 along @xmath36 .
then , as described at the end of this section , @xmath63 must be changed to allow reconstruction past this conjugate point . as observed above , the coordinate vectors in the @xmath67 coordinates are jacobi fields along @xmath80 ; in particular , if @xmath253 ( recall that @xmath254 are the coordinate vectors for the euclidean coordinates ) then the matrix @xmath255 satisfies @xmath256 supplemented with the initial data @xmath257 in fact , here we are simply adding the dependence on @xmath73 and @xmath63 to the same quantities already considered in the previous section . since parallel
translation preserves the metric , we also have the relation @xmath258 by taking determinants of the matrices on each side and then the natural log , we obtain the following formula @xmath259 recall that @xmath260 are the components of the parallel frame with respect to the cartesian coordinate vectors ( cf .
( [ fdef ] ) ) .
also , combining some of the previous formulas we have @xmath261 away from conjugate points along @xmath167 , we may invert to obtain @xmath262 which may further be combined with ( [ vrel ] ) to show @xmath263 where we use the notation @xmath264 [ rem ] a stepping " recovery procedure in the spirit of dix original method may now be presented .
we introduce a step size @xmath265 in @xmath24 , and for @xmath266 we label @xmath267 .
if we know @xmath268 , @xmath269 , and @xmath270 , then we approximate the same quantities at @xmath271 by the following strategy .
first we numerically estimate the derivatives @xmath272 then we use ( [ coc ] ) to estimate the derivatives @xmath273 we then have an estimate of @xmath274 .
next , we use this estimate to perform a forward euler step in ( [ eq : fpartr ] ) and get an approximation of @xmath275 . then , finally , we use ( [ vrel ] ) to obtain the approximation for @xmath276 .
we note that @xmath277 may be obtained through a completely independent calculation using ( [ jacobisys ] ) . because @xmath278 , we can then approximate @xmath279 as well . while the technique described in remark [ rem ] may be a direct generalization of dix method , in fact we seek a single closed system of ordinary differential equations that could be solved using any numerical scheme to give all the desired quantities .
this is possible , although the method we present here works in three or more dimensions only .
the reason for this limitation , as explained above , comes from our use of formulas ( [ ric ] ) and ( [ ric2 ] ) to express the hessian of @xmath163 in terms of the ricci curvature and the first order derivatives of @xmath163 in cartesian coordinates .
since we actually know the ricci curvature in @xmath67 coordinates given by @xmath70 , which we will label as @xmath281 , we also need the formula for the tensorial change @xmath282 now we can describe how to get the closed system of ordinary differential equations .
we differentiate ( [ eq : fpartr ] ) and use ( [ coc1 ] ) to get @xmath283 where we have suppressed the dependence on @xmath284
. we may use ( [ ric ] ) , ( [ ric2 ] ) , ( [ dfdx ] ) , and ( [ cartric ] ) to express the right - hand side of ( [ dfsys ] ) only in terms of @xmath285 , @xmath286 , derivatives of @xmath286 , and the ricci curvature @xmath287 . combining all the previous equations we now have a closed system of ordinary differential equations that may be solved uniquely up to conjugate points . in the next paragraph , we summarize the entire method for convenience . as claimed above we
have now produced a closed system of ordinary differential equations that may be solved to obtain @xmath13 and @xmath288 in cartesian coordinates .
the system is nonlinear and contains @xmath289 equations .
it may be written as @xmath290 { \bf j}_{k}^l \\[0.1 cm ] \dot { \bf j}_{k}^l \\[0.1 cm ] f^l_k \\[0.1 cm ] \frac{\partial f^l_k}{\partial { \widehat{x}}^p } \end{matrix}\right ) = \left(\begin{matrix } w_\gamma^l(r,{\widehat{x}},{\bf j},\dot { \bf j } , f,\frac{\partial f}{\partial { \widehat{x } } } ) \\[0.1 cm ] w_{{\bf j};k}^l(r,{\widehat{x}},{\bf j},\dot { \bf j } , f,\frac{\partial f}{\partial { \widehat{x } } } ) \\[0.1 cm ] w_{\dot { \bf j};k}^l(r,{\widehat{x}},{\bf j},\dot { \bf j } , f,\frac{\partial f}{\partial { \widehat{x } } } ) \\[0.1 cm ] w_{f;k}^l(r,{\widehat{x}},{\bf j},\dot { \bf j } , f,\frac{\partial f}{\partial { \widehat{x } } } ) \\[0.1 cm ] w_{\frac{\partial f}{\partial { \widehat{x}}};kp}^l(r,{\widehat{x}},{\bf j } , \dot { \bf j},f,\frac{\partial f}{\partial { \widehat{x } } } ) \end{matrix}\right ) .\ ] ] we describe how each of the `` @xmath113 '' functions on the right - hand side are to be evaluated . @xmath291 and @xmath292 are the simplest .
recalling that @xmath293 and using ( [ jacobisys ] ) they are given by @xmath294 next , again according to ( [ jacobisys ] ) , @xmath295 is given by @xmath296 since @xmath297 , @xmath298 is given by @xmath299 finally , @xmath300 is given by ( [ dfsys ] ) where we calculate @xmath301 using ( [ dfdx ] ) , and calculate @xmath302 in several steps using the values of @xmath301 already calculated , ( [ cartric ] ) , ( [ ric ] ) , and then ( [ ric2 ] ) . system ( [ w sys ] ) gives a nonlinear system of odes which can be solved to recover @xmath145 and @xmath303 up to conjugate points .
however , at the conjugate points the matrix @xmath285 will become singular , and we will not be able to continue .
note that we do not explicitly solve for @xmath13 in this system , but after we have found @xmath304 , then @xmath163 and therefore @xmath13 may be calculated from ( [ vrel ] ) . indeed since @xmath305 , ( [ vrel ] ) becomes @xmath306 by its construction , the system ( [ w sys ] ) has a solution up to the first conjugate point of @xmath63 along @xmath36 , and since the functions on the right hand side of ( [ w sys ] ) are lipschitz continuous away from the conjugate points this solution is unique .
thus we can recover @xmath13 along the entire length of @xmath36 using ( [ w sys ] ) if there is no conjugate point to @xmath63 .
however , if there is a conjugate point , then we must follow a more sophisticated strategy .
note that once we are able to calculate the matrix @xmath307 after step 1 , then we can identify all of the conjugate points of @xmath63 .
since we are only concerned with a finite length along the geodesic @xmath36 , by the morse index theorem ( see e.g. ( * ? ? ?
* theorem 15.1 ) ) there are only a finite number of points conjugate to @xmath63 , and each of these conjugate points also has only a finite number of conjugate points on the finite interval of interest
. thus we can pick an alternate value of @xmath308 such that @xmath63 and @xmath308 do not have any of the same conjugate points , and there are no conjugate points to either on @xmath309 $ ] .
performing step 1 with @xmath63 replaced by by @xmath310 , we can also consider the system ( [ w sys ] ) with @xmath63 replaced by @xmath310
. now suppose that @xmath311 is the set of times conjugate to @xmath63 in the interval @xmath312 given in decreasing order ( i.e. so that @xmath313 for all @xmath98 ) .
then for each @xmath98 there is an open interval @xmath314 containing @xmath315 such that @xmath316 does not contain any times conjugate to @xmath308 along @xmath36 .
suppose that @xmath317 is another collection of open intervals so that @xmath318 covers @xmath312 and @xmath319 $ ] for all @xmath98 where we are setting @xmath320 for some @xmath321 sufficiently small .
now , let us set @xmath322 ( recall that @xmath62 is the domain of the inverse coordinate map @xmath323 ) . by solving ( [ w sys ] ) and possibly shrinking @xmath324 we can find the wave speed @xmath13 in cartesian coordinates on the set @xmath325 . by the continuity of @xmath326 we can find an open set @xmath327 and an open interval @xmath328 containing @xmath107 and intersecting @xmath329 nontrivially such @xmath330 .
therefore , since the wave speed is known in @xmath331 we can find the parallel fields and jacobi fields in cartesian coordinates corresponding to @xmath308 for @xmath332 and @xmath333 . now ,
after possibly shrinking @xmath334 , we can solve ( [ w sys ] ) with @xmath63 replaced by @xmath308 to find the wave speed in cartesian coordinates on @xmath335 .
thus we have reconstructed the wave speed on @xmath336 .
switching the roles of @xmath63 and @xmath308 we can reconstruct the wave speed on a set of the form @xmath337 for some open set @xmath338 . continuing in this way after a finite number of steps we can eventually reconstruct the wave speed on a set of the form @xmath339 which is a neighborhood of the geodesic @xmath36 .
this completes the reconstruction in dimension three or higher .
the method of the previous section , step 2 in the reconstruction procedure , will not work in two dimensions .
the basic reason for this is that we can not determine all of the second partial derivatives of @xmath163 from the curvature of @xmath3 , but rather can only obtain the laplacian of @xmath163 as shown in ( [ scal ] ) . therefore in the two - dimensional case
we are forced to recover @xmath163 by solving ( [ scal ] ) .
we discuss this in more detail below , but first we will also revisit step 1 in the two - dimensional case where the formulae can be simplified .
the main simplification comes from the fact that in the two dimensional case the trace of @xmath340 contains all of the same information as @xmath340 itself , and so it is actually easier to consider this trace .
indeed , let us define @xmath341 in this case we have the following simple method of calculating @xmath342 away from conjugate points .
if @xmath24 is not conjugate to @xmath26 along @xmath80 , then there is a distance function defined for @xmath343 in a neighborhood of @xmath67 by @xmath344 in the seismic context this is nothing other than the local travel time along rays close to @xmath80 from @xmath345 to points near @xmath346 .
by @xcite we have @xmath347 we continue to review step 1 in the two dimensional case .
we note that the @xmath348 which appear in equation ( [ eq a ] ) are actually all scalars and the only nonzero component of @xmath349 is @xmath350 which is what we recover in ( [ eq ainit ] ) .
note also that in the two dimensional case , @xmath350 contains the same information as the sectional curvature ( if @xmath351 is chosen to have unit length with respect to @xmath3 then in fact @xmath350 is the sectional curvature ) .
the other simplification occurs in substep 4 . in the two dimensional case
the metric must have the form @xmath352 now by @xcite , @xmath353 satisfies the equation @xmath354 where @xmath355 is already known .
since the metric is also known in a neighborhood of @xmath58 we may simply solve this equation with the known initial data @xmath356 and @xmath357 in order to find the metric in the @xmath67 coordinates defined by @xmath70 .
we see that it is not necessary in this case to compute the matrix @xmath285 corresponding to the jacobi fields .
now we continue to show how step 2 may be accomplished in the two dimensional case .
the method makes use of the scalar curvature rather than the ricci curvature . in step 2
we must take a different strategy for the two dimensional case .
the difference is that we can not express the second derivatives of @xmath163 which appear in ( [ dfsys ] ) in terms of @xmath285 , @xmath286 , derivatives of @xmath286 , and the ricci curvature .
instead we use a method inspired by the treatment from ( * ? ? ?
* section 4.5.6 ) of a different problem .
after we recover the metric @xmath3 in the coordinates @xmath67 given by @xmath70 we use the scalar curvature equation ( [ scal ] ) to directly solve for @xmath163 in these coordinates .
indeed , we assume that @xmath358 are known , and so in fact we have cauchy data for @xmath163 on @xmath58 . thus @xmath163 satisfies a cauchy problem for the elliptic operator @xmath191 ( see ( [ scal ] ) ) expressed in @xmath67 coordinates .
this equation is given explicitly by @xmath359 here @xmath360 is the inverse of the matrix given by formula ( [ hatg ] ) extended to values of @xmath73 other than @xmath107 and @xmath361 is the determinant of the matrix @xmath362 .
the scalar curvature , @xmath363 , can be computed from @xmath250 .
expressed in the coordinates this is an elliptic equation although it is degenerate when @xmath24 is conjugate to @xmath63 along @xmath80 . adopting the notation of the previous section , by the unique continuation principal ( for a modern review of cauchy problems for elliptic operators see @xcite ) we can first reconstruct @xmath364 for @xmath365 .
we note , however , that this reconstruction is generally unstable ( once again see @xcite for a detailed review of the stability of this type of problem ) .
now once we have recovered @xmath163 in @xmath67 coordinates for @xmath365 the system ( [ w sys ] ) can be replaced by a significantly simpler system . indeed ,
if we combine the equation ( [ jacobisys ] ) for the jacobi field matrix with ( [ eq : fpartr ] ) and ( [ coc ] ) , then we have a closed system of ordinary differential equations which may be solved just like ( [ w sys ] ) for the higher dimensional case . for convenience
we write this system down explicitly .
the system is @xmath366 { \bf j}_{k}^l \\[0.1 cm ] \dot { \bf j}_{k}^l \\[0.1 cm ] f^l_k \\[0.1 cm ] \end{matrix}\right ) = \left(\begin{matrix } w_\gamma^l(r,{\widehat{x}},{\bf j},\dot { \bf j } , f ) \\[0.1 cm ] w_{{\bf j};k}^l(r,{\widehat{x}},{\bf j},\dot { \bf j } , f ) \\[0.1 cm ] w_{\dot { \bf j};k}^l(r,{\widehat{x}},{\bf j},\dot { \bf j } , f ) \\[0.1 cm ] w_{f;k}^l(r,{\widehat{x}},{\bf j},\dot { \bf j } , f ) \\[0.1 cm ] \end{matrix}\right ) .\ ] ] here @xmath367 , @xmath368 , and @xmath369 are given by the same formulae shown below ( [ w sys ] ) .
the last entry on the right hand side is given , according to ( [ eq : fpartr ] ) and ( [ coc ] ) , by @xmath370 solving these equations we can recover @xmath163 , and therefore also @xmath13 , in cartesian coordinates on @xmath371 .
once we have this we can find cauchy data on the surface @xmath372 for some @xmath373 for the scalar curvature equation in the coordinates given by @xmath326 , and repeat the process described above . thus we can introduce a stepping procedure as in the higher dimensional case and this completes the reconstruction in the case of two dimensions .
this theorem follows from the recovery procedure that we have presented in sections [ step1 ] through [ 2d ] .
indeed , from results in @xcite applying step 1 , which is described in section [ step1 ] , for any @xmath64 we can recover the metric @xmath3 in @xmath67 coordinates corresponding to @xmath70 for any @xmath374 .
then , in dimension @xmath375 or higher , the argument of section [ step2 ] completes the proof , while in dimension @xmath123 we must use the scalar curvature equation as described in section [ 2d ] .
the proof of theorem [ thm2 ] involves reduction to the case of theorem [ thm1 ] .
we also perform this reduction more explicitly in the case that @xmath2 is flat and @xmath13 is constant in a neighborhood of @xmath2 in the appendix .
let us begin by taking any @xmath376 .
by the hypotheses there exists a point @xmath377 .
let @xmath378 be a set of local coordinates defined on an open subset of @xmath379 containing @xmath380 and denote by @xmath381 the coordinate mapping which we suppose has range given by @xmath382 .
we construct @xmath381 such that @xmath383 .
finally , let @xmath384 denote the outward pointing unit normal , with respect to @xmath3 , vector field for @xmath2
. we will now write @xmath385 for the exterior derivative on @xmath2 . for @xmath64 and @xmath386
let us consider the set @xmath387 and mapping @xmath388 defined by @xmath389 this set and mapping can be found from the data .
now , if @xmath390 is dense in @xmath391 for all @xmath73 in a neighborhood of @xmath107 and @xmath392 then the geodesic represented by @xmath393 ( see the next paragraph ) is contained in @xmath157 and intersects @xmath2 transversally . in this case we continue .
otherwise we do not use this choice of @xmath394 .
intuitively , @xmath395 is a representation of the portion of a geodesic passing through @xmath396 which lies inside @xmath0 , and the range of @xmath397 parametrizes this geodesic segment . indeed , for every @xmath398 ,
let @xmath399 be given by @xmath400 \
g_z^{jk}(\phi^{-1}_z(\hat{x } ) ) \frac{\partial}{\partial \hat{x}^k } \\ & & \hskip.5 in + \left ( \sqrt{1 - \left | \frac{\partial}{\partial \hat{x}^j } \rho_\lambda \left ( \phi_z^{-1}(\hat{x } ) \right ) \right |_{g_z}^2 } \right ) \nu ( \phi_z^{-1}(\hat{x } ) ) \nonumber\end{aligned}\ ] ] where @xmath401 is the metric @xmath3 restricted to @xmath2 expressed in the @xmath73 coordinate frame .
then @xmath395 represents a segment of the geodesic @xmath402 .
we now begin relating the constructions we have made so far with the data required to apply theorem [ thm2 ] .
first , let us take a sufficiently small constant @xmath403 and set @xmath404 these can be constructed from the given data because @xmath405 for @xmath406 sufficiently small .
we can also construct the map @xmath407 defined by @xmath408 taking @xmath406 sufficiently small , and possibly shrinking @xmath62 we can make @xmath409 a diffeomorphism onto its image , and so @xmath410 is a coordinate map on the image of @xmath409 which , using the notation introduced earlier in the paper , we take to be @xmath58 with @xmath411 . also following the earlier notation we let @xmath412 be the outward pointing unit normal vector for @xmath58 at @xmath413 .
suppose now that @xmath64 , and @xmath414 $ ] .
if @xmath415 and @xmath26 are not conjugate along @xmath416 then there exists a @xmath398 such that @xmath417 .
now we have that @xmath418 is defined on a neighborhood @xmath419 of @xmath73 , and the set @xmath420 is then given by @xmath421 since we can recover this set and we know @xmath422 we can calculate the shape operator @xmath423 provided it exists .
therefore we may calculate @xmath119 .
since there are only a finite number of @xmath414 $ ] that are conjugate to @xmath415 along @xmath416 we may in fact obtain @xmath119 for all @xmath414 $ ] for which it is defined by continuity . finally since @xmath424
we can also find @xmath119 for all @xmath425 $ ] for which it is defined .
now suppose that @xmath426 .
provided that @xmath308 and @xmath406 are not conjugate along @xmath36 , there exists @xmath427 such that @xmath428 .
we may now repeat the previous construction with @xmath63 replaced by @xmath308 to obtain @xmath429 for all @xmath430 $ ] where it is defined , and this gives all of the data necessary to apply theorem [ thm1 ] .
therefore we can recover the wave speed @xmath13 in a neighborhood of @xmath431 . by the hypotheses @xmath157
can be covered by sets of this form , and so we can recover the wave speed @xmath13 on all of @xmath157 as claimed .
we generalized the method of dix for reconstructing a depth varying velocity in a half space , where depth is the cartesian coordinate normal to the boundary , to a procedure for reconstructing a conformally euclidean metric on a region of @xmath1 from expansions of diffraction travel times generated by scatterers in the region and measured on its boundary .
our procedure consists of two steps : in the first step , we reconstruct the directional curvature operator along geodesics as well as the metric in riemannian normal coordinates .
riemannian normal coordinates can be thought of as `` time '' coordinates as they appear in so - called seismic time migration .
we note that the directional curvature operator did not appear in the method of dix because of the class of velocity models he considered . in the second step ,
the velocity and the geodesics on which the velocity is reconstructed are obtained through a transformation to cartesian coordinates ; this can be thought of as a generalization of the `` time - to - depth '' conversion in the framework of dix original formulation . in dimension three or more
both steps are essentially formulated in terms of solving a closed system of nonlinear ordinary differential equations , for example , by application of a runge - kutta method . in dimension two
the second step requires the solution of a cauchy problem for an elliptic operator which may suffer from stability issues . through the associated discretization
, we accommodate the case of a finite number of scatterers in the manifold .
we admit the formation of caustics .
in this appendix we show more explicitly how the somewhat abstract procedure described in the proof of theorem [ thm2 ] to move from travel times , or generalized distance functions , to the shape operators for wavefronts can be done in a particular case .
we assume that @xmath0 is a lower half space and @xmath13 is constant in a neighborhood of the boundary and thus can be extended smoothly as a constant . here
we have the application in mind , and so we use the term travel time for the distance in the metric @xmath3 .
we use cartesian coordinates @xmath433 and assume that we have a flat boundary at @xmath434 so that @xmath435 and @xmath436 .
we focus on a single diffraction point , @xmath437 , and measure the travel times from @xmath437 as a function of the transverse coordinates at @xmath438 which we label as @xmath73 .
we assume that the wave speed @xmath13 is equal to a constant @xmath439 in a neighborhood of @xmath440 and write the travel time as @xmath441 which is a function of @xmath73 for @xmath73 in some open set @xmath382 .
first we can determine the tangent vector of the ray passing through a certain point @xmath73 of the boundary and corresponding to the travel time @xmath442 by @xmath443 @xmath444 ( these equations should be compared with ( [ eq : v ] ) ) .
we are using the prime notation here to indicate that the indices only run from @xmath445 to @xmath446 .
in practice it would be possible to calculate @xmath447 by finding the normal to the surface @xmath448 which is proportional to the vector @xmath449 also note that if we look at multiple wavefronts coming from different diffraction points , then we can identify contributions from diffraction points along the same ray by the fact that they must satisfy @xmath450 ( compare with ( [ eq : wtgamma ] ) ) where @xmath451 and @xmath452 correspond to the two different wavefronts .
now we choose @xmath453 and consider the surface of points with travel time @xmath63 from the given diffraction point . in the body of the paper this surface is labeled @xmath58 .
we may parametrize the surface @xmath58 by @xmath73 using the map ( labeled @xmath454 in the body of the paper ) @xmath455 it is through this map that we define the coordinates on @xmath58 . for @xmath456 we can also write a formula for the inverse of the coordinate map for the coordinates @xmath67 ( @xmath70 in the body of the paper ) .
this is @xmath457 now we may calculate the matrices @xmath458 and @xmath459 which give respectively the parallel translated fields @xmath460 and the coordinate vector fields @xmath166 ( which are also jacobi fields ) in the cartesian coordinates ( note that @xmath461 differs from @xmath462 in the body of the paper which represents the jacobi fields with respect to the parallel frame ) . to take advantage of the summation notation
we also write @xmath463 and @xmath464
. then we have @xmath465 where we may calculate ( using the notation @xmath466 ) @xmath467 the parallel fields are the same as the coordinate vectors at @xmath468 , and since for @xmath456 the wave speed is constant they are in fact constant in the cartesian frame on the same set .
thus we have , for @xmath456 , @xmath469 these matrices may also be written directly in terms of the travel time function @xmath441 . indeed if we write @xmath470 for the hessian matrix of @xmath441 , and @xmath471 for the gradient then @xmath472 and @xmath473 now we calculate the matrix @xmath474 which corresponds with the shape operator of @xmath475 with respect to the parallel frame . to begin , recall that @xmath476 where @xmath477 is the dual frame for @xmath478 .
now we can use the fact that the christoffel symbols in cartesian coordinates are zero where the wave speed is constant to calculate @xmath479 therefore @xmath480 once again , this can also be changed to an expression in terms of @xmath441 and its derivatives as @xmath481 and we may use this formula to calculate @xmath482 which we invert to give the initial conditions for the system ( [ eq a ] ) . to calculate the derivatives with respect to @xmath26 we must consider multiple wavefronts , and using the comment above identify singularities in the wavefront corresponding to diffraction points along the same ray . | we consider a region @xmath0 in @xmath1 with boundary @xmath2 and a metric @xmath3 on @xmath0 conformal to the euclidean metric .
we analyze the inverse problem , originally formulated by dix @xcite , of reconstructing @xmath3 from boundary measurements associated with the single scattering of seismic waves in this region . in our formulation the measurements
determine the shape operator of wavefronts outside of @xmath0 originating at diffraction points within @xmath0 .
we develop an explicit reconstruction procedure which consists of two steps . in the first step
we reconstruct the directional curvatures and the metric in what are essentially riemmanian normal coordinates ; in the second step we develop a conversion to cartesian coordinates .
we admit the presence of conjugate points . in dimension @xmath4
both steps involve the solution of a system of ordinary differential equations . in dimension @xmath5 the same is true for the first step , but the second step requires the solution of a cauchy problem for an elliptic operator which is unstable in general .
the first step of the procedure applies for general metrics . |
quantum computer offers great advantages over the classical computer@xcite , such as tremendous speedup , efficiently searching for unordered database , and factorizing large numbers .
quantum gates are the fundamental elements for the quantum computer .
quantum computation , quantum networks , and quantum algorithms can be decomposed into a set of universal gates and single - qubit gates@xcite .
controlled - not ( cnot ) gate or controlled - phase gate is the most popular nontrivial universal gate@xcite , and considerable progresses have been made on them@xcite . in recent years , constructions of the toffoli gate@xcite or
fredkin gate@xcite has been addressed from various perspectives , and these two universal multiqubit - control gates are the key operations in quantum networks , quantum phase estimate , fault - error quantum computing , quantum - error correction , quantum shor s algorithm@xcite , quantum grover s algorithm@xcite , and optimal long s algorithm@xcite
. a toffoli or fredkin gate can be implemented by series of nontrivial two - qubit gates along with single - qubit gates .
however , this strategy requires a number of gate steps , additional qubits , and increases the noise channels .
the synthesis of a @xmath0-qubit toffoli gate requires @xmath1 two - qubit gates@xcite .
the minimal cost of a three - qubit fredkin gate is five two - qubit gates@xcite .
therefore , it is significant to directly and efficiently implemented toffoli and fredkin gates without resorting to two- and single - qubit gates .
finding an efficient quantum computing with minimal quantum resources is the central work in quantum information process .
encoding qubits in multiple degrees of freedom ( dofs ) increases the capacity of the quantum channel , reduces the quantum resources , and are robust against the decoherence caused by the environments .
single photon is nowadays recognized as an excellent candidate for quantum information process in multiple dofs as it carries many qubit - like dofs@xcite including momentum ( linear , orbital , spatial ) , frequency , time - bin , polarization , and availability single - qubit manipulation . utilizing cross - kerr nonlinearity , sheng _ _ et al.__@xcite and liu _ _ et al.__@xcite proposed schemes for complete hyperentangled - bell - state analysis in two and three dofs , respectively . in 2012 , ren _ _
et al.__@xcite theoretically implemented hyperentangled - bell - state analysis in polarization - spatial dof via quantum dot . in 2014 , wang _ _ et al.__@xcite proposed a hyperdistillation scheme in two dofs . in 2015 , wang _
et al_.@xcite demonstrated quantum teleportation of a single photon with polarization - orbit dof .
recently , scholz _
et al_.@xcite and zhang _
et al_.@xcite demonstrated two - qubit deutsch - jozsa algorithm with a single photon in polarization - spatial dof , respectively .
et al_.@xcite constructed one - photon three - qubit universal gates .
the weak interaction in single photon level is the limitation of photonic quantum computing .
fortunately , in 2001 , knill , laflamme , and milburn@xcite proposed an interesting scheme for implementing linear optical controlled - phase gate with maximal success probability of 3/4 .
the strategy adopted to overcome undeterministic is to employ cross - kerr nonlinearity@xcite or photon - matter platforms@xcite . up to now , the giant cross - kerr nonlinearity is still a challenge in experiment . in recent years
, wide attention has been drawn to couple flying photonic qubit to stationary matter qubit as it provides a platform for scalable photonic or matter quantum computing .
ren _ _ et al.__@xcite proposed a spatial - polarization hyper - controlled - not gate with coupled quantum dot .
ren _ _ et al.__@xcite , wei _ _ et al.__@xcite , and wang _ _ et al.__@xcite presented schemes for multiphoton hyper - parallel quantum computing assisted by matter qubits coupled with resonators .
isolated electron - spin in diamond nitrogen - vacancy ( nv ) center has been recognized as an unique and promising candidate for solid - state quantum computing .
the appeal feature of the diamond nv center is its urtralong coherence time ( @xmath2s ) even at room temperature@xcite , and diamond nv center has been received strong interest in recent years .
the initialization@xcite , manipulation@xcite , storage@xcite , readout@xcite , and the nontrivial interactions between two diamond nv spins@xcite , which are the necessary operations for quantum computing , have been well experimentally demonstrated .
there are also a number of hallmark other demonstrations , such as the quantum deutsch - jozsa algorithm@xcite , controlled - rot gate@xcite , decoherence - protected controlled - rotation gates@xcite , and holonomic cnot gate@xcite . based on photon - nv entangled platform@xcite , hanson _
et al_.@xcite experimentally demonstrated quantum teleportation and bell inequality between distant diamond nv centers , and several proposals for implementing parallel and hyper - parallel quantum computing were also proposed@xcite . in this article , we investigate the implementations of the four - qubit toffoli and fredkin gates acting on two - photon systems via the diamond nv centers coupled to resonators .
the toffoli gate flips the polarization or spatial state of the target qubit conditional on the polarization and spatial states of the control qubits .
the fredkin gate swaps the polarization and spatial states of the target photon depending on the polarization and spatial states of the control photon . in our schemes ,
the two gates are heralded by the spins of the electrons in diamond nv centers .
additional photons , necessary for the cross - kerr - based and parity - check - based optical quantum computing , are not required .
compared to the ones in one dof , our schemes only require the half quantum resource , and are robust against the decoherence .
the estimations of our devices show that our schemes are possible in experiment with current state of technology .
* the optical property of an nv - cavity platform . * the toffoli gates we considered are equivalent to the controlled - controlled - controlled - phase flip ( cccpf ) gate up to two hadamard ( @xmath3 ) gates acting on the target qubits , i.e. , @xmath4 . here
, we employ the nv - cavity entangled platform to overcome the weak interactions between two single photons involved in the toffoli gate . as shown in fig .
1(a ) , @xmath5 and @xmath6 ( marked by @xmath7 ) are the ground triple states of the diamond nv center . in absence of a magnetic field , @xmath8 and @xmath9
are split with 2.87 ghz due to the spin - spin interaction , and the corresponding transitions , @xmath10 , are in the microwave regime .
the six electron excited states include@xcite @xmath11 , @xmath12 , @xmath13 , @xmath14 , @xmath15 , and @xmath16 .
@xmath17 , and @xmath18 , @xmath19 are orbital states with angular momentum projections @xmath20 and @xmath21 , respectively .
the energy gap arising from the spin - spin and spin - orbit interactions protects state @xmath22 against small strain and magnetic filed .
the transitions , @xmath23 , are driven by the resonator optical pluses with @xmath24 and @xmath25-polarized at 637 nm , respectively ( see fig .
1(b ) ) . therefore ,
the device shown in fig .
1 can be used as a quantum interface between optical and solid - state systems for scalable quantum information process .
-type electron energy structure of a trapped diamond nv center .
transitions @xmath26 are driven by the @xmath24 and @xmath25 photons , respectively .
, width=8 ] compared with the coupling of an unconfined diamond nv center , trapping a diamond nv center into an optical cavity enhances the coupling between the incident photon and the diamond nv center . nowadays , various of optical microcavities , such as photonic crystal cavity@xcite , fiber - based fabry - perot cavity@xcite , microsphere@xcite , toroidal resonators@xcite , coupling to the diamond nv centers have been experimentally demonstrated .
suppose that the diamond nv center coupling with a whispering - gallery - mode microresonator with two tapered fibers ( see fig .
1 ) . if the diamond nv in the state @xmath27 , the incoming @xmath25-polarized photon via port @xmath28 ( port @xmath29 ) senses a cold cavity , and departs through port @xmath30 ( @xmath31 ) with phase shift @xmath32 , while the incoming @xmath24-polarized photon via port @xmath28 ( port @xmath29 ) senses a hot cavity , and leaves from the cavity via port @xmath31 ( @xmath30 ) with phase shift @xmath33 . in the same way , if the diamond nv center in the state @xmath34 , the input @xmath24-polarized photon senses a cold cavity and it is transmitted , while the @xmath25-polarized photon senses the hot cavity and it is reflected . the reflection and transmission coefficients of the nv - cavity system
can be obtained by solving the heisenberg equations for the cavity mode @xmath28 with frequency @xmath35 and the diamond nv lowering operator @xmath36 with frequency @xmath37 , that is@xcite , @xmath38a - g\sigma_{- } - \sqrt{\kappa}(a_{\rm{in}}+a'_{\rm{in}})-g\sigma_{-}-h , \\ % & \dot{\sigma}_{- } = -\left[i(\omega_{0}-\omega_p)+\frac{\gamma}{2}\right]\sigma_{-}-g\sigma_z a- f. \end{split}\end{aligned}\ ] ] here , @xmath39 and @xmath40 are the decay rates of the cavity field and the diamond nv dipole , respectively .
@xmath41 is the cavity intrinsic loss rate ( side leakage ) .
@xmath42 is the coupling rate between the diamond nv center and the resonator .
@xmath43 is the pauli operator .
@xmath44 and @xmath45 are the noise operators .
the input field is connected with the output filed by @xmath46 and @xmath47 .
taking the weak excitation limitation , i.e. , @xmath48 , through the operation , the transmission and reflection coefficients of the nv - cavity system can be respectively written as@xcite @xmath49}{\big[i(\omega_0-\omega_p)+\frac{\gamma}{2}\big]\big[i(\omega_c-\omega_p)+\kappa+\frac{\kappa_s}{2}\big]+g^2},\\ % & r(\omega_p ) = \frac{\big[i(\omega_0-\omega_p)+\frac{\gamma}{2}\big]\big[i(\omega_c-\omega_p)+\frac{\kappa_s}{2}\big]+g^2}{\big[i(\omega_0-\omega_p)+\frac{\gamma}{2}\big]\big[i(\omega_c-\omega_p)+\kappa+\frac{\kappa_s}{2}\big]+g^2}. \end{split}\end{aligned}\ ] ] under the resonator condition @xmath50 , if the purcell factor @xmath51 with @xmath52 is ignored , the cavity mode of the hot cavity is reflected , i.e. , @xmath53 and @xmath54 . if @xmath55 , the cavity mode of the bare cavity is transmission , i.e. , @xmath56 and @xmath57 .
that is , the input - output relations between the single photon and nv center can be summarized as @xmath58 here , @xmath59 and @xmath60 with @xmath61 , @xmath62 and @xmath63 with @xmath64 , the polarization of the photon is defined with respect to the @xmath65 axis ( the diamond nv center axis ) .
* two - photon four - qubit toffoli gate acting on the polarization and spatial dofs . * the quantum circuit we designed , as shown in fig .
2 , performs the following operation on a two - photon four - qubit state @xmath66(\gamma_{1}|r_2\rangle + \gamma_{2}|l_2\rangle)(\delta_{1}|b_1\rangle + \delta_{2}|b_2\rangle ) \\ & + \alpha_{2}\beta_{2}\gamma_{1}|l_1\rangle |a_2\rangle|r_2\rangle ( \delta_{1}|b_1\rangle + \delta_{2}|b_2\rangle)+ \alpha_{2}\beta_{2}\gamma_{2}|l_1\rangle|a_2\rangle|l_2\rangle ( \delta_{1}|b_1\rangle - \delta_{2}|b_2\rangle ) , \end{split}\end{aligned}\ ] ] where @xmath67 , @xmath68 , @xmath69 , @xmath70 , @xmath71 , @xmath72 , @xmath73 , and @xmath74 are arbitrary complex parameters satisfying @xmath75 , @xmath76 , @xmath77 , and @xmath78 .
the indices 1 and 2 of @xmath79 ( or @xmath80 ) denote the @xmath79- ( or @xmath80- ) polarized wave - packets of the photon 1 and photon 2 , respectively .
@xmath81 ( @xmath82 ) and @xmath83 ( @xmath84 ) represent photon 1 ( photon 2 ) emitted from the spatial modes 1 and 2 , respectively .
( @xmath85 ) , transmits the @xmath79-polarized and reflects the @xmath80-polarized components , respectively .
single - qubit operations @xmath86 can be achieved by employing two half - wave plates oriented at @xmath87 and @xmath88 , respectively .
the local single - qubit operations , heralded by the outcomes the diamond nv center , via a feed - forward process are not shown in the sketch .
, width=6 ] in fig .
2 , we first use the circularly polarized beam splitter , cpbs@xmath89 , to depart the wave - packets , @xmath79 and @xmath80 , of photon 1 emitted from the spatial mode @xmath83 , i.e. , the @xmath90-component is transmitted and the @xmath91-component is reflected . the @xmath90-component does not interact with the diamond nv center and arrives at cpbs@xmath92 directly . before and after the @xmath91-component interacts with the block composed of cpbs@xmath93 , nv , and cpbs@xmath94 , the two single - qubit rotations @xmath95 and @xmath96 are respectively applied on photon 1 simultaneously with two hadamard operations @xmath97s applied on the diamond nv center .
subsequently , the @xmath90-component and @xmath91-component mix at cpbs@xmath92 . the operations ( @xmath98 )
transform the system composed of the photon 1 , photon 2 , and diamond nv center from the initial state @xmath99 into @xmath100 . here , @xmath101
@xmath102 the hadamard transformation @xmath97 can be achieved by employing @xmath103 pulse and it completes the following transformations @xmath104 can be achieved by employing two half - wave plates oriented at @xmath87 and @xmath88 , respectively . in the standard basis @xmath105 , @xmath106 can be written as @xmath107 next , photon 2 in the spatial mode @xmath84 is injected into the block . before and after photon 2
interacts with the block , @xmath108 are applied on it , respectively .
the above operations ( @xmath109 ) transform @xmath100 into @xmath110|+\rangle .
\end{split}\end{aligned}\ ] ] third , we measure the electron spin state of the diamond nv center in the basis @xmath111 .
if the diamond nv center in the state @xmath112 , we perform a phase shift @xmath113 on the photon emitted from the spatial mode @xmath84 , i.e. , @xmath114 and @xmath115 .
if the diamond nv in the state @xmath116 , classical feed - forward operations , phase shift @xmath113 and @xmath117 , are performed on the photons emitted from the spatial modes @xmath84 and @xmath83 , respectively .
after the above operations , the state of the system composed the two photons becomes @xmath118 .
\end{split}\end{aligned}\ ] ] it is obvious that the quantum circuit shown in fig .
2 determinately implementing a cccpf gate which flips the phase of the input state if and only if the system initially is in the state @xmath119 , and has no change otherwise . therefore ,
a two - photon four - qubit toffoli gate which performs a not operation on the spatial ( or polarization ) state of the photon 2 depending on the states of other qubits can be achieved by the scheme depicted by fig . 2 up to two hadamard gates .
* two - photon four - qubit fredkin gate acting on the polarization and spatial dofs . * the fredkin gate swaps the states of the two target qubits conditional on the states of the control qubits .
the quantum circuit depicted by fig .
3 implements a fredkin gate which completes the transformation @xmath120 as illustrated in fig . 3 , such fredkin gate can be implemented by three steps .
we first inject the photon 1 ( encoded as the control photon ) emitted from the spatial mode @xmath83 into the block composed of cpbs@xmath93 , phase shifters @xmath121 and @xmath122 , nv , and cpbs@xmath94 . before and after the photon interacts with such block , hadamard operations , @xmath123 and @xmath124 , are respectively applied on it with the half - wave plates oriented at 22.5@xmath125 simultaneously with two hadamard operations @xmath97s applied on the diamond nv centers .
the above operations ( @xmath126 ) transform the system composed of the two single photons and one diamond nv center from the initial state @xmath127 into @xmath128 . here , @xmath129
@xmath130 ( \gamma_{1}|r_2\rangle + \gamma_{2}|l_2\rangle ) ( \delta_{1}|b_1\rangle + \delta_{2}|b_2\rangle ) .
\end{split}\end{aligned}\ ] ] in the standard basis @xmath131 , hadamard operation @xmath132 can be expressed as @xmath133 phase shifter , @xmath121 or @xmath122 , introduces a phase shift @xmath113 to the passing photon , i.e. , @xmath134 and @xmath135 . can be achieved by employing a half - wave plate oriented at @xmath88 .
phase shifter @xmath136 completes the operations @xmath134 and @xmath135 .
the local classical feed - forward single - qubit operations are not shown in the sketch .
, width=7 ] subsequently , photon 2 emitted form the spatial mode @xmath82 or @xmath84 is injected and then arrives at cpbs@xmath137 .
cpbs@xmath137 transforms @xmath128 into @xmath138 \\&\otimes ( \gamma_{1}\delta_{1}|r_2\rangle|b_2\rangle + \gamma_{1}\delta_{2}|r_2\rangle|b_1\rangle + \gamma_{2}\delta_{1}|l_2\rangle|b_1\rangle + \gamma_{2}\delta_{2}|l_2\rangle|b_2\rangle ) .
\end{split}\end{aligned}\ ] ] the wave - packets emitted from the spatial mode @xmath82 pass through the block composed of cpbs@xmath93 , phase shifters @xmath121 and @xmath122 , nv , and cpbs@xmath94 , and such block transforms the state of the whole system into @xmath139 after the wave - packets of photon 2 mix at cpbs@xmath140 , the state of the whole system becomes @xmath141 we lastly measure the output state of the diamond nv center in the basis @xmath111 .
if the diamond nv center in the state @xmath112 , a single - qubit operation @xmath43 is performed on the photon emitted from the spatial mode @xmath83 . if the diamond nv center in the state @xmath116 , no feed - forward operation is applied . after the above operations ,
the state of the composite system is projected onto @xmath142(\gamma_{1}|r_2\rangle+\gamma_{2}|l_2\rangle)(\delta_{1}|b_1\rangle+\delta_{2}|b_2\rangle)\\ & % + \alpha_{2}\beta_{2}|l_1\rangle|a_2\rangle(\delta_{1}|r_2\rangle+\delta_{2}|l_2\rangle)(\gamma_{1}|b_1\rangle+\gamma_{2}|b_2\rangle ) .
\end{split}\end{aligned}\ ] ] therefore , the quantum circuit shown in fig .
3 completes a two - photon four - qubit fredkin gate which swaps the polarization and spatial states of the target photon when the control photon is in the state @xmath143 , and remain in their initial states otherwise .
the most popular universal libraries for multiqubit system are @xmath144toffoli or fredkin gate and hadamard gates@xmath145 .
implementations of the toffoli and fredkin gates have been attracted strong interesting in recent years , and most of them focus on the three - qubit system with one dof .
the traditionally three - qubit toffoli and fredkin gates have been proposed in various systems , such as hybrid photon - matter@xcite , linear optics@xcite , single - photon@xcite , nuclear magnetic resonance@xcite , ion trap@xcite , atom@xcite , quantum dot@xcite , diamond nv center@xcite and superconducting@xcite . in this work ,
we have proposed two schemes for implementing two - photon four - qubit toffoli and fredkin gates via nv - cavity interactions .
the individual interactions between computational qubits are achieved by employing linear optical elements and diamond nv centers .
the diamond nv center can be created with excellent optical coherence property , i.e. , the zero - phonon line ( zpl ) exhibiting the spectral diffusion close to the optical line width ( @xmath146mhz)@xcite . the enhancement of the zpl emission for the diamond nv center is particularly important as the diamond nv center can be served as a photon - emitter for quantum information process , and it has been enhanced from 3 - 4% to 70% of the total emission@xcite . in summary , we have designed two compact quantum circuits for determinately implementing two - photon four - qubit toffoli and fredkin gates through cavity - assisted interactions . compared to the traditional toffoli and fredkin gates with one dof , the gates presented are acting on both the polarization dof and the spatial dof , are robust against the decoherence caused by their environments , and reduce the photons from four to two .
the traditional two - body interactions in one dof do not appear in naturally physical systems and can not be implemented by solely using linear optical elements .
the two - body interactions involved in one single photon , which necessary for our toffoli and fredkin gates , can be achieved by employing linear optics .
additional photons , necessary for the cross - kerr - based , cluster - based , or parity - check based quantum computing , are not required in our schemes , and the diamond nv center only provides a platform for photon - photon interactions . compared to the ones synthesized by cnot and single - qubit gates , our schemes are not only robust against the environment noises but also reduce the complexity in experiment .
the length of the synthesis of a four - qubit toffoli gate is 16 nontrivial two - qubit gates @xcite .
the average fidelities and efficiencies show that our schemes may be possible with current technology .
* the average fidelities and efficiencies of the toffoli and fredkin gates . *
the key element of our schemes is the nv - cavity emitter . in the idealistic case ,
the optical transition rules of the emitter is perfect , however , the phase and amplitude of the incident photon are not perfect in experiment , and they are the functions of the @xmath39 , @xmath147 , and @xmath42 . if the side leakage is taken into account , there may be the reflection ( transmission ) part with coefficient @xmath148 ( @xmath149 ) of the incident photon in the cold ( hot ) cavity .
therefore , in the realistic case , the optical transition rules of the incident photon described by eq .
( [ eq3 ] ) should be changed with @xmath150 , @xmath151 , @xmath152 , @xmath153 , @xmath154 , @xmath155 , @xmath156 , and @xmath157 .
it is note that the above undesirable parts will degrade the performance of the emitter .
hence , in order to quality the performances of our schemes , we should investigate the average fidelity , which is defined as @xmath158 in experiment . here
the overline indicates average over all possible input ( output ) states . @xmath159 and @xmath160 are the normalized output states of the system in the realistic and idealistic cases , respectively .
therefore , the average fidelity of our toffoli or fredkin gate , @xmath161 or @xmath162 , can be calculated by @xmath163 where @xmath164 , @xmath165 , @xmath166 , @xmath167 , @xmath168 , @xmath169 , and @xmath170 , @xmath171 are taken for eqs .
( [ eq4]-[eq18 ] ) . the average fidelity of our toffoli and fredkin gates as functions of @xmath172 and @xmath173 are plotted in fig .
4 , respectively . and
( b ) , fredkin gate @xmath174 . , title="fig:",width=6 ] and ( b ) , fredkin gate @xmath174 . ,
title="fig:",width=6 ] efficiency is another powerful and practical tool for qualifying the performance of the setup . in experiment ,
the efficiency is defined as @xmath175 , i.e. , the yield of the incident photons . @xmath176 and @xmath177 are the numbers of gate s output and input photons , respectively .
the average efficiencies of our toffoli and fredkin gates , @xmath178 and @xmath179 , are calculated by @xmath180 where @xmath164 , @xmath165 , @xmath166 , @xmath167 , @xmath168 , @xmath169 , and @xmath170 , @xmath171 . by calculating
, one can find that @xmath181 the date shows @xmath178 ( @xmath179 ) against @xmath172 and @xmath173 are presented in fig .
5 . and
fredkin gate @xmath182 . , width=6 ] from figure 4 - 5 , one can see that the normalized side leakage and the normalized cavity coupling strength impact the gate s fidelity and efficiency . the higher @xmath172 and the lower @xmath173 , the higher fidelity and efficiency . the high fidelities and efficiencies of our toffoli and fredkin gates can be achieved .
for example , in the case @xmath183 and @xmath184 , @xmath185 , @xmath1860.979516 with @xmath187 , and @xmath188 and @xmath184 , @xmath189 , @xmath1860.868208 with @xmath190 .
our schemes are deterministic if the unavoidable effect of photon loss and imperfections are neglected .
our schemes in practice working with photon loss , non - ideal single - photon sources , unbalanced cpbss and bss , timing jitter , spin flips during the optical excitation , the mix between state @xmath22 and other excited states , signal to noise in the zpl channel , imperfect spin initialization and detection .
these imperfections degrade the fidelities and efficiencies of our two gates , and the imperfections which are due to the technical imperfection will be improved with further technical advances .
given the current technology , our schemes are more efficient than the traditional ones with one dof , the synthesis - based ones which are rely on cnot gates and single rotations , the parity - check - based ones , and they may be possible in experiment .
this work is supported by the national natural science foundation of china under grant nos .
11604012 and 11547138 , and the fundamental research funds for the central universities under grant no . 06500024 .
liu , q. , wang , g. y. , ai , q. , zhang , m. & deng , f. g. complete nondestructive analysis of two - photon six - qubit hyperentangled bell states assisted by cross - kerr nonlinearity . _
rep . _ * 6 * , 22016 ( 2016 ) .
ren , b. c. , wei , h. r. , hua , m. , li , t. & deng , f. g. complete hyperentangled - bell - state analysis for photon systems assisted by quantum - dot spins in optical microcavities . _
express _ * 20 * , 2466424677 ( 2012 ) .
abouraddy , a. f. , giuseppe , g. d. , yarnall , t. m. , teich , m. c. & saleh , b. e. a. implementing one - photon three - qubit quantum gates using spatial light modulators .
rev . a _ * 86 * , 050303(r ) ( 2012 ) . hu , c. y. ,
young , a. , obrien , j. l. , munro , w. j. & rarity , j. g. giant optical faraday rotation induced by a single - electron spin in a quantum dot : applications to entangling remote spins via a single photon .
b _ * 78 * , 085307 ( 2008 ) .
ren , b. c. , wei , h. r. & deng , f. g. deterministic photonic spatial - polarization hyper - controlled - not gate assisted by a quantum dot inside a one - side optical microcavity .
_ laser phys .
* 10 * , 095202 ( 2013 ) .
wei , h. r. , deng , f. g. & long , g. l. hyper - parallel toffoli gate on three - photon system with two degrees of freedom assisted by single - sided optical microcavities . _
express _ * 24 * , 1861918630 ( 2016 ) .
bar - gill , n. , pham , l. m. , jarmola , a. , budker , d. & walsworth , r. l. solid - state electronic spin coherence time approaching one second .
* 4 * , 1743 ( 2013 ) .
gaebel , t. _ et al .
_ room - temperature coherent coupling of single spins in diamond .
* 2 * , 408413 ( 2006 ) .
fuchs , g. d. , dobrovitski , v. v. , toyli , d. m. , heremans , f. j. & awschalom , d. d. gigahertz dynamics of a strongly driven single quantum spin .
_ science _ * 326 * , 15201522 ( 2009 ) .
shim , j. h. , niemeyer , i. , zhang , j. & suter , d. room - temperature high - speed nuclear - spin quantum memory in diamond .
rev . a _ * 87 * , 012301 ( 2013 ) .
bernien , h. _ et al .
_ heralded entanglement between solid - state qubits separated by three metres
. _ nature _ * 497 * , 8690 ( 2013 ) .
hensen , b. _ et al .
_ loophole - free bell inequality violation using electron spins separated by 1.3 kilometres .
_ nature _ * 526 * , 682686 ( 2015 ) .
england , d. g. , fisher , k. a. g. , maclean , j. p. w. & bustard , p. j. storage and retrieval of thz - bandwidth single photons using a room - temperature diamond quantum memory .
lett . _ * 114 * , 053602 ( 2015 ) .
wang , c. , zhang , y. , jiao , r. z. & jin , g. s. universal quantum controlled phase gates on photonic qubits based on nitrogen vacancy centers and microcavity resonators .
express _ * 21 * , 1925219260 ( 2013 ) .
faraon , a. , santori , c. , huang , z. , acosta , v. m. & beausoleil , r. g. coupling of nitrogen - vacancy centers to photonic crystal cavities in monocrystalline diamond .
lett . _ * 109 * , 033604 ( 2012 ) . | it is desirable to implement an efficient quantum information process demanding fewer quantum resources .
we designed two compact quantum circuits for determinately implementing four - qubit toffoli and fredkin gates on single - photon systems in both the polarization and spatial degrees of freedom ( dofs ) via diamond nitrogen - vacancy ( nv ) centers in resonators .
the gates are heralded by the electron spins associated with the diamond nv centers .
in contrast to the ones with one dof , our implementations reduce the quantum resource and are robust against the decoherence .
evaluations of fidelities and efficiencies of our gates show that our schemes may be implemented with current technology . |
na@xmath6coo@xmath1 was originally studied as a kind of rechargeable battery material@xcite and thermoelectric material@xcite .
more recently , the discovery of the superconductivity in its hydrated x=0.35 compound@xcite , where the effect of hydration is found to be limited to lattice expansion@xcite , makes it receive a renewed interest .
na@xmath6coo@xmath1 has a crystal structure consisting of 2d triangular lattice co sheets , octahedrally coordinated with o above and below the co planes , and layers of na ions sandwiched between the coo@xmath1 sheets .
the phase diagram of na@xmath6coo@xmath1 has been determined by changing the na content @xmath7 using a series of chemical reactions@xcite .
their electronic and magnetic properties are found to be strongly dependent on the number of charge carriers introduced by deviations from stoichiometry in the na sublattice .
theoretical studies on the electronic structure of na@xmath6coo@xmath1 are very active .
the early work by singh @xcitefocused on the magnetic properties .
he predicted a weak instabilities of itinerant ferromagnetic character in range @xmath8 to @xmath9 , with competing but weaker itinerant antiferromagnetic solution .
he used the virtual crystal approximation , instead of supercell model , to describe the partially occupation of na sites . as a result
, the possibility of sodium ion and co@xmath10/co@xmath4 charge ordering is not allowed . using pseudopotential plane wave method , ni _ _
et al.__@xcite have optimized the geometries for na@xmath6coo@xmath1 ( @xmath11 , 0.5 , 0.75 and @xmath12 ) with a @xmath13 supercell .
they found that the itinerant ferromagnetism is strongly suppressed by the local distortions of the oxygen around the cobalt . with optimized geometries , they identified a phase transition from wide - band ferromagnetic to narrow band paramagnetic metals with @xmath7 increasing .
by carefully examining the doped electron density , marianetti _ _ et al.__@xcite found that the rigid band model is not suitable for this na doped system . a rehybridization driven by a competition between the on - site coulomb interaction and the @xmath14-oxygen hybridization
is identified .
they have also used a modified hubbard model to study this kind of rehybridization . within the framework of density functional theory ( dft ) ,
the on - site coulomb correlation effects can be dealt with dft+u method@xcite . using this method , zou _ _
et al.__@xcite have studied the electronic structure of na@xmath6coo@xmath1 with varying @xmath15 ratio .
they did not find critical change of the band structure near @xmath16 , except a notable decrease of the dos at @xmath16 . to consider the relations between correlation and charge ordering , pickett and coauthors@xcite have applied the dft+u method to a @xmath17 supercell of na@xmath6coo@xmath1 at @xmath18 and @xmath19 .
they found that the parameter @xmath3 is critical to charge and spin orderings ( for small supercell , the so called charge disproportionation always leads to charge ordering ) .
they suggested that the phase diagram of na@xmath6coo@xmath1 is characterized by a crossover from effective single - band character with @xmath20 for @xmath21 into a three - band regime for @xmath22 , where @xmath23 and correlation effects are substantially reduced .
na@xmath0coo@xmath1 is of special interests among the na@xmath6coo@xmath1 family . in the recent experimental phase diagram of na@xmath6coo@xmath1 ,
a novel sodium ion ordered insulating state at @xmath24 was found@xcite , with the most strongly developed superstructure@xcite .
this insulating state separates two kinds of metallic phase , _
i.e. _ curie - weiss metal for @xmath21 and paramagnetic metal for @xmath22 .
it is suggested that the occurence of the insulating state indicates a strong interaction between the ions and holes even though they occupy separate layers , and a charge ordering is thought to accompany the sodium ion ordering@xcite .
therefore it is interesting to compare the electronic structure based on this experimental sodium ion ordered structure with previous theoretical results to identify the effect of na ordering .
for example , it will be interesting to see if the na ordering will produce a co charge ordering at dft level . on the other hand , the correlation effects of na@xmath0coo@xmath1
is considered to be a betweenness comparing to small values for @xmath21 and relatively large values for @xmath22@xcite , which demands a study at the dft+u level to identify the effect of correlation .
unfortunately , there is no theoretical work in the literature for this important @xmath24 case with both na ordering and correlation effects being properly considered . in this paper , we report the electronic structure of na@xmath0coo@xmath1 focusing on the possibility of charge ordering and based on the experimental superstructure at both dft and dft+u level .
the calculations in this work were performed with the vienna _ ab initio _ simulation package ( vasp)@xcite , which is a first - principles plane - wave code , treating the exchange and correlation in the dft scheme . in this study ,
the perdew - wang functional form@xcite of generalized gradient approximation ( gga ) was used . for spin - polarized calculations , the spin interpolation of vosko _ _ et al.__@xcite was also adopted .
the projector augmented wave ( paw)@xcite method in its implementation of kresse and joubert@xcite was used to describe the electron - ion interaction .
the kohn - sham equations were solved via a davidson - block iteration scheme@xcite .
a @xmath25 supercell was introduced to describe the experimental sodium ion ordering .
brillouin zone ( bz ) integrations were performed on a well converged grid of @xmath26 monkhorst - pack @xcite special points .
the total energy and density of states ( dos ) were calculated using the linear tetrahedron method with blochl corrections@xcite .
the plane wave kinetic energy cutoff was fixed to 500 ev .
s approach@xcite was adopted for dft+u calculations . since this dft+u
functional depends only on the difference of hubbard parameter @xmath3 and screened exchange parameter @xmath27 , @xmath27 was fixed to 1 ev during all dft+u calculations .
the hubbard on - site coulomb interaction is applied to co @xmath2 orbitals only .
geometry relaxation is performed at a spin unpolarized gga level .
a @xmath25 superstructure of na@xmath0coo@xmath1 has been found by powder neutron diffraction experiment@xcite .
this superstructure has the @xmath28 space group symmetry , with a cell formula : na@xmath29co@xmath30o@xmath31 . in this experimental structure model ,
the na1 and na2 sites ( @xmath32 and @xmath33 wyckoff sites for original @xmath34 space group , and @xmath32 and @xmath35 sites for this @xmath28 space group respectively ) are equally occupied , and the ordered na ions form one - dimensional zigzag chains .
two types of co ions ( @xmath36 and @xmath37 ) , which differ subtly in their coordination by oxygen , are also in chains .
the existence of two kinds of co sites allows the emergence of two inequivalent co ions , namely co@xmath10 and co@xmath4 , in the process of self - consistency .
therefore this structure model can be used to study the possibility of charge ordering .
the geometry optimization , starting from the experimental structure , reaches its convergence until all forces vanished within 0.01 ev / . the optimized geometry is shown in fig .
[ geometry ] . during the optimization ,
the cell parameters are fixed to the experimental values ( @xmath38 , b=5.630 , and @xmath39 ) , which results in only a neglectable stress less than 20 kb .
as listed in table [ tbl : geo ] , relaxations of atomic positions from the experimental ones are also found to be very small .
the largest relaxation comes from the first two na atoms along @xmath7 direction ( 0.06 ) , typical relaxations are about [email protected] .
the calculated two kinds of co - o distances ( 1.91 and 1.88 ) are very similar to experimental values ( 1.90 and 1.86 ) . in an ideal structure model of na@xmath6coo@xmath1 , co and o may form layers of edge - shared coo@xmath41 octahedra in a triangular lattice , but distortion of the octahedra with the variation of the o hight @xmath42 from its ideal value is always exists in real system .
our optimized @xmath42 varies from 0.0845 @xmath43 to 0.0892 @xmath43 for different o sites , resulting in the co - o - co angles varying from 96.2@xmath44 to 97.0@xmath44 . comparing to the 90@xmath44 angle in the undistorted octahedra
, the optimized structure shows a considerable distortion of the oxygen octahedra . for @xmath45 supercell , we get a slightly smaller distortion , but ni _ _ et al.__@xcite get a significantly larger one .
we notice that they have used a relatively small kinetic energy cutoff in their study .
we can clearly identify three well separated manifolds in the paramagnetic gga band structure and dos of na@xmath0coo@xmath1 , as shown in fig .
[ ggaband ] .
the lowest in energy is the o @xmath46 bands , with a bandwidth about 5.5 ev ( only part of them are shown in the figure ) .
the other two manifolds come from the co @xmath2 orbitals , which are splitted into a lower lying @xmath47 manifold and an upper lying @xmath14 manifold , separated by approximately 2.5 ev according to the oxygen octahedral crystal field .
although these three manifolds are well separated in energy , the hybridizations of o @xmath46 states with both the @xmath47 and @xmath14 states are notable .
the fermi energy locates at the upper edge of the @xmath47 manifold , indicating a metallic gga ground state .
the low energy properties of na@xmath0coo@xmath1 will be determined by the @xmath47 manifold , which has a bandwidth of 1.6 ev .
the trigonal symmetry of the co sites further splits the @xmath47 states into one with @xmath48 symmetry and a doubly degenerate @xmath49 pair . in fig .
[ ggaband ] , we also plot the @xmath50 ( co @xmath51 ) partial dos . as found by lee _ _
et al.__@xcite , the @xmath50 states have a bandwidth almost identical to that of the whole @xmath47 manifold , which indicates that the widely used single - band model may not be sufficient enough to describe the gga electronic structure of this material .
fermi surface often plays an important role in the electronic structure of materials .
our gga fermi surfaces for the five bands across fermi level are shown in fig .
[ fs ] , which are obtained by a @xmath52 bz sampling and viewed by xcrysden package@xcite .
lda / gga band structure is expected to give valuable and accurate information about the fermi surface even in system with strong correlations@xcite . in our supercell structure , there will be band folding from the first bz of the original hexagonal lattice ( black hexagon in the fig .
[ fs]a ) , and the fermi surface thus becomes very complex .
we notice that , on the other hand , the band structure of the superstructure geometry is far from simply band folding , for example , the fermi surface of the fourth band ( band 99 ) shows a prominent 3d character , which has not been reported in previous virtual crystal study@xcite . according to the complexity of the present fermi surfaces , nesting effect for na@xmath0coo@xmath1 may not be very strong , and we thus do nt expect to find corresponding spin density wave ( sdw ) state .
spin - polarized calculations are carried out in order to address the possibility of magnetic ordering . when the ordering is constrained to be ferromagnetic ( fm ) ,
a stable low spin solution is obtained , with magnetic moments of 0.4 and 0.45 @xmath5 for co1 and co2 respectively .
but the corresponding energy gain from ferromagnetic instability is only 26 mev / co .
the moment values are somewhat smaller than that ( 0.5 @xmath5 ) reported by singh@xcite under virtual crystal approximation , which may due to the difference between their structural model and ours . in our structure , the triangular symmetry is slightly broken .
antiferromagnetic ( afm ) ordering is frustrated on the 2d triangular lattice .
but if the magnetic moments on a sublattice vanish , the left other sublattice may geometrically permit the presence of afm ordering .
here we get an afm model ( afm1 ) by fixing the magnetic moments at co1 sites to be zero . as shown in fig .
[ geometry ] , the co2 sites form a 1d chain in the @xmath7-@xmath53 plane , and it is easy to establish afm ordering within this chain . in a unit cell , the two afm chains are set to be out of phase .
the afm1 model results in a magnetic moments of co2 ions about 0.39 @xmath5 , which is much larger the result ( 0.21 @xmath5 ) for the singh s afm model@xcite .
we notice that in that model each co ion has also two nearest neighbors of like spin in addition to four of opposite spin .
another afm model ( afm2 ) is constructed by alter the direction of magnetic moments of neighboring fm co layers .
although the interlayer exchange interaction is expected to be small , we get lower energy for afm2 state than that of fm state . until now
, we get a metallic ground state with intra - layer fm ordering and inter - layer afm ordering . comparing with the results of singh@xcite
, we may conclude that na ordering alone will not qualitatively affect the gga electronic structure to generate the band gap and charge ordering observed in experiments .
for the afm1 state , where charge ordering is enforced , the magnetic energy ( 7 mev / co ) is much smaller than those of afm2 and fm states , which indicates the enforced charge ordering under gga level is unfavorable in energy .
this result is consistent with the calculations for na@xmath54coo@xmath1 , where attempts using lda to obtain self - consistent afm spin ordering always converge to the fm or nonmagnetic solution@xcite .
this discrepancy between theory and experimental observation naturally leads to a further consideration of correlation effect .
dft+u@xcite is a kind of method aiming to deal with the on - site coulomb interaction within the framework of dft .
it treats the local electrons in a hartree - fock manner , which drives local orbital occupations to integral occupancy as @xmath3 increases . for transition metal oxide
, one may find a larger split of @xmath55 orbitals for larger @xmath3 , as shown in fig .
[ dosfm]-[dosafm2 ] , where partial dos of cobalt @xmath2 orbitals and oxygen @xmath46 orbitals are plotted for fm , afm1 and afm2 states respectively . the cobalt @xmath2 partial dos for co1 and co2 sites are separated .
we do not present the results for spin restricted paramagnetic states in this section , because the energy differences between these states and corresponding magnetic states increase rapidly with @xmath3 ( 0.4 ev / co for @xmath56 ev ) .
the paramagnetic states are still metallic even with @xmath3 as large as 7.0 ev . in fig .
[ dosfm ] , the top panel is the gga result , and the three below panels represent three typical electronic structure behaviors for different @xmath3 values . for small @xmath3 ( see panel b ) , the electronic structure is similar to the above gga one , except for a strong trend towards gap opening and charge ordering . as shown in fig .
[ moment ] , the moments on the two inequivalent co sites are nearly equal and also similar to the gga values ( which is the @xmath57 limit ) until @xmath583 ev . above @xmath59 , the unoccupied part of the @xmath48 states weight more and more for co2 states with the increase of @xmath3 .
disproportionation from co@xmath60 into @xmath61 co@xmath4 and @xmath62 co@xmath10 ions is nearly complete at @xmath63 ev and is accompanied by a metal insulator ( mott - like ) transition from conducting to insulating .
this insulating phase results from the splitting of the @xmath48 states , which lead to an unoccupied narrow band containing one hole for each co2 ions and an occupied band on the co1 ion , as shown in fig .
[ dosfm]c .
we judge the formation of charge ordering from local magnetic moments instead of from charge population directly , because the charge difference between co1 and co2 are relatively small ( only [email protected] electron ) .
this is a result of the hybridization of the o @xmath46 and co @xmath2 orbitals .
in fact , oxygen may directly contribute to the charge ordering in some oxides , as suggested by coey@xcite .
we also notice that the charge ordering and gap opening with a moderate @xmath3 is far from obvious , since zou _ _ et al.__@xcite have gotten a metallic ground state with a relatively large coulomb @xmath3 ( 5 ev ) but without adopting any supercell structure .
therefore the sodium ion ordering is critical to the insulating ground state .
there is no clear boundary for moderate and large values of @xmath3 , but a relatively large @xmath3 value will feature several characters as shown in fig .
[ dosfm]d .
first , the empty @xmath48 band are pushed further higher , which make it strongly mix with the upper @xmath14 bands . secondly , the large splitting of the co @xmath2 orbitals even squeeze some co2 @xmath2 states to the bottom of the o @xmath46 manifold , and consequently the o @xmath46 character at the edge of valence band near fermi energy has exceeded the co @xmath2 characters .
thirdly , the magnetic moments of cobalt are exceed 1 @xmath5 in this region ( up to 1.42 @xmath5 per co for @xmath56 ev ) , which seems curious at first sight .
we now give two possible rationalizations for this observation .
one interpretation comes from the population analysis process .
we notice that oxygen is also spin polarized in our calculations for large @xmath3 , which makes the net magnetic moment of the whole supercell is still about 1.0 @xmath5 per co. therefore the nominal magnetic moment being larger than 1 @xmath5 per co may be just an artifact of paw population analysis as a result of the strong hybridization between co and o. another possibility is that the large on - site coulomb interaction may really promote a transition from low spin state to a intermediate spin state as demonstrated by korotin _
et al._@xcite for co ions in vertex sharing coo@xmath41 octahedra .
this type of transition is consistent with the strong mixing of empty @xmath48 and @xmath14 bands .
the evolution of electronic structures with @xmath3 for afm states is generally similar to that for fm state .
the most significant difference is that the charge ordering occurs well before gap opening for afm1 state . as shown in fig .
[ moment ] , the co@xmath4 moment grows immediately for afm1 state as @xmath3 increases from 1 ev , but the gap opens until @xmath64 ev , which is different from the results of lee _ _ et al.__@xcite for na@xmath6coo@xmath1 with @xmath18 and @xmath19 . the energy difference between the fm and afm phases as a function of @xmath3 is plotted in fig .
[ moment]c .
the general trend gives lower energy afm state for small @xmath3 and lower energy fm state for large @xmath3 , except the relatively higher afm1 energy for small @xmath3 , which is caused by the artificial charge ordering as previously discussed .
for large @xmath3 , the lower energy of fm state is a natural result of the weakening of superexchange effect , and it is also consistent with the prominent dos of the majority spin component at deep energy as shown in fig .
[ dosfm]d .
the energy difference between fm and afm2 is generally small according to the relatively small exchange interaction between different layers . according to our result
, the magnitude of on - site coulomb interaction @xmath3 in this material can be probed by experimental magnetic measurement . from fig .
[ dosfm]-[dosafm2 ] , we can find an interesting splitting and ordering of the @xmath14 orbitals . with the increase of @xmath3 ,
the lower part of the @xmath14 orbitals becomes mainly from co1 sites and the upper part from co2 sites .
we notice that this kind of orbital ordering of unoccupied @xmath14 orbitals will not affect the low energy properties of na@xmath0coo@xmath1 . although it has been found in lacoo@xmath65@xcite , orbital ordering of valence band orbitals is not very possible in this structure , since they are not doubly degenerated .
a rehybridization process is proposed to describe the competition between the on - site coulomb interaction and the @xmath14-oxygen hybridization in doped coo layers@xcite .
this process can be seen clearly from the partial dos of o @xmath46 orbitals as shown in fig .
[ dosfm]-[dosafm2 ] . at gga level ,
hybridization of oxygen @xmath46 states and @xmath14 states are more significant than the @xmath47-oxygen hybridization . with the increase of @xmath3 ,
the on - site coulomb interaction becomes stronger , which can be minimized by unmixing the oxygen and @xmath14 orbitals in order to decrease the occupation of the @xmath14 orbitals . in fig .
[ dosfm]-[dosafm2 ] , we can clearly see this kind of unmixing by the decrease of oxygen @xmath46 partial dos in the @xmath14 manifold . for large value of @xmath3 ,
the hybridization of oxygen @xmath46 states and @xmath47 states is somewhat enhanced , which is not included in the modified hubbard model study by marianetti _ _
et al.__@xcite .
we have studied the effect of hubbard @xmath3 on the electronic structures of na@xmath0coo@xmath1 in the previous part of this section , but the actual value of @xmath3 in this material is not determined yet , of which very different values are proposed in the literature@xcite .
based on the folowing two experimental observes , we estimate that the vaule of @xmath3 should be moderate , as also suggested by lee _ _ et al.__@xcite .
first , the insulating ground state suggests that the value of @xmath3 should not be too small .
secondly , na@xmath6coo@xmath1 with @xmath7 away from 0.5 is always metallic indicates a narrow energy gap for na@xmath0coo@xmath1 , which prefer a not too large value of @xmath3 .
since there was experimental report@xcite on electron mass enhancement for small @xmath7 of na@xmath6coo@xmath1 , people have compared the experimental electronic specific heat coefficient @xmath66 with the band value to estimate the strength of correlation@xcite .
but in the na@xmath0coo@xmath1 case , where hubbard type correlation drives system to a mott insulator , @xmath66 will reduced by the correlation effects .
in fact , the experimental @xmath66 is only 3 mj / mol - k@xmath67,@xcite which is much smaller than our gga band value ( 10.5 mj / mol - k@xmath67 ) , but very near to the dft+u value ( zero ) . at a plausible value of @xmath3 , 4
ev for example , intralayer afm coupling is favored over fm and interlayer afm coupling ( ref fig .
[ moment]c ) .
we notice that the magnetic state in na@xmath6coo@xmath1 is far from being clearly determined in experiments .
there are some experiments supporting afm coupling@xcite , while there are also some recent experiments supporting fm coupling@xcite . with this intermediate value of @xmath3 ,
the ground state ( afm1 ) magnetic moment for co2 is 1.0 @xmath5 , and the magnetic energy is 0.123 ev / co .
based on dft+u method , we have studied the effects of na ordering and on - site coulomb correlation on the geometrical , electronic and magnetic properties of na@xmath0coo@xmath1 . without including on - site coulomb interaction @xmath3 ,
the experimental insulating phase can not be produced by dft . for small value of @xmath3 ,
ground state is still metallic . at this time , no charge ordering occur for fm coupling , and symmetry enforced charge ordering for intralayer afm coupling is unfavorable in energy . increasing parameter @xmath3 to a moderate value can open an energy gap and form charge ordering for both fm and afm states .
large @xmath3 value will largely split the co @xmath2 orbitals , which leads to squeezing co @xmath2 states to the bottom of o @xmath46 manifold , mixing of empty @xmath48 and @xmath14 states , and large co magnetic moments .
accompanying with the early work on na@xmath6coo@xmath1 ( @xmath18 and @xmath19)@xcite , our study will shed some insights into understanding of the complex interplay of na content , superstructure , correlation effects and tendency of various type of ordering in na@xmath6coo@xmath1 .
however , a complete understanding on this issue , for example , why the @xmath19 case with stronger commensurability effect and maybe also stronger correlation is not a charge ordered insulator in experiment , will need much further work .
this work is partially supported by the national project for the development of key fundamental sciences in china ( g1999075305 , g2001cb3095 ) , by the national natural science foundation of china ( 50121202 , 20025309 , 10074058 ) , by the foundation of ministry of education of china , and by icts , cas .
99 c. fouassier , c. delmas , and p. hagenmuller , mater .
bull . * 10 * , 443 ( 1975 ) i. terasaki , y. sasago , and k. uchinokura , phys .
b * 56 * , 12685 ( 1997 ) i. terasaki , physica b * 328 * , 63 ( 2003 ) y. wang , n. s. rogado , r. j. cava and n. p. ong , nature * 423 * , 425 ( 2003 ) k. takada , h. sakural , e. takayama - muromachi , f. izumi , r. a. dilanian , and t. sasaki , nature * 422 * , 53 ( 2003 ) c. a. marianetti , g. kotliar , and g. ceder , phys . rev . lett . * 92 * , 196405 ( 2004 ) m. d. johannes and d. j. singh , phys . rev .
b * 70 * , 14507 ( 2004 ) m. l. foo , y. wang , s. watauchi , h. w. zandbergen , t. he , r. j. cava , and n. p. ong , phys
lett . * 92 * , 247001 ( 2004 ) d. j. singh , phys .
b * 61 * , 13397 ( 2000 ) d. j. singh , phys . rev .
b * 68 * , 020503(r ) ( 2003 ) j. ni and g .-
zhang , phys .
b * 69 * , 214503 ( 2004 ) . v. i. anisimov , f. aryasetiawan and a. i. lichtenstein , j. phys . : condens .
matter * 9 * 767 ( 1997 ) .
zou , j .- l .
wang and z. zeng , phys .
b * 69 * , 132505 ( 2004 ) .
j. kunes , k .- w .
lee , and w. e. pickett , arxiv : cond - mat/0308388 .
lee , j. kunes , and w. e. pickett , phys .
b * 70 * , 45104 ( 2004 ) q. huang , m. l. foo , j. w. lynn , h. w. zandbergen , g. lawes , y. wang , b. h. toby , a. p. ramirez , n. p. ong , r. j. cava , arxiv : cond - mat/0402255 . h. w. zandbergen , m. l. foo , q. xu , v. kumar , r. j. cava , arxiv : cond - mat/0403206 . g. kresse and j. furthmuller , comput . mater .
* 6 * , 15 ( 1996 ) g. kresse and j. furthmuller , phys .
b * 54 * , 11169 ( 1996 ) .
j. p. perdew , j. a. chevary , s. h. vosko , koblar a. jackson , mark r. pederson , d. j. singh , and c. fiolhais , phys .
b * 46 * , 6671 ( 1992 ) .
s. h. vosko , l. wilk and m. nusair , can .
* 58 * , 1200 ( 1980 ) p. e. blochl , phys .
b * 50 * , 17953 ( 1994 ) g. kresse and d. joubert , phys . rev .
b * 59 * , 1758 ( 1996 ) h. j. monkhorst and j. d. pack , phys . rev .
b * 13 * , 5188 ( 1976 ) .
p. e. blochl , o. jepsen , and o. k. andersen , phys .
b * 49 * , 16223 ( 1994 ) s. l. dudarev , g. a. botton , s. y. savrasov , c. j. humphreys , a. p. sutton , phys .
b * 57 * , 1505 ( 1998 ) a. kokalj , j. mol . graphics mod . *
17 * , 176 ( 1999 ) .
code available from http://www.xcrysden.org/. m. coey , nature , 430 , 155 ( 2004 ) m. a. korotin , s. yu .
ezhov , i. v. solovyev , v. i. anisimov , d. i. khomskii and g. a. sawatzky , phys .
b * 54 * , 5309 ( 1996 ) h. sakurai , k. takada , s. yoshii , t. sasaki , k. kindo , and e. takayama - muromachi , phys . rev .
b * 68 * , 132507 ( 2003 ) t. fujimoto , g .- q . zheng , y. kitaoka , r. l. meng , j. cmaidalka , and c. w. chu , phys .
* 92 * , 047004 ( 2003 ) a. chainani , t. yokoya , y. takata , k. tamasaku , m. taguchi , t. shimojima , n. kamakura , k. horiba , s. tsuda , s. shin , d. miwa , y. nishino , t. ishikawa , m. yabashi , k. kobayashi , h. namatame , m. taniguchi , k. takada , t. sasaki , h. sakurai , e. takayama - muromachi , arxiv : cond - mat/0312293 .
f. c. chou , j. h. cho , p. a. lee , e.t .
abel , k. matan , and y. s. lee , arxiv : cond - mat/0306659 .
a. t. boothroyd , r. coldea , d. a. tennant , d. prabhakaran , l. m. helme , and c. d. frost , phys .
lett . * 92 * , 197201 ( 2004 ) t. motohashi , r. ueda , e. naujalis , t. tojo , i. terasaki , t. atake , m. karppinen , and h. yamauchi , phys .
b * 67 * , 64406 ( 2003 ) | we report a first - principles projector augmented wave ( paw ) study on na@xmath0coo@xmath1 . with the sodium ion ordered insulating phase being identified in experiments ,
pure density functional calculations fail to predict an insulating ground state , which indicates that na ordering alone can not produce accompanying co charge ordering , if additional correlation is not properly considered . at this level of theory ,
the most stable phase presents ferromagnetic ordering within the coo@xmath1 layer and antiferromagnetic coupling between these layers .
when the on - site coulomb interaction for co @xmath2 orbitals is included by an additional hubbard parameter @xmath3 , charge ordered insulating ground state can be obtained .
the effect of on - site interaction magnitude on electronic structure is studied . at a moderate value of @xmath3 ( 4.0 ev for example ) ,
the ground state is antiferromagnetic , with a co@xmath4 magnetic moment about 1.0 @xmath5 and a magnetic energy of 0.12 ev / co .
the rehybridization process is also studied in the dft+u point of view . |
the first stars in the universe almost certainly originated under conditions rather different from those characterizing present - day star formation . because elements heavier than lithium are thought to be produced exclusively through stellar nucleosynthesis , the primordial gas must have been chemically pristine , presumably resulting in stars of unusually low metallicity .
the recent discovery of an ultra metal - poor star by christlieb et al .
( 2002 ) suggests that stellar relics from this era exist even today in our own galaxy .
such old stars offer invaluable information about the history of structure formation and the chemical composition of the gas in the very early universe .
the study of the cooling of primordial gas and the origin of the first baryonic objects has a long history ( e.g. , matsuda , sato & takeda 1969 ; kashlinsky & rees 1983 ; couchman & rees 1986 ; fukugita & kawasaki 1991 ; tegmark et al .
1997 ) . within the framework of the currently favored paradigm for the evolution of structure , i.e. hierarchical growth by gravitational instability , low mass halos ( @xmath16 ) dominated by cold dark matter ( cdm ) seed the collapse of primordial gas within them by molecular hydrogen cooling .
numerical studies of the formation of primordial gas clouds and the first stars indicate that this process likely began as early as @xmath17 ( abel , bryan & norman 2002 ; bromm , coppi & larson 2002 ) . in these simulations ,
dense , cold clouds of self - gravitating molecular gas develop in the inner regions of small halos and contract into proto - stellar objects with masses in the range @xmath18 .
while these investigations support the notion that the first stars were unusually massive , the simulations to date have either mostly been limited to special cases or ignored the cosmological context of halo formation and collapse . in particular , the question of how the _ population _ of the first luminous objects emerged within a large cosmological volume has not been explored . planned observational programs will exploit future instruments such as _ jwst _ and _ alma _ to probe the physical processes which shaped the high - redshift universe . among the relevant scientific issues
are the star formation rate at high redshift ( e.g. , barkana & loeb 2001 ; springel & hernquist 2003a ) , the epoch of reionization ( e.g. , gnedin 2000 ; fan et al .
2001 ; cen 2002 ; venkatesan et al . 2003 ; sokasian et al . 2003 ; wyithe & loeb 2003a ) , and the fate of high - redshift systems ( white & springel 1999 ) .
the statistical properties of early baryonic objects are of direct relevance to understanding the significance of the first stars to these phenomena . in this context
, the key theoretical questions can be summarized as _ when and where did a large population of the first stars form ? _ and _ how and when did the universe make the transition from primordial to `` ordinary '' star formation ? _ semi - analytic modeling has often been used to address these questions qualitatively ( see , e.g. , loeb & barkana 2001 ) . using a spherical collapse model , tegmark et al .
( 1997 ) estimated the critical h@xmath5 mass fraction needed for cooling of primordial gas and a corresponding halo mass scale within which this gas can collapse .
abel et al . ( 1998 ) and fuller & couchman ( 2000 ) later used three - dimensional simulations to give refined estimates for the minimum collapse mass ( but only for a single or a few density peaks ) .
these results form the basis for `` minimum collapse mass '' models in which it is assumed that stars are formed only in halos with mass above a certain threshold .
using such a treatment , barkana & loeb ( 2000 ) and mackey et al .
( 2003 ) estimated the star formation rate and supernova rate at high redshift , and santos , bromm & kamionkowski ( 2002 ) computed the contribution to the cosmic infrared background from massive stars in the early universe .
predictions from these theoretical models are , however , quite uncertain , because of the relatively crude assumptions that are used to relate the attributes of luminous objects to those of dark matter halos .
a few attempts have been made to numerically model early structure formation in cosmological volumes .
jang - condell & hernquist ( 2001 ) simulated a cosmological box @xmath19 mpc across and found that low mass ( @xmath20 ) dark matter halos at @xmath21 are quite similar in their properties to larger ones at lower redshifts .
however , they did not include the gas component and hence could not directly address the nature of the first baryonic objects .
ciardi et al .
( 2000 ) used outputs from @xmath15-body simulations to locate star - forming systems in a cosmological volume .
more recently , ricotti et al . ( 2002a , b ) performed cosmological simulations including radiative transfer to compute the star formation history at high redshift .
feedback processes from the first stars likely played a crucial role in the evolution of the intergalactic medium and ( proto-)galaxy formation , but the detailed consequences of these effects remain somewhat uncertain .
radiation can produce either negative feedback , by dissociating molecular hydrogen via lyman - werner resonances ( dekel & rees 1987 ; haiman , abel & rees 2000 ; omukai & nishi 1998 , 1999 ) , or positive feedback from x - rays which can promote h@xmath14 production by boosting the free electron fraction in distant regions ( haiman , rees & loeb 1996 ; oh 2001 ) .
it is not clear whether negative or positive feedback dominates .
machacek , bryan & abel ( 2001 ) examined the the former using numerical simulations which included an h@xmath5 photo - dissociating radiation field of constant flux in the optically thin limit .
cen ( 2002 ) emphasized the positive impact of an early x - ray background on the formation of the first stars and discussed the possibility that the universe could have been re - ionized at an early epoch by population iii objects alone ( see also wyithe & loeb 2003a , 2003b ) .
using three - dimensional adaptive mesh refinement ( amr ) simulations , machacek et al .
( 2003 ) further argued that the net effect of an x - ray background on gas cooling is milder than one naively expects from simple analytic estimates .
the formation of the first stars , the evolution of early cosmic radiation fields , and the thermal properties of the high - redshift intergalactic medium ( igm ) are closely linked , and it is likely that semi - analytical studies of the formation of pre - galactic objects are limited by the complex _ cross - talk _ between them . clearly , high - resolution simulations in a proper cosmological context are needed to advance our understanding of the details of first structure formation in the early universe . in the present paper , the first in a series on early structure formation
, we study the cooling and collapse of primordial gas in dark halos using high resolution cosmological simulations . we evolve the nonequilibrium rate equations for 9 species and include the relevant gas heating and cooling in a self - consistent manner . from a large sample of dark halos ,
we determine conditions under which the first baryonic objects form .
we show that `` minimum collapse mass '' models are a poor characterization of primordial gas cooling and gas cloud formation , because these processes are significantly affected by the dynamics of gravitational collapse .
we also examine the influence of h@xmath5 dissociating radiation in the form of a uniform background field in the optically thin limit as well as by approximately accounting for gas self - shielding .
we quantify the overall negative effect of photo - dissociating radiation using both the numerical results and analytic estimates for the efficiency of gas cooling .
based on the simulation results , we develop a semi - analytic model to describe the formation of star - forming gas clouds in dark matter halos . to this end
we adopt a simple star - formation law and compute the global star - formation rate using a large @xmath15-body simulation .
the paper is organized as follows . in section 2
we describe our numerical simulations .
section 3 presents general results of the simulations , ignoring radiation .
sections 4 , 5 and 6 give basic properties of the dark matter halos found in our simulations .
we discuss the impact of far uv radiation on primordial gas cooling in section 7 .
we describe a semi - analytic model for early star formation and its applications in section 8 .
concluding remarks are given in section 9 .
ccccc run & @xmath22 & l ( @xmath23kpc ) & @xmath24 ( @xmath25 ) & @xmath26 ( @xmath23pc ) + a & 2@xmath27 & 600 & 100.0 & 54 + b & 2@xmath28 & 300 & 29.6 & 36 + c1 & 2@xmath29 & 300 & 100.0 & 54 + c2 & 2@xmath29 & 300 & 100.0 & 54 + dm & @xmath30 & 1600 & ( @xmath31 ) 10000.0 & 200 +
we use the parallel @xmath15-body / sph solver gadget ( springel , yoshida & white 2001 ) , in its `` conservative entropy '' formulation ( springel & hernquist 2002 ) .
we follow the non - equilibrium evolution of nine chemical species ( e@xmath1 , h , h@xmath2 , he , he@xmath3 , he@xmath4 , h@xmath5 , h@xmath6 , h@xmath1 ) using the method of abel et al .
( 1997 ) and we employ the cooling rate of galli & palla ( 1998 ) for molecular hydrogen cooling . the time stepping method employed in the sph simulations is described in the appendix .
the largest of our chemo - hydrodynamic simulations ( run a ) employs 48 million particles in a periodic cosmological box of @xmath32kpc on a side .
we consider a conventional @xmath0cdm cosmological model with matter density @xmath33 , cosmological constant @xmath34 and present expansion rate @xmath35 km s@xmath12mpc@xmath12 .
the baryon density is @xmath36 .
the initial power spectra for the baryonic and dark matter components are accurately computed from the boltzmann code of sugiyama ( 1995 ) , in which the pressure term of baryon perturbations is taken into account . the initial power spectrum is normalized by setting @xmath37 , and all of our simulations are started at @xmath38 . the initial ionization fraction was computed using recfast ( seager , sasselov & scott 2000 ) and was set to be @xmath39 for the @xmath0cdm universe we adopt .
details of the set - up of the initial conditions will be presented elsewhere ( yoshida , sugiyama & hernquist 2003 ) .
the basic simulation parameters are listed in table 1 .
there , @xmath40 is the simulation box side length , @xmath24 is the mass per gas particle , and @xmath26 is the gravitational softening length .
run b was carried out with a higher mass and spatial resolution to check the convergence of our numerical results .
it started from the same initial matter distribution as that of c1 on large scales .
runs c1 and c2 differ only in the assigned phase information and fluctuation amplitudes in the initial random gaussian fields .
they are used to test how sample variance of the initial density field affects the final results .
we carried out a simulation with dark matter only , denoted `` run dm '' , to construct halo merger histories to construct the semi - analytic model described in section 8 .
we continued the simulations until about 100 million years after the first bound object formed .
radiation from the first object(s ) should , in principle , be included because it affects the chemical and thermodynamic evolution of the surrounding igm in a large fraction of our simulated regions .
nevertheless , we do not take such radiative effects into account in the first series of our simulations , in order to isolate other dynamical effects on primordial gas cloud formation .
we carry out the same set of simulations with uv radiation in the lyman - werner bands and examine the global effect of photo - dissociation in section 7 . during the simulations ,
we save 64 snapshots of the particle data spaced logarithmically in cosmic expansion parameter from redshift @xmath38 to @xmath41 .
we use these outputs to identify cold , dense gas clouds , and to construct dark matter halo merger trees .
we locate dark matter halos by running a friends - of - friends ( fof ) groupfinder with linking parameter @xmath42 ( jenkins et al .
2001 ) in units of mean particle separation , and discard the groups which have less than 100 dark matter particles .
we define the virial radius @xmath43 of a halo as the radius of the sphere centered on the most bound particle of the fof group having overdensity 180 with respect to the critical density .
the virial mass @xmath44 is then the enclosed mass ( gas + dark matter ) within @xmath43 .
for the halos identified in the simulations , we measure the mass of gas which is cold ( @xmath45 ) and dense ( @xmath46@xmath47 ) . once the gas starts to cool , clouds of molecular gas grow rapidly at the centers of halos and their masses exceed the characteristic jeans mass @xmath48 for the typical temperature @xmath49 k and density @xmath50@xmath47 of the condensed primordial gas .
hence , they are expected to be sites for active star formation .
hereafter , we refer to such cold , dense gas clumps as `` gas clouds . ''
since a halo can host more than one gas clouds , we run a fof groufinder independently to the gas particles with a very small linking parameter @xmath51 . in this manner
we can separate groups of dense gas particles .
we then discard groups of gas particles that do not satisfy the above criteria of cold , dense gas . by checking the locations of the selected groups in all the outputs , however , we found no halos which host more than one gas clouds in this particular simulation .
figure [ plot1 ] shows the projected gas distribution in the simulation box of run a at @xmath52 .
the bright spots are the primordial gas clouds .
there are 31 gas clouds in the simulated volume .
it is important to note that , whereas most of them are strongly clustered in + high density regions , reflecting the clustering of the underlying dark matter , some gas clouds are found in less dense , isolated regions .
in figure [ plot2 ] we plot the minimum mass of the halos that host gas clouds at each output redshift .
it approximates the evolution of the minimum mass of the star - forming systems . in the figure
, we also show the evolution of the most massive halo in each run .
the apparent earlier formation epoch of the first bound object in run a is simply due to finite volume effects .
run a simulated a volume 8 times larger than the others , and thus it contained a higher-@xmath53 density fluctuation than runs c1 or c2 .
cosmic variance also explains the difference in the minimum mass between runs c1 and c2 .
the assigned initial power spectrum for run c1 had , by chance , somewhat larger amplitudes than for run c2 on the largest scales .
we checked that the dark halo mass functions are noticeably different at large mass scales between the two runs . on the other hand ,
excellent agreement is found in the minimum mass scale between the high resolution run b ( dot - dashed line ) and low resolution run c1 ( filled circles ) .
our result appears to be converged on mass scales which our simulations probe .
figure [ plot2 ] clearly shows that the minimum collapse mass scale lies at @xmath54 , with only a weak dependence on redshift in the range plotted .
our result agrees reasonably well with that of fuller & couchman ( 2000 ) , who carried out three - dimensional simulations for single high-@xmath53 density peaks .
the weaker redshift dependence found by us reflects the fact that we define the minimum collapse mass using a large sample of halos formed in various places in the simulation volume , rather than for a single or a few objects in high density regions .
our result is also roughly consistent with that of machacek et al .
( 2001 ; 2003 ) , who obtain a smaller value for the minimum mass , @xmath55 . an important quantity that determines the onset of gas cooling is the fraction of hydrogen molecules , @xmath56 .
in figure [ plot3 ] we plot @xmath56 against the virial temperature for halos + in run a at @xmath52 .
we compute the virial temperature for the halo mass using @xmath57 where @xmath58 is the mean molecular weight and @xmath59 is the collapse overdensity .
filled circles represent the halos harboring a cold dense gas cloud , while open circles are for the others .
the solid line is an analytical estimate of the h@xmath5 fraction needed to cool the gas , which we compute _ a l _ tegmark et al .
briefly , we compute the characteristic cooling time of a gas with density @xmath60 and temperature @xmath61 as @xmath62 where @xmath63 is the cooling rate due to molecular hydrogen rotational line transitions , and determine the critical molecular hydrogen abundance with which the gas can cool within a hubble time . note that we use the cooling function of galli & palla ( 1998 ) for our simulations and for this estimate .
the dashed line is the asymptotic molecular fraction of a gas in a transition regime when electron depletion makes the production of hydrogen molecules ineffective
. then the molecular fraction scales as @xmath64 ( see eq .
[ 17 ] of tegmark et al .
1997 ) in figure [ plot3 ] , halos appear to be clearly separated into two populations ; those in which the gas has cooled ( top - right ) , and the others ( bottom - left ) .
our analytic estimate indeed agrees very well with the distribution of gas in the @xmath56 - @xmath61 plane .
we emphasize , however , that this apparent agreement should not be interpreted as the model precisely describing the gas evolution .
also , the analytic model itself is expected to be accurate only to within some numerical factor .
although the h@xmath5 fraction primarily determines whether the gas in halos can cool or not , there are some halos within which gas clouds have not formed despite the high gas temperatures ( open circles with @xmath65 ) . at @xmath52 , about 30% of the massive halos are `` deficient '' in this manner .
similar features are also found in the result of the amr simulation of machacek et al .
( 2001 , their fig .
3 ) . in the next section
we further examine what prevents the gas from cooling and collapsing in these halos .
halos in cdm models grow hierarchically through merging and the accretion of smaller objects .
the complex and violent formation processes of dark halos affect the thermal and chemical evolution of the gas within them .
to addess this , we study the dynamical influence of dark matter on gas cloud evolution using halo merger histories .
we identified a total of 635 dark halos in run a at @xmath52 .
for all the halos , we traced their progenitors in earlier outputs to construct merger trees . in figure [ plot4 ]
we plot the mass evolution of a subset of halos that host gas clouds ( top - left panel ) and of another subset of halos that do not host gas clouds ( top - right panel ) . in the top - left panel ,
we mark trajectories by filled circles when they host gas clouds .
the figures show a clear difference between the two subgroups in their mass evolution .
most of the halos in the top - left panel experience a gradual mass increase since the time their masses exceeded @xmath66 , whereas those plotted in the top - right panel have grown rapidly after @xmath67 .
it appears that the gas in halos that accrete mass rapidly ( primarily due to mergers ) is unable to cool efficiently . rapid mass accretion and mergers dynamically heat the gas when halos form , causing it to become hot and rarefied , rather than allowing it to radiatively cool and condense
this is the situation first considered by white & rees ( 1978 ) in the context of hierarchical galaxy formation .
the simplest model to describe the evolution of radiative gas assumes that the gas cools when the characteristic cooling time is shorter than the dynamical time .
this scenario has been used to estimate the minimum mass scale for galaxies .
the gas cooling rate due to atomic hydrogen and helium associated with galaxy formation has a particular behavior in it decreases sharply below @xmath68 k by many orders of magnitude , which effectively prevents the cooling of gas in systems with low virial temperatures .
hence , the cooling criterion @xmath69 simply sets a definite minimum mass scale ( at a given redshift ) for galaxy formation .
the situation we consider here is clearly different , because molecular hydrogen cooling is a much _ less _ efficient process and has a weaker dependence on temperature .
more important , for this mechanism to be effective , a certain number of hydrogen molecules must first be produced , because the residual h@xmath5 abundance in the early universe is negligible .
figure [ plot4 ] illustrates these features . in the bottom panels , we plot the evolution of the molecular hydrogen fraction and the mean gas mass weighted temperature for the same halos as in the top panels . most of the trajectories in the bottom - right panels show a common feature : the temperature rises with little increase in the molecular hydrogen fraction .
we can understand this as follows .
consider an equal - mass merger where two halos each with a mass @xmath70 merge to form an object of mass @xmath71 at @xmath72 .
the h@xmath14 fraction of the gas in the two halos is + + + + predicted to be about @xmath73 ( see the dashed line in figure [ plot3 ] ) .
after virialization , the gas temperature becomes @xmath74k , and our estimate for the h@xmath14 fraction needed for the gas to cool is about @xmath75 , about a factor of 2.5 larger than the progenitor gas . in this case , the production of molecular hydrogen , the coolant , must precede cooling .
only when a further temperature increase or an increase in the molecular fraction brings the gas into the region above the critical line in the @xmath76 plane can the gas cool efficiently , unless significant heating occurs during this cooling phase .
the molecular hydrogen formation time scale for this typical halo is estimated to be @xmath77 where @xmath78 is the reaction coefficient of h@xmath1 formation via h + e@xmath1 @xmath79 h@xmath1 + @xmath80 .
the h@xmath14 formation timescale is comparable to the dynamical time for this halo .
it is interesting that the most massive halo plotted in the top - right panel of figure [ plot4 ] has a mass of @xmath81 at @xmath52 .
the mass growth of the halo is so rapid below @xmath82 , when it had a mass of @xmath70 , that the gas within it could not cool to form a dense gas cloud .
it has been instead continuously _ heated _ dynamically .
we can quantify this dynamical effect using our simulations .
we trace the progenitors of the massive ( @xmath83 ) halos in run a identified at @xmath52 , and compute recent mass accretion rates from @xmath84 , where we choose @xmath85 and @xmath86 . in figure [ plot5 ]
we plot the measured mass growth rates against the halo masses .
filled circles represent the halos that host gas clouds at @xmath52 , and open circles are for those that do not .
we derive a critical mass growth rate by equating the heating rate to the + molecular hydrogen cooling rate from @xmath87 and relate the increase in virial temperature to the mass growth rate by @xmath88 where the coefficient @xmath89 is computed from equation ( 1 ) at a given time . strictly speaking
, the temperature of the gas in a halo could be different from the halo s virial temperature .
we have carried out an additional simulation with non - radiative gas starting from the same initial conditions as run b ( the highest resolution simulation ) and found that the mean gas - mass weighted temperature is indeed close to the halo s virial temperature for almost all the halos , with deviations smaller than a factor of two ( see also figure 2 of machacek et al .
therefore , we may safely assume that the gas temperature is close to the virial temperature before cooling occurs . in figure [ plot5
] , the solid line is our analytic estimate for the critical mass growth rate . above the solid line ,
the mass growth rate is so large that dynamical heating acts more efficiently than cooling by molecular hydrogen .
thus , in such halos , gas cloud formation is effectively delayed or prevented .
note the steepness of the critical mass growth rate , which reflects the slope of the molecular hydrogen cooling rate @xmath90 in the temperature and density range of interest . for a halo with mass @xmath70 at @xmath91 , even a 20% mass increase per unit redshift results in net heating of the gas .
the critical mass growth rate + + therefore sets a natural lower limit for primordial gas cloud formation at a mass scale @xmath92 , in good agreement with the result shown in figure [ plot2 ] .
recent numerical studies by bromm et al .
( 2002 ) show that the initial angular momentum of primordial gas may determine the properties of the first star - forming clouds and possibly of the stars themselves . by pre - assigning the system s initial angular momentum and following its evolution , bromm et al .
( 2002 ) found that a disk - like structure is formed in high spin systems and that the gas subsequently fragments .
it is therefore interesting to ask whether such high spin systems are indeed produced in cosmological simulations .
we measure the spin parameters of the dark matter component and of the gas within them .
we follow the definition of bullock et al .
( 2001 ) : @xmath93 where @xmath94 is the specific angular momentum of each component ( dark matter or gas ) and @xmath95 is the circular velocity at the virial radius . in figure
[ plot6 ] , we plot the spin parameters of the dark matter and of the gas against the halo mass for run a at @xmath52 . for the gas component , we included all the gas ( hot + cold ) particles within the virial radius .
the distribution of the dark matter spin parameters is quite similar to that of both high mass halos ( van den bosch et al .
2001 ) and small halos ( jang - condell & hernquist 2001 ) . the spin parameter distribution for the dark matter is well fitted by the lognormal function @xmath96 \frac{d\lambda}{\lambda},\ ] ] with @xmath97 and @xmath98 .
we also note that the spin vectors of dark matter and the gas are not closely aligned , with a median deviation angle of @xmath99 degrees , in good agreement with the results of van den bosch et al .
( 2001 ) for higher mass halos . in figure [ plot6 ]
, the spin parameters appear relatively smaller for halos with gas clouds than for the entire halo population , because gas clouds form only in halos at the high - mass end .
bromm et al s result suggests that in systems with a spin parameter as large as 0.06 , gas clouds eventually flatten to form a rotationally supported disc .
we find only two star - forming halos in which either the gas or dark matter spin parameters are greater than 0.06 .
although rare , such objects do form in the cdm model .
it is important to point out , however , that the spin vectors of the gas and dark matter are _ not _ usually aligned .
this confirms the importance of setting up simulations in a proper cosmological context .
intriguingly , vitvitska et al .
( 2002 ) argue that halos which have experienced recent major mergers tend to have high spin parameters .
this may explain the overall trend in figure [ plot6 ] that gas clouds are preferentially found in low - spin halos .
it is still a difficult task to measure the spin parameters of the formed dense gas clouds accurately , because of resolution limitations . to address the question of the exact shape and the size of the final gas clump , we require a substantially higher resolution simulation .
the cooling and condensation of the gas within a dark halo can be qualitatively understood using a spherical collapse model .
following omukai ( 2001 ) , we consider a spherical gas cloud embedded in a dark halo .
we assume that the dynamics of the gas sphere is described by @xmath100 where @xmath101 is the gas density and the free fall time @xmath102 is given by @xmath103 we solve the energy equation @xmath104 together with the chemical reactions and the cooling rate computed in a consistent manner as in our simulation .
specifically , the net cooling rate @xmath105 includes cooling by molecular hydrogen @xmath106 , cooling by hydrogen and helium + atomic transitions @xmath107 , and the inverse compton cooling @xmath108 .
although the atomic line cooling is unimportant in the temperature range we consider , we include it for completeness . for our purposes , we follow the evolution of the gas after it is virialized .
we take the initial temperature of the collapsing gas cloud to be the virial temperature of the most massive halo in run c1 at @xmath109 , which has a mass of @xmath110 and a virial temperature @xmath111 k. we follow the thermodynamic evolution of the gas from @xmath109 to @xmath112 .
figure [ plot7 ] shows the distribution of the halo gas particles in the thermodynamic phase plane at @xmath112 .
we select the gas particles within @xmath113 pc ( physical ) of the center of the halo .
the trajectory computed by solving equations ( 7)-(9 ) is shown by the dashed line .
it describes the evolution of the gas from @xmath109 to @xmath112 reasonably well .
note that the cooling branch , appearing as dots in the right portion of the plot , does not exactly represent the evolutionary track of the gas particles .
the dots show the densities and temperatures of the gas particles _ at the output time _ , @xmath112 .
the cooled primordial gas piles up near the halo center with a characteristic temperature @xmath114k and number density @xmath115@xmath47 , in good agreement with the prediction of the spherical collapse model .
in the previous sections , we focused on the formation of the very first objects in the absence of an external radiation field . after the first stars form , they emit photons in a broad energy range .
radiation from the first stars affects not only the igm in the vicinity of the stars , but could also build up a background radiation field in certain energy bands .
photo - dissociation of molecular hydrogen due to radiation in the lyman - werner ( lw ) bands ( 11.18ev - 13.6ev ) is of primary importance , because the lw radiation can easily penetrate into the neutral igm ( dekel & rees 1987 ; haiman , abel & rees 2000 ; omukai & nishi 1999 ; glover & brand 2001 ) .
we first model the influence of lw radiation by including a uniform radiation background in the optically thin limit , as in machacek et al .
( 2001 ) . in section 7.3
, we also take into account gas self - shielding in the same set of simulations and compare the results with those for the optically thin cases . to begin
, we adopt a constant radiation intensity in the lw band of either @xmath116 or @xmath117 erg s@xmath12 @xmath13 hz@xmath12 str@xmath12 .
we use the photo - dissociation reaction coefficient given in abel et al .
( 1997 ) , @xmath118 hereafter , we describe the intensity by the conventional normalization @xmath119 .
we adopt the values @xmath120 by noting that the lw radiation with intensities @xmath121 does not significantly affect the abundance of molecular hydrogen in halos , whereas radiation with @xmath122 will quickly dissociate hydrogen molecules .
this can be easily seen by computing the dissociation time scale @xmath123 during run c1 , we turned on the background radiation at @xmath124 , slightly before the formation of the first gas cloud in the simulation box .
although it might seem more consistent to turn on the radiation only after the first object is formed , we start it slightly earlier , in order to study the evolution of the gas in the most massive halo as well .
figure [ plot8 ] shows the distribution of gas particles in the thermodynamic phase plane at @xmath112 .
the dashed lines in figure [ plot8 ] are computed by solving equations ( 7)-(9 ) including the lw radiation with @xmath125 and 0.01 . for these cases
we computed the evolution of the abundance of molecular hydrogen @xmath126 for @xmath127 and 0.01 . and then evaluated the gas cooling rate @xmath128 . for @xmath129 ,
the cooling branch now appears to lie on a shallower line than for the case with no radiation ( figure 7 ) .
the analytic model again describes the feature reasonably well .
it is interesting that under the influence of lw radiation the characteristic temperature of the gas clouds becomes higher than in the case with no radiation .
omukai ( 2001 ) argues that both the density and temperature of gas clouds in the regime before they start to undergo run - away collapse become higher for a stronger radiation field and the characteristic jeans mass becomes _
smaller_. although our simulation does not probe this regime owing to lack of resolution , it is intriguing that the presence of radiation affects the final mass of the collapsing gas clouds .
note also that , in a cosmological context , even a slight delay of gas cooling can make the subsequent evolution very different because of the rapid formation of dark matter halos .
the case with @xmath119=0.1 is noticeably different from the other two ( figure 7 , 8) models .
gas cooling and collapse are almost entirely prevented when the lw radiation dissociates molecular hydrogen .
for this case , we compute the h@xmath14 formation and dissociation timescales to be @xmath130 myrs , @xmath131 myrs , respectively , with the latter being much smaller than the former .
thus , we expect the h@xmath14 abundance to be close to the equilibrium value .
the equilibrium h@xmath14 abundance is then given by + + @xmath132 which yields the fraction @xmath133 in the present case .
it is nearly two orders of magnitude smaller than the critical fraction needed to cool the gas ( see figure 2 ) ; obviously , such gas can not cool .
we generalize the above argument using the larger box simulation run a. we turn on a uniform background radiation field with @xmath134 at @xmath124 and continue the run until @xmath52 .
we call this simulation `` run a - r ''
. figure [ plot9 ] shows the mean molecular hydrogen fraction against the virial temperature for the halos identified at @xmath52 in this simulation . as in figure
[ plot3 ] , the critical molecular hydrogen fraction to cool the gas is shown by the solid line , and the equilibrium h@xmath14 abundance computed by equation ( [ eq_eq ] ) is indicated by the dashed line .
for this plot , we computed the equilibrium h@xmath14 abundance by accounting for the fact that gas densities in small halos are smaller than the universal baryon fraction times the mean matter density within halos , because of gas pressure ( loeb & barkana 2001 ) .
the dashed line in figure [ plot9 ] thus appears to steepen toward the low temperature end .
the agreement with the measured mean molecular fraction is quite good , although the simulated halos show a large scatter at high temperatures .
since the analytic model we adopted + is expected to be accurate only to within some numerical factor , we plot a factor of two smaller critical h@xmath5 abundance in figure [ plot9 ] ( solid line ) than that in figure 3 .
this brings the critical curve into good agreement with the simulation results . in figure
[ plot9 ] , the critical temperature @xmath135 defined at the point where the critical @xmath136 is equal to the equilibrium h@xmath14 abundance also agrees reasonably well with the actual minimum temperature ( vertical dotted line ) found in our simulation .
overall , we find that radiative effects are quite substantial .
the mean molecular hydrogen fractions drop by nearly an order of magnitude for a radiation field with @xmath119 = 0.01 .
an order of magnitude higher radiative flux will make the mean molecular fraction even smaller and make primordial gas cooling very inefficient , as we have seen in figure [ plot8 ] .
although our simulation results in the previous section highlighted a negative aspect of radiative feedback , the _ true _ importance of this effect remains somewhat uncertain , because of our oversimplified treatment of the background field .
the optically thin assumption breaks down as dense gas clouds form , requiring that self - shielding be taken into account .
however , the strength of self - shielding is a difficult question to address . for a static gas
it is indeed significant , giving an effective shielding factor @xmath137 for molecular hydrogen column densities @xmath138 @xmath13 ( drain & bertoldi 1996 ) . for a gas with extremely large velocity gradients and disordered motion ,
the gas remains nearly optically thin even for column densities @xmath139@xmath13 ( glover & brand 2001 ) . since the full treatment of three - dimensional radiative transfer for 76 lw lines , even when only the lowest energy level transitions are included , is practically intractable ( see , however , ricotti et al .
2001 for one - dimensional calculations for a stationary gas ) , we adopt the following approximate method to estimate the maximum effect of self - shielding . we consider only shielding by gas in virialized regions and do not consider absorption by the igm in underdense + + regions , assuming that the amount of residual intergalactic h@xmath5 is negligible .
although this is not true initially , the intergalactic h@xmath5 fraction quickly decreases after the very first stars appear ( haiman , abel & rees 2000 ) . on the other hand , the so - called saw - tooth " modulation of background radiation owing to neutral hydrogen lyman series absorption ( haiman et al .
1997 ) is substantial in the lyman - werner band because neutral hydrogen is abundant at very high redshifts .
nevertheless , we do not consider the evolution and attenuation of radiation and we fix the radiation intensity as an input , rather than computing it consistently from actual star formation .
thus the assigned intensity may be regarded as that after the radiation is attenuated by intergalactic neutral hydrogen .
we consider the evolution and modulation of the radiation spectrum in section 8 . as for the simulations presented in section 7.1
, we apply a background radiation field in the lw band .
the radiation intensity at each position in the simulated region is computed by assuming that it is attenuated through surrounding dense gas clouds .
we define a local molecular hydrogen column density @xmath140 in a consistent manner employing the sph formalism .
we use the local molecular hydrogen abundance and + density to obtain an estimate for the column density around the @xmath141-th gas particle according to @xmath142 where @xmath143 is the position of the @xmath141-th gas particle and @xmath144 is the length scale we choose in evaluating the column density . in practice , we select an arbitrary line - of - sight and sum the contributions from neighboring gas particles within @xmath144 by projecting an sph spline kernel for all the neighboring particles whose volume intersects the sight - line .
we repeat this procedure in six directions along x , y , and z axes centered at the position of the @xmath141-th particle and take the _
minimum _ column density as a conservative estimate .
we mention that our method is similar in spirit to the local optical depth approximation of gnedin & ostriker ( 1997 ) .
we have chosen the length scale @xmath145 physical parsec by noting that the virial radius of a halo with mass @xmath146 is just about 100 parsec .
there are not significantly larger gas clumps than this scale in the simulated region .
the local column density estimates are easily computed along with other sph variables with a small number of additional operations .
figure [ plot10 ] shows the gas density and molecular hydrogen fraction profiles for the most massive halo in run a at @xmath147 .
the gas density profile is very close to an isothermal density profile , scaling as @xmath148 , except in the central 10 pc . in the bottom panel ,
we compare the estimated column densities at the position of each gas particle computed directly in the simulation ( dots ) with the analytic estimate ( dashed line ) .
we use the spherically averaged gas density and molecular hydrogen fraction profiles shown in the top panel and integrate from an arbitrary outer boundary as @xmath149 , to obtain the analytic estimate @xmath150 .
the agreement is quite good , assuring that our technique yields accurate estimates for the column density .
following drain & bertoldi ( 1996 ) , we parameterize the shielding factor as @xmath151 . \label{eq_shield}\ ] ] the photo - dissociation reaction coefficient is then given by @xmath152 the bottom panel of figure [ plot10 ] shows that , at the halo center , the column density is close to @xmath153 @xmath13 , and hence the radiation intensity is expected to be significantly reduced according to equation ( [ eq_shield ] ) .
it should be emphasized that the above expression for the shielding factor is derived for a stationary gas , and thus the actual effect of self - shielding could be substantially smaller because of gas velocity gradients and disordered motions , as discussed in machacek et al .
the results using the above estimate can , therefore , be regarded as describing the maximum possible effect of self - shielding .
we again use run a. similar to the optically thin case , we turn on a background radiation field with @xmath154 at @xmath124 , and compute self - shielding factors for all the over - dense gas particles ( hence for virtually all the gas particles in virialized regions ) .
we call this simulation `` run a - s '' ( for shielding ) . figure [ plot11 ] shows the mean molecular fraction against virial temperature for halos in run a - s at @xmath52 .
the mean molecular hydrogen fractions lie , with substantial scatter , on a steeper line than for the optically thin case ( compare with figure [ plot9 ] ) , indicating effective shielding . for large halos , indeed , the mean molecular hydrogen fractions are close to the values we found in run a ( with no radiation , see figure 3 ) . on the other hand ,
the gas in small halos is nearly optically thin , and their mean molecular fractions are close to those found in the optically thin limit ( figure [ plot9 ] ) .
we have found that we can quantify this trend by computing an effective shielding factor @xmath155 where the function @xmath156 is defined in equation ( [ eq_shield ] ) , @xmath136 is computed assuming no radiation , @xmath157 is the hydrogen number density taken to be 180 times the mean density , and @xmath43 is the halo s virial radius .
we have introduced a constant factor @xmath158 .
choosing @xmath159 and computing the equilibrium h@xmath14 abundance from equation ( [ eq_eq ] ) for the _ effectively attenuated _ radiation flux using equation ( [ eq_k31 ] ) , we have obtained molecular fractions which agree very well with the abundances found in the simulation .
the dashed line in figure [ plot11 ] shows the equilibrium h@xmath14 abundance calculated in this manner .
a value somewhat smaller than unity was chosen for @xmath158 by noting that a large fraction of the gas in the outer envelope of the halos remains nearly optically thin , as can be inferred from figure [ plot10 ] .
the net effect of photo - dissociating radiation is to raise the minimum collapse mass scale for primordial gas cooling ( haiman et al .
2000 ; machacek et al . 2001 ; wyse & abel 2003 ) . in figure
[ plot12 ] , we plot the minimum mass of halos that host gas clouds for 3 sets of simulations .
although there are small fluctuations , the minimum mass scales remain approximately constant , at @xmath160 and @xmath161 for run a , run a - s , and run a - r , respectively .
figure [ plot13 ] summarizes our findings in this section .
we compute the critical h@xmath5 fraction and the asymtotic h@xmath5 fraction for no radiation case in the same manner as described in section 2 .
we also show the equilibrium h@xmath5 abundance for two cases with the lw radiation with @xmath154 ( thick dotted line ) and with @xmath162 ( thin dotted line ) using equation ( [ eq_eq ] ) .
finally we compute the equilibrium abundance h@xmath5 by taking the gas self - shielding into account using equations ( 14)-(16 ) .
the effect of lw radiation is to reduce the fraction of molecular hydrogen , @xmath163 , for a given virial temperature @xmath164 . for optically thin radiation with @xmath154 , @xmath165 is more than an order of magnitude smaller than in the case with no radiation ( compare the dashed line denoted as case ( a ) with the thick dotted line denoted as ( b ) ) .
the case with maximal gas self - shielding in our implementation lies in between these two cases and bridges the low temperature ( @xmath166 k ) end of the optically thin case and the high temperature ( @xmath167 k ) end of the no radiation case .
we expect that the true effect of gas self - shielding should lie between these two extremes .
the minimum collapse mass scales for the three cases can be given , via the mass - temperature relation in equation ( 1 ) , by the crossing points of these three curves with the critical molecular fraction shown by the solid line in figure [ plot13 ] .
as can be inferred from figure [ plot13 ] , the minimum virial temperature for gas cloud formation by molecular hydrogen cooling monotonically increases for increasing radiation intensity . in the figure
we also show the equilibrium molecular hydrogen abundance for optically thin radiation with @xmath162 by the thin dotted line .
it does not cross the critical @xmath168 curve , and thus , under such high intensity radiation , molecular hydrogen cooling never becomes efficient in the entire temperature range plotted .
note , however , that in large halos with virial temperatures higher than @xmath169 k , the gas can cool by atomic hydrogen transitions . the overall cooling efficiency in large halos is then dominated by atomic cooling and will + not be critically affected by the h@xmath14 dissociating radiation regardless of its intensity . throughout this section
we have considered radiation only in the lyman - werner bands . while photons with energy above 13.6ev are likely
to be completely absorbed by abundant neutral hydrogen in the igm surrounding the radiation source ( but see below ) , those with energies below the lyman - werner bands can easily penetrate into the igm and so the relevant processes involving these low energy photons should , in principle , be taken into account .
the most important process in our context is the photodetachment of h@xmath1 : @xmath170 since h@xmath1 catalyzes the formation of molecular hydrogen , photodetachment by photons having sufficient energy ( @xmath171 ev ) could inhibit molecular hydrogen formation
. however , as discussed in machacek et al .
( 2001 ) , neglecting this process does not affect our results because of the high densities in the core regions where molecular hydrogen formation reactions ( via the h@xmath1 channel ) occur significantly faster than photodetachment for the radiation intensities we used . haiman et al . (
1996 ) and kitayama et al .
( 2001 ) argue that a moderate uv radiation field including photons with energy above 13.6ev can promote molecular hydrogen formation and thus enhance primordial gas cooling . since the overall strength of this _ positive _ effect depends on the intensity and spectral shape of the radiation field in a complicated manner , it is beyond the scope of the present paper to analyze this in detail .
it also appears that such positive effects are appreciable only in a restricted range of conditions .
we refer the reader to haiman et al .
( 1996 ) and kitayama et al .
( 2001 ) for a discussion .
ideally , we wish to carry out large , high - resolution @xmath15-body / hydrodynamic simulations with gas chemistry and radiation to evolve structure formation and the cosmic radiation field together , in a fully consistent manner .
such simulations should employ large enough volumes to contain a sufficient number of objects , while maintaining at least the same resolution as the runs described here .
however , computations such as this are still beyond our reach , in terms of computational power .
we attempt to overcome this obstacle by applying a semi - analytic method to a large dark matter @xmath15-body simulation which can be carried out at a substantially lower cost .
we employ a simulation with 324@xmath172 cdm particles in a box of 1600 @xmath23kpc ( run dm in table 1 ) .
the simulation parameters were chosen such that the mass per dark matter particle is @xmath173 , allowing us to resolve halos with masses @xmath70 by 50 particles . in galaxy formation semi - analytic models that utilizes a halo merger tree constructed from @xmath15-body simulations ( e.g. kauffmann et al .
1999 ) , the smallest halos are resolved by 10 simulation particles .
kauffmann et al .
( 1999 ) quote that almost all the 10-particle halos identified in one output are found as halos in later outputs in their high resolution simulations .
thus we justify that our choice of the simulation parameters allows us to robustly construct merger histories of halos with mass larger than @xmath70 .
we implement a set of simplified `` recipes '' to describe the thermodynamic and chemical evolution of the gas in dark matter halos , and calibrate the prescriptions against the numerical results presented in the previous sections .
we emphasize that our model significantly differs from , and improves upon , previous analytic methods in which no dynamical information is incorporated . since the efficiency of gas cooling in halos is primarily determined by the molecular hydrogen abundance , which is itself a function of the background radiation field as well as the gas density and temperature , we need to compute the evolution of the radiation flux coupled with the formation and evolution of the first stars .
we compute the frequency dependent radiation intensity at a given redshift from @xmath174 where @xmath175 is the total emissivity from all the sources at redshift @xmath176 .
we need to account for the `` saw - tooth '' modulation of the background radiation spectrum due to the lyman - series absorption by neutral hydrogen .
( note that hydrogen lyman-@xmath89 at 10.2 ev is outside the range relevant to h@xmath14 photo - dissociation . )
we follow haiman et al .
( 1997 ) and compute the saw - tooth modulation using a screening approximation . assuming that an effective screen due to abundant neutral hydrogen blocks photons in the lyman - series lines from all sources at redshift above @xmath177 , we write equation ( [ eq_jevol ] ) as + @xmath178 where the maximum redshift for the @xmath141-th lyman line at frequency @xmath179 is given by @xmath180 the total source emissivity @xmath175 is computed by multiplying the emissivity per single star by the number of active stars at redshift @xmath176 .
we explore a specific model in which a single massive population iii star is formed in each cold dense gas cloud .
we then use the pop iii sed computed by bromm , kudritzki & loeb ( 2001 ) , assuming that the stars are more massive than 100 @xmath181 . for such massive stars ,
the luminosity per unit stellar mass is @xmath182 erg s@xmath12 hz@xmath12 @xmath183 in the lyman - werner band , with only a weak dependence on the stellar mass .
omukai & palla ( 2003 ) argue that population iii stars with masses up to 600 @xmath181 could form by rapid accretion , whereas abel et al .
( 2002 ) claim , based on their simulation , that a reasonable estimate for the maximum mass is 300 @xmath181 . in view of the uncertainty in this quantity , we take the stellar mass to be a free parameter , keeping in mind that the total emissivity per unit volume scales as the stellar mass , provided that one massive star forms per gas cloud . we also assume that the mean lifetime of such massive stars is 3 million years ( e.g. schaerer 2002 ) .
then , the total emissivity is given by @xmath184 where @xmath185 is the emissivity per star and @xmath186 and @xmath187 are the number of active stars and the number of star - forming gas clouds at redshift @xmath188 , respectively .
our next task is to compute the number of star - forming gas clouds in the simulation box . using the outputs of run dm
, we locate dark halos in the same manner as described in section 2 and construct a halo merger tree by tracing the formation history of individual halos ( see yoshida et al .
2002 for details ) .
we then employ a simplified , yet physically motivated , prescription for the cooling of primordial gas within dark halos , which is also based on the results presented in the previous sections .
for all the halos at a particular redshift , we judge whether the gas within them can cool by determining if abundant numbers of hydrogen molecules have formed .
specifically , we adopt the following criteria for a halo to be star - forming : ( 1 ) the mean molecular hydrogen fraction is greater than the critical molecular hydrogen fraction , and ( 2 ) the recent mass growth rate of the halo is smaller than the critical mass growth rate
. it might seem that an even simpler criterion @xmath189 suffices for halos at a given mass .
while such a model could work approximately if there is no evolving background radiation , we prefer following the merger history explicitly .
it is important to specify _ when _ the gas in a halo cools , because the coupling of the increasing radiation flux with star formation makes the onset time of gas cooling critical .
the above cooling criteria can be formulated as @xmath190 and @xmath191 where @xmath192 is a function of the background radiation intensity @xmath193 , the virial temperature @xmath164 , and redshift @xmath188 .
we compute @xmath192 using the asymptotic molecular fraction ( equation [ 17 ] in tegmark et al . [ 1997 ] ) for no or negligible ( @xmath194 ) background radiation , while we use the equilibrium h@xmath5 abundance using equation ( 12 ) , ( 14 ) , ( 15 ) and ( 16 ) for background radiation with @xmath195 .
the critical molecular fraction , @xmath196 , is computed for the virial temperature @xmath164 and the density of the gas within a halo .
we approximate the instantaneous dynamical heating rate of a halo from its mass increase since the previous output time .
the choice of the time interval for measuring the mass increase remains somewhat arbitrary .
obviously , it should not exceed either the characteristic gas cooling time or the dynamical time .
we take a conservative value of 3 myrs for the time interval , which satisfies these requirements for most of the halos .
it should be noted that our model does not trace the accumulation of cold dense gas .
cold gas clouds can gradually grow in mass over a long time scale .
however , describing the evolution requires specifying the gas density profile and local temperatures in a halo at a given time . instead of following such a complicated procedure ,
our model simply specifies the time when an enough amount of cold gas ( @xmath197 ) is accumulated .
we applied this model to our run a by discarding gas particles and using only the dark matter particles with the particle mass scaled appropriately .
we followed the halo formation history and applied the model to it . in figure [ plot14 ]
we plot the computed number of star - forming halos against redshift . we compare it with the number of the gas clouds found in run a. the good agreement is encouraging .
we have also checked that not only do the total numbers of gas clouds agree , but that there is nearly a one - to - one match between the star - forming halos in our sph simulation and those identified by our semi - analytic model . for reference , the number of halos with mass greater than @xmath198 is also shown in figure [ plot14 ] ( dotted line ) .
the simple minimum - mass model over - predicts the number of star - forming clouds by up to a factor of three .
it is worth mentioning that , in some previous works ( e.g. mackey et al . 2003 ) , the critical virial temperature for primordial gas cloud formation was taken to be 400 k. in figure [ plot14 ] we plot the number of halos with virial temperature larger than 400 k. it predicts more than an order of magnitude larger ( nearly two orders of magnitude at @xmath199 ) number of star - forming regions .
this over - estimate of the number of star - forming regions would result in a substantial overestimate for the star - formation rate and associated supernova rate .
we now couple the formation of the first stars to the evolution of the background fuv radiation so that we can predict the global star formation rate .
starting from the earliest output at @xmath200 , we identify star - forming regions in the manner described in section 8.2 , assuming initially that the background radiation intensity is zero . in run
dm the first star - forming region appears at @xmath201 .
we then determine the total emissivity within the simulation box and compute a frequency - dependent radiation flux at the _ next _ timestep according to equation ( [ eq_jevol2 ] ) . at every timestep , we average the radiation intensities over frequency in the lyman - werner band .
then , a constant radiation field with the average intensity is ( assumed to be ) applied .
namely , the mean intensity @xmath202 is used to compute the molecular hydrogen fraction in halos using equations ( [ eq_eq ] ) and ( [ eq_k31 ] ) , and then the cooling criteria described in section 8.2 are checked for all the halos .
figure [ plot15 ] shows the evolution of the background radiation intensity @xmath202 and the spectrum of the processed radiation field @xmath203 in the lyman - werner band at @xmath204 .
the evolution of the radiation intensity is computed for two cases , one in the optically thin limit ( filled circles ) and the other with gas self - shielding ( open squares ) , as described in section 7 .
the difference between the two cases becomes noticeable when the mean radiation intensity exceeds @xmath205
@xmath13 hz@xmath12 str@xmath12 .
the dissociation time scale for radiation with intensity below this level is @xmath206 myrs , and thus its effect on gas cooling remains quite subtle even in the optically thin case .
when the radiation intensity is above this level , it dissociates hydrogen molecules quickly and suppresses primordial gas cooling .
it indeed acts more efficiently in the optically thin case and quenches star - formation more strongly .
suppressed star - formation makes the evolution of the background radiation slower than in the model with self - shielding , as seen in figure [ plot15 ] .
interestingly , our model predicts that the mean background radiation intensity reaches a value + + @xmath116 erg s@xmath12 @xmath13 hz@xmath12 str@xmath12 only at @xmath91 .
this is because , as is clearly seen in the bottom panel of figure [ plot15 ] , the hydrogen lyman - series absorption causes a substantial intensity decrease in the lyman - werner band . without the absorption ,
the mean intensity would have been more than an order of magnitude higher .
as we advance to lower redshifts using our model , some halos grow to have virial temperatures exceeding @xmath207 ( see figure [ plot13 ] ) . in principle , the gas should then cool via atomic hydrogen transitions even if hydrogen molecules are completely photo - dissociated .
it is conceivable that a large halo will be formed through successive mergers in which molecular hydrogen cooling has never become efficient ( hutchings et al .
2001 ) . however , we find that all the massive halos ( @xmath208 ) have a progenitor in which a star has already formed .
our model + assumes that star formation takes place only once in a halo and in its descendants during the redshift range we consider .
this crudely mimics the strong radiative feedback effect due to photo - dissociation and photo - heating ( omukai & nishi 1999 ; shapiro et al .
1997 ) in the vicinity of the first stars .
since the physical time between @xmath209 and @xmath210 is about 100 million years , the strong radiative feedback likely suppresses subsequent star formation over a significant fraction of this period ( yoshida , abel & hernquist 2003 ) , unless additional processes such as metal dispersal by supernovae are invoked .
finally , we use the number of stars formed to measure the conventional comoving star formation rate ( sfr ) density , @xmath181yr@xmath12mpc@xmath211 .
figure [ plot16 ] shows our model prediction for the comoving pop iii star formation rate .
we compare two cases by setting the mass of a pop iii star to be either @xmath212 ( filled circles , model m100 ) or @xmath213 ( open squares , model m600 ) .
we take a value of @xmath214 for the maximum pop iii star mass from omukai & palla ( 2003 ) . for both cases we included gas self - shielding in the model .
( thus model m100 in figure [ plot16 ] is the result from the same model as in figure [ plot15 ] . )
the star - formation rate density gradually increases from @xmath215 and reaches a value @xmath216yr@xmath12mpc@xmath211 at @xmath72 .
note that in figure [ plot16 ] the predicted sfr can be approximately scaled by the assumed stellar mass @xmath217 because of the straightforward unit conversion we used . in model m600 , the star - formation rate is then about 6 times larger than in model m100 .
the coupling of star - formation to the evolution of the background radiation causes each to be regulated by the other ( wyse & abel 2003 ) . as more stars form ,
the radiation intensity rises , which suppresses primordial gas cooling and hence the global star - formation rate .
then , the evolution of the radiation field slows , maintaining the star formation rate at a moderate level .
the flattening in the sfr for model m600 at @xmath218 is due to this regulation .
it is more prominent in model m600 than in model m100 because the total emissivity per unit volume is larger in model m600 .
we note that a characteristic feature of the population iii star - formation calculated by mackey et al .
( 2003 ) is not seen in our result .
their model assumes a sudden quenching of star - formation , which produces a prominent `` cliff '' in their figure 3 .
the actual regulation of the star - formation due to lyman - werner radiation occurs in a more complex way as we have just described , and thus the suppression of the pop iii star - formation is seen as a flattening of the star - formation rate .
intriguingly , the predicted sfr appears quite similar to the result of the highest resolution simulation for popiii stars by ricotti et al .
( 2002a ) , even though their simulations and model assumptions differ from ours .
ricotti et al .
( 2002b ) presented extensive tests using a series of simulations and examined the effect of various parameters on the global star formation rate .
they showed that an important parameter which influences the sfr at @xmath219 is , indeed , the size of the simulation volume .
therefore , it is worth asking whether the finite box size of our simulation affects the overall results .
while the mass resolution of run dm is enough for our purposes ( see the discussion at the beginning of this section ) , it is not clear if the simulated volume ( a cube of 1600@xmath23 kpc on a side ) contains a fair sample of star - forming regions .
we address this issue using the press - schechter halo mass function .
we have verified that the mass function of the dark matter halos is reasonably well - fitted by the press - schechter mass function in a mass range between @xmath220 , and in a redshift range @xmath221 , in agreement with jang - condell & hernquist ( 2001 ) at slightly lower redshifts .
since our analytical model relies on a one - star - per - halo assumption , we use the total number of halos with mass greater than @xmath222 to test whether the abundance of large halos is consistent with the press - schechter prediction .
figure [ plot17 ] shows the comoving number density of halos with mass greater than @xmath223 computed from the press - schechter mass function , in comparison with that found in run dm .
they agree reasonably well and the incomplete sampling of the halo mass function due to the finite box size is appreciable only at @xmath224 .
+ note also that the small discrepancy remaining at lower redshifts ( @xmath225 ) may be partly due to inaccuracies in the analytic mass function itself .
simulating a larger volume could make the discrepancy smaller at @xmath226 , but it would not significantly alter our results because the star formation rate is not dominated by rare massive halos in our model .
therefore , we conclude that the level of agreement shown in figure [ plot17 ] over the relevant redshift range is satisfactory .
springel & hernquist ( 2003a ) and hernquist & springel ( 2003 ) recently studied the star - formation history of a @xmath0cdm universe using a set of numerical simulations and an analytic model .
they found that the evolution of the star - formation rate can be well described by a simple functional form .
their simulations include only the normal - mode " of star - formation in larger mass systems , where gas cooling occurs via atomic hydrogen and helium transitions , and the regulation of star - formation is governed by supernovae rather than radiation ( see springel & hernquist 2003b ) . despite these differences ,
it is interesting to see how the two modes of star - formation compare at high redshift . in figure
[ plot18 ] , we compare the star - formation at very high redshifts ( @xmath225 ) computed from our model and that of hernquist & springel ( 2003 ) for @xmath227 .
we show the sfr for two cases with @xmath212 ( open circles ) and @xmath228 ( open squares ) .
we also provide a simple functional fit in a similar spirit to springel & hernquist ( 2003a ) , @xmath229 in units of @xmath230 .
the solid line in figure [ plot18 ] shows equation ( 23 ) for @xmath231 and the dot - dashed line is for @xmath212 .
this simple fit describes the results of our full semi - analytic model reasonably well , as clearly seen in figure [ plot18 ] .
the sfr of hernquist & springel ( 2003 ) begins just at @xmath232 , where our model prediction for the popiii star formation rate ends .
the overall shapes of the sfr of the two modes appear remarkably similar , although this may be just a coincidence since the physical mechanisms governing star formation and its regulation are very different in these two regimes .
we also note that feedback from the first stars formed at @xmath233 could affect small ( proto-)galaxy formation at @xmath234 by photo - heating , in the same way as we argued in section 8.3 .
therefore it remains unclear how the transition from pop iii to `` ordinary '' star - formation takes place .
this issue clearly merits further study .
we have carried out cosmological simulations of primordial gas cloud formation and determined the basic properties of the first baryonic objects for the standard @xmath0cdm cosmology .
the minimum collapse mass scale is set by the fraction of molecular hydrogen produced in the primordial gas in dark matter halos , which is primarily determined by the virial temperature of the system .
our simulations reveal that the merging process of dark halos disturbs the formation and evolution of primordial gas clouds , and we detailed how such dynamical processes delay the cooling and collapse of the gas .
we have also examined the impact of far ultra - violet radiation in the lyman - werner bands on the formation of primordial gas clouds .
we have followed the cooling and condensation of the gas within the most massive halos in one of our simulations ( run c1 ) for a few cases with a constant intensity radiation in the optically thin limit . due to photo - dissociation of molecular hydrogen , gas cooling
is suppressed for radiation with intensity @xmath235 .
we showed that the evolution of the gas cloud is qualitatively well - understood using a spherical collapse model .
we then implemented a technique to compute molecular hydrogen column densities in and around virialized regions in the simulations and computed the gas self - shielding factor in its maximum limit .
we found that , with the help of self - shielding , primordial gas in large halos cools efficiently even when irradiated by fuv radiation with intensity @xmath236 .
the overall effect of the external fuv radiation is to raise the minimum halo mass scale for efficient gas cooling .
we quantified this effect using a large volume simulation ( run a ) and found that , in the optically thin limit , the influence of fuv radiation can be formulated using an equilibrium h@xmath5 abundance .
furthermore , we found that gas self - shielding can also be described by defining an effective shielding factor in a simple manner .
while our experiments in the two extreme cases , the optically thin limit and with maximal gas self - shielding , should bracket the true effect of gas self - shielding , a more detailed study is needed to obtain an accurate estimate of this process .
a promising approach may be to discard high velocity gas elements in computing molecular hydrogen column densities ( simon glover , private communication ) .
ideally , one would like to perform @xmath15-body hydrodynamic simulations with gas chemistry and radiation using a large number of particles to study the coupling of star formation to the evolution of the radiation field .
unfortunately , to simulate a volume equivalent to our dark matter simulation run dm with the same resolution as our run a , would require at least @xmath237 particles .
as an alternative , we developed a semi - analytic model for primordial gas cooling and gas cloud formation . by comparing our model predictions with the results of sph simulations , we found that the model reproduces the number of primordial gas clouds in the simulated volume very well .
we then successfully applied it to outputs of a large @xmath15-body simulation . assuming a simple star formation - law , we have computed the star formation rate and the evolution of the background radiation at high redshift ( @xmath238 ) .
we believe that our model can be used to reliably estimate the rate of events associated with star formation , such as supernovae and gamma - ray bursts .
while our simulations include the physics needed to study basic properties of the first star - forming clouds , there are still a few additional mechanisms that could be important . an interesting possibility proposed by haiman et al .
( 2000 ) and oh ( 2001 ) is that an early x - ray background could increase the number of free electrons in the igm and promote the production of hydrogen molecules .
analytical estimates ( haiman et al . 2000 ) and numerical simulations ( machacek et al .
2003 ) indicate , however , that the net effect is mild and the positive feedback by an x - ray background does not entirely compensate the negative feedback effect due to photo - dissociating lyman - werner radiation unless unreasonably high x - ray intensities are assumed .
it thus appears that our model prediction for the star - formation rate should be robust , even without additional contributions from x - ray sources .
it is probably more important to include direct radiative transfer effects from the first stars . in the present model ,
we only crudely included the overall effect by quenching further star formation locally after the first star is formed .
in our simulations we find that many gas clouds are located close to one another , with physical separations smaller than 10 kpc .
these star forming clouds at high redshifts are , as we have shown , embedded in dark halos that are strongly clustered ( see figure [ plot1 ] ) .
a single massive population iii star can ionize a large volume of the surrounding igm and thus can have considerable impact on the formation of the nearby gas clouds and stars . in a forthcoming paper
we will study direct radiative transfer effects using ray - tracing simulations .
we thank frank van den bosch , volker springel , edmund bertschinger , chung - pei ma , volker bromm , tetsu kitayama , and particularly marie machacek for fruitful discussions .
we also thank the anonymous referee for giving us many constructive comments .
this work was supported in part by nsf grants aci 96 - 19019 , ast 98 - 02568 , ast 99 - 00877 , and ast 00 - 71019 .
the simulations were performed at the center for parallel astrophysical computing at the harvard - smithsonian center for astrophysics .
11 abel , t. , anninos , p. , norman , m. l. , & zhang , y. 1997 , new astronomy , 2 , 181 abel , t. , anninos , p. , norman , m. l. , & zhang , y. 1998 , apj , 2 , 181 abel , t. , bryan , g. l. , & norman , m. l. 2002 , science , 295 , 93 barkana , r. & loeb , a. 2000 , apj , 539 , 20 barkana , r. & loeb , a. 2001 , physics report bromm , v. , coppi , p. s. , & larson , r. b. 2002 , apj , 564 , 23 bromm , v. , kudritzki , r. p. , & loeb , a. 2001 , apj bullock , j. s. , dekel , a. , kollat , t. s. , kravtsov , a. v. , klypin , a. a. , porciani , c , & primack , j. r. 2001 , apj , 555 , 240 cen , r. 2003 , apj , in press couchman , h. m. p. & rees , m. j. 1986 , mnras , 221 , 53 ciardi , b. , ferrara , a. , governato , f. & jenkins , a. , 2000 , mnras , 314 , 611 christlieb , n. , bessel , m. s. , beers , t. c. , gustafsson , b. , korn , a. , barklem , p. s. , karlsson , t. , mizuno - wiedner , m. , & rossi , s. 2002 , nature , 419 , 904 dekel , a & rees , m. j. 1987 , nature , 326 , 455 drain , b. t. & bertoldi , f. 1996 , apj , 468 , 269 fan , x. , et al .
2000 , aj , 120 , 1167 fukugita , m. & kawasaki , m. 1991 , mnras , 269 , 563 fuller , t. m. & couchman , h. m. p. 2000 , apj , 544 , 6 galli , d. & palla , f. 1998 , a&a , 335 , 403 glover , s. c. o. & brand , p. w. j. l. , 2001 , mnras , 321 , 385 gnedin , n. 2000 , apj , 535 , 530 gnedin , n. & ostriker , j. p. 1997 ,
apj , 486 , 581 haiman , z. , rees , m. j. , & loeb , a. 1996 , apj , 467 , 522 haiman , z. , rees , m. j. , & loeb , a. 1997 , apj , 476 , 458 haiman , z. , abel , t. , & rees , m. j. 2000 , apj , 534 , 11 hernquist , l. & springel , v. 2003 , mnras , submitted ( astro - ph/0209183 ) hutchings , r. m. , santoro , f. , thomas , p. a. & couchman , h. m. p. 2002 , mnras , 330 , 927 jang - condell , h. , & hernquist , l. 2001 , apj , 548 , 68 jenkins , a. , frenk , c. s. , white , s. d. m. , colberg , j. m. , cole , s. , evrard , a. e. , couchman , h. m. p. , yoshida , n. , 2001 , mnras , 321 , 372 kashlinsky , a. & rees , m. j. 1983 , mnras , 205 , 955 kauffmann , g. , colberg , j. m. , diaferio , a. & white , s. d. m. , 1999 , mnras , 303 , 188 kitayama , t. , susa , h. , umemura , m. & ikeuchi , s. 2001 , mnras , 326 , 1353 loeb , a. & barkana , r. 2001 , ara&a , 39 , 19 machacek , m. e. , bryan , g. l. , & abel , t. 2001 , apj , 548 , 509 machacek , m. e. , bryan , g. l. , & abel , t. 2003 , mnras , 338 , 273 mackey , j. , bromm , v. , & hernquist , l. 2003 , apj , 586 , 1 matsuda , t , sato , h & takeda , h. 1969 , prog .
, 41 , 840 oh , s. p. 2001 , apj , 569 , 558 omukai , k. & nishi , r. 1998 , apj , 508 , 141 omukai , k. & nishi , r. 1999 , apj , 518 , 64 omukai , k. 2001 , apj , 546 , 635 omukai , k. & palla , f. 2003 , submitted to apj ricotti , m. , gnedin , n. y. , & shull , j. m. 2001 , apj , 560 , 580 ricotti , m. , gnedin , n. y. , & shull , j. m. 2002a , apj , 575 , 33 ricotti , m. , gnedin , n. y. , & shull , j. m. 2002b , apj , 575 , 49 santos , m. r. , bromm , v. & kamionkowski , m. 2002 , mnras , 336 , 1082 schaerer , d. 2002 , a&a , 382 , 28 seager , s. , sasselov , d. d. & scott , d. 2000 , apj , 128 , 407 shapiro , p. r. , raga , a. c. & mellema , g. 1997 , in proceedings of the 13th iap astrophysics colloquium , eds .
p. petitjean and s. charlot sokasian , a. , hernquist , l. , abel , t. & springel , v. 2003 , astro - ph/0303098 springel , v. , yoshida , n. & white , s. d. m. 2001 , new astronomy , 6 , 79 springel , v. & hernquist , l. 2002 , mnras , 333 , 649 springel , v. & hernquist , l. 2003a , mnras , 339 , 312 springel , v. & hernquist , l. 2003b , mnras , 333 , 289 sugiyama , n. 1995 , apj , 100 , 281 tegmark , m. , silk , j. , rees , m. , blanchard , a. , abel , t. , & palla , f. 1997 , apj , 474 , 1 van den bosch , f. , abel , t. , croft , r. , white , s. d. m , & hernquist , l. 2002 , apj , 576 , 21 venkatesan , a. , tumlinson , j. & shull , j. m. 2003 , apj , in press vitvitska , m. , klypin , a. a. , kravtsov , a. v. , wechsler , r. h. , primack , j. r. & bullock , j. s. , 2002 , apj , 581 , 799 white , s. d. m. & rees , m. j. 1978 , mnras , 183 , 341 white , s. d. m. & springel , v. 1999 , in _ first stars _ , proc . of the mpa / eso workshop , eds .
weiss , a. , abel , t. & hill , v. wyithe , j. s. & loeb , a. 2003a , apj , in press wyithe , j. s. & loeb , a. 2003b , apjl submitted ( astro - ph/0302297 ) wyse , j. & abel , t. 2003 , to appear in proceeding of the 13th october conference yoshida , n. , abel , t. & hernquist , l. 2003 , in preparation yoshida , n. , stoehr , f. , springel , v. & white , s. d. m. 2002 , mnras , 335 , 762 yoshida , n. & sugiyama , n. 2003 , submitted to mnras
we describe the time step criterion that incorporates the non - equilibrium nature of the physics in our simulations .
the code gadget employs an individual time step scheme and the usual time step criterion for the @xmath141-th gas particle is given by the courant condition @xmath239 where @xmath240 have their usual meanings ( see springel et al .
2001 ) , @xmath241 regulates the strength of the artificial viscosity , and @xmath242 is the courant factor . for each gas particle
, we supplement this criterion with two additional constraints so that the time step does not exceed the gas cooling time and the characteristic chemical reaction time .
we monitor the cooling time @xmath243 and use the rate of change of the electron number density to judge the characteristic chemical reaction time @xmath244 we set the tolerance parameter @xmath245 throughout the simulations . | we use cosmological simulations to study the origin of primordial star - forming clouds in a @xmath0cdm universe , by following the formation of dark matter halos and the cooling of gas within them . to model the physics of chemically pristine gas
, we employ a non - equilibrium treatment of the chemistry of 9 species ( e@xmath1 , h , h@xmath2 , he , he@xmath3 , he@xmath4 , h@xmath5 , h@xmath6 , h@xmath1 ) and include cooling by molecular hydrogen . by considering cosmological volumes , we are able to study the statistical properties of primordial halos and the high resolution of our simulations enables us to examine these objects in detail .
in particular , we explore the hierarchical growth of bound structures forming at redshifts @xmath7 with total masses in the range @xmath8 .
we find that when the amount of molecular hydrogen in these objects reaches a critical level , cooling by rotational line emission is efficient , and dense clumps of cold gas form .
we identify these `` gas clouds '' as sites for primordial star formation . in our simulations
, the threshold for gas cloud formation by molecular cooling corresponds to a critical halo mass of @xmath9 , in agreement with earlier estimates , but with a weak dependence on redshift in the range @xmath10 .
the complex interplay between the gravitational formation of dark halos and the thermodynamic and chemical evolution of the gas clouds compromises analytic estimates of the critical h@xmath5 fraction .
dynamical heating from mass accretion and mergers opposes relatively inefficient cooling by molecular hydrogen , delaying the production of star - forming clouds in rapidly growing halos .
we also investigate the impact of photo - dissociating ultra - violet ( uv ) radiation on the formation of primordial gas clouds .
we consider two extreme cases by first including a uniform radiation field in the optically thin limit and secondly by accounting for the maximum effect of gas self - shielding in virialized regions . for radiation with lyman - werner band flux @xmath11 erg s@xmath12 @xmath13 hz@xmath12 str@xmath12 ,
hydrogen molecules are rapidly dissociated , rendering gas cooling inefficient . in both the cases we consider ,
the overall impact can be described by computing an equilibrium h@xmath14 abundance for the radiation flux and defining an effective shielding factor . based on our numerical results ,
we develop a semi - analytic model of the formation of the first stars , and demonstrate how it can be coupled with large @xmath15-body simulations to predict the star formation rate in the early universe . |
the standard " inflationary model is based on a flat ( @xmath0 ) friedmann - robertson - walker ( frw ) geometry , motivated by the fact that inflationary expansion rapidly wipes out any original spatial curvature .
however , even though inflation drives the curvature term , @xmath1 towards zero , or equivalently , drives the total density parameter @xmath2 towards 1 , this does _ not _ imply that @xmath3 .
conditions leading to the open set of values @xmath4 are far less fine - tuned than those corresponding precisely to @xmath3 ( although closed models have other fine - tuning aspects ) . but irrespective of any arguments about fine - tuning , the spatial curvature of the real universe is in principle determined by observations ; theory will have to give way to data if the data clearly tell us that @xmath5 .
recent cosmic microwave background ( cmb ) and other data @xcite are not conclusive , but include the possibility that @xmath6 , with @xmath7 future experiments such as planck will reduce these error bars and give more accurate information about the curvature of the universe .
( note that the true global value of @xmath8 and its observed value in the local hubble volume may differ due to cosmological perturbations seeded by inflation . )
current data mean that we should take closed universes , and therefore closed inflationary models , seriously .
inflation in a closed universe is sometimes considered to be ruled out by unrealistic fine - tuning @xcite .
if this were true , then observational confirmation of @xmath9 would rule out inflation .
but in fact there is a wide range of closed inflationary models , and it is not too difficult to construct simple and consistent models without excessive fine - tuning ( see also ref .
it is also true that fine - tuned initial conditions can not be ruled out by purely scientific arguments , as we discuss below . for the models that we consider ,
the value of @xmath10 is set by choosing one parameter , while the magnitude of large - scale scalar perturbations fixes another parameter .
if @xmath11 is taken as 1.02 , then the power spectra of cmb anisotropies and matter can show testable differences from the standard flat model @xcite .
if @xmath10 is very small , then curvature is negligible as regards structure formation processes in the universe .
the smaller that @xmath12 is , the more closely the models approximate the standard flat model . but in the early inflationary universe , positive curvature can play a significant role and lead to novel features that do not arise when @xmath13 , for example allowing a minimum in the scale factor @xmath14 which is otherwise not possible , or putting limits on the number of possible e - foldings in the inflationary epoch @xcite .
primordial history is very different when @xmath6 than in the case @xmath15 , regardless of how small @xmath16 is , and a variety of possibilities arise , including universes with a minimum radius .
singularity theorems have been devised that apply in the inflationary context , showing that the universe necessarily had a beginning ( according to classical and semi - classical theory ) @xcite . in other words , according to these theorems the quantum gravity era can not be avoided in the past even if inflation takes place . however , the models we present escape this conclusion , because they do not satisfy the geometrical assumptions of these theorems . specifically , the theorems assume either ( a ) that the universe has open space sections , implying @xmath3 or @xmath17 , or ( b ) that the hubble expansion rate @xmath18 is bounded away from zero in the past , @xmath19 .
there are inflationary universes that evade these constraints and hence avoid the conclusions of the theorems ( this was also noted in ref .
it is also possible to find counter - examples that are open , i.e. , where assumption ( a ) is not violated , but this typically involves sophisticated constructions @xcite .
here we consider closed models in which @xmath6 and @xmath20 can become zero , so that both assumptions ( a ) and ( b ) of the inflationary singularity theorems are violated .
the models are simple , obey general relativity , and contain only ordinary matter and ( minimally coupled ) scalar fields .
previous examples of closed inflationary models @xcite are , to our knowledge , either bouncing models or models in which inflation is preceded by deceleration ( so that a singularity is not avoided ) .
the @xmath6 bouncing universe collapses from infinite size in the infinite past and then turns around at @xmath21 to expand in an inflationary phase .
the canonical model for such a bounce is the de sitter universe in the @xmath6 frame , with @xmath22 these coordinates cover the whole spacetime , which is geodesically complete @xcite .
however , the bouncing models face serious difficulties as realistic cosmologies .
the initial state is hard to motivate ( collapsing from infinite size without causal interaction ) , and it is also difficult to avoid nonlinearities in the collapse that prevent a regular bounce .
we show here that when @xmath23 there are closed inflationary models that do not bounce , but inflate from a static beginning , and then reheat in the usual way .
the inflationary universe emerges from a small static state that has within it the seeds for the development of the macroscopic universe , and we call this the emergent universe " scenario .
( this can be seen as a modern version and extension of the eddington universe . )
the universe has a finite initial size , with a finite amount of inflation occurring over an infinite time in the past , and with inflation then coming to an end via reheating in the standard way .
the redshift and the total number of e - folds remain bounded through the expansion of the universe until the present day , because the scale - factor is bounded away from zero in the past .
there is _ no _ horizon problem , since the initial state is einstein static .
since they start as einstein static , they avoid a singularity . the initial static state can be chosen to have a radius above the planck scale , so that these models can even avoid a quantum gravity regime , whatever the true quantum gravity theory may be . because they can undergo a large amount of inflation , these models can be effectively the same as the standard flat models as regards structure formation processes .
therefore they are not vulnerable to future reductions in the observed value of @xmath24 below 0.02 ( provided that it remains positive ) , and they can reproduce the successes of the standard inflationary cosmologies , but from very different primordial foundations .
we do not require exotic physics or matter .
the early universe contains a standard scalar field @xmath25 with energy density @xmath26 and pressure @xmath27 , and possibly also ordinary matter with energy density @xmath28 and pressure @xmath29 , where @xmath30 .
the cosmological constant is absorbed into the potential @xmath31 .
there are no interactions between matter and the scalar field , so that they separately obey the energy conservation and klein - gordon equations , @xmath32 the raychaudhuri field equation @xmath33 , \label{ray_fr}\ ] ] has first integral the friedmann equation , @xmath34 -{\frac{k}{a^{2}}}\ , , \label{friedmann}\ ] ] which together imply @xmath35 + { \frac{k}{a^{2}}}\ , .
\label{b1}\ ] ] the raychaudhuri equation gives the condition for inflation , @xmath36 for a positive minimum in the inflationary scale factor , @xmath37 , @xmath38 where the time @xmath39 may be infinite .
the only way to satisfy eq .
( [ 3 ] ) with non - negative energy densities is if @xmath6 .
closed inflationary models admit a minimum scale factor if inflation occurs for long enough , since curvature will eventually win over a slow - rolling scalar field as we go back into the past ( cf .
the inflationary singularity theorems mentioned above exclude this case , since they either only consider @xmath40 , or explicitly exclude the possibility @xmath41 .
closed models with a minimum scale factor @xmath42 include both bouncing and ever - expanding cases .
we do not proceed further with the bouncing models because of their acausal initial state and the difficulty of achieving a stable bounce .
ever - expanding models avoid these problems .
a simple ever - inflating model is the closed model containing radiation ( @xmath43 ) and cosmological constant @xmath44 , with scale factor @xcite @xmath45 ^{1/2}.\ ] ] in the infinite past , @xmath46 , the model is asymptotically einstein static , @xmath47 .
inflation occurs for an infinite time to the past , but at any finite time @xmath48 , there is a finite number of e - folds , @xmath49 where the last equality holds for @xmath50 .
the curvature parameter at @xmath48 is strongly suppressed by the de sitter - like expansion : @xmath51 the exact model eq .
( [ rad ] ) is a simple example of eddington - type solutions .
there are trajectories of this type in the classical phase space @xcite satisfying einstein s equations .
they are past - asymptotically einstein static and ever - expanding , and the fact that they exist already shows there are inflationary universes that evade the conditions of the singularity theorems presented in @xcite .
however , in these models inflation does not end .
below we discuss more realistic models , based on inflationary potentials , that do exit from inflation .
these universes are singularity - free , without particle horizons , and ever - expanding ( @xmath52 ) .
even though emergent universes admit closed trapped surfaces , these do not lead to a singularity , since @xmath6 and the weak energy condition is violated in the past @xcite .
the einstein static universe is characterized by @xmath6 and @xmath53const .
equations ( [ cons])([b1 ] ) then imply that @xmath54 where @xmath55 and @xmath56 is the primordial vacuum energy .
if the scalar field kinetic energy vanishes , i.e. if @xmath57 , then @xmath58 , so that there must be matter to keep the universe static .
if the static universe has only a scalar field , i.e. if @xmath59 , then the field must have nonzero ( but constant ) kinetic energy , so that it rolls at constant speed along the flat potential .
( dynamically , this case is equivalent to a stiff fluid , @xmath60 , plus cosmological constant @xmath61 @xcite . )
the radius @xmath62 of the initial static universe can be chosen to be above the planck scale , @xmath63 by suitable choice of @xmath64 , @xmath65 and @xmath66 ( with all of them @xmath67 ) .
thus these models can in principle avoid the quantum gravity era . a simple way to realize
the scenario of the emergent universe is the following .
consider a potential that is asymptotically flat in the infinite past , @xmath68 but drops towards a minimum at a finite value @xmath69 .
the scalar field kinetic energy density is asymptotic to the constant einstein static value , @xmath70 where we used eqs .
( [ es1 ] ) and ( [ es2 ] ) with @xmath59 . because @xmath71 , no matter is needed to achieve the initial static state .
the field rolls from the einstein static state at @xmath72 and the potential slowly drops from its einstein static original value . provided that @xmath73 decreases more rapidly than the potential , we have @xmath74 , so that the universe accelerates , by eq .
( [ 2 ] ) . since @xmath75 and @xmath76 , while @xmath77 , the klein - gordon equation ( [ kg ] ) shows that the universe is expanding ( @xmath19 ) .
plotted against @xmath78 .
, width=234,height=302 ] inflation ends at time @xmath48 , where @xmath79 .
then reheating takes place as the field oscillates about the minimum at @xmath69 . in the asymptotic past ,
@xmath80 , the primordial cosmological constant @xmath81 is given by eq .
( [ aes ] ) as @xmath82 so that @xmath61 is large for a small initial radius . at the minimum
, @xmath83 defines the cosmological constant that dominates the late universe , @xmath84 a typical example of a potential is shown in fig . 1 , which is based on the potential presented in ref .
@xcite : @xmath85 ^ 2,\ ] ] where @xmath86 is a constant energy scale .
the infinite time of inflation , from @xmath87 to @xmath88 ( @xmath89 ) , produces a finite amount of inflation ( and a finite redshift to the initial einstein static state ) .
the potential produces expansion that is initially qualitatively similar to the exact solution in eq .
( [ rad ] ) , so that the total number of e - folds from the initial expansion can be estimated , following eq .
( [ rad2 ] ) , as @xmath90 provided that @xmath62 is chosen small enough and @xmath48 large enough , a very large number of e - folds can be produced .
the parameters in the potential in eq .
( [ pot ] ) can be chosen so that the primordial universe is consistent with current observations @xcite .
as with standard inflationary models , fine - tuning is necessary to produce density perturbations at the @xmath91 level , and to fix @xmath92 so that @xmath93 .
these issues are faced by all inflationary universe models @xcite .
the specific geometrical fine - tuning problem in the emergent models is the requirement of a particular choice of the initial radius @xmath62 , or equivalently a specific unique choice of the primordial cosmological constant , eq .
( [ pcc ] ) .
this choice must then be supplemented by a further fine - tuning a choice of initial kinetic energy such that eq .
( [ aes ] ) holds .
both conditions are required to attain an asymptotic einstein static state , the one ensuring asymptotic validity of the friedmann equation , and the other , that of the raychaudhuri equation .
the model can be criticized because of its fine - tuned initial state .
there are two basic responses to this criticism .
* the criticism does not rule out these models as valid physical models , but rather claims they are not likely to occur in reality because they are improbable .
but the inflationary singularity theorems @xcite are not based on probabilities , and the emergent models do indeed show that the geometric conditions of those theorems need not be satisfied . * the force of the fine - tuning criticism is based partly in philosophy not only in physics .
there is no scientifically based proof that the unique physical universe has to be probable .
this points to a tension in cosmology between two viable but opposed views on how to explain the current state of the universe . *
the one view is that the present state of the universe is highly probable because physical processes make it very likely to have occurred @xcite . *
the opposite view is that nature prefers symmetry , and the universe is likely to have originated in a highly symmetric , and necessarily fine - tuned , state @xcite .
the key point is that there is no scientific proof that the one or the other of these approaches is the correct approach to use in cosmology .
a quantum gravity theory may resolve this and other issues relating to the origin of the universe , but in the absence of such a theory , it seems worthwhile to pursue the implications of both approaches .
the underlying problem within classical and semi - classical cosmology is the uniqueness of the universe , and all the scientific and philosophical difficulties that this entails @xcite .
one can not straightforwardly apply statistics or probability to a unique object . there may be a scientific basis for the use of probability in the context of an ensemble of universes a multiverse . but this is a complex and controversial proposal , and there is no evidence that it is correct ; indeed it may not be scientifically testable @xcite .
consequently the choice between these two fundamentally different approaches to cosmological origins remains of necessity a partly philosophical one , while we lack a more complete quantum gravity theory that explains what actually underlies the existence of the real physical universe .
furthermore , it is not simply a case of fine - tuning versus non - fine - tuning .
the standard @xmath3 inflationary models , which avoid a special initial state , are not free of fine - tuning , in particular , the fine - tuning entailed in the choice of @xmath3 .
there is no mechanism to attain @xmath3 ( as opposed to @xmath94 ) , and no obvious way to prove observationally that @xmath95 to the infinite accuracy required to establish that in reality @xmath3 .
this standard assumption would appear to involve the same kind of fine - tuning as occurs in the emergent universe , because of the exact balance required to set @xmath3 .
the standard inflationary models necessarily have a singularity in the past @xcite . although it is possible to evade the singularity when @xmath3 @xcite , this requires complicated modifications of the spacetime .
it also appears to invoke a fine - tuned initial state , not unlike the fine - tuning involved in the emergent models .
the choice can therefore also be seen as between a singularity and a fine - tuned non - singular initial state .
there are arguments for each choice .
the singular inflationary models have the strength of generality of initial conditions ( at some time after the classical singularity , i.e. , effectively after the planck era ) .
they have a disadvantage of starting at a spacetime singularity where all of physics breaks down and spacetime itself comes to an end ( though quantum gravity effects may be able to avoid this ) .
the emergent universe has the advantage of avoiding a singularity ( with or without a planck era ) .
there are also arguments that a highly symmetric start to the universe is necessary for the thermodynamic arrow of time to function as it does @xcite .
the emergent universe scenario gives a framework for investigating what are the implications if whatever process caused the universe to come into being , preferred the high - symmetry state of the einstein static universe to any less ordered state .
there are various arguments in support of the einstein static model as a preferred initial state . * it is neutrally stable against inhomogeneous linear perturbations when @xmath96 ( as in the simple emergent model above ) , or when @xmath97 and the sound speed of matter obeys @xmath98 , as shown in ref .
@xcite , extending the initial results of refs .
it is of course unstable to _ homogeneous _ perturbations , which break the balance between curvature and energy density .
this instability is crucial for producing an inflationary era . *
it has no horizon problem . *
it maximizes the entropy within the family of frw radiation models @xcite . *
it is the unique highest symmetry non - empty frw model , being invariant under a 7-dimensional group of isometries @xcite .
the einstein static model has a well - defined vacuum , and casimir effects ( see , e.g. , ref .
@xcite ) could play an important role in determining the primordial parameters of the emergent universe .
the realization of the emergent universe outlined above and elaborated in ref .
@xcite , illustrated qualitatively in fig . 1 , has the advantage of simplicity .
but the fact that the initial einstein static state is only achieved asymptotically in the infinite past could be seen as a disadvantage .
it is possible to find other realizations of the scenario in which the universe starts expanding from an einstein static state at a finite time in the past . for example
, one can consider potentials with a critical value @xmath99 , where @xmath100 , above a stable minimum at @xmath69 , i.e. , @xmath101 the scalar field is initially at the critical position @xmath102 ; since there is no kinetic energy ( @xmath57 ) , matter is necessary ( @xmath103 ) to provide an initial static state .
specifically , by eqs .
( [ es1 ] ) and ( [ es2 ] ) , we have @xmath104 if the initial position is an unstable ( tachyonic ) maximum , i.e. , @xmath105 , then the field rolls down to the true minimum , similar to natural inflation potentials .
if the initial position is a false vacuum minimum , then the field needs to tunnel towards the true minimum .
another alternative is an inflationary potential that depends on temperature in much the same way as in the first inflationary universe models . in the initial einstein static state , with matter at some temperature , the field starts at a minimum , which becomes unstable after a perturbation .
this leads to expansion of the universe , and consequently the temperature falls , which results in the appearance of a lower minimum .
we have shown that inflationary cosmologies exist in which the horizon problem is solved before inflation begins , there is no singularity , no exotic physics is involved , and the quantum gravity regime can even be avoided .
these emergent universe models can be constructed with simple potentials ( illustrated schematically in fig . 1 ) , giving past - infinite inflation from an asymptotically initial einstein static state , with a bounded number of e - folds and redshift , followed by reheating in the usual way .
explicit and simple forms of the potential can be found that are consistent with present cosmological observations for suitable choice of parameters in the potential @xcite .
other realizations exist in which the universe starts inflating from an einstein static state at a finite time in the past .
the emergent models illustrate the potentially strong primordial effects of positive spatial curvature , leading to a very different early universe than the standard models , while producing a late universe that can be observationally distinguished from the standard case only by high - precision observations .
if one requires , within classical and semi - classical general relativity theory , a non - singular universe with standard fields and matter , then this is only possible if @xmath6 .
if one further rules out bouncing models because of problems in achieving a regular bounce , then one is led to the emergent universe scenario , with its fine - tuned initial einstein static state .
this state is preferred by its appealing stability , entropy and geometric properties , and provides , because of its compactness , an interesting arena for investigating the casimir and other effects .
the emergent universe scenario with @xmath6 provides an interesting alternative to the standard inflationary scenario with @xmath3 , in the case that @xmath8 is close to , but above , 1 .
this may be seen as a choice between a fine - tuned non - singular initial state , and a singularity that precedes generic initial conditions .
we do not claim that the emergent scenario is superior to the standard scenario , but simply that it is worthwhile investigating , since neither it nor the standard scenario are yet ruled out by scientific arguments . in the absence of a quantum gravity theory that can explain the true nature of the origin of the universe , both approaches remain valid to investigate and test .
the two approaches lead to observationally viable models , at least until there is firm observational evidence determining the curvature parameter @xmath106 of the universe .
if future observations turn out to provide strong evidence for positive curvature , then the standard models will be ruled out , and the emergent models and other closed models with a singularity and deceleration preceding inflation @xcite , would be contenders to describe the universe .
of course if @xmath24 is shown to be negative , then all closed models and the standard models will be ruled out .
however , if the universe is in fact exactly flat , observations will not be able to rule out a very tiny positive curvature ( and the existence of perturbations prevents a very accurate measurement ) .
thus if the standard is model is correct , the emergent universe with perturbatively small curvature can not be falsified by observations , although it does produce a very different early universe .
we thank anthony aguirre , john barrow , bruce bassett , david coule , jeff murugan , christos tsagas , tanmay vachaspati and david wands for helpful comments , and emily leeper for assistance with the figure .
ge is supported by the nrf ( sa ) and rm by pparc ( uk ) .
rm thanks the university of cape town for hospitality while part of this work was done .
a. borde and a. vilenkin , phys .
lett . * 72 * , 3305 ( 1994 ) ; ibid .
rev . d*56 * , 717 ( 1997 ) ; a.h .
guth , astro - ph/0101507 ; a. borde , a.h .
guth , and a. vilenkin , gr - qc/0110012 ; a. vilenkin , gr - qc/0204061 .
starobinsky , sov .
. lett . * 4 * , 82 ( 1978 ) ; j.d .
barrow and r.a .
matzner , phys .
d*21 * , 336 ( 1980 ) ; s.w . hawking , nucl . phys .
b*239 * , 257 ( 1984 ) ; a.v .
toporensky , grav . cosmol . * 5 * , 40 ( 1999 ) ; n. kanekar , v. sahni , and y. shtanov , phys .
d*63 * , 083520 ( 2001 ) ; c. gordon and n. turok , hep - th/0206138 ; r.h .
brandenberger , s.e .
joras , and j. martin , phys .
rev . d*66 * , 083514 ( 2002 ) . this approach can be seen as an extension of comments by dingle in the 1930 s and the pioneering work of misner in the 1960 s , e.g. , c.w .
misner , astrophys .
* 151 * , 431 ( 1968 ) ; ibid .
* 22 * , 1071 ( 1969 ) .
this view was central to einstein s paper on the einstein static model in 1917 , and was developed into the cosmological principle used by mccrea , bondi , and others to justify the robertson - walker metrics ; see , e.g. , h. bondi , _ cosmology _ ( cambridge university press , cambridge , 1960 ) .
a more informal version was used by einstein and de sitter to justify universe models with @xmath3 ; see the discussion of this often used simplicity principle " by , r.h .
dicke and p.j.e .
peebles , in _ general relativity : an einstein centenary survey _ , ed .
s. w. hawking and w. israel ( cambridge university press , cambridge , 1979 ) , p. 504 | observations indicate that the universe is effectively flat , but they do not rule out a closed universe .
the role of positive curvature is negligible at late times , but can be crucial in the early universe . in particular
, positive curvature allows for cosmologies that originate as einstein static universes , and then inflate and later reheat to a hot big bang era .
these cosmologies have no singularity , no beginning of time " , and no horizon problem .
if the initial radius is chosen to be above the planck scale , then they also have no quantum gravity era , and are described by classical general relativity throughout their history . |
information theoretic security has received a significant attention recently .
one mainstream direction has been on the wireless transmission of confidential messages from a source to a destination , in the presence of internal and/or external eavesdroppers . toward achieving that goal
, the communicating pair exploits the stochasticity and the asymmetry of wireless channels between the communicating pair and the eavesdroppers .
a stochastic encoder at the transmitter makes use of the available channel state information ( csi ) in a way for the mutual information leaked to the adversaries remain arbitrarily small .
it is designed in a way that , even when the adversaries have access to the full csi of the main channel , i.e. , between the transmitter and the receiver as well as the eavesdropper channel , i.e. , between the transmitter and itself , it still will obtain an arbitrarily low rate of information on the message .
likewise , the adversary relies on csi to make decisions .
for instance , a half - duplex hybrid adversary , capable of jamming or eavesdropping at a given time ( but not both simultaneously ) decides between jamming vs. eavesdropping , based on the available csi .
the assumption that the adversaries have full csi of all channels is typical in the literature @xcite-@xcite .
while this assumption leads to robust systems in terms of providing security as it makes no assumptions on the adversaries , it can be too conservative in some cases .
for example , to obtain main csi , an adversary relies on the same resource as the transmitter : feedback from the legitimate receiver .
hence , from the perspective of the receiver , there is a tradeoff between revealing csi and keeping it secret : if the legitimate receiver chooses not to reveal csi , it will sacrifice some achievable rate of reliable communication , but the adversary will have to make decisions under a higher uncertainty . in this paper , we study the tradeoffs involving obtaining csi .
we ask the questions , should a legitimate pair obtain csi and if so , what should be their strategy ? to that end , we focus on the system depicted in figure [ fig : sysmdl ] .
we assume all three channels to be block fading . in each block ,
based on the available csi , the half - duplex adversary can choose to do jamming at a fixed transmission power or eavesdropping , but not both . our objective is to maximize the rate of reliable communication over the main channel , subject to full equivocation @xcite ( weak secrecy ) at the adversary .
the adversary can follow an arbitrary strategy in its choice of jamming vs. eavesdropping at any given block . in the case in which the receiver feeds back main csi , it may do so in two ways : directly by sending back the exact state of the channel or by sending reverse pilots , trying to exploit channel reciprocity ( similar to @xcite ) .
while the former method completely reveals the main csi , it eliminates the possibility of the adversary to learn the jammer csi . on the other hand , while the pilot feedback successfully hides the main csi , it enables the adversary to estimate the jammer csi . in terms of the secrecy encoding strategies ,
we address the possibilities under two general scenarios : the ergodic and the delay - limited . in the former case ,
one message is encoded across infinitely many blocks and in the latter case , a separate message is encoded over each block , to be decoded immediately . thus , in the delay - limited scenario , we also impose an additional probabilistic constraint on the decoding and secrecy outage events . in the delay limited scenario ,
we show that by revealing the main csi , the receiver achieves a higher secrecy rate under the outage constraint , compared to transmission with no csi .
furthermore , we show that main csi is more valuable for the adversary than the jamming csi in both delay - limited and ergodic scenarios . in the ergodic scenario
, we observe that the transmitter may not need the csi to achieve higher secrecy rates .
there is a recent research interest on hybrid adversaris . in @xcite ,
the authors formulate the mimo wiretap channel as a two player zero - sum game in which the payoff function is an achievable ergodic secrecy rate .
the strategy of the transmitter is to send the message in a full power or to utilize some of the available power to produce artificial noise .
the conditions under which pure nash equilibrium exists are studied . in @xcite , the authors consider fast fading main and eavesdropper channels and static jammer channel . under this channel configuration
, they propose a novel encoding scheme which is called block - markov wyner secrey encoding . in @xcite
, the authors introduce a pilot contamination attack in which the adversary jams during the reverse training phase to prevent the transmitter from estimating the main csi correctly . as a result , the transmitter incorrectly designs precoder which will increase the signal strength at the adversary that eavesdrops the main channel during the data transmission phase .
the rest of this paper is organized as follows . in section [ chap :
system ] , we first describe the system model .
we then explain the channel model and csi feedback models . at the end of the section
, we explain the problem formulations . in section [ chap : result ] , we present the results for both delay limited and ergodic scenarios . in section [ chap : numeric ] , we present our numerical results and conclude the paper in section [ chap : conc ] .
in this paper , we focus on a block fading channel model .
time is divided into discrete blocks and there are @xmath0 channel uses in each block .
channel state is assumed to be constant within a block and varies randomly from one block to the next .
we assume all parties are half - duplex , thus the adversary can not jam and eavesdrop simultaneously .
the observed signals at the legitimate receiver and the adversary in @xmath1-th block are as follows : @xmath2 where @xmath3 is the transmitted signal , @xmath4 is the transmission power , @xmath5 is the signal received by the legitimate receiver , @xmath6 is the signal received by the adversary , @xmath7 , @xmath8 , and @xmath9 are noise vectors distributed as complex gaussian , @xmath10 , @xmath11 , and @xmath12 , respectively , and @xmath13 is the jamming power .
indicator function @xmath14 , if the adversary is in a jamming state in the @xmath1-th block ; otherwise @xmath15
. channel gains , @xmath16 , @xmath17 , and @xmath18 are defined to be the independent complex gains of transmitter - to - receiver channel , transmitter - to - adversary channel , and adversary - to - receiver channel , respectively ( as illustrated in figure [ fig : sysmdl ] ) .
associated power gains are denoted with @xmath19 , @xmath20 , and @xmath21 .
we assume that channel reciprocity principle is valid for all channels , i.e. , reverse channels and forward channels have identical gains .
we also assume that joint probability density function of instantaneous power gains , @xmath22 , where @xmath23 $ ] , is well defined and known by all entities .
the legitimate pair may choose to obtain main csi or communicate without it .
we call the latter strategy the _ no csi case_. if they choose to obtain main csi , at the beginning of each time block , the transmitter sends training symbols to the legitimate receiver . in this paper
, we ignore the overhead associated with this training process .
we assume that , using the training symbols sent at the beginning of block @xmath1 , the legitimate receiver obtains perfect knowledge of @xmath16 and the adversary obtains perfect knowledge of @xmath17 . once the receiver observes main csi , it uses two possible methods for feeding back this information .
the first one is directly feeding back the observed channel state : the value of @xmath16 is encoded at the receiver and sent to the transmitter in a feedback packet .
thus , we call this feedback method the _ packet feedback_. we assume that the legitimate receiver and the adversary both decode this packet successfully and learn @xmath16 .
the second method is using pilot based csi feedback in which the receiver sends training symbols to the transmitter .
we call this second method the _ pilot feedback_. we assume that by using these reverse training symbols , the legitimate transmitter obtains perfect knowledge of @xmath16 and the adversary obtains the perfect knowledge of @xmath18 . thus , in the first method , the adversary obtains the knowledge of @xmath16 , but not @xmath18 , whereas the reverse is true in the second method .
the goal of the adversary is to minimize the achieved secrecy rate .
the strategy space of the adversary in each block is binary : jamming or eavesdropping .
the transmitter does not observe the strategy of the adversary in any given block , whereas we assume that the adversary knows the strategy of the transmitter a priori . from one block to the next , the adversary chooses its strategy based on the transmitter s strategy and the obtained channel power gains , i.e. , @xmath24 for the no csi case , @xmath24 , @xmath25 for the packet feedback case , and @xmath24 , @xmath26 for the pilot feedback case .
we denote the vector of channel power gains observed by the adversary at the beginning of the @xmath1th block with @xmath27 .
the secrecy level of a transmitted message is measured by the equivocation rate at the adversary .
the equivocation rate at the adversary is defined as the entropy of the transmitted message conditioned on the channel output and the available csi at the adversary .
if the equivocation rate is equal to the secrecy rate , the message is said to be transmitted with _
perfect secrecy_. in our system , we consider two levels of encoding . at the higher level
, secrecy is realized wyner code introduced in @xcite .
there , @xmath28 represents a wyner code of size @xmath29 that bears a confidential message set @xmath30 , where @xmath31 is the codeword length in number of bits . a message , @xmath32 is mapped to @xmath33 bits by a secrecy encoder @xcite and these @xmath33 bits
are then mapped to channel encoded bits at the lower level of encoding using a sequence of codes , @xmath34 , one for each block @xmath35 . here , the code rate , @xmath36 , is chosen based on the main csi , obtained at the transmitter .
the sequence of codewords is denoted with @xmath37 , and the decoder , @xmath38 maps the received sequence , @xmath39 to @xmath40 .
the average error probability of the sequence of codes @xmath41 is denoted with the associated sequence @xmath42 . in this paper , we focus on two scenarios as to how secrecy encoding and channel encoding are applied : delay limited and ergodic . in the delay - limited scenario ,
the transmitter encodes a separate secret message , @xmath43 in each block @xmath1 .
consequently , we use a separate secrecy encoder @xmath44 for each block , where @xmath45 is chosen to be identical to @xmath36 to meet the channel rate .
the channel encoder merely maps these @xmath46 bits to @xmath0 gaussian random variables , @xmath47 , forming a gaussian codebook @xmath48 . for the delay - limited scenario ,
we define the secrecy outage and connection outage events @xcite as : @xmath49 respectively , where @xmath50 in the ergodic scenario , the transmitter has one secret message @xmath51 and encodes it over @xmath52 blocks using the @xmath53 encoder . in the ergodic scenario ,
secrecy rate @xmath54 is said to be achievable if , for any @xmath55 , there exists sequence of channel codes @xmath41 for which the following are satisfied : @xmath56 for sufficiently large n and m and for any @xmath57 such that @xmath58=1 $ ] . here , we consider two possible channel encoding strategies : * 1 .
encoding across blocks : * we use this strategy in the no main csi case .
in this strategy , the @xmath33 bits at the output of the secrecy encoder is channel encoded via a single @xmath59 gaussian codebook and the @xmath31 symbols are transmitted over the channel over @xmath52 blocks .
block - by - block encoding : * we use this strategy when the main csi is available . to utilize the main channel knowledge
, the transmitter chooses some @xmath60 and encodes the information using a gaussian codebook with the rate @xmath61 over block @xmath1 . with this choice , the transmitted codeword
will not be decoded successfully if the adversary is in the jamming state and @xmath62 . to handle this possibility
, we use a plain arq strategy in each block ( similar to @xcite ) .
transmissions that receive a negative acknowledgement ( nak ) are retransmitted until they are decoded successfully .
one can notice that , in both the delay - limited and the ergodic scenarios , we use a constant secrecy rate @xmath54 .
the goal of the transmitter is to maximize the secrecy rate , @xmath54 over the strategy space of secrecy encoding and channel encoding rates . to that end , we consider the worst case scenario , in which the adversary perfectly tracks the strategy of the transmitter in each block .
furthermore , since the transmitter does not know the real state of the adversary , the strategy pair of the transmitter should satisfy the constraint for any arbitrary strategy of the adversary .
thus , we choose the secrecy rate using : @xmath63 for all @xmath1 in both the delay - limited and the ergodic scenarios , subject to the following outage constraint in the delay - limited scenario only : @xmath64 with probability 1 , where @xmath65 and @xmath66 are indicator functions that take on a value @xmath67 in case of a connection and a secrecy outage , respectively . we evaluate @xmath68 under the no csi , packet feedback , and pilot feedback cases .
note that the constraint enforces that the fraction of packets that are not in the both secrecy outage and the connection outage should be larger than a threshold as the number of blocks , @xmath52 , goes to infinity .
note that , the reason why we focus on the specific encoding strategies specified under delay limited and ergodic scenarios in section [ subsec : encoding ] is that , the solution , @xmath69 , of the maximin problem stated in ( [ problem ] ) is unknown @xcite .
the outage constraint in ( [ threshold ] ) is formulated as @xmath70 note that in ( [ constraint2 ] ) , the secrecy outage event is represented with @xmath71 and the connection outage events are given as @xmath72 and @xmath73 . for the delay limited scenario ,
the solution of ( [ problem ] ) , subject to constraint ( [ constraint2 ] ) leads to the following ordering of the achievable rate with respect to the type of feedback : @xmath74 [ thm : main ] the above theorem implies that main csi , which is obtained with packet feedback is more valuable for the adversary than the jammer csi , which is obtained with pilot feedback .
we now give an outline for the proof .
the details of the proof can be found in section [ chap : proof ] .
the basic idea is to compare the feasible set of the problem ( [ problem ] ) for three cases .
the feasible set is defined as @xmath75 where @xmath76 is the set of channel power gains , @xmath77 , such that the adversary is in the jamming state if @xmath78 . the equivalent form of ( [ problem ] ) is as follows : @xmath79 .
we observe that the solution to ( [ problem ] ) is directly related to the size of the feasible set .
the strategy pair @xmath80 is the element of the feasible set , @xmath81 if @xmath82
\geq \alpha\text { w.p . 1}$ ] . we can write @xmath83 for the no csi and packet feedback cases as follows : @xmath84\nonumber\\ & c_{min}^{\text{no csi}}=\nonumber\\ & p\left[r_s + \log\left(1+ph_e\right)\leq r\leq \log\left(1+\frac{ph_m}{1+p_j h_z}\right)\right]\nonumber\end{aligned}\ ] ] we can observe that @xmath85 so we have @xmath86 . by assuming the adversary is full - duplex , we find a lower bound , @xmath87 for @xmath88 at csi feedback cases .
then , we observe @xmath89 which shows @xmath90 and this concludes the proof .
we first present a secrecy rate that is achievable under the no csi case .
[ thm : nocsi ] the achieved secrecy under no csi is : @xmath91\right]^{+}\label{nofeedback}\end{aligned}\ ] ] the proof of ( [ nofeedback ] ) can be found at section [ section : nocsi ] .
next , we present an upper bound for the secrecy rates achieved with the block - by - block encoding strategy .
[ upperbound ] under the block - by - block encoding strategy , the achievable secrecy rate is upper bounded by @xmath92^+ \\
\times p[h_z \leq h_z^ * ] \end{gathered}\ ] ] to find this upper bound , we employ the following strategy .
when the transmitter receives a nak signal , the transmitter sends an independent group of bits on the next block instead of retransmitting the previous packet @xcite . in @xcite ,
the authors use this scheme for the secret key sharing .
the crucial observation is that an upper bound for the achievable secret key rate is also an upper bound for the achievable secrecy rates .
the details of the proof can be found at section [ chap : upperbound ] .
note that with the plain arq strategy described in section [ chap : ergodicformulation ] , the achievable rate is identical to the expression provided in ( [ eq : block_upper_bnd ] ) , with @xmath93 $ ] replaced with @xmath94)^2 $ ] , and @xmath95 in the expectation term replaced with @xmath60 as shown in @xcite . by comparing the upper bound given in theorem [ upperbound ] ,
we gain understanding on the performance of no csi case . in particular , whenever the achievable rate with the no csi case exceeds this bound , we know for sure that encoding across blocks ( no csi ) is preferable over block - by - block encoding with csi .
the main difference between block - by - block encoding and encoding across blocks is that , in the former , the unsuccessfully received packets are discarded , whereas in the no csi case , all the information received by the receiver is used to decode the message .
the set of parameters for which this is the case is illustrated in section [ chap : numeric ] in an example .
one can also write a general relationship between the performance with packet feedback and pilot feedback : [ comparison ] any secrecy rate achievable with packet feedback strategy is also achievable with pilot feedback strategy .
theorem [ comparison ] , shows that main csi is more valuable for the adversary than the jamming csi in the ergodic scenario , as was the case in the delay limited scenario .
the proof can be found at section [ chap : comparison ]
we first analyze the delay - limited case .
we assume that both main and eavesdropper channels are characterized by block rayleigh fading , where the main channel and eavesdropper channel power gains follow an exponential distribution with a mean 10 and 1 , respectively .
we also assume the jamming channel does not experience fading , where power gain is equal to 1 .
the transmission power , @xmath4 , and the jamming power , @xmath13 are identical and chosen to be @xmath96 . in figure
[ fig : limited ] , we plot the secrecy rate , @xmath54 , as a function of the outage constraint threshold , @xmath97 under the no csi , the packet feedback , and the pilot feedback cases .
the achievable rates in figure [ fig : ergodic ] follow the same ordering as given in theorem ( [ thm : main ] ) .
next , we simulate the ergodic scenario and compare the two strategies , encoding across blocks without csi and block - by - block encoding with packet csi feedback .
we used the same power parameters as in the simulations for the delay - limited case .
all three channels are assumed to be block rayleigh - fading wtih @xmath98 = 1 $ ] we select the encoding parameter , @xmath60 such that @xmath99 = 0.75 $ ] . in figure [
fig : ergodic ] , we illustrate the region where the encoding across blocks outperforms the block - by - block encoding on the @xmath100,e[h_m]\right)$ ] space .
the region to the left of the border , given in the plot contains the set of @xmath100,e[h_m]\right)$ ] for which the encoding across blocks results in a higher secrecy rate .
the intuition behind this observation is that , when @xmath101 $ ] is much larger than @xmath102 $ ] , the positive operator inside the upper bound , @xmath103 loses its significance .
we consider the wiretap channel model under the presence of half duplex adversary that is capable of either jamming or eavesdropping at a given time .
we analyzed the achievable rates under a variety of scenarios involving different methods for obtaining transmitter csi .
in particular , we considered no csi , csi with packet based feedback , and csi with pilot based feedback .
each method provides a different grade of information not only to the transmitter on the main channel , but also to the adversary on all channels .
we show for the delay limited scenario that , the highest secrecy rate is achieved with the pilot based feedback .
similarly , in the ergodic case , we prove that the pilot - based csi feedback outperforms the packet - based csi feedback , however interestingly , in certain cases no csi may lead to a higher achievable secrecy rates than with csi .
feasible set for the problem ( [ problem ] ) is defined to be @xmath104 where @xmath76 is the set of channel power gains such that the adversary is in the jamming state if @xmath78 .
we have the following lemma . if @xmath105 , the solution to ( [ problem ] ) is identical for all strategies of the adversary .
the equivalent form of ( [ problem ] ) is as follows : @xmath106 as seen from ( [ equivalent ] ) , @xmath68 does not depend on strategy of the adversary .
however , the size of feasible set directly depends on the csi feedback scheme . from ( [ equivalent ] )
, we can see that the solution to ( [ problem ] ) is directly proportional with the size of the feasible set . in the rest of the proof
, we will use this observation to show the ordering .
@xmath107 in the no feedback case , constraint term ( [ constraint2 ] ) is reduced to @xmath108 \text { , w.p .
1}\label{sll}\\ & = \int_{h_e } e\left[i_{\log\left(1+\frac{ph_m}{1+p_j h_z}\right ) \geq r } i_{h_e \in a_p}\right.\nonumber\\ & \;+\left.i_{\log\left(1+ph_e\right ) \leq r - r_s}i_{\log\left(1+ph_m\right ) \geq r } i_{h_e \notin a_p}\mid h_e = h_e\right]\nonumber\\ & \qquad\qquad\qquad \qquad \qquad\qquad\qquad f_{h_e}(h_e)\,d h_e \text { , w.p .
1}\\ & = \int_{h_e } e\left[i_{\log\left(1+\frac{ph_m}{1+p_j h_z}\right ) \geq r}\right ] i_{h_e \in a_p}\nonumber\\ & \qquad\qquad+e[i_{\log\left(1+ph_m\right ) \geq r } ] i_{\log\left(1+ph_e\right ) \leq r - r_s } i_{h_e \notin a_p}\nonumber \\ & \qquad\qquad \qquad \qquad\qquad \qquad \qquad
f_{h_e}(h_e)\,d h_e \text { , w.p . 1 } \label{strategy}\end{aligned}\ ] ] where ( [ nocsi ] ) follows from the fact that we have @xmath109 since the adversary only knows the instantaneous power gain of the eavesdropper channel , ( [ sll ] ) follows from the strong law of large numbers theorem and ( [ strategy ] ) follows from the independence of @xmath110 , @xmath111 , and @xmath95 .
@xmath112 set that minimizes @xmath113 is as follows : @xmath114 the minimized @xmath113 for a given rate pair ( @xmath115 ) is @xmath116\nonumber\end{aligned}\ ] ] feasible set for the transmitter for the no csi feedback case is as follows @xmath117 we now show that @xmath85 . for the packet feedback case ,
channel encoding rate is a function of the main csi , @xmath111 .
the adversary knows the instantaneous power gains of the eavesdropping channel and the main channel .
constraint term in ( [ constraint2 ] ) is reduced to @xmath118,\nonumber\\ & \qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad \text{w.p .
1}\label{sll2}\\ % & = \int_{h_e } e\left[i_{\log\left(1+\frac{ph_m}{\sigma^2+p_j h_z}\right ) \geq r(h_m ) } i_{h_m , h_e \in a_p } + i_{\log\left(1+\frac{ph_e}{\sigma^2}\right ) \leq r(h_m)-r_s } i_{h_m , h_e \notin a_p}\mid h_e = h_e\right]\nonumber\\ % & f_{h_e}(h_e)f_{h_m}(h_m)\,d h_e d h_m \text { , w.p .
1}\\ & = \int_{h_m , h_e } e\left[i_{\log\left(1+\frac{ph_m}{1+p_j h_z}\right ) \geq r(h_m)}\right ] i_{h_m , h_e \in a_p } \nonumber\\ & + i_{\log\left(1+ph_e\right ) \leq r(h_m)-r_s } i_{\log\left(1+ph_m\right ) \geq r(h_m)}i_{h_m , h_e \notin a_p } \nonumber\\ & \qquad\qquad \qquad f_{h_e}(h_e)f_{h_m}(h_m)\,d h_e d h_m \text { , w.p . 1 } \label{strategy2}\end{aligned}\ ] ] where ( [ packetcsi ] ) follows from the facet that @xmath119 since the adversary knows both the instantaneous power gain of the eavesdropper channel and the main channel .
( [ sll2 ] ) follows from the strong law of large numbers theorem and ( [ strategy2 ] ) follows from the independence of @xmath110 , @xmath111 , and @xmath95 .
@xmath112 set that minimizes @xmath113 is as follows : @xmath120 when we combine ( [ strategy2 ] ) and ( [ setmin2 ] ) , the minimized can be @xmath121 written as @xmath122\nonumber.\end{aligned}\ ] ] feasible set for the transmitter for the packet feedback case is as follows @xmath123 it is easy to see that @xmath85 then we have @xmath107 .
@xmath124 we will show that for any given @xmath125 , @xmath121 is a lower bound for ( [ constraint2 ] ) in the channel feedback cases .
constraint ( [ constraint2 ] ) can be written as @xmath126\label{newconst}%= \lim_{m\to \infty}1/m \sum_{i=1}^{m } \left [ i_{(h_m(i),h_e(i),h_z(i ) \in \overline{\mathcal{o}}}\right ] \label{newconst}\ ] ] where @xmath127 constraint @xmath113 decreases as the siz of set @xmath128 increases .
let s define an upper bound for @xmath128 : @xmath129 where @xmath130 .
then , we find a lower bound for @xmath113 by putting @xmath131 in ( [ newconst ] ) : @xmath132 since @xmath133 , we have @xmath134 .
we employ the _ encoding across block _ strategy explained in section [ subsec : encoding ] , where @xmath135 $ ] . if @xmath136 $ ] , the adversary prevents the reliable communication by jamming at every block .
note that in no csi case , the adversary only obtains @xmath24 so equivocation rate is defined as @xmath137 .
equivocation analysis for the _ encoding across block _ is as follows .
@xmath138 where @xmath139 follows from the fact that codeword @xmath140 is uniformly distributed over a set of size @xmath29 .
we continue with the following steps .
@xmath141-\epsilon_1\nonumber \\ & \qquad\qquad\qquad\qquad-\frac{h(x^{nm}|z^{nm } , w , h_e^m)}{nm } , \nonumber\\ & \stackrel{(c)}{\geq}r_m- e\left[\log(1+ph_e)\right]-\epsilon_1-\epsilon_2\nonumber\\ & = r_s -\epsilon,\nonumber\end{aligned}\ ] ] where @xmath142 . here
, @xmath143 is satisfied for any @xmath144 and @xmath145 with @xmath146=1 $ ] and @xmath147 since @xmath148\ ] ] , @xmath149 follows from the fano s inequality .
let s define @xmath150 and @xmath151 $ ] where @xmath152 is the estimation of the codeword @xmath140 .
@xmath153 here , any @xmath154 , ( [ fano ] ) is satisfied for sufficiently high @xmath0 and @xmath52 .
the reason is that since @xmath155 , @xmath156 for the sequence of codes ( @xmath157 ) as @xmath158 .
we first show that if @xmath54 is an achievable secrecy rate under the packet feedback strategy , we have for any @xmath159 .
note that here , the message @xmath160 is conditioned on random vectors , @xmath161 and @xmath162 .
@xmath163 where @xmath164 is the set defined in section [ chap : ergodicformulation ] .
here , ( [ definition ] ) follows from the definition of achievability , and ( [ pr1])follows from the fact that @xmath58=1 $ ] .
we consider the following case for the rest of the proof .
when the transmitter receives a nack signal , on the next block , the transmitter sends an independent group of bits instead of retransmitting the previous packet @xcite . in @xcite
, the authors use this scheme for the secret key sharing .
the crucial observation is that an upper bound for an achievable secret key rate is also an upper bound for achievable secrecy rates .
we define index set f that contains the indexes of blocks on which the transmitted codeword is successfully decoded .
suppose that size of f is @xmath165 , @xmath166 , and @xmath167 .
we now prove that @xmath168 .
@xmath169^+ i(i)\nonumber\\ & \qquad \qquad \qquad \qquad \qquad \qquad\qquad \qquad \qquad \qquad + n\delta_n \nonumber\end{aligned}\ ] ] where @xmath170 . here , ( a ) follows from fano s inequality ( b ) follows from the independent choice of the codeword symbols transmitted in each block that does not allow the eavesdropper to benefit from the observations corresponding to the previous nacked blocks , ( c ) results from the data processing inequality , ( d ) follows from the independence of @xmath171 and @xmath172 , ( e ) follows from @xcite , and ( f ) follows from @xcite .
@xmath173^+\nonumber\\ & \qquad \qquad \qquad \qquad \qquad \qquad \times\frac{1}{m}\sum_{i=1}^m \mathbf{i(i)}+\delta_n\\ & r_e \stackrel{(g)}{\leq } \alpha\mathbf{e}\left[\log\left(1+\frac{ph_m}{1+p_jh_z}\right)-\log(1+ph_e)\right]^+\nonumber\end{aligned}\ ] ] ( g ) follows from the fact that @xmath174 and from the ergodicity of channels such that @xmath175 \text { as } m \to \infty$ ] .
suppose that @xmath54 is a secrecy rate achieved with the packet based strategy and @xmath176 .
notice that the equivocation rates for the pilot and packet feedbacks are defined as @xmath177 and @xmath178 , respectively .
since @xmath54 is an achievable rate with the packet based strategy , by definition , for any @xmath55 there exists @xmath179 and @xmath180 such that for @xmath181 and @xmath182 , we have @xmath183 , @xmath184 with @xmath185 .
we define @xmath186 .
since @xmath162 and @xmath161 are independent random vectors and @xmath187 = 1 $ ] , we have @xmath188 = 1 , \forall h_e^m \in \mathcal{a}_m$ ] . to observe that @xmath189 f_{h_e^m } ( h_e^m ) \;dh_e^m = 1\end{aligned}\ ] ] we can see that @xmath190 = 1 $ ] , @xmath191 such that @xmath192=1 $ ] .
we now prove the lemma with following inequalities .
@xmath193=1.\end{aligned}\ ] ] @xmath139 follows from the fact that conditioning reduces the entropy , @xmath143 follows from the fact that since @xmath194 , @xmath195 , and @xmath149 follows from the fact that @xmath190 = 1 $ ] .
1 y. liang , h. poor , and s. shamai , `` secure communication over fading channels , '' _ ieee transactions on information theory _
24702492 , june 2008 .
p. gopala , l. lai , and h. el gamal , `` on the secrecy capacity of fading channels , '' _ ieee trans inf .
54 , no . 11 , pp . 50595067 , nov .
x. tang , r. liu , p. spasojevic , and h. v. poor , `` on the throughput of secure hybrid - arq protocols for gaussian block - fading channels,''__ieee trans .
inf . theory__,vol .
55 , no . 4 , pp .
15751590 , apr .
a. d.wyner , `` the wire - tap channel '' .
_ bell syst .
j. _ , 54(8):13551387 , october 1975 t. l. marzetta and b. m. hochwald , `` fast transfer of channel state information in wireless systems , '' _ ieee trans .
signal process .
4 , pp . 12681278 , apr . 2006 .
g. amariucai and s. wei , `` half - duplex active eavesdropping in fast fading channels : a block - markov wyner secrecy encoding scheme , '' _ ieee trans .
inf . theory _ , vol .
58 , no . 7 , pp .
46604677 , july 2012 .
a. mukherjee and a. l. swindlehurst , `` jamming games in the mimo wiretap channel with an active eavesdropper,''__ieee trans .
signal process .
_ _ , vol .
1 , jan . 2013 .
x. zhou , b. maham , and a. hjrungnes , pilot contamination for active eavesdropping , " _ ieee trans .
wireless commun .
903907 , mar .
z. rezki , a. khisti , and m. alouini , `` on the ergodic secret message capacity of the wiretap channel with finite - rate feedback , '' in _ proc .
inf . theory _
, 2012 , pp . 239243 .
s. k. leung - yan - cheong and m. e. hellman , `` the gaussian wire - tap channel , '' _ ieee trans .
inf . theory _ ,
it-24 , no . 4 , pp . 451456 , jul .
y. liang , g. kramer , h. v. poor , and s. shamai , `` compound wiretap channels , '' _
eurasip j. wireless commun .
_ , pp . 113 , 2009 , article 142374 .
y. abdallah , m. a. latif , m. youssef , a. sultan and h. ei - gamal , `` keys through arq : theory and practice , '' _ ieee transactions on information forensics and security _
, vol . 6 , no .
3 , pp . 737751 , sep | we consider the wiretap channel model under the presence of a hybrid , half duplex adversary that is capable of either jamming or eavesdropping at a given time .
we analyzed the achievable rates under a variety of scenarios involving different methods for obtaining transmitter csi .
each method provides a different grade of information , not only to the transmitter on the main channel , but also to the adversary on all channels .
our analysis shows that main csi is more valuable for the adversary than the jamming csi in both delay - limited and ergodic scenarios .
similarly , in certain cases under the ergodic scenario , interestingly , no csi may lead to higher achievable secrecy rates than with csi . |
in statistical physics one basically considers a large set of reversible and conservative ordinary differential equations for the description of particle dynamics .
the temporal evolution for this cumbersome problem , even for a modest number of particles , requires a statistical description which introduces the concept of probability distribution function ( pdf ) for the phase space of the system .
irreversibility , equilibrium , and more important non - equilibrium properties , surge from this probability conception of systems ( with a large number of degrees of freedom ) and its deterministic evolution .
briefly , the methodology reduces ( under some assumptions ) to a kinetic description which displays an irreversible behavior to equilibrium observed in macroscopic systems .
the assumptions for this approach are : i ) macroscopically , a system is described by a finite set of observables ; ii ) the robust instability of the microscopic motions which is at the basis of the sensibility to initial conditions and the ergodic assumption .
iii ) a _ stosszahlansatz _ that introduces explicitly a broken before - after symmetry of the probability distribution evolution .
about 20 years ago , nicolis _
@xcite introduced a systematic coarse - graining approach to the treatment of the macroscopical variables .
as a consequence , this coarse - graining breaks naturally the past - future symmetry in time , leading to an irreversible master equation , for a reduced probability distribution function of the system . in the present paper , we apply this systematic approach to a conservative and explicit reversible cellular automata . in particular , we consider the q2r model , introduced by vichniac in the mid 80s @xcite , which is a cellular automata that runs on a two dimensional grid of finite size , and is reversible in a physical sense , that is , the automata rule is not only invertible , but the backward rule reads exactly the same as the forward one .
moreover , it was shown by pomeau @xcite that the q2r automata possesses a conserved energy - like quantity . the main reason to apply the coarse - graining approach to a cellular automata instead to a coupled system of ordinary differential equations ,
is because a cellular automata is a discrete model with boolean entities as microscopic variables , thus , the system is numerically reversible and conservative . in consequence , q2r seems to be a good benchmark to test the principles of statistical physics .
but , the phase space is finite , hence the dynamical system only possesses fixed points and periodic orbits , therefore it can not be ergodic , at least in the usual sense of continuos dynamics .
however , for large enough systems , the phase space becomes huge , and the periodic orbits may be , as we show , exponentially long , so in practice , of infinite period .
further , if the initial state is random , the temporal behavior may be quite random and it possesses many aspects of chaotic systems , as sensibility to initial conditions , mixing , etc . for any purpose , the observation of a short periodic orbit is really improbable for large enough systems with random initial conditions . in general , there is a huge number of initial conditions that are `` almost '' ergodic .
numerical studies show that the premises of statistical physics are valid , in particular the observables may be computed using the standard methods of statistical physics .
the study of the dynamics and properties of the q2r model has a long history . soon after the seminal works of vichniac @xcite and pomeau @xcite , herrmann@xcite implemented the q2r algorithm to study the two space dimensional ising model in the frame of the micro - canonical ensemble .
he studied the global magnetization , obtaining an excellent representation for the magnetization as a function of the initial conserved energy , displaying a coherent picture for the phase transition of the ising model .
later , herrmann , carmesin and stauffer @xcite studied numerically the probability to reach an `` infinitely '' long period for some energies .
moreover , if the energy is large enough this probability tends to unity @xcite .
next , takesue @xcite focused on the possible realization of statistical mechanics for reversible cellular automata .
his studies concerned explicitly all class of rule in the one dimensional case , the q2r being only a special case . however , the q2r ( 90r in his terminology ) , is the analogue of an ideal gas of particles with speeds + 1 or -1 , a system that can not reach equilibrium in practice .
but , it is ergodic only in thermodynamical equilibrium .
more recently , in ref .
@xcite , one of us ( sr ) has studied numerically the irreversible behavior and the existence of a spontaneous transition from a non - coherent state to a coherent state .
the present article is organized as follows , the q2r model , as well as its main features and findings are presented in section [ model ] .
this section is sub - divided as follows : we briefly report the numerical studies of ref .
@xcite in sub - section [ longtime ] ; the phase space properties , in particular some results on the distribution of periods of the dynamics is reported in subsection [ phasespace ] ; and , finally , the scope of the paper is presented in sub section [ scope ] .
section [ mastereqn ] , introduces the notion of a master equation for the statistical description of the dynamics .
finally , in section [ results ] we provide some precise examples , where a coarse graining is realized , in order to get an adequate and tractable master equation .
we provide an exhaustive validation of the technique and we discuss different coarse - graining over the phase space . finally , sec .
[ conclusions ] shows our conclusions .
for simplicity we shall consider a regular two dimensional lattice with @xmath0 nodes , in which each node is only seen by its four closest neighbors ( the von neuman neighborhood ) , finally we use periodic boundary conditions .
each node @xmath1 possesses a discrete value @xmath2 that may take a value + 1 or -1 .
the q2r model , introduced by vichniac @xcite , is based upon the following two step rule : @xmath3 where the function @xmath4 is such that @xmath5 and @xmath6 if @xmath7 . in the sum @xmath8 corresponds to the von neuman neighbor in the site @xmath1 .
the reversibility follows directly from the inverse relation @xmath9 which is the backward rule ( notice that @xmath10 in all cases ) .
this two step rule may be naturally re - written as an one step rule by introducing a second dynamical variable @xcite : @xmath11 the rule ( [ q2r ] ) is complemented with the initial condition @xmath12 and @xmath13 . as shown by pomeau @xcite , the energy @xmath14 = -\frac{1}{2 } \sum_{\left <
i , k\right > } x^{t}_k y^{t}_i , \label{energy}\end{aligned}\ ] ] is preserved , @xmath15 = e[\left\{x^{t=0},y^{t=0}\right\ } ] $ ] under the dynamics defined by the q2r rule ( [ q2r ] ) .
moreover , the energy is bounded by @xmath16 . despite the existence of an energy - like quantity ,
it is not possible to speak about a hamiltonian discrete dynamic because , the variables @xmath17 and @xmath18 and the energy ( [ energy ] ) are discrete quantities @xcite .
moreover , supported by the existence of a large number of periodic orbits ( see next ) , it is believed that q2r possesses a large number of other invariants , however , up to date , other conserved quantities are not known yet .
numerical simulations of the q2r model in two space dimensions for large system sizes , _ eg . _ @xmath19 , and random initial conditions , shows that the dynamics displays a fluctuating spatio - temporal pattern showing regions with states + 1 and sectors with states @xmath20 , as well as , zones with chessboard - like pattern @xcite . the full patterns will be characterized by the global magnetization : @xmath21 = \sum_{k } x^{t}_k , \label{magnetization}\end{aligned}\ ] ] naturally , the function @xmath22 is restricted to the interval @xmath23 , and the available values of @xmath22 are ranged in uniform steps of @xmath24 .
a detailed characterization of the evolution , as well as , of the fluctuations of the magnetization has been treated in detail in ref .
briefly , after a transient the average magnetization depends mainly on the initial energy .
if the energy is low , one sees that the average magnetization evolves slowly in time to an `` equilibrium '' state with an almost constant value plus weak fluctuations . for larger energies , the fluctuations enter to play an important role .
one may observe that the system is in an almost stable state , but then suddenly jumps into a metastable state with zero average magnetization , and then jumps into an opposite magnetization state @xcite .
the plot of the temporal average of the global magnetization versus the energy is done in fig .
one sees that the magnetization spontaneously increases below a critical energy per site around @xmath25 , close to the critical energy of the ising model @xmath26 .
@xcite . moreover , in refs .
@xcite it is compared the magnetization as a function of the internal energy of the system showing a close agreement with the numerical values .
magnetization curves as a function of initial energy for three different system sizes : @xmath27 , @xmath28 , and @xmath19 , as indicated in the inset .
each point corresponds to a different initial condition , in this case we sample different energies .
as it can be noticed , there is a finite system dependence on the critical behaviour of the system , indeed the critical behavior disappear for small system sizes @xmath27 , @xmath28 , whereas for large system the magnetization curve reach a critical behavior .
the continuous line represents the well known statistical mechanics calculation for the ising model @xmath29.,width=264 ] the configurational space of all states is defined through all possible values of the state @xmath30 .
the resulting space is composed by the @xmath31 vertices of a @xmath32-dimensional hypercube .
the smallest possible system corresponds to a @xmath33 lattice . in this case
there are @xmath34 states and the phase space is a hypercube in dimension 8 , however the dynamics is too simple , it does contain cycles of period 4 at most . the phase space of a @xmath35 system is the largest possible one that can be studied exactly case by case . in this case
the system possesses @xmath36 states and it contains a rich variety of cycles ( see table [ example4x4 ] ) .
this case will be studied deeply as a good benchmark for conjectures in larger dimensional systems . as an example , from this case , it is observed that the total number of cycles , @xmath37 , of period @xmath38 and energy @xmath39 , would be bounded by @xcite @xmath40 from the data one has that for @xmath41 and @xmath35 , @xmath42 , but this value varies as the lattice size increases . here one notices a dramatic difference among the cases @xmath43 greater or smaller than @xmath44 , if @xmath43 is greater than @xmath45 the probability to see a long period is exponentially small , but on the contrary , for @xmath46 this probability reaches the unity .
higher lattice sizes confirm this scenario but modify slightly the value of @xmath47 .
this behavior is consistent with the numerical simulations of ref .
@xcite .
though the q2r model is quite simple its dynamics is usually very rich as it has been documented extensively in literature .
moreover , this conservative and reversible system appears to behave as a typical macroscopic system , as the number of degrees of freedom increases , showing , in particular , a typical irreversible behavior , sensibility of the initial conditions , a kind of mixing , etc .
it is believed that this q2r is a good representation of an ising model in thermodynamical equilibrium .
the phase space of the q2r system of @xmath48 sites possesses @xmath49 states , which is partitioned in different sub - spaces of constant energies , which itself are partitioned in a large amount of smaller subspaces of periodic orbits or fixed points .
notice that , because the system is conservative , there are neither attractive nor repulsive attractors , all attractors are fixed points or cycles .
this feature of the phase space is schematized in fig .
[ cycledraft ] ( a ) , where the constant energy subspace shares in principle many cycles and fixed points . an arbitrary initial condition of energy @xmath39 falls into one of these cycles , and it runs until it comes back to the initial configuration after a time @xmath38 , which could be exponentially long and it displays a complex behavior ( not chaotic stricto - sensu , see for instance @xcite ) .
more important the probability that an initial condition possesses such a complex behavior is finite @xcite .
moreover , q2r manifests `` sensitivity to initial conditions '' , that is , if one starts with two distinct , but close , initial conditions , then , they evolve into very different cycles as time evolves @xcite . in some sense ,
an initial state explores vastly the phase space justifying the grounds of statistical physics . in conclusion ,
the overall picture is : although for a finite size system the deterministic automata q2r possesses a periodic dynamics so it is not ergodic , there is a huge number of initial conditions that explore vastly the configurational space ( this is particularly remarkably for initial conditions of random structure ) .
therefore , one expects that a master equation approach my be successful .
\(a ) ( a ) scheme of a sub - space of constant energy composed by a number of cycles and fixed points .
( b ) cartoon of a cycle of period @xmath38 , for which the cycle is composed of @xmath38 states .
, title="fig:",width=264 ] \(b ) ( a ) scheme of a sub - space of constant energy composed by a number of cycles and fixed points . ( b ) cartoon of a cycle of period @xmath38 , for which the cycle is composed of @xmath38 states .
, title="fig:",width=264 ]
given a set of initial conditions with a fixed energy @xmath39 , the probability distribution @xmath50 evolves following a perron - frobenius like - equation @xmath51 which , in principle , can be computed after the microscopic rule of evolution ( [ q2r ] ) .
indeed @xmath52 is easy to build : if the state @xmath53 at time @xmath54 goes to @xmath55 , at a time @xmath56 , then one sets the @xmath57 components to 1 , that is @xmath58 .
checking all available elements , @xmath59 , for a given energy we built the huge , @xmath60 , linear operator , @xmath52 .
this matrix possesses a large number of blocks and zeroes revealing the existence of a large number of cycles in the q2r model ( in some sense , @xmath52 is a kind of adjacency matrix of a graph , the graph being the total number of cycles shown for a given energy ) .
however , this description is impractical because of the typical magnitude of @xmath61 .
therefore , the full phase space is reduced to a description using gross or macroscopic variables , namely the total magnetization of the system ( [ magnetization ] ) , instead of microscopic variables .
let us denote by @xmath62 , the reduced probability distribution in terms of the variable @xmath22 .
the final master equation will be of the form @xmath63 the linear operator , @xmath64 , acts only in the subspace of constant @xmath39 , but is spanned over arbitrary values of the magnetization .
as the original perron - frobenius equation , @xmath64 depends explicitly of the q2r rule and it may be computed in principle . however , in practice , it is not possible because of the complex structure of the dynamics , in particular because of the existence of a myriad of different periods for a given @xmath39 .
empirically , it seems that the probability to reach a long period is finite , therefore it is probable that any random initial condition would reach a extremely long cycle and the sequence of magnetizations would represent a markov chain @xmath65 . moreover , it is possible to reduce again the information via a coarse - graining partition of the possible values of the markov chain .
this partition is defined through a finite number of sets of no overlapping intervals : @xmath66 $ ] . in this context , we denote the original distribution function @xmath67 by a discrete vector of dimension @xmath68 , that is : @xmath69 . finally , we obtain a coarse - grained like master equation @xcite @xmath70 where @xmath71 is the probability transition matrix defined via the following conditional probability:@xmath72 here @xmath73 is at interval @xmath74 at time @xmath54 , and @xmath75 will be at interval @xmath76 at @xmath56 . finally , @xmath77 denotes the transpose of @xmath71 , and , the matrix @xmath71 does not depend on time , which is a feature of a markov process .
the coarse - graining method is schematized in the following fig .
[ scheme ] .
the distribution @xmath78 at a time @xmath54 is schematized in the left distribution .
the fraction inside the interval @xmath74 , is distributed , after the evolution into a new distribution @xmath79 schematized in the left diagram .
the normalized distribution provides the @xmath80-th element of the @xmath80-th column : @xmath81 .
, width=302 ] important features of the master equation ( [ mastereqn ] ) are : \1 ) the probability vector @xmath82 should be positive and normalizable .
let @xmath83 be a @xmath68-dimensional vector , then we set @xmath84 more important , because of normalization , @xmath85 , one has @xmath86 .
this implies that the probability is conserved under the evolution @xmath87
\2 ) the perron - frobenius equation maybe solved exactly , provided an initial given distribution @xmath88 : @xmath89 .
\3 ) because of the frobenius theorem , it exists an eigenvalue which is one , @xmath90 , while others eigenvalues are inside the unitary circle @xmath91 for @xmath92 .
let @xmath93 be the eigenvector associated with the eigenvalue @xmath90 ; this is an invariant vector @xmath94 .
\4 ) in the following , we denote by @xmath95 the eigenvectors of @xmath77 corresponding to @xmath96 . naturally one has @xmath97
\5 ) the existence of an equilibrium state : @xmath98 .
\6 ) because all elements in the @xmath99-matrix are positive , any non negative initial distribution remains non negative .
one may wonder if one realizes the same process but instead to look at the system at times @xmath54 and @xmath56 , one looks at time @xmath54 and @xmath100 or more generally at @xmath54 and @xmath101 .
let us call @xmath102 the resulting probability transfer matrix after @xmath103 iterations , therefore it is easy to show that this matrix should verify the chapman - kolmogorov or compatibility condition @xcite , @xmath104 in particular , for @xmath105 one should satisfy @xmath106 which is true for conditional probabilities , because of the relation @xmath107 which is equivalent to the right hand side .
other compatibility conditions are @xmath108 finally , le us state an important result due to pomeau @xcite .
the @xmath68-time correlation functions imposes some restrictions on the @xmath99-matrix .
because of the time reversal symmetry , for all indices @xmath109 , the symmetry relation @xmath110 must be satisfied . in the following we apply this coarse graining approach to compute the probability transfer matrix for some particular cases
in this section we shall apply the coarse graining approach to the q2r dynamics in the case of small lattice size . in ref .
@xcite we have fairly explored the computation of the probability transition matrix , in particular , in the case of extended systems ( @xmath19 ) , however in this case the cycles are usually huge , therefore this general approach is not really complete . in this sense
, we focus our effort in treating moderate system sizes , namely @xmath111 , @xmath27 , @xmath28 , having all of them tractable cycles . in general for a system of small size ,
one is able to find some cycles for a given energy . building a time series for the magnetization @xmath112
, then one defines a partition on the possible values of the magnetization , as explained in a previous sec .
[ mastereqn ] . in the cases considered here ,
it is always possible to use the finest possible partition , that is , for the exact available values of the magnetization ( something impractical in large systems ) . in this case
the partitions are composed by a set of @xmath113 ( @xmath48 is assumed to be even ) well defined values @xmath114 .
that is for @xmath35 the partition has a maximum of 17 elements , for @xmath27 there are 65 elements , and for @xmath28 the partition possesses a maximum of 257 elements . a first result concerns the equivalence of the probability density function of magnetization obtained via the time series of the magnetization and the equilibrium distribution resulting from the eigenvectors of the probability transition matrix @xmath71
hence , the results arising from temporal averages and transition probability matrix in the configurational space are consistent themselves .
this fact ensures a first validation of the method .
however , the probability transition matrix provides extra information of a system , among them , the non - equilibrium properties , given by the spectrum of @xmath71 .
next , we shall describe the methodology for the case of a @xmath115 lattice size for an orbit of @xmath116 and a period @xmath117 .
the probability transition matrix @xmath71 is constructed following the steps of the previous section [ mastereqn ] . but
first , we shall verify that the master equation does not strongly depend on the length of the time series for the magnetization .
it is important to underline that we think that this is a crucial step , because it allows us to compare explicitly the dependence of the results on the partial length of the cycles . something not possible in larger systems , because in these cases we shall never be able to build the complete period for the time series . to test that
, we shall use again the finest partition . in this case the transition matrix has dimension @xmath118 ( so we shall not provide them explicitly ) and we shall characterize it by the equilibrium distribution , and the full set on eigenvalues of @xmath71 .
[ method ] ( a ) and ( b ) plot the equilibrium distribution @xmath93 and the set of 257 eigenvalues for the same sequence , @xmath119 , but for four different lengths of the time series : @xmath117 ( the complete cycle ) , and @xmath120 ( partial sequences ) .
visually one sees that there is not substantial difference among the different values of @xmath121 .
moreover , the following table [ tabletest ] compares quantitatively the mean square difference measuring @xmath122 and @xmath123 .
here @xmath68 is the number of partitions .
\(a ) ( a ) plot of the equilibrium distribution @xmath93 for the case of a @xmath124 system with @xmath116 ( @xmath125 ) and a cycle of a period @xmath126 .
the computation of @xmath93 is compared with shorter sequences of the same time series of length @xmath120 .
( b ) the set of 257 eigenvalues of the @xmath71-matrix for the same conditions of ( b ) .
naturally , @xmath127 is present by the definition of the probability transfer matrix @xmath71 .
, title="fig:",width=264 ] \(b ) ( a ) plot of the equilibrium distribution @xmath93 for the case of a @xmath124 system with @xmath116 ( @xmath125 ) and a cycle of a period @xmath126 .
the computation of @xmath93 is compared with shorter sequences of the same time series of length @xmath120 .
( b ) the set of 257 eigenvalues of the @xmath71-matrix for the same conditions of ( b ) .
naturally , @xmath127 is present by the definition of the probability transfer matrix @xmath71 .
, title="fig:",width=264 ] we notice that the case @xmath128 is showed up in the finest partition . actually , in the finest partition
this case the probability transfer matrix has a rank @xmath129 .
but it has three zero eigenvalues and two complex one .
we interpret that the fine coarse graining is not a good partition .
this partition must no be an invariant measure as required in ref.@xcite .
why does this happen in the present case ? , and it does not happen in other needs to be elucidated . | a master equation approach is applied to a reversible and conservative cellular automata model ( q2r ) .
the q2r model is a dynamical variation of the ising model for ferromagnetism that possesses quite a rich and complex dynamics . the configurational space is composed by a huge number of cycles with exponentially long periods .
following nicolis and nicolis [ _ phys .
rev . _ * a 38 * , 427 - 433 ( 1988 ) ] , a coarse - graining approach is applied to the time series of the total magnetization leading to a master equation that governs the macroscopic irreversible dynamics of the q2r automata .
the methodology is replicated for various lattice sizes . in the case of small systems
, it is shown that the master equation leads to a tractable probability transfer matrix of moderate size which provides a master equation for a coarse - grained probability distribution .
the method is validated and some explicit examples are discussed . |
holographic models of nonperturbative physical systems have been more successful quantitatively than should have been expected .
the most developed applications of holographic model building are to quantum chromodynamics ( qcd ) @xcite , electroweak symmetry breaking @xcite and condensed matter systems , especially superconductors @xcite .
arguments based on insensitivity to model details @xcite , approximate conformal invariance @xcite and decoupling of high - dimension states and operators @xcite have been put forward in an attempt to understand the unreasonable effectiveness of some of these models .
holographic models of 3 + 1 dimensional systems are 4 + 1 dimensional theories in which the behavior of fields near the boundary of the spacetime , typically anti - de sitter ( ads ) space , determines the properties of the corresponding lower - dimensional system .
however , gauge theories in more than 3 + 1 dimensions are generally nonrenormalizable .
deconstruction of extra dimensions provides a gauge - invariant completion of higher - dimensional gauge theories @xcite .
a deconstructed extra - dimensional model is a lower - dimensional theory which , below some energy scale , has an effective description in which one or more extra dimensions are latticized .
deconstruction is useful for model building in that it is sometimes possible to reduce the number of `` lattice sites '' to just a few while maintaining the interesting phenomenology of a higher - dimensional model , yielding a relatively simple model of the system of interest .
for example , in the context of electroweak symmetry breaking , deconstruction provides one route to little higgs models @xcite .
more recently , this approach has been used to construct models with some of the properties of holographic superconductors , even though defined in the natural dimension of the superconducting system @xcite . among the successful predictions of holographic models
are certain features of high - temperature superconductors such as an enhanced ratio of the superconducting gap @xmath1 to the critical temperature @xmath2 @xcite .
bottom - up holographic models of finite - temperature systems typically begin with an ads - schwarzschild or ads - reissner - nordstrom black hole geometry , the latter taking into account the backreaction of charge density on the geometry .
these geometries are chosen mainly for simplicity , but in holographic models of superconductors derived from string theory @xcite , the spacetime geometries may be more complicated and depend on the fluxes of fields associated with d - brane configurations .
other geometric backgrounds in holographic models arise as the induced metric on a brane embedded in a higher - dimensional spacetime , such as on the flavor branes in the holographic qcd model of sakai and sugimoto @xcite , and these induced geometries are not derived as the solution to einstein s equation with a specified energy - momentum tensor .
it is the goal of this paper to explore the sensitivity of observables to variations in the details of holographic models of superconductors and in deconstructed variations of those models . as such
, we consider holographic superconductors in generalizations of the 3 + 1 dimensional ads - schwarzschild metric .
we find certain generic features in the phenomenology of these models , but details such as the ratio of the superconducting gap to the critical temperature are sensitive to the model details , which suggests that successful quantitative predictions in prototypical models are likely accidental .
this is not to say that those models will not prove valuable in explaining the puzzling properties of unconventional superconductors , only that quantitative predictions are more model dependent than one might have hoped .
it has already been noted that there are quantitative and even qualitative distinctions between superconducting models , for example between those which take into account the backreaction of the charge density on the metric and those that do nt @xcite .
the work presented here focuses on sensitivity to the extra - dimensional spacetime , parametrizing the ads black - hole metric in a particular way in order to quantify the variability of superconducting observables in a class of holographic models and in deconstructed versions of those models . here
we briefly review the construction of holographic superconductor models and the calculation of observables in those models . in a holographic superconductor ,
a charged field condenses in an extra - dimensional black - hole background whose hawking temperature is below some critical temperature @xmath0 . the temperature of the lower - dimensional system is identified with the hawking temperature of the higher - dimensional black hole @xcite .
the charged condensate spontaneously breaks the electromagnetic u(1 ) gauge group and gives rise to superconducting phenomenology @xcite . in this work
we ignore the backreaction of the charge density on the spacetime geometry , and for now we consider an abelian higgs model in a 3 + 1 dimensional ads - schwarzschild spacetime background .
this is meant to describe a system which is superconducting in two spatial dimensions , _
e.g. _ the copper - oxide planes of cuprate superconductors .
we can choose coordinates such that the lengths are normalized to the ads scale and the metric has the form @xmath3 , \label{eq : metric}\ ] ] where @xmath4 with @xmath5 corresponding to the 3 + 1 dimensional ads - schwarzschild metric .
with the euclidean time @xmath6 compactified with period @xmath7 , the hawking temperature associated with the modified black - hole metric follows from the condition that there be no conical singularity at the horizon . in the absence of a conical singularity ,
if @xmath8 is the proper distance from the horizon @xmath9 to a nearby point displaced only in the radial ( @xmath10 ) direction and @xmath11 is the proper circumference of the euclidean - time circle at that fixed radial position , then @xmath12 . for metrics of the form ( [ eq : metric ] ) , for small proper displacements from the horizon , @xmath13 and with @xmath14 given by eq .
( [ eq : f ] ) , the hawking temperature is then : @xmath15 in the continuum model , observables are independent of the choice of coordinates .
however , away from the continuum limit , the deconstructed models are sensitive to the latticization of the extra dimension , which in turn depends on the coordinate choice . in the continuum model ,
the action for the scalar field @xmath16 and u(1 ) gauge field @xmath17 ( @xmath18 ) is @xmath19,\ ] ] where @xmath20 is the metric defined by eq .
( [ eq : metric ] ) . for definiteness
we take @xmath21 in ads units , as in refs .
@xcite and @xcite . near the boundary @xmath22 ,
the field @xmath16 has solutions @xmath23 in this model both independent solutions for @xmath24 have finite action , so the ads / cft interpretation of the two solutions is ambiguous .
here we choose the interpretation that @xmath25 is the condensate of the cooper pair operator , while @xmath26 would then be the external source for that operator , which we assume vanishes .
hence , @xmath27 is a boundary condition for the solutions of interest .
the bulk u(1 ) gauge field , @xmath17 , is dual to the electric current and the background electromagnetic field . in order to allow for nonvanishing chemical potential and charge density
, we consider solutions in which the time component , @xmath28 , is nonvanishing .
the equations of motion have solutions for which @xmath28 behaves near the boundary as @xmath29 where @xmath30 is the coefficient of the non - normalizable solution and is identified with the chemical potential , which is a source for @xmath31 , the charge density .
the phenomenology of the model is determined by fixing the temperature @xmath32 as it appears in the ads black - hole metric , solving the coupled equations of motion for @xmath16 and @xmath17 subject to the ultraviolet ( _ i.e. _ @xmath22 ) boundary conditions @xmath27 , @xmath33 , and the infrared ( _ i.e. _ @xmath34 ) boundary conditions @xmath35 and @xmath36 . the last condition follows from the equations of motion , but is enforced as a regularity condition on the numerical solutions . the cooper pair condensate @xmath37 and background charge density are then determined by @xmath25 ( _ c.f . _ eq .
( [ eq : psisol ] ) ) and @xmath31 ( _ c.f . _
( [ eq : asol ] ) ) , respectively .
varying the temperature @xmath32 then allows for a determination of the phase structure of the model , as @xmath38 for @xmath39 . to analyze the frequency - dependent conductivity we fix the background for @xmath16 and
instead solve the equations of motion for @xmath17 in a background with @xmath40 , @xmath41 , corresponding to a uniform oscillating background electric field @xmath42 , polarized in the @xmath43 direction .
solutions are chosen to be ingoing at the horizon in order to enforce causal behavior of the current two - point function @xcite .
the solution for @xmath44 as @xmath45 then determines the electric current @xmath46 , from which the conductivity , @xmath47 follows .
a generic feature of superconductors is the existence of a frequency gap @xmath48 below which there are no modes available to excite and generate a current , so that @xmath49 for @xmath50 for @xmath51 .
for nonvanishing temperature , even as @xmath52 the current may be nonvanishing , where for small enough temperature @xmath53 , where @xmath54 is the superconducting gap . from the weakly coupled bcs theory , we would expect @xmath55 , which also appears to be satisfied in the original model of ref .
. we will study a class of models , based on the models of ref .
@xcite , in which the extra dimension of the holographic superconductor is deconstructed .
models with certain similarities to these were also considered in ref .
the higher - dimensional u(1 ) gauge theory is replaced by a u(1)@xmath56 gauge theory in one fewer dimension , where @xmath57 in the continuum limit .
scalar link fields charged under `` neighboring '' pairs of u(1 ) gauge groups are arranged to have prescribed expectation values , breaking the u(1)@xmath56 gauge group in such a way that the resulting action is that of the latticized higher - dimensional theory .
the massive gauge fields replace the kaluza - klein modes in the continuum model .
the fluctuations of the link fields do not correspond to degrees of freedom in the continuum theory , so we assume that they are heavy compared to the scales of interest in our analysis and disregard them in our analysis . expanding the fields in components ,
the action of the holographic model is , @xmath58 \ , , \end{aligned}\ ] ] where the lower - case latin indices @xmath59 are summed over the @xmath60 and @xmath61 coordinates .
we now latticize the spacetime in one dimension by replacing the @xmath10 coordinate by a discrete set of @xmath62 points : @xmath63 where @xmath64 , @xmath65 is the lattice spacing in @xmath10-coordinates , and @xmath66 is a uv cutoff .
the lagrangian for the deconstructed theory is of the form , @xmath67+\sum_{j=1}^{n-1 } \left [ |d_\mu \sigma_j |^2 - z_j v_j \right],\ ] ] where @xmath68 is the scalar potential for link field @xmath69 , and the coefficients @xmath70 and metric factors @xmath71 by which indices are contracted vary with lattice position @xmath72 . the parameters in the model may be chosen ( see ref .
@xcite for more details ) such that the effective theory below the scale set by the link fields is given by the lagrangian , @xmath73 + \sum _ { j=2}^{n-1 } a_j \left [ \frac{1}{2 f_j } ( f_{0a})_j^2 -\frac{1}{4 } ( f_{ab})_j^2 \right ] \nonumber \\ & + & \sum _ { j=2}^{n-1 } a_j \left [ \frac{1}{z_j^2 f_j } | \partial_0 \psi_j -i { \phi}_j \psi_j |^2 -\frac{1}{z_j^2 } |\partial_a \psi_j -i { a_a}_j \psi_j|^2 - \frac{1}{z_j^4 } m^2 |\psi_j|^2 \right ] \ , , \end{aligned}\ ] ] where @xmath74 , @xmath75 , and the primes correspond to discretized derivatives , for example @xmath76 the u(1 ) gauge group at the first lattice site ( the uv boundary site ) is identified with the electromagnetic gauge group .
solutions to the equations of motion with discretized versions of the boundary conditions on the fields @xmath16 and @xmath17 allow for the calculation of observables by analogy with the holographic analysis in the continuum model @xcite . in the case of a small number of lattice sites there is no _ a priori _ reason to expect phenomenology similar to that of the continuum model .
indeed , we find significant deviation from the predictions of the continuum model , though certain qualitative features remain .
more complete details of these computations are presented below .
to consider the sensitivity of observables to the spacetime geometry , we allow the power @xmath77 in eq .
( [ eq : f ] ) to deviate from its value @xmath5 corresponding to the ads - schwarzschild spacetime . for generic @xmath77
the metric is not a solution to einstein s equations with a prescribed energy - momentum tensor .
however , the initial choice of ads - schwarzschild geometry was made for simplicity and is equally arbitrary , and we can imagine either fluxes of fields that would give rise to the requisite energy - momentum tensor , or we can imagine that the class of spacetimes described here corresponds to the induced metric on a brane embedded in a particular higher - dimensional spacetime . the goal here is to parametrize a class of deviations from the prototypical spacetime in order to analyze the sensitivity of observables to the detailed form of the spacetime metric .
our particular choice of parametrized metric is mostly arbitrary , though the class of spacetimes considered here remains asymptotically ads near the boundary at @xmath22 .
we first analyze the continuum theory for @xmath78 and @xmath79 . in our numerics
, we cut off the spacetime in the uv at @xmath80 and near the horizon at @xmath81 .
we impose the boundary conditions discussed previously , and we fix @xmath82 , which by a scaling relation in the model also fixes @xmath83 @xcite . the superconducting condensate and the real part of the conductivity are plotted in fig .
[ fig : contin ] . the delta function in the real part of the conductivity may be inferred from a pole in the imaginary part , not shown in the figure , by the kramers - kronig relation for the conductivity . at low temperature
the conductivity features a sharp gap below which the real part of the conductivity nearly vanishes . at the gap frequency ,
re@xmath84 ) display a step - function type behavior , while im@xmath84 ) has a sharp local minimum . even at larger temperatures ,
we define the gap frequency @xmath48 as the location of the local minimum of im@xmath85 . in fig .
[ fig : cgap ] , we plot the conductivity with respect to frequency scaled in units of @xmath86 .
note that the three plots are nearly identical .
in particular , the ratio @xmath87 at the minimum of im@xmath85 is nearly idependent of @xmath77 in this range . however , as we will see there are important quantitative distinctions at small @xmath88 .
it was noted in the original model of ref .
@xcite that the gap - to-@xmath0 ratio is larger in the holographic model than in the weakly - coupled bcs theory , in rough quantitative agreement with experimental results in high - temperature superconductors
. the normal component of the dc conductivity is defined as @xmath89 $ ] . for low enough temperatures
, we find that @xmath90 in which @xmath91 , for some constant @xmath92 .
the coefficient @xmath54 in the exponent is the superconducting energy gap . in order to compare with
@xmath48 found previously , we fit our data for a range of @xmath93 around @xmath94 , which gives a good exponential fit for @xmath95 in that range , with relatively large @xmath96 .
the results are summarized in table i. & @xmath97 & & & @xmath98 & & @xmath99 & & @xmath100 & + & @xmath101 & & & @xmath102 & & @xmath103 & & @xmath104 & + & @xmath105 & & & @xmath106 & & @xmath107 & & @xmath107 & + [ tab : cgap ] we next examine the deconstructed model with @xmath78 and @xmath79 for @xmath108 .
we generally set the uv cutoff at @xmath109 , except for the case @xmath110 , for which we set @xmath111 to better match the continuum model .
the lattice spacing at the horizon is fixed at @xmath112 , decoupled from the lattice spacing in the bulk which varies as the horizon moves with temperature .
we again use a scaling relation to set @xmath113 so that @xmath0 is fixed .
as discussed in @xcite , we have the following discretized version of the boundary conditions : @xmath114 where the primes are discretized derivatives as in eq .
[ eq : deriv ] .
electromagnetism is defined as the u(1 ) interaction at the uv boundary site , _ i.e. _ the lattice site closest to @xmath22 .
we find solutions for which the @xmath60-component of the bulk gauge fields oscillate while the other components do not fluctuate , @xmath115 where on the right - hand side of eq .
( [ eq : axi ] ) , @xmath116 is time - independent .
the conductivity @xmath117 is found to be given by a discretized version of the holographic calculation for @xmath118 in the continuum model : @xmath119 to obtain reasonable phenomenology we find that an ingoing - wave type boundary condition is necessary even in the deconstructed models . due to the behavior of the metric near the horizon
, we find it beneficial to impose a discretized version of the ingoing - wave boundary condition a bit away from the horizon in order to better mimic the continuum solutions .
in particular , we impose the frequency - dependent boundary condition of ref .
@xcite : @xmath120 the shift into the bulk , given by @xmath121 , is chosen to be @xmath122 and @xmath123 for @xmath124 and @xmath125 , respectively . in fig .
[ fig : dec ] we plot the condensate and the real part of the conductivity for @xmath126 .
it was suggested in ref .
@xcite that the large resonances in the conductivity may correspond to exciton - polariton resonances due to the broken u(1 ) gauge groups in the model .
the @xmath127 and @xmath79 cases are qualitatively similar , and some examples are given in fig .
[ fig : gap ] .
the critcal temperatures at which the condensate starts to form are listed in table [ tab : tc ] .
& & & @xmath126 & & @xmath128 & & @xmath129 & & @xmath130 & & @xmath131 & + & @xmath0 & & @xmath132 & & @xmath133 & & @xmath128 & & @xmath134 & & @xmath135 & + & & & @xmath79 & & @xmath136 & & @xmath137 & & @xmath138 & & @xmath139 & + [ tab : tc ] to further analyze observables in the deconstructed models we mimic the analysis of the continuum model .
it can be seen directly from the locations of the minimum of im@xmath85 in fig .
[ fig : gap ] that @xmath140 , but its value is not sensitive to @xmath77 in the range examined , even in the 5-site model .
the relation ( [ eq : gap ] ) continues to be well satisfied and defines the gap @xmath54 as in the continuum model .
the pole in the imaginary part of the conductivity is manifest in fig .
[ fig : gap ] , and is related to the delta - function in the real part via a kramers - kronig relation . the results for observables are listed in table iii .
& & & @xmath126 & & @xmath141 & & @xmath142 & & @xmath143 & & @xmath144 & + & @xmath97 & & @xmath132 & & @xmath145 & & @xmath146 & & @xmath147 & & @xmath148 & + & & & @xmath79 & & @xmath149 & & @xmath150 & & @xmath151 & & @xmath152 & + & & & @xmath153 & & @xmath154 & & @xmath154 & & @xmath155 & & @xmath156 & + & @xmath157 & & @xmath132 & & @xmath158 & & @xmath158 & & @xmath159 & & @xmath160 & + & & & @xmath79 & & @xmath161 & & @xmath161 & & @xmath162 & & @xmath163 & + & & & @xmath153 & & @xmath164 & & @xmath165 & & @xmath166 & & @xmath167 & + & @xmath168 & & @xmath132 & & @xmath169 & & @xmath170 & & @xmath171 & & @xmath172 & + & & & @xmath79 & & @xmath173 & & @xmath174 & & @xmath175 & & @xmath176 & + [ tab : dgap ]
we have analyzed the dependence of the charged condensate and the complex conductivity on the form of the black - hole metric in holographic superconductors and in deconstructed versions of those models .
we found that certain model predictions are relatively insensitive to the details of the spacetime .
for example , the approximate relation between the gap frequency and the superconducting condensate , @xmath177 persists while the metric is varied in the continuum model . in the deconstructed model this ratio differs from 1 , but remains insensitive to the deconstructed metric . on the other hand ,
we have seen relatively strong dependence of other observables on the details of the metric , such as the ratio of the superconducting gap @xmath54 to @xmath0 .
furthermore , in deconstructed models we found that this ratio can be significantly smaller than in the continuum model .
the model sensitivity supports the conclusion that the quantitative success of the simplest holographic models of superconductors is accidental .
however , qualitative features of these and related models continue to suggest the possibility of explaining some of the unusual properties of unconventional superconductors . for example , in addition to a large gap , related models describe a strange metallic phase @xcite in which the resistivity does not vary like @xmath178 as in the fermi liquid description of metals . in an effort to make contact with physical systems , it remains important to continue to investigate which aspects of the holographic models and their deconstructed cousins are responsible for the nonconventional behavior of the superconductors described by these models .
we are grateful to patrick king for collaboration at early stages of this work .
we also thank dylan albrecht and chris carone for useful communication .
this work was supported by the nsf under grant phy-1068008 .
je also thanks edward coco for his generous support .
g. f. de teramond and s. j. brodsky , phys .
lett . * 94 * , 201601 ( 2005 ) [ hep - th/0501022 ] ; j. erlich , e. katz , d. t. son and m. a. stephanov , phys .
lett . * 95 * , 261602 ( 2005 ) [ hep - ph/0501128 ] ; j. erlich , int .
j. mod .
a * 25 * , 411 ( 2010 ) [ arxiv:0908.0312 [ hep - ph ] ] .
t. sakai and s. sugimoto , prog .
phys . * 113 * , 843 ( 2005 ) [ hep - th/0412141 ] ; t. sakai and s. sugimoto , prog .
phys . * 114 * , 1083 ( 2005 ) [ hep - th/0507073 ] . c. csaki , c. grojean , h. murayama , l. pilo and j. terning , phys .
d * 69 * , 055006 ( 2004 ) [ hep - ph/0305237 ] ; j. hirn and v. sanz , phys . rev .
lett . * 97 * , 121803 ( 2006 ) [ hep - ph/0606086 ] ; m. piai , hep - ph/0609104 ; c. d. carone , j. erlich and j. a. tan , phys . rev .
d * 75 * , 075005 ( 2007 ) [ hep - ph/0612242 ] .
s. s. gubser , phys .
d * 78 * , 065034 ( 2008 ) [ arxiv:0801.2977 [ hep - th ] ] .
s. s. gubser , c. p. herzog , s. s. pufu and t. tesileanu , phys .
lett . * 103 * , 141601 ( 2009 ) [ arxiv:0907.3510 [ hep - th ] ] .
s. a. hartnoll , c. p. herzog and g. t. horowitz , phys .
* 101 * , 031601 ( 2008 ) [ arxiv:0803.3295 [ hep - th ] ] ; s. a. hartnoll , c. p. herzog and g. t. horowitz , jhep * 0812 * , 015 ( 2008 ) [ arxiv:0810.1563 [ hep - th ] ] .
j. erlich and c. westenberger , phys .
d * 79 * , 066014 ( 2009 ) [ arxiv:0812.5105 [ hep - ph ] ] .
s. j. brodsky and r. shrock , phys .
b * 666 * , 95 ( 2008 ) [ arxiv:0806.1535 [ hep - th ] ] ; g. f. de teramond and s. j. brodsky , phys .
* 102 * , 081601 ( 2009 ) [ arxiv:0809.4899 [ hep - ph ] ] .
a. l. fitzpatrick , j. kaplan , e. katz and l. randall , arxiv:1304.3458 [ hep - th ] . c. t. hill , s. pokorski and j. wang , phys .
d * 64 * , 105005 ( 2001 ) [ hep - th/0104035 ] .
n. arkani - hamed , a. g. cohen and h. georgi , phys .
* 86 * , 4757 ( 2001 ) [ hep - th/0104005 ] .
n. arkani - hamed , a. g. cohen and h. georgi , phys .
b * 513 * , 232 ( 2001 ) [ hep - ph/0105239 ] .
d. albrecht , c. d. carone and j. erlich , phys .
d * 86 * , 086005 ( 2012 ) [ arxiv:1208.2700 [ hep - th ] ] .
e. witten , adv .
* 2 * , 505 ( 1998 ) [ hep - th/9803131 ] .
d. t. son and a. o. starinets , jhep * 0209 * , 042 ( 2002 ) [ hep - th/0205051 ] .
e. kiritsis and v. niarchos , jhep * 1107 * , 112 ( 2011 ) [ erratum - ibid .
* 1110 * , 095 ( 2011 ) ] [ arxiv:1105.6100 [ hep - th ] ]
. s. a. hartnoll , j. polchinski , e. silverstein and d. tong , jhep * 1004 * , 120 ( 2010 ) [ arxiv:0912.1061 [ hep - th ] ] . | holographic models of superconductors successfully reproduce certain experimental features of high - temperature superconductors , such as a large gap - to-@xmath0 ratio compared to that of conventional superconductors . by deconstructing the extra dimension of these holographic models ,
similar phenomenology is described by a class of models defined in the natural dimension of the superconducting system .
we analyze the sensitivity of certain observables in holographic and deconstructed holographic superconductors to details of the extra - dimensional spacetime .
our results support the notion that certain quantitative successes of simple models of this type are accidental .
however , we also find a certain universal relationship between superconducting observables . |
the interaction of driven particles , flexible lines and membranes etc . with disorder is an important topic in condensed matter physics @xcite .
usually , this disorder is taken to be _ quenched _ , or frozen , such that its properties do not change within the relevant time scales . however , under certain conditions , this changes as in the case of the diffusion of solute atoms in metallic alloys @xcite or oxygen vacancies in superconductors @xcite .
the mobile impurities play an important role in the dynamics of such systems , as evidenced for example by the portevin - le chatelier ( plc ) effect in solid solutions @xcite .
there , within a certain range of temperatures and applied strain rates , the dynamic interaction of lattice dislocations and diffusing solute atoms result in phenomena such as negative strain rate sensitivity of the flow stress , giving rise to macroscopic serrations in the stress strain curve and strain localization in the form of bands of activity of various types @xcite . here , we consider the simple test problem of a single particle interacting with a cloud of diffusing impurities , with the dynamics constrained in one dimension ( a line ) . we restrict ourselves to the region of the parameter space
in which the impurities have a vanishingly small probability to escape from the vicinity of the particle . despite its apparent simplicity ,
such a system exhibits rich dynamics , but at the same time has features that make the problem analytically tractable . in the absence of external forces , we consider the statistics of _ monotonic excursions _ @xmath0 of the particle , i.e. the distances the particle moves to a particular direction ( here `` left '' or `` right '' along the one - dimensional line ) without changing direction , see also figure [ fig : trajectories ] .
we find that these obey power law distributions @xmath4 with the exponent @xmath5 .
the same is true for the monotonic changes of the external force @xmath2 when the particle is driven with a slow constant velocity .
this paper is organized as follows : in the next section , we consider the interaction of a particle with a single mobile impurity , in the absence of external forces .
then we generalize this to the case with more impurity particles . in section
[ driven ] , the effect of external drive is studied .
finally , section [ concl ] finishes the paper with conclusions .
as a starting point of our analysis , we consider the dynamics of a single particle interacting with one diffusing impurity particle .
the equations of motion for the system are @xmath6 where @xmath7 and @xmath8 are the positions of the particle and the impurity particle , respectively .
@xmath9 is the interaction force between the particle and the impurity particle , @xmath10 defines the relative mobilities of the impurity and the particle and @xmath11 is gaussian white noise with standard deviation @xmath12 mimicking the effect of temperature .
the dynamics of the particle can be analyzed by considering the stochastic process for the velocity @xmath13 . by differentiating the equation of motion of the particle with respect to time and using the equation of motion for @xmath8 , one obtains @xmath14.\ ] ] close to @xmath15
, the force @xmath9 can be taken to be linear in @xmath16 and thus the derivative of the force can be approximated by a constant , @xmath17 , with @xmath18 . with @xmath19 and @xmath20 , equation ( [ eq : x1 ] ) can then be rewritten in the form of an ornstein - uhlenbeck process for @xmath13 , @xmath21 for the process @xmath13 , equation ( [ eq : x2 ] ) describes brownian motion pushed toward the origin by a linear damping term .
this problem has been considered e.g. in @xcite , and the scaling exponents are known . in particular , the first return times @xmath22 to origin of @xmath13 have a probability distribution scaling as @xmath23 with @xmath24 and the cut - off scale @xmath25 .
similarly , the average shape of an excursion , @xmath26 , scales for @xmath27 as @xmath28 , with @xmath29 .
then , by using the scaling relation @xmath30 @xcite , one obtains the distribution of the lengths @xmath31 of monotonic excursions of the particle , @xmath32 with @xmath1
. the cut - off scaling can be found as follows : the cut - off of the first return time distribution of equation ( [ eq : x2 ] ) is given by @xmath25 @xcite . here ,
both @xmath33 and @xmath34 in equation ( [ eq : x2 ] ) depend on @xmath35 and @xmath10 . due to the relation @xmath36 with @xmath29 , @xmath37
is expected to scale like @xmath38 we check this result in numerical simulations , where we take for simplicity @xmath9 to be of the form @xmath39 , corresponding to @xmath40 , i.e. @xmath41 .
we integrate the equations of motion ( [ eq : motion ] ) with the euler algorithm .
the strength of the thermal noise was chosen to be sufficiently weak such that the impurity can not escape from the neighborhood of the particle .
figure [ fig : trajectories ] shows an example of the trajectories of the particle and the impurity .
figure [ fig : onesolute ] displays the distribution of @xmath0 for various values of @xmath42 . for a weak enough interaction strength a , the distributions display the expected scaling with @xmath1 .
the cut - off is found to scale as @xmath43 , in agreement with our results above .
this implies that the motion of the particle is reminiscent of _ truncated _ levy flight , with a step length distribution given by equation ( [ eq : dist ] ) .
however , the steps are not instantaneous - their durations @xmath22 exhibit power law scaling as well , equation ( [ eq : pt ] ) . by interpreting these step durations as waiting times between instantaneous steps
, one would obtain for early times the scaling @xmath44 . however , here @xmath0 and @xmath22 are not independent ( due to the relation @xmath45 ) . as the early time behavior
is dominated by a single large step , one must consider instead the effect of a single step given its duration , @xmath46 ^ 2 p(t ) dt = \int_0^t t^{2\gamma-\tau_t } dt \sim t^{2\gamma-\tau_t+1},\ ] ] corresponding to @xmath47 . due to the truncated nature of the levy flight ,
one expects a cross - over to diffusive behavior with @xmath48 for long times .
this is verified in figure [ fig : diffusion ] .
the case of a fixed number @xmath49 of impurity particles is a straightforward generalization of that presented in the previous subsection .
the equations of motion become @xmath50 from these equations , with the same procedure as above , one obtains an equation of the form of equation ( [ eq : x2 ] ) by setting @xmath51 and @xmath52 .
thus , the same scaling , i.e. @xmath53 is expected .
one should notice , however , that the cut - off scale @xmath37 is getting smaller with increasing @xmath54 . with similar arguments as above
, one finds that @xmath55 this is verified by the numerical results presented in figure [ fig : manysolutes ] .
next we proceed to study the effect of a weak external force @xmath56 on the dynamics of the particle . in this context
a _ constant velocity _ drive is perhaps the more interesting form of driving as a small _ constant force _ with @xmath57 does not change the dynamics from the non - driven case : equation ( [ eq : x1 ] ) remains the same even if a constant force term is introduced in equation ( [ eq : motion ] ) .
in particular , we consider a particle driven by a force given by @xmath58 , where @xmath59 is the driving velocity and @xmath60 is a spring constant characterizing the response of the driving mechanism .
the equations of motion read @xmath61 in systems like this driven with a constant velocity , the interesting quantity is the statistics of the external force fluctuations . to this end , we consider the stochastic process @xmath62 . with a similar approach as above ,
one can write @xmath63\partial_t f + \frac{kc}{\mu}\sum_i \eta_i \nonumber \\ & & + \frac{kc}{\mu}\left[v(n+\mu)-f\right],\end{aligned}\ ] ] where the relation @xmath64 has been used . in the steady state , the last term in equation ( [ eq : f1 ] ) has a zero mean , as one can write for the average steady state force @xmath65 , where @xmath66 is the magnitude of the average retarding force acting on the particle due to a single impurity . in the steady state
the condition @xmath67 holds , and thus @xmath68 . assuming that the fluctuations @xmath69 = \delta f$ ]
are small compared to those of the white noise term in equation ( [ eq : f1 ] ) , i.e. @xmath70 , equation ( [ eq : f1 ] ) can be approximately written as @xmath71\partial_t f + \frac{kc}{\mu}\sum_i \eta_i,\ ] ] which is again of the same form as equation ( [ eq : x2 ] ) .
thus , the monotonic changes of the external force @xmath72 are expected to be distributed according to a power law @xmath73 , with the exponent @xmath74 and the cut - off scale @xmath75 scaling as @xmath76^{3/2}}.\ ] ] notice that the condition @xmath70 implies that @xmath77^{3/2 } } \ll 1.\ ] ] for most of the relevant parameter values condition ( [ eq : cond ] ) is fulfilled , only for @xmath78 this is not the case .
+ next we check these predictions numerically for different values of the parameters satisfying the condition ( [ eq : cond ] ) .
figure [ fig : force ] shows an example of the behaviour of the force as a function of time . after an initial transient ,
the system reaches a steady state in which the force fluctuates around a constant average value . in figure
[ fig : forcedists ] , we show the distributions of the monotonic changes of the external force @xmath2 in the steady state , for different values of the various parameters .
power law scaling of the distributions consistent with the exponent value @xmath74 is observed , with a cut - off of the distributions in agreement with equation ( [ eq : cutoff ] ) .
the different @xmath2-values appear to be uncorrelated in time .
in this paper we have studied the dynamics of a single particle interacting with a cloud of diffusing impurities . in the absence of external forces
, the problem can be mapped to brownian motion in a potential within the harmonic approximation of the attractive particle - impurity interaction .
thus , the monotonic excursions of the particle are distributed as a power law , @xmath79 , with @xmath1 . for a particle driven with a small constant velocity ( such that the particle is dragging the impurity cloud without escaping from it )
, the external force fluctuations follow the same dynamics , which makes it possible to derive the probability distribution for the monotonic changes of the external force scaling as @xmath80 , again with @xmath74 . while a typical experimentally relevant scenario would correspond to either a higher dimensional object such as a flexible line or a large number of interacting particles interacting with mobile impurities , the simple setup of the present study serves as a convenient starting point for such considerations illustrating the relevant phenomena in a transparent manner .
one interesting observation is that already at the level of a single particle interacting with one or more mobile impurities the dynamics has scale free features , arising from the properties of simple random walks .
physical situations in which these kind of considerations could be relevant include dislocations interacting with solute atoms , a subject that has recently attracted considerable attention @xcite . in many of these studies ,
more realistic interaction forces between dislocations and solute atoms have been used , but the focus has been on different quantities such as the _ average _ velocity of the dislocation .
the most intriguing phenomena associated with the presence of diffusing impurities are the collective effects arising from the simultaneous interaction of large number of entities with each other and with mobile impurities .
an example of such a system is provided by interacting dislocation ensembles interacting with diffusing solute atoms in solid solutions , giving rise to phenomena such as the portevin - le chatelier effect . in the plc effect ,
large numbers of dislocations synchronize their motion to form macroscopic deformation bands of various kinds . as this
is widely believed to be due to the dynamic interaction of the dislocations with the diffusing solute atoms , a natural future line of research would be to study such effects by considering numerically the dynamics of a large number of interacting dislocations interacting with diffusing impurities .
one further motivation for such studies could be the recent observation that purely stochastic effects can induce switching between collective motion states @xcite .
ll and mja gratefully thank the financial support of the european commissions nest pathfinder programme trigs under contract nest-2005-path - com-043386 .
they also acknowledges the financial support from the center of excellence program of the academy of finland .
10 nattermann t 1983 _ phys .
status solidi _ b * 119 * 209 huse d a and henley c l 1985 _ phys .
lett . _ * 54 * 2708 kardar m and zhang y c 1987 _ phys .
lett . _ * 58 * 2087 cahn j w 1962 _ acta metall . _
* 789 blavette d , cadel e , fraczkiewicz a , and menand a 1999 _ science _ * 286 * 2317 chudnovsky e m 1998 _ europhys . lett . _ * 43 * 445 portevin a and le chatelier f 1923 _ c. r. acad .
paris _ * 176 * 507 ananthakrishna g , noronha s j , fressengeas c , and kubin l p 1999 _ phys . rev . _ e * 60 * 5455 lebyodkin m a , brechet y , estrin y , and kubin l p 1995 _ phys .
lett . _ * 74 * 4758 hhner p , ziegenbein a , rizzi e , and neuhuser h 2002 _ phys .
_ b * 65 * , 134109 colaiori f , baldassaarri a , and castellano c 2004 _ phys .
_ e * 69 * 041105 lubeck s 2004 _ int j. mod .
_ b * 18 * 3977 wang y , srolovitz d j , rickman j m , and lesar r 2000 _ acta mater . _ * 48 * 2163 rickman j m , lesar r , and srolovitz d j 2003 _ acta mater . _ * 51 * 1199 deo c s , srolovitz d j , cai w , and bulatov v 2005 _ phys .
_ b * 71 * 014106 kolpas a , moehlis j , and kevrekidis i g 2007 _ pnas _ * 104 * 5931 | the dynamics of a test particle interacting with diffusing impurities in one dimension is investigated analytically and numerically . in the absence of an applied external force
, the dynamics of the particle can be characterized by a distribution of monotonic excursions @xmath0 , which scales as a power law with an exponent @xmath1 . when the particle is driven at a slow constant velocity
, there is again a power law distribution for the monotonic changes of the force @xmath2 , which is characterized by a similar exponent @xmath3 .
these results can be understood from the theory of random walks . |
the general problem that this article is about is the following : consider a nonlinear equation system and an outer iteration method to solve it that consists of solving a subproblem at each step using a second , inner iteration method .
now we want to answer the following question : how can we efficiently control the iteration error of the outer iteration method ? or otherwise put : how accurate do we need to solve the inner systems to obtain a certain iteration error for the outer nonlinear equation ? for the case of the outer iteration being newton s method , this problem has been successfully solved . the inner problem is a linear system and the concept of an inexact newton s method was introduced in @xcite . there
, at each newton step the inner iteration is terminated when a relative tolerance criterion in the linear residual is satisfied .
based on this , it is possible to give conditions on the sequence of relative tolerances to obtain linear , superlinear or quadratic convergence of the inexact newton s method . essentially , the sequence of tolerances has to converge to zero fast enough as the newton scheme progresses and then quadratic convergence is obtained .
following up , a strategy that has this property and leads to a very efficient scheme was suggested in @xcite .
there , the point is that the initial systems are solved quite coarsely and these schemes are part of widely used software packages , for example of sundials @xcite . note that with this knowledge , the choice of iterative solver for the inner iteration obtains a better basis : if most of the systems are solved very coarsely and thus very few iterations are needed , it is more important that the method is cheap per iteration than how fast we can reach machine accuracy . in this
setting , when looking at unsymmetric linear systems and krylov subspace methods , gmres @xcite beats bicgstab @xcite , since it needs only one matrix vector product per iteration instead of two .
now when looking at fixed point iterations , an iteration typically consists of evaluating a function and not of solving a system .
however , two prominent and important examples where this happens are the picard iteration and fluid - structure interaction and thus we call these inexact fixed point schemes . surprisingly , the problem framed in the first paragraph has not been analyzed for these .
for the picard iteration , which is a common tool in the context of the incompressible navier - stokes equations , the evaluation of the right hand side corresponds to solving a linear system .
strategies for choosing a termination criterion for the inner iteration are empirically discussed for example in @xcite . in fluid - structure interaction , a standard approach
are partitioned coupling schemes , where existing solvers for the subproblems are reused @xcite .
commonly in the form of a dirichlet - neumann iteration , this consists of subsequently solving the fluid and the structure problem with appropriate boundary conditions and reasonable tolerances .
it is common to formulate the coupling condition at the interface in the form of a fixed point equation .
recently , it was suggested to use a time adaptive implicit time integration scheme for fluid structure interaction @xcite , where the time step is chosen based on an error tolerance . as is common in this setting ,
the tolerances for the solvers for the appearing nonlinear equation systems are chosen such that the iteration error does not interfere with the error from the time integration scheme @xcite , but nevertheless as large as possible to avoid unnecessary computations .
thus , it is imperative to be able to control the iteration error . to solve our problem
, we proceed in the following way .
first , we will review well known results on perturbations of fixed point schemes @xcite .
the general idea is then to quantify the iteration errors based on the termination criterion of the inner iteration such that the existing perturbation results can be applied . for the dirichlet - neumann iteration
, we first have to extend the perturbation results to the case of a nested fixed point equation of the form @xmath0 .
as it turns out , the type of termination criterion chosen in the inner solver is crucial for the answer to our problem in that when using a nonstandard relative criterion , we obtain convergence of the fixed point iteration to the exact solution independently of how accurate we solve the inner systems , leading to an efficient way of controlling the outer iteration error . on the other hand , no general statement can be made on a standard relative or absolute termination criterion , but the analysis suggests that these do not have favorable properties .
all of this is confirmed by numerical results , which show that the latter criteria cause convergence to a solution that is farther away from the exact one the less accurate we solve the linear systems or otherwise put , the less error we want in the fixed point equation , the more accurate we have to solve the linear systems .
consider the fixed point equation @xmath2 with @xmath3 , @xmath4 closed and where we assume that @xmath5 is lipschitz continuous with lipschitz constant @xmath6 .
this implies by the banach fixed point theorem that has a unique solution @xmath7 .
furthermore , we consider the perturbed fixed point iteration @xmath8 where @xmath9 is a perturbation that could originate from an iterative solver and for simplicities sake we denote the norm of @xmath9 by @xmath9 as well .
we furthermore assume for simplicities sake that @xmath10 is also a self - map on @xmath4 .
thus this iteration obtains a solution @xmath11 of the perturbed fixed point equation @xmath12 the question is now : how far is the solution @xmath11 of that equation away from @xmath7 ?
the answer is giving by the following theorem , see for example @xcite .
[ ortrhetheorem ] for the solutions @xmath11 and @xmath7 of problems and we have :
@xmath13 this means that the error is of the order @xmath9 as is to be expected , but interestingly , it becomes larger , the closer the lipschitz constant of @xmath14 is to one or otherwise put , the less contractive the function is .
this implies that in these cases , the error will be much larger than @xmath9 and thus a much smaller tolerance would have to be supplied to acchieve the desired error .
note that in practice , we typically do not know the lipschitz constant and that in the nonlinear case , it depends on the definition of the domain @xmath4 .
thus the important lipschitz constant is the local one in the solution . if we instead consider a sequence of perturbations @xmath15 , respectively a nonconstant perturbation , and thus the iteration @xmath16 the first question is when this sequence converges to @xmath7 .
the answer is given by the next theorem , also from @xcite : [ perturbedtheorem ] the iteration converges to the solution of the unperturbed problem if and only if @xmath17 .
a specific case is @xmath18 with @xmath19 , which we call the adaptive strategy . as an application of the above theorems
, we now analyze the convergence of the picard iteration .
this is often employed in the context of the incompressible navier - stokes equation and corresponds to a fixed point iteration for the equation @xmath20 where @xmath21 is an approximation of a jacobian in @xmath22 @xcite .
thus , the fixed point iteration @xmath23 is implemented by solving @xmath24 for @xmath25 up to a certain tolerance using an iterative scheme .
the scheme can be analyzed either as a fixed point scheme , which results in linear convergence provided that the lipschitz constant @xmath26 of @xmath27 can be bounded from above away from one or as a method of newton type where @xmath28 is an approximation of the exact jacobian and we have linear convergence as long as this approximation is good enough . when solving , either the relative termination criterion @xmath29 the relative termination criterion @xmath30 or the absolute criterion @xmath31 are used , where @xmath32 and @xmath33 are relative and adaptive tolerances .
to analyze the consequences of choosing one of these using theorems 1 and 2 , we need to quantify the perturbation in the form .
thus , we define @xmath34 to obtain @xmath35 and we can write @xmath36 in the case of the relative termination criterion , we can estimate the norm of the right hand side in by @xmath37 we furthermore have @xmath38 thus , @xmath39 all in all , we obtain with the condition number @xmath40 @xmath41 with the additional and reasonable assumption that @xmath42 is bounded , this is a perturbation of the form and @xmath15 converges to zero independent of the choice of @xmath32 !
thus by theorem [ perturbedtheorem ] this iteration converges to the exact solution independently of how accurate we solve the linear equation systems .
we now formulate this as a theorem .
[ picardtheorem ] let @xmath43 and the function @xmath28 be given that maps the closed set @xmath44 onto quadratic regular matrices .
assume that the function @xmath45 is lipschitz continuous with lipschitz constant @xmath6 and correspondingly has a unique fixpoint @xmath7 .
furthermore assume that @xmath46 is bounded on @xmath4 and that the inexact fixedpoint iteration defined by converges to a limit @xmath11 .
then @xmath47 , independent of the choice of @xmath32 .
in case of the relative criterion , the estimate @xmath48 for the norm of the left hand side in holds which is bounded away from zero provided that @xmath49 is .
thus , it is not clear if this iteration satisfies theorem [ perturbedtheorem ] , but in the general case , the iteration will not converge to @xmath7 .
similarly if we use the absolute termination criterion , we obtain @xmath50 which is also bounded away from zero if @xmath49 is .
numerical results that confirm theorem [ picardtheorem ] and demonstrate that the other two iterations behave like being of the form can be found in section [ picardresultssection ] .
we would like to point out that the criterion is sometimes suggested in the literature on the picard iteration , e.g. @xcite , but that an absolute termination criterion is suggested in @xcite .
there it is suggested to just `` gain one digit '' , meaning to use a tolerance of 0.1 .
now consider two functions @xmath51 and @xmath52 with @xmath53 closed and the fixed point equation @xmath54 again with solution @xmath7 .
we now consider an iteration where both the evaluation of @xmath55 and of @xmath56 are perturbed , namely @xmath56 is perturbed by @xmath57 and @xmath55 by @xmath15 : @xmath58 again , assume that this iteration is well defined and that this sequence has the limit @xmath11 .
then , we obtain the following theorem .
[ fsitheorem ] let @xmath55 and @xmath56 be lipschitz continuous with lipschitz constants @xmath59 and @xmath60 , respectively .
assume that @xmath61 .
then we have , if @xmath62 for all @xmath63 , that @xmath64 in the case @xmath65 and @xmath66 , we obtain @xmath67 finally , @xmath47 if and only if both @xmath57 and @xmath15 converge to zero . proof : the proof is technically identical to the one of theorem [ ortrhetheorem ] .
we have due to the lipschitz continuity @xmath68 and thus in the limit @xmath69 , @xmath70 for a constant perturbation overall , e.g. @xmath62 for all @xmath63 , we obtain in the limit @xmath71 which proves the inequality .
if we have constant but separate perturbations @xmath9 and @xmath72 of @xmath56 and @xmath55 , we obtain from @xmath73 in the general case , due to positivity , the right hand side of is zero if and only if both @xmath15 and @xmath57 are such that for @xmath74 or @xmath75 , @xmath76 by an identical proof to theorem 2 , this is the case if and only if both @xmath15 and @xmath57 converge to zero . note
that this implies that the sequence @xmath15 perturbing the inner function @xmath55 is less important by a factor of the lipschitz constant @xmath60 of the outer function .
thus , a possible strategy is to define @xmath77 meaning that we solve the fluid part less accurate by a factor of @xmath60 .
unfortunately , @xmath60 has to be known for this . as an application of the theory from section 2.3
, we consider a problem that is a basic building block in fluid structure interaction , namely the transmission problem , where the laplace equation with right hand side @xmath78 on a domain @xmath4 is cut into two domains @xmath79 using transmission conditions at the interface @xmath80 : @xmath81 we now employ a standard dirichlet - neumann iteration to solve it .
using any linear discretization , this corresponds to alternately solving the problems @xmath82 and @xmath83 were problem corresponds to a discretization of the transmission problem on @xmath84 only with dirichlet data on @xmath85 given by @xmath86 on the coupling interface and problem corresponds to a discretization of on @xmath87 only with neumann data on @xmath85 given by the discrete normal derivative of @xmath88 on @xmath85 .
it can be shown that convergence of the approximate solutions on the whole domain is equivalent to the convergence of the solution on the interface only @xcite . by considering - as one iteration , we obtain a fixed point formulation @xmath89 where @xmath90 is @xmath91 on the interface , @xmath92 and @xmath93 . hereby @xmath94 is the matrix that computes the discrete normal derivatives in @xmath84 on @xmath85 and @xmath95 is the discrete trace operator with respect to @xmath85 .
otherwise put , @xmath95 is the projection of the space that @xmath91 is in onto the space of discrete unknowns on @xmath85 . in practice ,
the linear equation systems are solved iteratively , typically using the conjugate gradient method ( cg ) up to a relative tolerance of @xmath96 .
thus , we obtain a perturbed nested fixed point iteration of the form and the question is now again if we can quantify this perturbation .
we have @xmath97 and @xmath98 for the iteration we obtain @xmath99 again , the second factor is what is tested in the termination criterion of cg . in the case of the relative criterion , here
stated as @xmath100 we obtain @xmath101 @xmath102 now the point is that since the whole iteration is linear , we can write down a linear mapping that maps @xmath103 onto @xmath104 for arbitrary @xmath63 .
let this have lipschitz constant @xmath105 , then we have @xmath106 thus the perturbation has limit zero if @xmath107 .
this is the case if and only if the sequence @xmath108 is convergent , which is in fact the case provided that @xmath78 is sufficiently harmless , as can be seen from the literature on domain decomposition methods , e.g. ( * ? ? ?
analagously to the picard iteration , if we choose the absolute termination criterion or the relative one based on the right hand side , we obtain a bound of the form @xmath109 respectively @xmath110 again , we can not make a statement on the limit of @xmath15 . in the second case , meaning the iteration with neumann data , we obtain @xmath111 and analogous arguments produce the same results for @xmath57 . thus by theorem [ fsitheorem ] , we have that when using the relative criterion we obtain convergence to the exact solution for any @xmath32 .
it is important to note that under the assumptions , all sequences considered , wether perturbed or not , are convergent and therefore , the fixed point iteration will terminate when using the standard criterion @xmath112 however , as just shown , the perturbed iteration converges to an approximation of the unperturbed fixed point and thus , the algorithm can terminate when we are in fact not @xmath113-close to the solution .
a further difference between the different iterations that should be stressed is that the iterations perturbed by a constant are fixed point iterations , wheras the schemes with a variable perturbation are in fact , not .
thus , the convergence speed , which is otherwise linear with constant @xmath26 , is not clear and numerical evidence suggests that it is in fact slower than for the other iteration .
thus , we could argue to employ the schemes with constant perturbation , measure the lipschitz constant numerically after a few iterations and then adjust the tolerance based on theorem [ ortrhetheorem ] or [ fsitheorem ] .
unfortunately , it is not clear what the @xmath9 from these theorems is , respectively , it is based on quantities that are hard to measure like @xmath114 . thus , we can not guarantee a certain iteration error in this way , to do this we must employ the nonstandard relative termination criterion . finally , it is important to note that this analysis is mostly relevant to the time independent case . otherwise , when considering this inside an implicit time integration scheme , additional requirements on the solutions appear , namely that the solutions in the subdomains have a certain accuracy .
for all numerical experiments , the fixed point iteration is terminated when the norm @xmath115 is smaller than @xmath116 . furthermore , with the exception of the results on the picard iteration , all computations were performed in matlab , where matlab 2012a was used for all computations with the exception of the results in section 3.2.3 , where matlab 2013a was employed .
.[problem2xepsminusxstar]@xmath117 ( left ) and @xmath118 ( right ) for different values of @xmath9 and @xmath26 for the solution of the scalar nonlinear equation [ cols="^,^,^,^",options="header " , ] the results are depicted in table [ fsiproblemefficiency ] . for the computation with a - divergence
was observed .
otherwise , the schemes roughly obey the desired behavior that the error is proportional to tol , although all are above the desired accuracy .
otherwise , the scheme corresponding to is about a factor of five faster than that corresponding to and up to a factor of two faster than that corresponding to while providing more accurate results .
we considered perturbed fixed point iterations where the perturbation results from inexact solves of equation systems by iterative solvers .
thereby , we extended a perturbation result for fixed point equations to the case of a nested fixed point equation . applying these results to the picard- and the dirichlet - neumann iteration for steady states
, we showed that these converge to the exact solution indepently of the tolerance in the subsolver , if a specific relative termination criterion is employed .
this justifies extremely coarse solves in the inner solvers and suggests the use of gmres as krylov subspace solver for unsymmetric systems .
if an absolute or standard relative criterion is used , the theory indicates that we will not converge to the exact solution .
numerical results demonstrate this behavior .
thus , to obtain a certain accuracy in the fixed point solution when using a standard relative or absolute criterion , we have to solve the inner systems more accurate the tighter the tolerance , whereas for the nonstandard relative criterion we can solve the inner systems very coarsely independent of desired accuracy .
numerical results show that this is the most efficient way to treat these systems .
part of this work was funded by the german research foundation ( dfg ) as part of the collaborative research area sfb trr 30 , project c2 . furthermore
, i d like to thank gunar matthies for performing the numerical experiments on the picard iteration in section 3.1.2 .
, _ cfd - based nonlinear computational aeroelasticity _ , in encyclopedia of computational mechanics , e. stein , r. de borst , and t. j. r. hughes , eds .
3 : fluids , john wiley & sons , 2004 , ch . 13 , pp . 459480 . | we analyze inexact fixed point iterations where the generating function contains an inexact solve of an equation system to answer the question of how tolerances for the inner solves influence the iteration error of the outer fixed point iteration .
important applications are the picard iteration and partitioned fluid structure interaction .
we prove that the iteration converges irrespective of how accurate the inner systems are solved , provided that a specific relative termination criterion is employed , whereas standard relative and absolute criteria do not have this property . for the analysis ,
the iteration is modelled as a perturbed fixed point iteration and existing analysis is extended to the nested case @xmath0 .
= 0.9 _ @xmath1 numerical analysis , centre for the mathematical sciences , lund university , box 118 , 22100 lund , sweden + institute of mathematics , university of kassel , heinrich - plett - str .
40 , 34132 kassel , germany + department of mathematics / computer science , university of osnabrck , albrechtstr .
28a , 49076 osnabrck , germany + email : philipp.birkenna.lu..de_ _ keywords : fixed point iteration , picard iteration , transmission problem , dirichlet - neumann iteration , termination criteria _ |
most research on fault tolerant control sits within the context of large manned aircraft . with regards to unmanned aerial vehicles ( uavs )
, the majority of the literature describes the application of fault tolerant control ( ftc ) to rotorcraft rather than fixed wing aircraft . in this paper
we develop a nonlinear model predictive control ( nmpc ) based controller for a fixed wing aircraft model suitable for the purposes of fault
tolerant flight control .
the controller is first tested and analysed on a lower order 2d robot model . then assuming that fault information is available we successfully demonstrate for the first time the use of an nmpc based fault tolerant flight control system for a fixed wing aircraft .
+ the main characteristic of a fault tolerant control ( ftc ) system is its ability to automatically cope with system faults before they turn into a serious system failure .
the integration of an ftc scheme significantly increases the ability of the system to maintain overall stability in the presence of a fault @xcite .
model predictive control , a highly promising approach to re - configurable and fault tolerant control ( @xcite , @xcite , @xcite and @xcite ) , focuses on _ what to control _ , instead of _ how to control_. this is a subtle , but illuminating , difference allowing for the exploitation of inherent system characteristics such as nonlinearities and cross - coupling effects by the controller , rather than their influence being minimised .
this is possible due to the underlying daisy chaining capability of mpc @xcite .
for example @xcite , the primary function of the rudder of an aircraft is to provide yaw or sideways control .
however , the rudder can also have some effect on the roll of the aircraft . therefore
, in the event of the failure of an aileron actuator , the primary control surface for roll , it is still possible to execute a limited roll manoeuvre with the rudder .
this degree of fault - tolerance in the flight control system requires a suitable re - configurable architecture to be purposefully designed and implemented @xcite , which is able to re - establish control , albeit with limited capacity , and execute the required manoeuvres .
the mission can then either be continued with the failed component or aborted , the primary objective being to avert a catastrophic failure or the loss of the aircraft , and to ensure that it is brought back to ground safely .
the aim of this paper is to design a controller for exactly this purpose .
+ much of the research in model predictive control ( mpc ) is based on linear mpc where a linear model of the plant is utilised .
this paper develops an nmpc controller with the ultimate goal of application to fault tolerant flight control for uavs .
the next section ( section [ sec : nmpc ] ) will detail the workings of mpc , followed by a discussion of nmpc .
the implementation of nmpc requires the solution of an optimal control problem at each time step , and in section [ sec : nmpc ] several methods of achieving a solution are investigated . in section [ sec : brach ] a small selection of these optimal control methods , namely direct single shooting , direct multiple shooting and two collocation methods the first based on euler integration and the second on pseudospectral discretisation are applied to the brachistochrone problem , with the results aiding in the selection of the best technique to integrate into the nmpc controller design .
the results of this section showed the pseudospectral collocation method to be the best choice .
section [ sec:2drobotmodel ] applies a pseudospectral based linear and nonlinear mpc controller to a 2d robot model in both open loop and closed loop scenarios .
the results show that the nonlinear controller outperforms the linear controller particularly as perturbations increase and linear assumptions are violated as is the case when faults occur .
section [ sec : flightcontrol ] applies the final nmpc controller design to the longitudinal motion of a generic aircraft model assuming fault detection information is available .
the results successfully demonstrate the capability of our controller as a fault tolerant flight control system .
many fault tolerant control system designs require two controllers one for the nominal case and one for the fault case .
our system requires the design of only one controller which can handle both nominal and fault cases .
finally section [ sec : conclusion ] gives the analysis followed by the conclusion .
model predictive control ( mpc ) , also referred to as receding horizon control , is an advanced control technology developed by practitioners within the industrial process industry that has had considerable impact on industrial process control .
mpc , being capable of handling equipment and safety constraints @xcite , allows systems to operate at or near constraints , yielding a more efficient and profitable operation . unlike many other control system designs , where the model of the plant is used only for design and analysis purposes , in mpc
the model is an integral part of the control algorithm and is used to predict future behaviour of the plant in order to calculate the optimal control trajectory .
+ linear mpc , where the internal model is linear , is a thoroughly researched area and is commonly used in practice .
the internal prediction model predicts the behaviour of the plant over a future prediction horizon , @xmath0 . the idea is to select the best input that will produce the best predicted behaviour . a number of coincidence points are placed over the horizon with the aim of bringing the predicted output as close as possible to the reference trajectory .
this is achieved by optimising a cost function , commonly a quadratic cost via quadratic programming in the case of linear mpc .
only the first input of the calculated trajectory is applied to the plant and the prediction window slides along by the sampling time .
once the window slides to the next time step and the calculated input is applied to the plant , the new plant states are fed back to the controller and the whole cycle begins again .
the length of the prediction window remains fixed but slides forward by one sampling interval at each step ; a process referred to as the receding horizon strategy . to reduce the computational burden a control horizon , @xmath1 ,
can be defined which is smaller than @xmath0 .
the control inputs are calculated only along the control horizon , beyond which point the value of @xmath2 remains constant .
performance stability increases as the length of @xmath1 approaches the length of @xmath0 .
+ whilst nmpc uses the same structure as linear mpc the advantage of the former is the incorporation of a nonlinear process model for highly nonlinear systems .
these nonlinear models are based on `` first principles '' and are obtained from an understanding of the physical nature of the system @xcite . increased ability to handle the computing demands of solving nonlinear optimization problems
has lead to a rise in interest in nmpc within the control community .
+ the interest in nmpc , which began in the 90s , has been driven by the fact that today s processes need to be operated under tighter performance specifications , with more environmental and safety constraints that can only be met when process nonlinearities and constraints are explicitly considered in the controller design @xcite .
the major limitation of linear mpc is that plant behaviour is described by a linear dynamic model , making it unsuitable for both moderately as well as highly nonlinear processes with large operating regimes @xcite .
nmpc is more frequently used in the process industry because the time scales encountered are in the order of minutes , making real - time requirements less severe than in , for example , aerospace applications . + while nmpc has the potential to improve process operation , it poses theoretical and practical problems that are more challenging than those associated with its linear counterpart , mainly due to the nonlinear program that must be solved online at each sampling period @xcite . however , the inherent robustness of nmpc that allows it to deal with input model uncertainties without taking them directly into account , is a definite advantage , which is what makes it the focus of this paper .
the next sub - section discusses optimal control techniques that can be used to solve nmpc problems .
for the nmpc methodology to be practically feasible the optimisation must be performed within the time constraints governed by the sampling period of the application @xcite . hence , when designing and implementing nmpc strategies , consideration must be given to computational efficiency @xcite .
globally optimal nmpc methods can provide benefits over local techniques and can be successfully used for online control @xcite .
nmpc methods are generally based around tailoring nonlinear programming algorithms @xcite to fit the structure of the online optimization or parametrising the predictions in terms of degrees of freedom .
this directly affects the size of the online optimisation problem and in turn the computational burden of the nmpc strategy .
+ methods used to solve optimal control problems commonly fall under two categories : direct and indirect @xcite .
direct methods have better convergence properties than indirect methods and can be used quickly to solve a number of practical trajectory optimisation problems , hence only direct methods are considered in this paper . + the path - constrained trajectory optimisation problem as detailed in @xcite , is an area that has been heavily researched and forms an integral part of the design of an nmpc controller .
the main aim of optimal control is to determine the state and control pair that minimises a cost functional .
that is , if a state and control pair is represented by @xmath3 then the aim is to minimise : @xmath4 + \int_{t_0}^{t_f } \!\left [ \mathcal{l}\left(\mathbf{x}(t),\mathbf{u}(t),t\right)\right ] \ , \mathrm{d}t,\ ] ] subject to the nonlinear state equations : @xmath5 , \ ] ] the initial and terminal constraints @xmath6 = 0 , \\
\psi_0\left[\mathbf{x}\left(t_f\right)\right ] = 0,\end{aligned}\ ] ] the mixed state - control path constraints @xmath7\leq \mathbf{g}_u,\ ] ] and the box constraints @xmath8 here : @xmath9 are the state variables , @xmath10 the control inputs , @xmath11 the time , @xmath12 the terminal non - integral cost ( also known as mayer component ) .
@xmath13 is integral cost ( known as the bolza component ) , @xmath14 represents the initial point conditions , @xmath15 the final point conditions , @xmath16 the lower bounds on the path constraints , and @xmath17 the upper bounds on the path constraints .
+ solving the problem defined by equations to is difficult .
the direct methods detailed here solve this problem by applying a discretisation process and using standard algorithms to solve the resulting discrete optimisation problem .
+ a multitude of discretisation techniques exist for converting the continuous time problem to discrete time . in direct methods
the mathematical programming problem , equations to , is solved by considering either discretised inputs , or a combination of discretised inputs and states , as decision variables .
the most common direct methods use the controls and states as optimisation parameters .
+ when deciding on a discretisation process many important factors need to be considered @xcite , such as the accuracy of the solution for a particular discretisation method given a number of optimisation variables , the computational expense of a particular discretisation method and the robustness of the discretisation method to the initial guess .
the methods chosen for implementation and analysis are direct single shooting @xcite , direct multiple shooting @xcite , direct collocation using euler integration @xcite and direct collocation using pseudospectral discretisation @xcite . - are used to solve the given problem with initial and terminal conditions .
they convert two point boundary value problems ( bvps ) into initial value problems ( ivps ) by guessing the value of the derivative at the initial boundary . every time
a guess is made a `` shot '' ( using de solvers to find a solution ) , is fired in an attempt to hit the end boundary .
it is an iterative process where `` shots '' are made until the end boundary is reached within a desired tolerance .
+ the main difference between direct single and direct multiple shooting methods is that in the latter a bvp is converted into multiple ivps .
the interval of computation , @xmath18 $ ] is divided into @xmath19 subintervals and an ivp is solved over each subinterval .
all of the solutions over the subintervals are pieced together to form a continuous trajectory / solution , whenever the solutions to the ivps match at the beginning and end of each subinterval , the matching conditions .
these matching conditions introduce algebraic equations which must be satisfied along with the boundary conditions .
- involves fully discretising the problem ( all controls and all states ) and then solving the discrete problem numerically .
the discretisation method , either integration or differentiation based , used to approximate the state equations must be combined with a method for approximating the integral in the generalised bolza problem .
pseudospectral methods , for example , are differentiation based methods relying on differentiating lagrange polynomial expansions of the approximating polynomials for the states , while hermite - simpson based techniques are often thought of as integration methods .
a comparison of various methods can be found in @xcite .
we investigated both an integration based method as well as a differentiation based method .
+ derivative based ( or pseudospectral ) methods provide a better convergence rate known as spectral accuracy ( @xcite , @xcite ) .
the underlying idea is to represent the solution @xmath20 via a truncated series expansion and to use analytic differentiation of the series to obtain spatial derivatives of @xmath20 .
the spectral differentiation matrix , @xmath21 , is a linear mapping of a vector of @xmath22 function values @xmath23 to a vector of @xmath22 derivative values @xmath24 , and its calculation depends on the choice of the approximating series and the location of the points @xmath25 .
+ one advantage pseudospectral methods have over finite element or finite difference methods is that the underlying polynomial space is spanned by orthogonal polynomials that are infinitely differentiable global functions ( @xcite , @xcite , @xcite ) .
the choice of collocation points is crucial in pseudospectral methods @xcite and the legendre - gauss - lobatto ( lgl ) points are implemented here because they provide maximum accuracy for quadrature approximations while at the same time avoiding the runge phenomenon during interpolation @xcite . in the case of lgl
, the nodal points , which are the zeros of the derivatives of the legendre polynomials , lie in the interval @xmath26 $ ] with the end points of the interval being included in the discretisation .
+ finally an nlp solver is required as it forms an integral part of nmpc .
different discretisation methods are affected by the choice of nlp solvers in terms of the speed and robustness of the solution obtained @xcite . for this work snopt @xcite
is the solver of choice due to its popularity and because it is readily available .
snopt solves the quadratic programming subproblem with a quasi - newton approximation to the hessian , via a large - scale sparse sequential quadratic programming algorithm .
the brachistochrone problem is a nonlinear , nontrivial problem with an analytic solution very similar to the 2d robot problem to be addressed later .
the analytical solution allows determination of the accuracy of the implementation of each method and provides a benchmark in choosing a numerical method to continue development of an nmpc controller .
the brachistochrone problem , simply stated , is to find the shape of a wire such that a bead sliding on the wire without friction , in uniform gravity , will reach a given horizontal displacement in minimum time .
+ the analytical solution is given by : @xmath27 where @xmath28 , @xmath29 and @xmath30 are the horizontal and vertical displacements of the bead in the @xmath31-plane , @xmath32 is the gravitational force and @xmath33 is the final x - displacement . the optimal control problem is to minimise the cost function : @xmath34 subject to the equations of motion of the bead : @xmath35 and the initial and terminal constraints : @xmath36 here @xmath37 is the time taken to reach @xmath33 and @xmath38 is the speed of the bead .
+ the number of discretisation points was varied for each method to investigate their effect and to determine the method most suitable for developing the fault tolerant controller .
the value of @xmath33 is set to @xmath39 and the value of @xmath32 for this work is @xmath40 . for the direct single shooing method the control points , @xmath41 , were varied from 5 to 500 and for each value of @xmath41 the state points were varied from 10 to 1000 . similarly the number of sections for the direct multiple shooting method was varied from @xmath42 to @xmath43 and the control points chosen for each section went from @xmath44 to @xmath45 .
the coincidence points for both the collocation methods ( euler integration and pseudospectral ) varied from @xmath46 to @xmath47 .
+ the accuracy of the different numerical methods was assessed by comparing the solutions produced by them and the analytical solution given in .
the comparison was performed by producing plots of the magnitude of the errors in @xmath29 and @xmath30 .
the results showed that the pseudospectral method produced the most accurate solution and increasing the number of coincidence points beyond @xmath48 did not increase the accuracy of the solution .
hence for this reason the pseudospectral method with @xmath48 was used as the nominal solution in the analysis of the cpu time .
+ the time taken to reach @xmath49 in the solution of the analytical problem given by equation is @xmath50 .
plots of the cpu time taken to reach an optimal solution as a percentage of the nominal were produced .
the error plots show the percentage error between the optimal solution produced by each method and the nominal solution .
+ an example plot for the direct single shooting method is given in figure [ fig : chap3_dss_time_nu5 ] for @xmath51 .
the plots showed that for @xmath51 the cpu time is less than the nominal for all @xmath52 however the correct final time is unattainable .
the cpu times for the pseudospectral method are given in figure [ fig : chap3_pseud_time ] .
as expected the cpu time increases for increasing @xmath22 and in general the cpu time taken by the pseudospectral method is higher when compared with the other methods .
+ , @xmath51 ] ] overall the results showed that the pseusdospectral method can produce more accurate results with fewer discretisation points consequently requiring less time . while , for large values of @xmath22 , the pseudospectral method results in a greater cpu time , larger values of n are deemed unnecessary to obtain a high level of accuracy . for this reason only the pseudospectral method with @xmath54 points is used in design of the nmpc controller .
in this part of the design phase of the nmpc controller the pseudospectral method is integrated into an mpc framework .
an nmpc controller is designed , implemented and tested for a 2d robot model for trajectory following . both the open and closed loop problems are addressed and comparisons are made to linear mpc .
the next section details the 2d robot model .
the 2d robot model given in figure [ fig : chap3_robot eom ] is used for both the linear and nonlinear implementations of mpc .
the equations of motion for this robot model are : @xmath55 where @xmath56 is the @xmath56-coordinate of the point @xmath57 , @xmath58 the @xmath58-coordinate of the point @xmath57 , @xmath59 the heading angle , @xmath60 the right wheel angular velocity , @xmath61 the left wheel angular velocity and @xmath38 the speed given by @xmath62 .
+ the next few sub - sections detail the development of the linear and nonlinear mpc controllers . for a fair comparison
the pseudospectral method with 50 collocation points is chosen as the method for discretisation for both controllers . in mpc
an open loop problem is solved at each time step .
hence it was important to implement and test the controller on the open loop problem .
+ many tuning parameters can be used to determine the performance of the controller ; the weighting factors on the cost function , the design of the cost function , the length of the prediction horizon , the initial condition , the integration time step , and the number of discretisation points required for an acceptable solution . from the previous analysis the numerical technique chosen for the application of nmpc is the pseudospectral method with 50 discretisation / coincidence points . as a part of this work the effect of the choice of the cost function , the prediction window length ,
the integration time step and the effect of the initial condition on the solution were all considered .
+ the 2d robot is required to follow the path : @xmath63 travelling with a velocity of @xmath64 and constraints of @xmath65 on the wheel speeds @xmath60 and @xmath61 .
the objective is to drive the robot back to the reference path , from @xmath66 to @xmath67 .
five different cost functions were developed : + * cost type 1 : * errors between the reference / nominal path and the robot path are minimised : @xmath68 * cost type 2 : * errors between the robot path and the nominal path , plus the error between the actual wheel speeds , @xmath60 and @xmath61 and the nominal wheel speeds are minimised : @xmath69 *
cost type 3 : * errors between the robot path and the nominal path , plus the difference between the wheel speeds are minimised : @xmath70 * cost type 4 : * errors between the nominal speed and robot speed as well as the errors between the nominal angular acceleration and the robot s angular acceleration are minimised : @xmath71 * cost type 5 : * errors between the nominal speed and robot speed as well as the errors between the nominal angular acceleration and the robot s angular acceleration along with the errors between the nominal path and the robot path are minimised : @xmath72 each cost type was tested using both linear and nonlinear mpc .
the robot initial @xmath73 and @xmath59 was set to @xmath74^\intercal$ ] for both controllers .
the prediction window length was varied between @xmath75 , @xmath76 and @xmath77 . through trial and error
the weights were set to : @xmath78 the optimal trajectories produced by all the different cost functions for the varying window lengths were plotted for both the linear and nonlinear mpc cases .
results showed that for a window length of 1 sec cost types 1 , 2 , 3 and 5 were able to drive the robot back onto the desired path by the end of the window for both the linear and nonlinear cases , however cost type 4 was unsuccessful in doing so .
+ the error plots for all the prediction window lengths were generated showing the magnitude of the error in the y - direction between the nominal path ( @xmath79 ) and the actual robot path .
overall the results showed that for a path following scenario , it is best to not only minimise the path errors , but to also follow a velocity profile to obtain a smoother non oscillating solution .
for this reason cost type 5 ( equation ) was chosen for the final controller design and is used in the remainder of this analysis .
in addition , the error plots show that the difference in errors produced by the linear and nonlinear controllers are the least for this cost function , making it the best candidate for comparison purposes .
+ the effect of the integration time step was also investigated .
the analysis given above considers only the optimal solution produced by the controller , however in an mpc framework only the first output is applied to the plant , with the plant then providing sensor information to the navigation subsystem , for example , ( on an aircraft ) to calculate location and orientation information .
hence it is important to understand the effect of the integration time step in conjunction with the optimal control input .
the integration time step was varied as follows ; @xmath80 $ ] for varying @xmath0 lengths , namely 1 sec , 5 secs and 10 secs , and the optimal trajectories were plotted .
all results showed that the integration time step has very little effect on the results . upon further investigation
the results showed that the smallest integration time step of 0.001secs was able to give a solution closest to the optimal .
the length of the prediction horizon was seen to have the greatest effect on the integrated solution with the integrated solution getting closer to the optimal solution as the look ahead increased .
another point to note is that the integrated output produced with the nonlinear controller more closely matched the optimal solution compared to the linear integrated output .
+ varying the length of the prediction window showed that it is always best to have a longer window as this produced the lowest errors , particularly in the linear controller case .
in addition , the longer window allowed the robot to reach the nominal path more quickly .
the accuracy of the integrated solution increased as the window length increased , but , unfortunately a longer window resulted in an increase in computation time . + from the results obtained a window length of 5 secs was chosen for the final nmpc controller design as it was a good compromise between efficiency and accuracy .
a window length of 1 second proved to be too short to produce an accurate solution particularly in the case of the integrated solution . while a window length of 10 secs produced an integrated solution closely matching the optimal solution in the nonlinear controller case , the solution produced by a window length of 5 secs , while not as precise , still managed to develop a solution closely resembling the optimal solution .
hence for the rest of this research a window length of 5 seconds is used along with cost type 5 and an integration time step of 0.01secs .
+ the initial condition is another factor that must be considered in the design and selection of the controller .
the sensitivity of the starting point on the overall solution is critical particularly in the case of linear techniques . in this analysis
the robot is required to follow the same path as above , @xmath67 , and the initial @xmath58 is varied from 0 m to 10 m in steps of 0.1 m .
the errors between the nominal path and the actual robot position are calculated at various points along the prediction horizon namely at 1 sec , 2 secs , 3 secs , 4 secs and 5 secs for all initial @xmath58 values .
plots of errors versus initial @xmath58 for the different times were obtained and figures [ fig : chap3_inity_1secs ] and [ fig : chap3_inity_5secs ] show the results at @xmath81 sec and @xmath82 secs .
the results show the errors between the optimal solution and the nominal path as well as the errors between the integrated output and the nominal path for both the linear and nonlinear controllers .
the errors arising from the integrated output of the linear controller are shown on a separate plot underneath the main plots as these errors were much higher compared to the others and by plotting all errors on the one graph the errors produced by the other solutions were not as clearly visible .
the results show that as the time increases from 1 second to 5 secs the errors decrease as the robot approaches the nominal path .
the results clearly show that the further away the robot is from the nominal path ( i.e. the greater the perturbation ) the higher the error in the case of the linear controller .
+ at the 1 second mark along the prediction window ( figure [ fig : chap3_inity_1secs ] ) the errors between the solution produced by the nonlinear controller and the nominal path ( @xmath67 ) were seen to be linear as a function of initial @xmath58 .
moving further along , the prediction window shows that these errors decrease and are very close to zero for any @xmath83 .
there is only a small region around the nominal path , @xmath67 , during which the errors produced by the linear controller are zero and match those produced by the nonlinear solution at any time along the prediction window . ] ] the next subsection investigates the closed loop problem and compares the output produced by both linear and nonlinear mpc .
the aim of these simulations is to implement and investigate the behaviour of both linear and nonlinear mpc regarding the closed loop problem .
the fault tolerant problem is essentially closed loop , hence to apply nmpc to fault tolerant control the pseudosepectral nmpc controller design is tested on the 2d robot model where the robot is required to travel on a circular trajectory .
+ based on the analysis from the previous subsections a prediction window length of 5 seconds is used with 50 collocation points , cost type 5 and an integration time step of 0.01secs
. constraints of @xmath65 are placed on the control inputs which are the angular velocities produced by the right and left wheels .
three different scenarios were set up : scenario 1 the robot begins on the path with initial conditions @xmath84^\intercal$ ] , scenario 2 the robot begins slightly off the path with initial conditions @xmath85^\intercal$ ] and scenario 3 the robot begins well off the path with initial conditions @xmath86^\intercal$ ] .
+ for stability @xmath87 @xcite . in many mpc\nmpc formulations
@xmath1 is less than @xmath0 . while this greatly reduces the computational expense it does however produce a suboptimal solution @xcite ,
hence for the purposes of this research the control horizon is equal to the prediction horizon . for scenario 1 where the robot begins on the path the trajectories produced by both linear and nonlinear controllers were the same .
the calculated optimal inputs produced by both controllers were also identical . in the case of scenario 2 where the robot begins slightly off the path , both
the linear and nonlinear controllers managed to bring the robot back onto the path .
the plots of the optimal inputs showed that initially both controllers work at the maximum constraint to drive the robot back onto the path .
once the path is reached ( i.e. perturbations are small ) both controllers exhibit the same performance . + the trajectory plots for scenario 3 show that only the nonlinear controller is able to bring the robot back onto the path with the linear controller unable to drive the robot back to the path .
the control inputs produced by both controllers clearly showed that the linear controller worked very hard to take the robot back onto the path by consistently working at the constraint limits however it was still unable to return the robot back to the path .
+ the results clearly showed that the pseudospectral nmpc solution to the nonlinear model predictive controller is a viable choice outperforming its linear counterpart when the perturbations are large . in the next section
we apply the solution developed thus far to a generic aircraft model .
note that fdi is assumed for the simulation exercise carried out in section [ sec : flightcontrol ] .
the nmpc controller developed in the previous sections was applied to the longitudinal motion of an aircraft to demonstrate fault tolerant control . the generic aircraft model developed here for control
law design and validation is based on the mcdonnell douglas f-4 aircraft @xcite .
it is a fixed wing aircraft equipped with throttle , elevators , ailerons and a rudder for control .
longitudinal motion is predominantly controlled via the throttle and elevators which is used to pitch the aircraft nose up and down and hence the remaining controls will not be considered here .
figure [ fig : chap5_angles_sl_paper1 ] presents a sketch of a generic aircraft which identifies the location of the elevators and defines the coordinate frames in which the equations of motion are defined .
these equations are all carried out in the body axis which has its origin at the centre of gravity ( c.g . ) on the body of the aircraft however position and velocity information are commonly presented in the ned frame which is an earth fixed coordinate system .
the generic aircraft model has the aerodynamic characteristics of the mcdonnell douglas f-4 aircraft however the dimensional and mass properties are those given in tables [ table : chap5_acdims ] and [ table : chap5_mass_props_si ] respectively .
.aircraft dimensional properties [ cols="^,^",options="header " , ] [ table : chap5_constraints3dof ] the weighting matrices are diagonal matrices with the following values along the diagonal for each state which were set through trial and error : @xmath88 . note : the control surface rates given in table [ table : chap5_constraints3dof ] are realistic for a high performance unstable airframe or for a lower weight aircraft with a stable airframe ; in either case a feasible fictional aircraft model has been produced for simulation purposes to demonstrate proof of concept . to illustrate the concept of ftc ,
seven different scenarios were set up , with the first one being the no fault case .
the next five scenarios had the throttle stuck at @xmath89 and @xmath90 for the entire duration of the flight .
scenario seven simulated the throttle getting stuck at @xmath90 80 secs into the flight .
+ it is also assumed that fault detection information is available .
this includes the time at which the throttle becomes stuck and and the position at which it is stuck .
this information is used to update the constraint values of the nmpc controller .
providing the controller with the most accurate and up to date information enables it to make better use of the healthy actuators .
as previously mentioned the force and moment coefficients are valid for @xmath91 . for all scenarios @xmath92 was checked to ensure that @xmath93 was never exceeded .
the plots of @xmath92 vs time ( they have not been provided here due to space constraints ) showed @xmath92 to remain below @xmath93 hence the equations of motion were never violated .
another means of avoiding this would be to place a constraint on @xmath92 in the nmpc controller .
- the demanded airspeed was @xmath94 true airspeed .
the plot given in figure [ fig : chap5_st_ta ] shows the aircraft true airspeed for each of the scenarios .
the results show that in a fault free case the aircraft is able to fly at the demanded true airspeed .
however , when the throttle is stuck at @xmath95 or even @xmath96 there is too much power continually being provided to the aircraft resulting in a large airspeed response .
when the throttle is stuck at @xmath97 the aircraft is able to maintain the demanded @xmath98 for only a short period of time , at the beginning of the flight mission .
however at @xmath99 throttle the maximum deviation from the demanded airspeed is approximately @xmath100 at any given time .
when the throttle drops below @xmath99 the aircraft is unable to maintain the true airspeed which drops to approximately between @xmath101 and @xmath102 .
the results for scenario 7 show that once the fault occurs at 80 secs the true airspeed immediately begins to drop , as expected .
one of the main points to note is that the stall speed was never reached ; the controller was able to avoid the aircraft stalling regardless of the severity of the fault .
- the vertical speed response of the aircraft was also analysed and plots for scenarios 1 - 4 are given in figure [ fig : chap5_st_vd ] .
the plots show the aircraft response along with the constraints ( in red ) placed on the vertical speed . for high values of throttle ( @xmath95 and @xmath96 )
the vertical speed is continuously bouncing between the constraints in an attempt to maintain the true airspeed demand .
for the case when the throttle is stuck at @xmath97 the vertical speed profile is seen to be similar to the no fault case , except in the descent phase . during this phase ,
when the aircraft is descending and gaining speed , the vertical speed response can be seen to continuously move between the constraints to regulate the speed .
results showed that for throttle values less than @xmath99 there is insufficient power to maintain a climb hence the vertical speed is seen to operate at the lower constraint or at zero .
in the case of scenario 7 it was found that once the fault occurred at 80 secs the vertical speed moved between the constraints , working hard to maintain the true airspeed .
- in regards to fault tolerance , the elevator activity is of the most interest . if the throttle is stuck the elevator provides a level of redundancy to maintain the aircraft speed .
figure [ fig : chap5_st_ea ] shows plots of elevator activity for the different scenarios .
the plots clearly show that any change in throttle increases the elevator activity when compared to the no fault case .
the elevator activity increases in an attempt to regulate the airspeed of the aircraft . in the case of the high throttle values ( @xmath95 and @xmath96 ) the elevator activity is the highest because a higher level of power is continually being provided to the aircraft exceeding the amount required to fly at the demanded speed .
hence the elevator constantly jumps between the constraints in an attempt to compensate for the excess power . for the @xmath99
stuck throttle case the elevator activity does increase compared to the no fault case ; however @xmath97 throttle was found to be closer to the amount required to maintain the given height profile , hence the elevator does not need to work as hard compared to the @xmath95 and @xmath96 cases .
for the lower throttle values activity increases during the climb and descent phases . in the climb phase of the mission
there is not enough power available to the aircraft , so it compensates by erratically deflecting the elevator . during the descent phase
however there is too much power ; to regulate this and to stay within the velocity constraints the activity increases .
the last scenario shows that at the fault occurrence time of 80 secs the elevator increases activity to compensate for the faulty throttle .
the plots indicate large oscillations in the elevator activity however the data presented shows 200 seconds of flight .
actuator dynamics have been modelled in both the controller and the plant model . zooming in on the @xmath95
stuck throttle case ( figure [ fig : zoomedea ] ) it can be seen that the rate dynamics are respected .
stuck throttle - elevator activity ( zoomed ) , constraints ( red lines ) , aircraft response ( blue ) ] - the trajectory flown by the aircraft during the different scenarios is given in figure [ fig : chap5_st_hp ] .
the no fault case , as expected , follows the reference height profile perfectly .
the @xmath97 case is also able to closely maintain the profile . in the @xmath95 and @xmath96 cases the aircraft continually tries to regulate the airspeed to compensate for the excess power
the solutions produced by both scenarios show the aircraft overshooting followed by an undershoot , so the solution oscillates around the reference .
when the throttle becomes stuck at @xmath99 the aircraft begins the climb phase of the mission but is only able to continue climbing for 20 secs before it begins gliding towards the ground . in the @xmath90
stuck throttle case the aircraft completes the straight and level phase of the mission but does not have enough power to begin climbing , and descends to the ground .
the final scenario shows that the elevator is able to compensate for the stuck throttle in mid - flight and successfully finish the mission .
this paper was dedicated to the theoretical and practical implementation aspects of nmpc .
a viable design for a nonlinear mpc controller applicable to ftc was sought .
a number of discretisation methods were implemented using the well known brachistrochrone problem and the pseudospectral numerical method was found to have the best performance .
this method was applied in an nmpc framework to a 2d robot model in both open and closed loop settings .
comparisons were made between linear mpc and the nonlinear mpc solutions . for small perturbations
the two controllers produced the same results .
however , for large perturbations where nonlinearity effects are more significant , the pseudospectral nonlinear mpc controller was found to produce more accurate results than the linear mpc controller .
the final design was also applied to a generic aircraft model where it was assumed that fault detection information was provided .
the motivation behind this was to demonstrate the concept of fault tolerant flight control using nmpc .
it was found that the nonlinear nmpc controller designed in this work is a viable controller for fault tolerant flight control .
this was evident from the results which show that in the event of a stuck throttle the controller is able to manipulate the movement of the elevator to compensate for the fault .
kale mm , chipperfield aj .
_ robust and stabilized mpc formulations for fault tolerant and reconfigurable flight control_. proceedings of the 2004 ieee international symposium on intelligent control 2004 ; 222 - 227 .
boskovic jd , sai - ming l , mehra rk .
_ on - line failure detection and identification ( fdi ) and adaptive reconfigurable control ( arc ) in aerospace applications_. proceedings of the 2001 american control conference 2001 ; 4:2625 - 2626 .
doi:10.1109/acc.2001.946269 gopinathan m , boskovic jd , mehra rk , rago c. _ a multiple model predictive scheme for fault - tolerant flight control design_. proceedings of the 37th ieee conference on decision and control 1998 ; 2:1376 - 1381 .
doi:10.1109/cdc.1998.758477 .
long c.e , polisetty pk , gatzke ep . _ nonlinear model predictive control using deterministic global optimization_. journal of process control 2006 ; 16(6):635 - 643 .
doi : http://dx.doi.org/10.1016/j.jprocont.2005.11.001 .
diehl m , bock hg , diedam h , wieber pb . _ fast direct multiple shooting algorithms for optimal robot control_. fast motions in biomechanics and robotics - lecture notes in control and information sciences 2006 ; 340:6593 .
doi:10.1007/978 - 3 - 540 - 36119 - 0 - 4 .
diehl m , bock hg , schlder jp , findeisen r , nagy z , allgwer f. _ real - time optimization and nonlinear model predictive control of processes governed by differential - algebraic equations_. journal of process control 2002 ; 12(4):577585 .
doi : http://dx.doi.org/10.1016/s0959 - 1524(01)00023 - 3 .
williams p. _ a comparison of differentiation and integration based direct transcription methods ( aas 05 - 128)_. advances in the astronautical sciences american astronautical society by univelt 2005 ; 120(1):389409 .
gill pe , murray w , saunders ma .
_ users guide for snopt version 7 : software for large - scale nonlinear programming_. citeseer 2006 ; http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.217.5929 [ last accessed : 20 - 01 - 2016 ] . | this paper describes the design process for developing a nonlinear model predictive controller for fault tolerant flight control . after examining and implementing a number of numerical techniques
, this paper identifies pseudospectral discretisation as the most suitable for this design . applying the controller to a 2d robot model shows that the nonlinear controller performs much better than the linear controller , especially in the closed loop scenario . assuming fault detection information , applying the technique to the longitudinal motion of a generic aircraft model shows the design to be eminently suitable for flight control .
keywords : nonlinear model predictive control , pseudospectral , optimal control , 2d robot , flight control |
low mass stars in star - forming regions produce strong x - ray emission from coronal magnetic activity , as evidenced by high temperature ( @xmath310 mk ) emission from flares and active regions ( feigelson & montmerle 1999 ; gagn et al .
2004 ; preibisch et al .
the role of accretion in the production of x - rays from young stars is less well understood .
magnetic structures on stars can extend all the way to the inner accretion disk , as suggested from models of decaying stellar flares ( favata et al .
extended x - ray emission from jets also occurs , as observed in dg tau , a system oriented so that the circumstellar disk plane aligns with our line of sight ( gdel et al .
soft x - ray emission may arise from shocks at the stellar base of this extended x - ray jet ( gdel et al .
2008 ; schneider & schmitt 2008 ) .
x - rays can also be produced in accreting systems when the accreting material accelerates to supersonic velocities and shocks near the stellar surface , heating the gas to a few mk . in the standard model of accreting classical t tauri stars ( ctts )
, the magnetic field of the star connects to the accretion disk near the corotation radius , channeling the accretion stream from the disk to a small area on the star ( calvet & gullbring 1998 ) .
x - rays from the shock may be difficult to detect if the shock is formed too deep in the photosphere ( hartmann 1998 ) or if they are absorbed by the stream of preshocked neutral or near - neutral gas .
furthermore , the x - ray shock signature may be difficult to distinguish from soft coronal emission without the high resolution spectroscopy necessary to determine temperature and density and identify the emitting regions .
shock models predict high electron density @xmath1 ( @xmath8 @xmath6 ) at relatively low electron temperature @xmath0 ( a few mk ) , features that can be tested with x - ray line ratio diagnostics .
tw hydrae ( tw hya ) was the first ctts found to show these characteristics of high density and low temperature ( kastner et al .
2002 ; hereafter , k02 ) , based on a 48 ks _ chandra _ high energy transmission grating ( hetg ) spectrum .
tw hya is one of the oldest known stars ( @xmath3 10 myr ) still in the ctts ( accreting ) phase .
it is uniquely poised in an interesting state of evolution when it will soon stop accreting , lose its disk , and perhaps form planets ( e.g. , calvet et al .
2002 ) . at a distance of only 57 pc
, it is also one of the brightest t tauri stars in x - rays .
interstellar absorption is minimal , and the observed absorption can be considered intrinsic to the system ( k02 ) .
tw hya s circumstellar disk is nearly face - on with an inclination angle of 7@xmath9 ( qi et al .
thus the star appears pole - on .
key to k02 s argument for accretion is the electron density ( @xmath10 @xmath6 ) determined from he - like emission line ratios , and confirmed by subsequent observations with the _ xmm - newton
_ reflection grating spectrometer ( stelzer & schmitt 2004 ) and the _ chandra _ low energy transmission grating ( letg ; raassen 2009 ) . while @xmath1 for some active cool star coronae approaches such high values at high @xmath0 ( @xmath11 mk ) , e.g. , 44 boo , ux ari , and ii peg ( sanz - forcada et al . 2003 ; testa et al .
2004 ) , the high @xmath1 of tw hya is produced at the significantly lower @xmath0 ( @xmath33 mk ) expected for the accretion shock . bp tau ( schmitt et al . 2005 ) , v4046 sag ( gunther et al . 2006 ) , ru lup ( robrade & schmitt 2007 ) , and mp mus ( argiroffi et al . 2007 ) also show high values of the electron density . on the other hand , t tau and ab aur have a low @xmath1 ( gdel et al .
2007a ; robrade & schmitt 2007 ) ; and , the binary hen 3 - 600 appears to lie at an intermediate value of @xmath1 ( huenemoerder et al . 2007 ) .
despite large differences in @xmath1 , all of these stars show an excess ratio of soft x - ray emission to hard x - ray emission , manifested by their large / ratios as compared with active main sequence stars ( gdel & telleschi 2007 ; robrade & schmitt 2007 ) .
additional soft x - ray diagnostics available with a deep _ chandra _
exposure might be expected to shed light on the accretion process and test the shock models .
since accretion is expected to result in higher densities than found in an active stellar corona , it is critical that the density determination be secure .
interpretation of the line ratio in terms of high density is not unique , since photoexcitation by ultraviolet radiation mimics the effects of high density . determining the density from is not subject to the same degeneracy , since the relevant ultraviolet radiation , observed with the far ultraviolet explorer ( fuse ; dupree et al .
2005 ) , is too weak to photoexcite its metastable level .
the three he - like systems ( , , and ) together can provide independent determinations not only of the electron density , but also of the electron temperature and hydrogen column density .
deep spectroscopy can , first , put the accretion interpretation on firm ground and second , allow us to determine the structure of the accretion region itself .
we were awarded a _
chandra _ large observing project to establish definitively whether the x - ray emission in tw hya is associated with accretion , using high signal - to - noise ratio spectroscopy with the hetg , and , if so , to determine the emission measure distribution and elemental abundances , search for x - ray signatures of infall , outflow , and turbulence , and constrain the properties of the accretion hot spot . during the _ chandra _ observations we also conducted a ground - based observing campaign , which will be reported elsewhere ( dupree et al .
2010 , in preparation ) . here
, we describe the x - ray observations in section 2 and report analysis results in section 3 , where the densities , temperatures , and column densities can be extracted from this rich spectrum .
section 3 also presents elemental abundances derived from fits to the spectrum , as well as the measurement of turbulent velocity .
section 4 describes spectral predictions from a model of the tw hya shock . in section 5
we discuss evidence for a new type of coronal structure .
section 6 gives a summary of results and conclusions .
_ chandra _ observed tw hya with the hetg in combination with the acis - s detector array for a total of 489.5 ks intermittently spanning the period between 2007 feb 15 and 2007 mar 3 .
the observation consists of 4 segments , with exposure times as follows : 153.3 ks ( obsid 7435 ) , 157.0 ks ( obsid 7437 ) , 158.4 ks ( obsid 7436 ) , and 20.7 ks ( obsid 7438 ) .
the summed spectrum has about ten times the exposure time of the spectrum analyzed by k02 .
figure 1 shows the total first - order light curve , with flux variations of about a factor two during the observation . assuming a distance of 57 pc , the observed x - ray luminosity @xmath12 , measured using the medium energy grating between 2.0 and 27.5 , is @xmath13 erg s@xmath14 .
one clear flare , with total x - ray luminosity of 2.1 @xmath15 erg s@xmath14 , persists for @xmath315 ks , producing a total energy of 4 @xmath16 ergs .
the flare is retained in the following analysis since its contribution to line fluxes is negligible ( see section 3.4 ) .
we have extracted spectra from the event files , and produced calibration files using the _ chandra _ interactive analysis of observations ( ciao ver .
3.4 ) software package ( fruscione et al . 2006 ) . with response matrices combined from the four observations , _
sherpa_(freeman et al .
2001 ) was used to extract individual emission line fluxes from the co - added observations , maintaining separate first order spectra for the four grating arms ( plus and minus ; meg and heg ) during the fitting procedure . a model for the continuum plus background
was found from a global fit to line - free regions and used in addition to the line models .
background rejection is high with acis order sorting .
thus the background contributes relatively little to this model except at the long wavelength end of the spectrum , where the effective area is low . in the o vii region , for example , the background contains about half as many counts as the continuum . for the relatively weak but important line , background , continuum , and line
have about 1 , 2 , and 16 counts , respectively .
line models are gaussian functions with negligible or small widths relative to the width of the instrumental line response functions ( @xmath30.012 and 0.023 , for heg and meg , respectively ) .
table 1 gives the identified lines and their observed fluxes .
for most lines the fit widths are less than half the instrumental width and are not statistically significant .
we discuss the widths of the strongest lines in section 3.5 .
lrccl & 6.182 & 1.87 @xmath17 0.19 & 163.0 & corona + & 6.648 & 2.72 @xmath17 0.17 & 146.6 & corona + & 6.688 & 0.74 @xmath17 0.14 & 138.3 & corona + & 6.740 & 1.66 @xmath17 0.15 & 151.6 & corona + & 8.421 & 2.24 @xmath17 0.20 & 162.4 & corona + & 9.169 & 2.25 @xmath17 0.22 & 135.4 & shock + & 9.231 & 0.90 @xmath17 0.17 & 133.2 & shock + & 9.314 & 1.27 @xmath17 0.19 & 130.1 & shock + & 9.481 & 1.64 @xmath17 0.20 & 114.5 & corona + & 9.708 & 2.75 @xmath17 0.27 & 86.2 & corona + & 10.239 & 9.23 @xmath17 0.53 & 81.1 & corona + & 10.765 & 3.54 @xmath17 0.43 & 67.9 & shock + & 11.001 & 8.48 @xmath17 0.63 & 67.4 & shock + & 11.544 & 23.7 @xmath17 0.99 & 55.3 & shock + & 11.783 & 3.28 @xmath17 0.50 & 52.3 & corona + & 11.932 & 3.01 @xmath17 0.50 & 49.6 & corona + & 12.134 & 76.1 @xmath17 1.9 & 48.5 & corona + & 12.266 & 4.3 @xmath17 0.6 & 46.6 & mixed + & 13.447 & 177 .
@xmath17 4.2 & 22.3 & shock + & 13.553 & 114 .
@xmath17 3.3 & 22.7 & shock + & 13.699 & 58.7 @xmath17 2.3 & 25.0 & shock + & 14.208 & 4.59 @xmath17 0.76 & 29.0 & corona + & 14.256 & 1.63 @xmath17 0.53 & 28.5 & corona + & 14.820 & 5.04 @xmath17 0.80 & 23.7 & shock + & 15.014 & 36.5 @xmath17 1.8 & 21.9 & shock + & 15.176 & 10.4 @xmath17 1.2 & 20.8 & shock + & 15.261 & 17.4 @xmath17 1.4 & 20.0 & shock + & 16.006 & 30.6 @xmath17 2.3 & 15.9 & shock + & 16.071 & 5.17 @xmath17 1.0 & 16.3 & corona + & 16.780 & 25.5 @xmath17 3.0 & 12.1 & shock + & 17.051 & 27.9 @xmath17 2.3 & 11.1 & shock + & 17.096 & 26.3 @xmath17 2.3 & 11.2 & shock + & 17.207 & 4.68 @xmath17 1.4 & 10.7 & shock + & 17.396 & 4.68 @xmath17 1.5 & 9.8 & shock + & 17.768 & 3.66 @xmath17 1.6 & 7.7 & shock + & 18.627 & 17.9 @xmath17 1.4 & 7.3 & shock + & 18.969 & 213 .
@xmath17 8.4 & 6.6 & shock + & 20.910 & 3.8 @xmath17 2.5 & 3.3 & shock + & 21.601 & 117 .
@xmath17 10.0 & 2.8 & postshock + & 21.804 & 72.4 @xmath17 9.1 & 2.7 & postshock + & 22.098 & 15.2 @xmath17 4.4 & 2.1 & postshock + & 24.781 & 68.5 @xmath17 7.6 & 2.6 & shock + all line fluxes are within 3@xmath18 of the values reported by k02 except for @xmath1913.437 , which has slightly more than twice as much flux in our observation .
our line fluxes are also within 3@xmath18 of the values reported by stelzer & schmitt ( 2004 ) for lines observed with both instruments , except for @xmath1913.553 , for which our flux is about half . for lines in common with those reported by raassen ( 2009 )
all fluxes agree within 3@xmath18 . with the higher signal - to - noise ratio of our data , new diagnostic information is available .
a major goal of this study is the determination of electron density from the he - like ions .
we first discuss the diagnostics , and then model the global spectrum .
figure 2 shows the high quality in the he - like and line regions .
interestingly , the region shows no apparent lines , which can contaminate the diagnostics ( e.g. , ness et al .
upper limits on the strongest line at 13.52 imply that contributes 1% or less to the line fluxes .
the triplet line region is more complex ( fig .
3 ) , with lower signal - to - noise ratio and apparent line blending between 9.2 and 9.4 . for this region ,
the fit is constrained such that all known lines are fixed in wavelength relative to the resonance line @xmath199.169 , which is fit separately for each grating arm .
lines are gaussian functions with widths set to zero ( i.e. , only the instrumental line response function is used ) .
line fluxes of the lyman series lines @xmath20 to @xmath21 are scaled according to their oscillator strengths , since collision strengths are not available in the literature .
we note that systematic errors from line blends are not included in the errors given in table 1 .
one of the advantages of x - ray spectroscopy with high signal - to - noise characteristics is the availability of numerous line ratio diagnostics to assess conditions in the radiating plasma . in particular , he - like systems provide valuable diagnostic opportunity , as discussed extensively in the literature ( porquet et al .
2001 ; smith et al . 2001 ; and references therein ) .
the g - ratio , defined as the flux ratio of the forbidden ( @xmath22 ) plus intercombination ( @xmath23 ) lines to the resonance ( @xmath24 ) line , is sensitive to the electron temperature ( @xmath0 ) , while the r - ratio ( flux ratio of the forbidden to intercombination lines ) is sensitive to @xmath1
. ratios of emission lines in the resonance line series ( transitions from 2p , 3p , etc .
to ground , also known as he@xmath25 is the resonance line sometimes labeled @xmath26 or @xmath27 . ] , he@xmath28 , and so on ) are sensitive both to @xmath0 , through their relative boltzmann factors , and to the absorption by an intervening hydrogen column density ( @xmath2 ) through their energy separation .
this analysis emphasizes the use of line ratio diagnostics , including those discussed above .
we use the astrophysical plasma emission code ( apec ) with the atomic data in atomdb v1.3 ( smith et al .
2001 ) to calculate line emissivities , except for he - like , for which we use the more accurate calculations of chen et al .
the model for absorption by gas with cosmic abundances is taken from morrison & mccammon ( 1983 ) . in this section
we present the physical conditions of the emitting plasma as determined from the line ratios of specific he - like and other ions , in particular @xmath1 , @xmath0 , and @xmath2 .
we use information determined from line ratios to constrain an empirical model of the emission measure distribution ( emd ) , and then allow additional parameters of the model to be constrained by global fits to the spectrum .
we give the elemental abundances determined from these fits .
finally we present velocity constraints from line centroids and widths .
we use g - ratios from multiple ions to determine the electron temperature .
figure 4 shows the observed g - ratios of he - like o , ne and mg , overplotted on their respective theoretical functions of @xmath0 . for comparison ,
the emissivities of their resonance lines peak at @xmath29 2 , 4 , and 6 mk , respectively , as shown in the figure .
table 2 gives the derived @xmath0 from these ions .
the g - ratio indicates that @xmath30 mk .
the @xmath0 determination is highly reliable , given that recent theoretical calculations , which include the resonance contributions to the collisional excitation cross sections , agree with experimental measurements to within 7% ( chen et al . 2006 ; smith et al . 2009a , 2009b ) . on the other hand , the theoretical calculations for and have higher systematic uncertainties , and the observational errors are larger , such that the @xmath0 values derived from and carry less weight ( but are consistent with ) . both and are found far below their temperature of maximum emissivity @xmath31 , supporting an accretion shock scenario for their formation .
lcc & [email protected] & 2.4@xmath32 + & [email protected] & 2.5@xmath33 + & [email protected] & 2.1@xmath34 we use r - ratios from multiple ions to determine the electron density .
figure 5 shows the observed r - ratios for , , and , overplotted on the theoretical functions of electron density @xmath1 from apec .
table 3 presents the @xmath1 values derived from these r - ratios .
sensitivity to @xmath0 is negligible near the observed r - ratio values .
smith et al .
( 2009 ) show that the apec r - ratio for in the high density range is indistinguishable from the r - ratio calculated by chen et al .
the theoretical r - ratios from apec for all three ions are in excellent agreement with porquet et al .
( 2001 ) over the density range reported here .
lcc & [email protected] & 5.7@xmath35 + & [email protected] & 3.0@xmath36 + & [email protected] & 5.8@xmath37 from we derive @xmath38 @xmath6 , while gives @xmath39 @xmath6 . the 3@xmath18 range from , between 2.5 and 3.9 @xmath40 @xmath6 ,
provides a tight constraint .
0.033 , quite close to the value of 0.446 @xmath17 0.124 reported by k02 .
the difference of nearly a factor of 2 in the derived @xmath1 appears to be due to the difference in atomic data used , with their models possibly based on the raymond & smith code ( raymond 1988 ) . ]
the best - fit value for , using all the lines in the region as described in section 2 , is @xmath41 @xmath6 . for comparison
, we have also fit only the 3 lines of mg xi , again with line widths and relative positions fixed , obtaining @xmath42 7 @xmath17 2 @xmath40 @xmath6 , in good agreement with the best - fit value .
this comparison suggests that line blending does not unduly affect the result . at low @xmath1
the population of the metastable @xmath43 level builds up since the decay rate is extremely low . at high @xmath1 electron impact excites the metastable level population up to the @xmath44 levels . in principle , low r - ratios may also be produced at low @xmath1 by photoexcitation .
stelzer & schmitt ( 2004 ) and k02 dismiss photoexcitation based on the stellar temperature ( @xmath3 4000 k ) , but drake ( 2005 ) suggests that a hot spot above the stellar surface , where the accretion stream shocks , could produce sufficient photoexciting radiation .
we have measured the flux at the excitation wavelengths 1647 , 1270 and 1034 for , , and , respectively , from the _ hubble space telescope _ stis ( herczeg et al .
2002 ) and the fuse spectra ( dupree et al . 2005 ) .
the intensity at 1034 is an order of magnitude lower than at the longer wavelengths , constraining the black body temperature of the hot spot to be less than 10,000 k , in good agreement with _ international ultraviolet explorer _ analysis showing a blackbody at 7900 k covering about 5% of the stellar surface ( costa et al .
2000 ) . thus the ultraviolet measurements and the r - ratio rule out photoexcitation and support the interpretation of high @xmath1 . together with the temperature derived from the g - ratio
, the density supports an accretion origin for .
the ratio of @xmath1917.096 to @xmath1917.051 becomes sensitive to @xmath1 above @xmath8 @xmath6 ( mauche et al .
. the measured ratio of 1.06 @xmath17 0.13 is at the low @xmath1 limit , ruling out the high density from this line ratio suggested by ness & schmitt ( 2005 ) .
the hetg spectrum contains several series of resonance lines that indicate absorption by neutral gas along the line of sight .
the most useful of these diagnostics are the and he@xmath25/he@xmath28 ratios because their formation temperatures are independently determined from the g - ratio .
table 4 presents the observed line ratios , and their use as diagnostics for @xmath2 .
figure 6 shows the theoretical he@xmath25/he@xmath28 ratios as functions of @xmath0 for and , for both unabsorbed and absorbed cases .
the observed ratios are overplotted using @xmath0 determined from their g - ratios ( which , as noted above , is more reliable for than for ) . for the best - fit @xmath0 , gives @xmath45 @xmath46 and gives @xmath47 @xmath46 .
these two values show a meaningful difference . for at 2.5 mk , the observed ratio is 11@xmath18 below the unabsorbed model , making the presence of an absorbing column quite definitive .
the column density from is consistent with the lower limit of @xmath48 @xmath46 determined from the absence of and lines longward of 40 in the _ chandra _ letg spectrum ( raassen 2009 ) .
lcrrrr he@xmath49/he@xmath50 & [email protected] & 8.54 & 6.52 & 5.22 & 2.49 + ly@xmath49/ly@xmath50 & [email protected] & 9.63 & 8.07 & 6.82 & 4.55 + he@xmath49/he@xmath50 & [email protected] & 11.14 & 10.07 & 9.23 & 7.34 + ly@xmath49/ly@xmath50 & [email protected] & 12.93 & 11.89 & 10.91 & 8.93 + ly@xmath49/ly@xmath50 & [email protected] & 7.061 & 6.49 & 5.96 & 4.88 ratios of the h - like series lines ( ly@xmath51 , ly@xmath52 , etc . )
are also sensitive to both absorption and temperature , but unlike the he - like ions , h - like ions do not present an independent temperature diagnostic . if the absorption column depends on the ion , as implied by the inconsistency between and , then we need to assume a model @xmath0 to obtain a column density from the h - like line ratios .
assuming 2.5 mk , the observed ratio ly@xmath25/ly@xmath28 indicates @xmath53 @xmath46 .
this value is consistent within the errors with @xmath2 from and thus a third absorber is not required . for at 2.5 mk , the column density would be larger than the column density for , as shown in table 4 ; however , table 4 also shows that no absorption is required if the lines are formed at 10 mk .
we must further consider the effects of temperature on line ratios .
we can also estimate the temperature using the charge state balance .
table 5 presents the he@xmath25/ly@xmath25 ratios for oxygen and neon .
first , the observed ratios are corrected for different column densities as found above , and then @xmath0 is derived from the respective corrected ratios . for oxygen , good consistency
is found for the temperatures derived from the g - ratio and the he@xmath51/ ly@xmath51 ratio , given a column density of about 8 @xmath54 @xmath46 . for neon , on the other hand , all the derived temperatures are larger than @xmath0 derived above from the g - ratio .
thus it appears that some of the emission originates at higher temperature . using @xmath55 @xmath46 and @xmath56 mk , and the observed he@xmath51 flux , we predict the flux of the ne x ly@xmath51 to be only 18% of the observed value . in this case
, the rest of the emission must be formed at higher temperature .
if the trend of increasing @xmath2 with charge state continues , then @xmath2 for would be higher than @xmath57 @xmath46 , and the fraction of ne x formed at 2.5 mk would be somewhat larger .
lcccc he@xmath49/ ly@xmath49 & [email protected] & 0.70 & 0.85 & 1.67 + he@xmath49/ ly@xmath49 & [email protected] & 2.48 & 2.60 & 3.04 + + identification & + + he@xmath49/ ly@xmath49 & 3.03 & 2.83 & 2.66 & 2.23 + he@xmath49/ ly@xmath49 & 4.16 & 4.08 & 4.01 & 3.87 the emissivity functions for lines of and both peak near 5 mk , suggesting the possibility of comparing their formation temperatures .
the ratio of @xmath1915.014 to @xmath1915.261 is sensitive to @xmath0 , due to a blend of an inner shell excitation line of at 15.261 (brown et al .
the temperature sensitivity essentially comes from the charge state balance of and . using the reported experimental dependence of the line ratio on the electron beam energy of the electron beam ion trap at lawrence livermore national laboratory ( brown et al .
2001 ) , the observed hetg value gives @xmath58 mk . despite similar peak temperatures , and
are not formed together .
apparently , the long tail of the ly@xmath51 emissivity function toward high temperature contributes significantly to its emission .
the lines are predominantly formed in the accretion shock , while the lines are formed both in the shock and in the hotter corona .
one might expect that the ions formed at low temperature ( , , and ) should all experience similar absorption .
using @xmath2 derived from , we can derive characteristic emitting @xmath0 for and of 1.4 mk and @xmath59 mk , respectively .
these temperatures are far lower than @xmath0 derived from the g - ratios .
they are both also far lower than @xmath60 mk derived from the charge state ratio of oxygen represented over the range of @xmath2 found here , as shown in table 5 .
we conclude that an increase in @xmath2 with increasing charge state ( presumably temperature ) up through is consistent with all the data .
we note that argiroffi et al .
( 2009 ) claim evidence for resonance scattering of the strongest lines in the xmm - rgs spectrum of mp mus , though their rgs spectrum of tw hya is too weak to show the signatures of absorption we find here .
we rule out resonance scattering in tw hya using the strong series of .
an optical depth @xmath61 of about 0.4 for he@xmath51 would be required to model the observed he@xmath51/he@xmath52 ratio . for a single ion ,
the optical depth @xmath61 scales as @xmath62 , such that we can then predict the other line ratios .
we find then that he@xmath51/he@xmath63 ratio is overpredicted by 50% .
since the oscillator strengths decrease rapidly with increasing principal quantum number @xmath64 , only the he@xmath51 line might be affected by resonance scattering .
we find instead , that all the lines are affected by absorption , and rule out resonance scattering as the absorption mechanism .
we conclude that the absorber is neutral or near neutral , consistent with the preshock accreting gas , as suggested by theoretical studies .
for example , lamzin ( 1999 ) considers the absorption by the preshock gas of the x - ray emitting plasma for different geometries and orientations .
gregory et al . ( 2007 ) use realistic coronal magnetic fields coupled with a radiative transfer code to calculate the obscuration of x - ray emission by accretion columns , and suggest that this effect can explain the observed low x - ray luminosities of accreting young stars relative to non - accretors .
we present here four models for the emission measure ( em ) distribution , where the factor 0.8 accounts for the hydgrogen to electron density ratio and the integral is taken over the volume @xmath65 .
the intensity of a spectral feature @xmath66 em @xmath67 where @xmath68 is the emissivity in units of ph @xmath69 s@xmath14 and @xmath70 is the distance to the source .
the function @xmath68 depends on @xmath0 , and is given in steps of log [ @xmath0 ( k ) ] = 0.1 in atomdb ( smith et al .
2001 ) . ] and elemental abundances ( table 6 ) .
these models serve to illustrate how values derived from the line ratios affect the global fit to the spectrum and to show whether abundance determinations are robust . for all four models ,
the first - order heg and meg spectra are fit to a set of variable abundance apec models , using _
sherpa _ and applying different constraints to the fitted parameters .
all models have acceptable goodness - of - fit values .
model a has two @xmath0 components and a single absorber .
model a should be appropriate for comparison with other x - ray spectra of cool stars obtained at lower spectral resolution .
lccccc @xmath711 ( mk ) & [email protected] & 2.5 & 2.5 & 2.5 + em1 ( @xmath72 @xmath6 ) & [email protected] & [email protected] & [email protected] & [email protected] + @xmath712 ( mk ) & [email protected] & [email protected] & 12.6 & 12.6 + em2 ( @xmath72 @xmath6 ) & [email protected] & [email protected] & [email protected] & [email protected] + @xmath2 ( @xmath73 @xmath46 ) & [email protected] & 1.0 & 1.0 & 1.0 + n ( 8.05 ) & [email protected] & [email protected] & [email protected] & [email protected] + o ( 8.93 ) & [email protected] & [email protected] & [email protected] & [email protected] + ne ( 8.09 ) & [email protected] & [email protected] & [email protected] & [email protected] + mg ( 7.58 ) & [email protected] & [email protected] & [email protected] & [email protected] + si ( 7.55 ) & [email protected] & [email protected] & [email protected] & [email protected] + fe ( 7.67 ) & [email protected] & [email protected] & [email protected] & [email protected] model b is also a two-@xmath0 model , but with constraints imposed from the line ratios . the lower @xmath0 is fixed at 2.5 mk , and the single absorber @xmath2 is fixed at 1.0 @xmath74 @xmath46 ( a rough average of the values obtained from line ratios ) .
the @xmath1-sensitive forbidden and intercombination line regions are excluded from the fit .
table 6 also gives results from multicomponent models c and d , each with the low @xmath0 and a single @xmath2 fixed , as in model b. assuming that the hotter component is coronal in nature , a broad distribution of @xmath0 is reasonable , although the shape is not well constrained .
we broaden the hot component using a gaussian - shaped function centered around 12.5 mk . for model
c the entire spectrum is fit simultaneously , with only the forbidden and intercombination line regions excluded as for model b. for model d the emission measures for the two components are fit to line - free regions , identified visually and using apec models .
the thermal continuum emission fit to line - free regions is strongly temperature - dependent and thus constrains the emission measure distribution .
the abundances are then fit to narrow regions centered on the strong lines .
these separate fitting procedures are iterated until the values stop changing .
the resulting model d em distribution is shown in figure 7 .
the larger statistical error for the low @xmath0 emission measure in model d compared with model c ( table 6 ) stems from the smaller number of bins used in the fits ; however , we expect the systematic errors for model d to be lower , given that we are only using bins whose information content is secure , and thus we choose model d to illustrate the model spectrum . by almost a factor of 2 .
we attribute these difficulties primarily to the complexity of the absorption , and secondarily to the few constraints on the shape of the em distribution . ]
figure 8 compares the model d spectra predicted by the broad `` coronal '' component and the soft `` accretion '' component with the observed spectra .
table 1 lists the origin of each line as accretion or corona based on model d. an additional component with @xmath0 lower than 2.5 mk can be added to both models c and d but is not required , as it does not improve the fit significantly .
similar results are obtained with somewhat different choices of width and peak @xmath0 for the hot component . since the line ratios indicate increasing @xmath2 with higher charge state , we tried adding a second absorber to the high @xmath0 gaussian component for models b , c , and d. robust solutions are difficult to obtain , as the absorption , abundances , and normalizations of the emission measures do not independently determine the global x - ray spectrum .
we also explored the possibility that the low and high @xmath0 components have different metal abundances but again , without robust results .
using @xmath0=2.5 mk , as derived from the line ratio , instead of @xmath0=3.58 mk as found in model a , has important implications for interpreting the formation of some of the emission lines .
the ratio of the strong he@xmath51 and ly@xmath51 line emission drives the model a fit to more than 3.0 mk with very little absorption . specifically , only @xmath315% of the ly@xmath25 emission arises from the 2.5 mk component , whereas a 3.58 mk component can produce more than half of the emission .
the 2.5 mk component produces essentially all of the , , , and emission and more than half the emission from .
the fluxes from these lines formed at 2.5 mk show no increase over their average value during the flare , as shown in figure 1 , indicating that the flare is associated with the hotter corona rather than the accretion shock , and justifying its retention in our analysis .
lines from , , ( fig . 1 ) , and are produced by the hotter component and may participate in the flare . while k02 found that was overpredicted by their model by a factor of about 3 , our models b , c , and d have no such problem .
in fact , for models c and d , the small flux of the resonance line forces the emission measure to be negligible between @xmath33 and 5 mk ( fig .
the relative elemental abundances given in table 6 are reasonably consistent for the different models .
the abundances are also similar to those found by k02 , stelzer & schmitt ( 2004 ) , and drake et al .
it is important to note that the continuum emission arises primarily from the hot component , except at the longest wavelengths ( fig .
thus absolute abundances in the low @xmath0 component for ne , o , fe and other species can not be determined reliably .
nevertheless , the extremely large ne / o abundance ratio appears to be a very robust result .
differences in the absolute abundances derived here reflect both differences in @xmath0 and the degeneracy between emission measure and @xmath2 .
for example , nitrogen and oxygen are formed entirely in the low temperature component and thus the abundance differences among the models reflect differences among the low temperature emission measure and @xmath2 , as does the difference in the neon abundance between models a and b. the difference in the neon abundance between models c and d also reflects the differences in their emission measures at 2.5 mk .
weak emission is apparent from the s and ar complexes , but we have not included line fluxes from these elements in table 1 because the individual lines ( in particular the he - like resonance line ) can not not be cleanly determined in the four grating arms .
instead we have used the model c global fit to obtain 0.21 @xmath17 0.09 and 1.41 @xmath17 0.49 times solar ( anders & grevesse 1989 ) for the s and ar abundances , respectively .
a high ar / o abundance would add weight to drake et al.s ( 2005 ) suggestion that the accretion stream is depleted of grain - forming elements ; however , the high ar abundance is unfortunately not statistically significant . in this section ,
we discuss velocity constraints from the line measurements .
the measurements of the emission line fluxes given in table 1 use gaussian lines in addition to the standard hetg calibration line spread function .
table 7 gives the lines with the highest signal - to - noise ratios , for which gaussian widths are measured .
lines not listed do not show statistically significant widths , but are consistent with the range of widths found here .
for these fits the continuum level is allowed to vary to provide a better local fit . assuming the lines form in the same region , and using a flat continuum flux level from 13.45 to 13.75 , we force the 3 triplet lines to have the same gaussian width to improve the significance of the velocity measurement .
figure 2 shows the best fit for the lines . as before ,
the centroids of the lines measured in each grating arm are fit independently .
we thus obtain a velocity of 183 @xmath17 16 km s@xmath14 .
the thermal doppler fwhm velocity for neon at 2.5 mk is 75 km s@xmath14 .
the error in the calibration is estimated to be about 5% .
thus we determine a turbulent velocity of 165 @xmath17 18 km s@xmath14 , including calibration error in quadrature with the 1@xmath18 statistical error .
the line profiles are consistent with each other within the errors and suggest a range of velocities from zero up to about the sound speed of the gas ( 185 km s@xmath14 at 2.5 mk ) .
the turbulent velocity is well below the preshock gas velocity of @xmath3500 km s@xmath14 .
line centroids are consistent with reference wavelengths from atomdb to within the calibration uncertainty on the absolute wavelength scale .
lrcccc & 10.239 & 0.0042 @xmath17 0.0030 & 122 @xmath17 87 & 75 & 96 + & 11.544 & 0.0024 @xmath17 0.0024 & 62 @xmath17 62 & 75 & 0 + & 12.134 & 0.0082 @xmath17 0.0009 & 203 @xmath17 23 & 75 & 189 + & 13.447 & 0.0103 @xmath17 0.0010 & 229 @xmath17 22 & 75 & 216 + & 13.553 & 0.0057 @xmath17 0.0014 & 126 @xmath17 32 & 75 & 101 + & 13.699 & 0.0073 @xmath17 0.0019 & 159 @xmath17 41 & 75 & 140 + & 15.014 & 0.0065 @xmath17 0.0040 & 129 @xmath17 80 & 45 & 121 + & 18.969 & 0.0074 @xmath17 0.0026 & 116 @xmath17 41 & 85 & 79 + we summarize here the physical conditions of the x - ray plasma from a consistent analysis of the spectra .
the new diagnostic measurements strongly support k02 s argument that the low temperature x - ray component ( @xmath75 2.5 mk ) arises in the accretion shock . while the density we derive from , @xmath76 @xmath6 , is within the range found in active stars at higher @xmath0 ( @xmath3 6 mk ) , it is at least an order of magnitude larger than the few r - ratios reported ( huenemoerder et al .
2001 ; ness et al . 2002 ; osten et al .
the @xmath1 derived from the r - ratio is also more than an order of magnitude larger than derived from in other cool stars ( see sanz - forcada et al .
2003 ; ness et al . 2004
; testa et al . 2004 ) .
we also measure the r - ratio and rule out photoexcitation , since the observed r - ratio can only be produced by high @xmath1 , namely @xmath77 @xmath6 . the g - ratios from , , and give similar @xmath78 mk , significantly lower than the @xmath0 of peak emissivity for ( 4 mk ) and ( 6.3 mk ) . the g - ratio from is particularly reliable because of good statistics and accurate atomic data . in our multi-@xmath0 models ( models c and d ) , this low @xmath0 component is isolated and does not connect continuously with the hotter , coronal component .
active main - sequence stars tend to have emission measures rising up to 6 mk and above , such that , if biased at all , @xmath0-sensitive line ratios should be biased toward higher rather than lower temperature .
thus the @xmath0 measurements also support the accretion shock scenario .
more accurate g - ratio models for and are required to establish any @xmath0 differences among these ions .
we also measure strong absorption using line ratios from several ions , and require at least two different absorbing column densities , with the higher charge states experiencing more absorption .
it seems likely that the gap between the two components of the em distribution is caused by neutral h absorption of the soft x - ray spectrum that is produced by the hot coronal component .
the pattern of absorption is reminiscent of the `` two - absorber x - ray sources '' reported by gdel et al .
( 2007b ) , where the coronal component has ten times more absorption than the soft component ; however , unlike tw hya , these sources have high accretion rates and are believed to drive jets .
we note that the absorption in tw hya is identified here from line ratio diagnostics , whereas the absorption in gdel et al .
( 2007b ) was determined from ccd spectra .
we were not able to determine multiple absorbers using standard global fitting methods , even with the high resolution spectrum presented here , because the continuum spectrum has low signal - to - noise ratio .
on the other hand , it seems likely that with low spectral resolution , multiple absorbers could also be easily missed .
thus the presence of more than one absorber may be a more universal characteristic than we can presently establish without reliable line ratio diagnostics .
all the values of @xmath2 found here are higher than those found using the hydrogen ly@xmath49 profile , for which a conservative upper limit is @xmath79 @xmath46 ( herczeg et al .
2004 ) , indicating that the absorption is not due to the interstellar medium but is intrinsic to the stellar system .
our column densities are consistent with the range of values previously reported from rosat and asca ( kastner et al .
in fact the two different values from rosat and asca are consistent with our different values and may reflect the complex absorption coupled with the different instrument responses rather than time - variable absorption .
as noted also by gdel et al .
( 2007b ) for the highly absorbed coronae of dg tau a , gv tau , dp tau , our x - ray measurements are higher than expected from the optical extinction assuming standard gas - to - dust ratios . for x - ray emission that is highly localized compared with the stellar ly@xmath25 , and preferentially absorbed by the accretion stream directly above it ,
a difference in @xmath2 derived by the two methods seems entirely reasonable .
we can test the hypothesis that some of the measured x - ray emission comes from the accretion shock by constructing a standard one - dimensional ( 1d ) model of the magnetospheric infall , shock heating , and postshock cooling .
this model is a simplified version of similar 1d models in the literature ( calvet & gullbring 1998 ; gnther et al .
2007 ) , and is tailored to specific measured properties of tw hya .
the magnetospheric accretion is assumed to follow a set of dipole magnetic field lines that thread the accretion disk .
we use knigl s ( 1991 ) expression for the inner truncation radius of the disk to determine the distance at which parcels of gas are launched from rest .
this expression depends on the mass accretion rate @xmath80 , the surface magnetic field strength @xmath81 , and the stellar mass @xmath82 and radius @xmath83 . for tw hya , we take @xmath84 and @xmath85 ( batalha et al .
2002 ) . as in cranmer ( 2008 ) , we assume a canonical t tauri star magnetic field strength of @xmath86 g. a range of accretion rates for tw hya has been reported , from @xmath87 yr@xmath14 ( muzerolle et al .
2000 ) to @xmath88 yr@xmath14 ( batalha et al . 2002 ) .
the low end of this range produces the best agreement with the observations presented here , so the muzerolle et al .
accretion rate will be adopted for the cooling - zone models below . with the above values , the knigl ( 1991 ) truncation radius of the disk
is found to be approximately @xmath89 , which implies a ballistic free - fall velocity at the stellar surface of @xmath90^{1/2 } \ , \approx 509 \,\
, \mbox{km } \,\ , \mbox{s}^{-1 } \,\ , .\ ] ] since the preshock gas is often assumed to have a sound speed only of order @xmath11 km s@xmath14 ( e.g. , muzerolle et al .
2001 ) , the highly supersonic flow should produce a strong shock at the stellar surface , heating the plasma to a postshock temperature of @xmath91 , where @xmath92 is the boltzmann constant and @xmath93 is the mean atomic mass .
for the above parameters , this value is @xmath94 3.4 mk . [ we neglect any preshock heating due to photoionization since the x - ray luminosity is relatively low . see calvet & gullbring ( 1998 ) . ]
the preshock mass density @xmath95 in the accretion stream can be estimated via mass flux conservation at the surface , i.e. , @xmath96 where the @xmath97 is the filling factor of the stellar photosphere heated by the accretion stream .
the value of @xmath97 was estimated to be 1.1% for the dipole magnetic geometry assumed above ( see also cranmer 2008 ) .
assuming complete ionization ( with 10% he by number ) , the preshock electron number density @xmath98 @xmath6 .
because the shock is strong , we obtain an immediate postshock density @xmath99 @xmath6 ( i.e. , a factor of four increase across the shock , with the assumed adiabatic index of @xmath100 ) .
the postshock velocity @xmath101 is thus approximately 130 km s@xmath14 .
as one proceeds deeper through the cooling zone , the density increases and the velocity and temperature decrease , until a depth is reached at which there is no longer any influence from the shock .
the `` bottom '' density @xmath102 can be estimated using ram pressure balance between the accretion stream and the unperturbed atmosphere . using equation 2 the pressure of the accretion stream
is given by @xmath103 ( e.g. , hartmann et al .
1997 ; calvet & gullbring 1998 ) . assuming the accretion is stopped in the first few scale heights above the photosphere ( at which radiative equilibrium drives the temperature to @xmath104 ) , the pressure of the atmosphere is @xmath105
by equating the two pressures , the density can be found by solving for @xmath106 with @xmath107 k for tw hya ( batalha et al . 2002 ;
see also cranmer 2008 )
. converting to electron density , this corresponds to @xmath108 @xmath6 , a factor of 1000 higher than the immediate postshock density . finally , we computed the physical depth and structure of the postshock cooling zone using the analytic model of feldmeier et al .
( 1997 ) , which assumes that the radiative cooling rate @xmath109 has a power - law temperature dependence ( @xmath110 ) .
the thickness of the cooling zone is proportional to @xmath111 , and for the case of the tw hya parameters used here , it has a value of 385 km .
the spatial dependence of density , temperature , and velocity in the cooling zone is specified by equations ( 8)(11 ) of feldmeier et al .
( 1997 ) , and the shapes of these curves resemble those of gnther et al.(2007)which were computed using a more detailed radiative cooling rate reasonably well ( see fig .
note that choosing the stronger mass accretion rate of batalha et al.(2002 ) would have led to a truncation radius closer to the star ( @xmath112 ) , a lower freefall velocity ( @xmath113 km s@xmath14 ) , and thus a lower postshock temperature ( @xmath114 k ) .
the densities would all have been higher than their counterparts above by about an order of magnitude , and the thickness of the cooling zone would have been smaller ( 32 km ) .
using the temperature and density in the model , we compute the line intensities as functions of distance from the accretion shock .
figure 9 shows line intensities through the region for the strongest lines of , , and .
we compare the integrated intensities to the observations . assuming the abundances determined from model d , and using @xmath115 @xmath46 , the predicted line intensities for the , and resonance lines are 0.25 , 0.16 , and 0.40 times the observed fluxes , respectively .
higher abundances improve the agreement , as do lower column densities .
nevertheless , agreement within a factor of 4 seems quite reasonable , given the difficulty of obtaining absolute abundances and the complexity of the absorption .
the predicted g - ratios give @xmath29 2.5 and 2.2 mk for and , respectively .
the @xmath1 calculated for is @xmath116 @xmath6 , in excellent agreement with the observation ; however , for , the calculated @xmath1 is @xmath117 @xmath6 , seven times larger than observed .
the density is clearly discrepant with the standard model .
the physical conditions predicted at the shock front are in excellent agreement with the observations ; however , the predictions for the postshock cooling plasma are in stark disagreement with the observations .
most problematic is that the electron density derived from is _ lower _ than that for , when the postshock gas should be compressing as it cools and slows and hence the density from should be larger than from .
furthermore , @xmath2 determined from is four times _ smaller _ than that determined from , whereas in the standard postshock model should be observed through the same or larger column density of intervening material . in this section ,
we explore the properties of the postshock cooling plasma based on our new results . using the apec code to calculate emissivities for a relevant grid of @xmath0 and @xmath1
, we construct a set of two - component shock models for comparison with the diagnostic lines from the shock .
the two components represent region 1 , the higher @xmath0 region near the shock front , and region 2 , the lower @xmath0 region of postshock plasma .
we do not discuss the hot coronal plasma here , but consider the diagnostic lines from ions that are formed below 3 mk , in particular , , , and .
each component in the model has four parameters : @xmath0 , @xmath1 , @xmath2 , and volume @xmath65 .
the two regions experience different absorption with a factor of 2.5 larger @xmath2 at higher @xmath0 , as required by the resonance series lines .
the abundances are taken from the model d fits .
the predicted fluxes from the two regions are summed for comparison with the observations .
table 8 gives the parameters for the best - fitting model , the predicted fluxes for each of the two regions , and the ratios of predicted to observed line fluxes .
for , , and the predictions agree to better than 30% with the observed line fluxes .
for the predictions are low by a factor of less than 2 , but the line ratios agree within 30% .
the diagnostic line fluxes place tight constraints on the model .
lcccc region 1 & @xmath118 & 3.00 & @xmath119 & @xmath120 + region 2 & @xmath121 & 1.75 & @xmath122 & @xmath123 + + & r & i & f & ly@xmath49 + o & 260 . & 166 . & 3.6 & 512 .
+ ne & 447 . & 260 . & 73.9 & + mg & 3.6 & 1.3 & 1.8 & + + & r & i & f & ly@xmath49 + o & 124 . & 70.8 & 36.7 & 25.1 + ne & 16.1 & 4.8 & 10.1 & + mg & 0.0 & 0.0 & 0.0 & + + & r & i & f & ly@xmath49 + o & 1.00 & 0.69 & 0.53 & 0.69 + ne & 1.31 & 0.85 & 0.75 & + mg & 1.28 & 1.14 & 1.13 & in this simple model , all of the emission from and most of the emission from come from region 1 , which we associate with the shock front . if we assume a cooling length of 200 km , based on the accretion shock model described above ( see fig .
9 ) , and use the derived emitting volume of @xmath124 @xmath125 , we obtain an area filling factor for region 1 of 1.5% of the surface area of the star .
region 2 has a larger volume than region 1 by a factor of 300 as determined primarily from .
if we assume that the cooling zone is the continuation of region 1 , we again obtain a length of about 200 km ; however , this short length implies a filling factor larger than the stellar surface area and is not plausible .
instead , we use the scale height of 0.1 @xmath126 for 1.75 mk plasma to estimate a filling factor of 6.8% for region 2 , a value in good agreement with estimates from the ultraviolet ( costa et al . 2000 ) and optical continuum veiling ( batalha et al .
2002 ) . scaling the mass as @xmath127 implies that the mass of region 2 is 30 times the mass of region 1 .
romanova et al . (
2004 ) have used three - dimensional magnetohydrodynamic ( mhd ) simulations to show that the hot spots formed on the surface by accretion are inhomogeneous , with the lower density and temperature regions filling a larger surface area than the denser central region , seemingly suggestive of region 2 .
however , the larger mass of region 2 also implies a higher accretion rate than for region 1 , and thus a larger density , in contradiction to our diagnostics .
therefore , we do not interpret region 2 as a second accretion shock with a different accretion rate , but rather as a new accretion - driven phenomena .
the impact from the accretion presumably provides the energy to heat additional material in the stellar atmosphere to over 1 mk .
very recently , two - dimensional mhd simulations have shown that violent outflows of shock - heated material can propagate from the base of the accretion shock for cases with a high thermal to magnetic pressure ratio @xmath50 ( orlando et al .
2009 ) .
one of the key concepts of the standard one - dimensional accretion model suggests a simple picture ( fig .
the field line that truncates the disk separates the open magnetic field lines of the polar region from the closed magnetic field loops in the equatorial regions .
the region heated by the shock ( our region 2 ) provides a large supply of ionized material available to form the soft x - ray emitting corona presented here , a warm wind ( dupree et al .
2005 ) , and soft x - ray jets ( gdel et al .
this corona exists because magnetic loops confine it , but it is fed by the accretion process and is thus unlike stellar coronae in dwarf stars on the main sequence .
figure 10 illustrates the shock and its surroundings .
this model provides an explanation for the soft x - ray excess discovered in the _ xmm - newton _ extended survey of the taurus molecular cloud ( xest ; gdel et al .
2007c ; telleschi et al .
grating spectra show that this excess emission manifests itself in ctts as enhanced relative to as compared with weak - line t tauri ( wtts ) and main - sequence stars ( gdel & telleschi 2007 ; robrade & schmitt 2007 ) .
our measurements also demonstrate that the postshock cooling gas can , in some cases , dilute the signature of high @xmath1 and explain why not all accreting systems show high densities .
having carefully distinguished between the accretion and corona in section 3 , we reconsider the assumption that these two components arise from different processes .
gdel & telleschi ( 2007 ) find that the soft x - ray excess depends both on the level of magnetic activity as measured by the total ( including coronal ) @xmath12 and on the presence of accretion . while we suggest that the role of accretion is to heat and ionize the surrounding stellar atmosphere to create a soft x - ray ( postshock cooling ) plasma , it seems possible that this plasma could be further heated to @xmath128 mk , either by the usual coronal heating ( mhd ) processes , or possibly by the accretion energy .
cranmer ( 2009 ) finds that in some systems , accretion can provide the energy for coronal heating to such high temperatures .
studies of star forming regions have found that young stars show strong x - ray activity , with energetic flares and extremely high x - ray emitting temperatures ; however , the increase in activity with rotation rate breaks down for the accreting systems , leading to the conjecture that the stellar dynamo might operate differently in these systems ( preibisch et al . 2005 ) .
we suggest that the accretion process may play a more prominent role than has generally been believed . while x - ray signatures of the accretion shock itself may not usually dominate the emission , the energy from accretion may in fact contribute to feeding and heating a new type of corona .
we have analyzed the deep _ chandra _ hetg spectrum of tw hya to assess the relative contributions of accretion and coronal emission .
we summarize the most important of our findings : * the spectrum shows line and continuum emission from @xmath128 mk coronal plasma , which may have a broad emission measure distribution . unlike active main sequence stars , the coronal em distribution abruptly cuts off below @xmath129 mk , probably due to absorption by the accretion stream . *
the spectrum also shows emission lines from plasma at @xmath130 mk , as predicted by standard models of the accretion shock . @xmath0 and @xmath1 diagnostics from he - like are in excellent agreement with accretion models for tw hya . *
we measure the flux of the forbidden line in tw hya for the first time to better than 3@xmath18 , allowing a determination of electron density . this density is lower than @xmath1 from by a factor of four and lower than @xmath1 from the standard postshock cooling model by a factor of seven , and leads to a new model of an accretion - fed corona . *
the high @xmath0 ions suggest a stellar corona .
by contrast , line ratio diagnostics from the lower @xmath0 ions , indicate two regions with different densities . in our model , the accretion shock itself has @xmath131 mk and @xmath132 @xmath6 , while the postshock cooling region has @xmath133 mk and @xmath134 @xmath6 .
the postshock plasma has 30 times more mass than the shock itself .
the surface area filling factor of the shock is 1.5% while the postshock filling factor is approximately 6.8% .
* line ratio diagnostics require at least two different absorbers , with higher column density ( @xmath2 ) for higher charge state .
absorption by the accretion stream is larger by a factor of 2.5 for the shock than for the postshock region . *
a range of models for the em distribution and elemental abundances provides acceptable fits to the global spectrum .
while absorption , emission measure , and abundances are not independently determined by such methods , the high ne / o abundance seems to be a robust result . * from line profile analysis of lines
we determine a subsonic turbulent velocity of about 165 km s@xmath14 .
this high resolution x - ray spectrum of tw hya presents a rich set of diagnostic emission lines for characterizing the physical conditions of the high energy plasma and understanding the dominant physical processes .
the observations strongly support the model of an accretion shock producing a few mk gas at high density , as first suggested by kastner et al .
( 2002 ) from an earlier hetg spectrum .
the diagnostics also support the role of accretion in producing the soft x - ray excess at previously discovered ( gdel & telleschi 2007 ; robrade & schmitt 2007 ) .
this excess emission requires that the accretion shock heat a large volume in the surrounding stellar atmosphere . in tw hya the filling factor at is larger than the filling factor of the shock at , and is consistent with the filling factors of the ultraviolet and optical continua . both open and closed magnetic field lines may emerge from this surrounding area to channel this ionized gas back into the corona .
while it is not yet clear whether the energy from the impact of accretion at the star can heat a hot ( 10 mk ) corona , accelerate a highly ionized stellar wind , or drive hot jets , we suggest that accretion - fed coronae may be ubiquitous in accreting young stars .
we acknowledge support from nasa to the smithsonian astrophysical observatory ( sao ) under _ chandra _ go7 - 8018x for gjml .
nsb and sjw were supported by nasa contract nas8 - 03060 to sao for the chandra x - ray center .
src s contribution to this work was supported by the sprague fund of the smithonian institution research endowment , and by nasa grant nng04ge77 g to sao .
we thank randall smith for supporting the customized apec code runs .
we thank the _ chandra _ mission planning team for their efforts to accommodate the ground - based observing campaign . | we present x - ray spectral analysis of the accreting young star tw hydrae from a 489 ks observation using the _ chandra _ high energy transmission grating .
the spectrum provides a rich set of diagnostics for electron temperature @xmath0 , electron density @xmath1 , hydrogen column density @xmath2 , relative elemental abundances and velocities and reveals its source in 3 distinct regions of the stellar atmosphere : the stellar corona , the accretion shock , and a very large extended volume of warm postshock plasma .
the presence of , , and emission lines in the spectrum requires coronal structures at @xmath310 mk .
lower temperature lines ( e.g. , from , , and ) formed at 2.5 mk appear more consistent with emission from an accretion shock . he - like line ratio diagnostics indicate that @xmath4mk and @xmath5 @xmath6 in the shock .
these values agree well with standard magnetic accretion models .
however , the _ chandra _ observations significantly diverge from current model predictions for the postshock plasma .
this gas is expected to cool radiatively , producing as it flows into an increasingly dense stellar atmosphere .
surprisingly , indicates @xmath7 @xmath6 , five times lower than @xmath1 in the accretion shock itself , and @xmath3 seven times lower than the model prediction .
we estimate that the postshock region producing has roughly 300 times larger volume , and 30 times more emitting mass than the shock itself .
apparently , the shocked plasma heats the surrounding stellar atmosphere to soft x - ray emitting temperatures and supplies this material to nearby large magnetic structures which may be closed magnetic loops or open magnetic field leading to mass outflow .
our model explains the soft x - ray excess found in many accreting systems as well as the failure to observe high @xmath1 signatures in some stars .
such accretion - fed coronae may be ubiquitous in the atmospheres of accreting young stars . |
recent results from reverberation - mapping of the broad emission - line regions ( blr ) in agn indicate that the blr distance from the central radiation source roughly scales as @xmath1 ( peterson 1995 ) .
recently we have elaborated a different method for estimating the blr distance , using the emission - line photoionization model . in order to explain the anticorrelation between the h@xmath2 line width and the soft x - ray spectral slope in narrow - line seyfert 1 galaxies ( nls1 ) ( boller , brandt and fink 1996 ; wang , brinkmann and bergeron 1996 ) , wandel and boller ( 1997 ) showed that a steeper spectrum would produce a blr at a relatively larger distance from the central source ,
so the keplerian velocity is lower .
this modified blr distance calculated from the photoionization relation agrees well with the reverberation distance ( wandel 1996 ; 1997 ) . in this work
we connect this model to the r(blr ) @xmath3 relation by independently determining the central mass from the x - ray variability .
if the emission lines are doppler - broadened by keplerian motion in the gravitational potential of the central mass , the full width at half maximum is given by : @xmath4 where m is the mass of the central black hole and r the radius of the emission line region .
the physical conditions in the ionized gas emitting the broad lines are characterized by the ionization parameter u , the ratio of ionizing photons to electrons @xmath5 where @xmath6 is the luminosity of the central source , per unit energy , @xmath7 is the electron density .
the radius of the blr may then be written as @xmath8 where @xmath9 is the ionizing luminosity , and @xmath10 is the mean energy of the ionizing photons .
typical values in agn clouds give @xmath11 ( cf .
rees , netzer & ferland 1989 ) . for @xmath12
( 2 ) gives @xmath13 where @xmath10 is in rydbergs .
assuming the ionizing spectrum has a power - law shape with the spectral index of the soft x - ray band ( found e.g. from the rosat data ) , @xmath14 , gives the observed anti - correlation between the line width and the spectral index ( wandel and boller 1997 ) .
wandel ( 1996 ; 1997 ) has shown that the blr radius calculated from eq .
( 2 ) above does indeed agree well with the radius estimates available from reverberation calculations for about a dozen objects ( see kaspi et.al .
1996 and these proceedings ) .
we present an independent derivation of the radius - luminosity relation , determining the mass from x - ray variability .
an upper limit for the black hole mass is given by ( e.g. wandel & mushotzky 1986 ) @xmath15 where @xmath16 is in seconds .
this relation assumes that the bulk of the x - ray continuum is emitted within 5 schwarzschild radii . using the doubling time as the characteristic time for variability we determine upper limits for the black hole mass for those objects whose continuum luminosities
have been observed to vary significantly .
1 gives the doubling time ( we have extrapolated amplitude variations linearly to a factor of 2 to determine @xmath16 ) versus the observed fwhm .
although there is some scatter , there is a strong indication for an increase of the doubling time with observed fwhm .
this may indicate a relation between the central mass and the line width .
we may use eq .
( 1 ) to establish an estimate for the blr radius calculated from the observed line width and x - ray variability , independent of the luminosity , which can test the r - l@xmath17 relation : @xmath18 note that this is an upper limit , as is eq . ( 4 ) .
the diamonds in figure 2 show the blr radius calculated from eq .
( 5 ) vs. the x - ray luminosity for the objects in our sample with significant x - ray variability and a few objects from wandel and mushotzky ( 1986 ) .
we see that there is no correlation between r and l. on the other hand , as expected , the blr radius calculated from the photoionization model ( eq .
3 ) shows a clear @xmath19 correlation ( circles in figure 2 ) , which is confirmed independently by reverberatiom mapping .
looking at figure 2 we may conclude that the x - ray variability method systematically overestimates the central mass for the low luminosity nls1 galaxies . | we developed a new method of determination of the size of the broad emission - line region ( blr ) in active galactic nuclei .
this method relates the radius of the broad - line region of agn to the soft x - ray luminosity and spectral index . comparing the blr distances calculated from our model to the blr distances determined by reverberation mapping
shows that our scaling law agrees with the @xmath0 empirical relation .
here we investigate a complimentary method of estimating the blr distance - based on the keplerian broadening of the emission lines and the central mass estimated from x - ray variability . |
Subsets and Splits