article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
in the coming years , one of the main tasks in particle physics is the understanding of the mechanism of the electroweak symmetry breaking .
after the experimental determination of the higgs boson mass , the higgs potential is fully fixed in the standard model .
however , it is very important to independently measure the self - coupling of the higgs boson , which can be obtained from studying the production of higgs boson pairs . since the corresponding cross section is @xmath2 times smaller than the one for single higgs boson production , higgs boson pair production poses a challenging problem for the lhc , even after the luminosity upgrade around 2020 .
there are a number of phenomenological analyses which investigate the possibility to extract the self coupling from cross section measurements .
first studies have been performed more than 15 years ago @xcite .
since the discovery of the higgs boson there has been an increasing interest in this topic and a number of refined analyses have been performed , see , e.g. , refs .
@xcite .
higgs boson pairs can be produced via the fusion of two partons or vector bosons , via the radiation off vector bosons , or in association with heavy quarks .
similar to single higgs boson production , the numerically dominant mechanism is gluon fusion although the leading order ( lo ) contribution is loop - suppressed . due to the larger yukawa coupling
, the dominant contribution comes from top quark loops in the standard model . for this reason we concentrate in this paper on such contributions . for the lo order corrections the exact dependence on the top quark mass and the center - of - mass energy
is known @xcite . at
next - to - leading ( nlo ) qcd corrections have been computed for the first time more than 15 years ago @xcite in the infinite top quark mass limit using an effective theory .
finite top quark mass effects have been investigated in ref .
@xcite where a systematic expansion in the inverse top quark mass has been applied and a quantitative estimate of the quark mass effects has been provided .
it has been estimated that they do not exceed @xmath3 of the nlo contribution .
finite top quark mass effects have also been considered in ref .
@xcite where the exact real radiation contribution is combined with the effective - theory virtual corrections . as a result , a reduction of about @xmath4 of the cross section is obtained .
we will comment in section [ sec::nlo ] on this issue . within the effective theory
also next - to - next - to - leading ( nnlo ) contributions are available @xcite . in this context
it is interesting to note that the three - loop matching coefficient of the effective operator for two higgs bosons and two , three or four gluons is different from the one for single higgs boson production @xcite .
the results for the virtual corrections obtained in ref .
@xcite have been cross - checked in ref .
@xcite where the calculation has been performed without reference to the effective theory .
the resummation of threshold - enhanced logarithms to next - to - next - to - leading logarithmic ( nnll ) accuracy has been performed in refs .
@xcite .
the remainder of this paper is organized as follows : in the next section we review the construction of the soft - virtual approximation for the production cross section and discuss two options to factorize the exact lo result from the higher order contributions .
we argue that a factorization at the level of the differential cross section w.r.t .
the higgs boson pair invariant mass leads to more stable results .
afterwards we reconsider in section [ sec::nlo ] the top mass corrections at nlo .
virtual nnlo corrections including finite top quark mass effects are computed in section [ sec::nnlo ] .
they are used in section [ sec::impr_nnlo ] to present phenomenological results for higgs pair production up to nnlo .
section [ sec::con ] contains our conclusions .
we write the perturbative expansion of the partonic cross section for the production of higgs boson pairs in the form @xmath5 where @xmath6 , @xmath7 is the partonic center - of - mass energy and @xmath8 . since the quark - induced channels are numerically small @xcite we consider in this paper only the @xmath9 channel .
we use the variable @xmath10 to parametrize the dependence of the cross section on the higgs boson and top quark mass .
we renormalize the top quark mass in the on - shell scheme .
furthermore , we set the factorization and renormalization scale equal to each other and write @xmath11 . for later convenience
we introduce for @xmath12 @xmath13 and denote the sum of the first two terms on the right - hand side by @xmath14 .
finite top quark mass effects to @xmath15 at nlo have been considered for the first time in ref .
the applied method is based on `` reversed unitarity '' @xcite where , with the help of the optical theorem , the imaginary part of forward scattering amplitudes are computed to obtain the total cross section .
the virtual corrections have also been computed by directly considering the @xmath15 amplitude .
expansion terms up to order @xmath16 of the nlo contribution to @xmath17 have been computed @xcite .
the factorization of the exact lo corrections has then been implemented at the partonic level for the total cross section using @xmath18 where @xmath19 contains the exact dependence on @xmath20 and @xmath21 . in ref .
@xcite , where the nlo result has been computed for the first time in the heavy top quark limit , a different approach has been applied .
actually , the exact lo result has been factorized before the integration over the higgs pair invariant mass . in this approach
the ( total ) nlo cross section reads @xmath22 where exact and exp reminds whether exact or ( in @xmath21 ) expanded results are used .
@xmath23 is the invariant mass squared of the higgs boson pair .
expressed in terms of @xmath24 it is possible to re - write eq .
( [ eq::nlo_fact1 ] ) as @xmath25 from the comparison of eqs .
( [ eq::nlo_fact2 ] ) and ( [ eq::nlo_fact3 ] ) one expects that ( [ eq::nlo_fact2 ] ) leads to better results since the differential factorization ( df ) in eq .
( [ eq::nlo_fact2 ] ) results in a better - behaved integrand , in particular for @xmath26 . for the virtual corrections , which are proportional to @xmath27 , one has immediate access to the @xmath23 dependence and the df of eq .
( [ eq::nlo_fact2 ] ) can be applied . ) and ( [ eq::nlo_fact2 ] ) are equivalent . ]
the real corrections , however , are obtained with the help of the optical theorem which directly leads to the total cross section and thus eq .
( [ eq::nlo_fact2 ] ) can not be used . for the construction of the soft - virtual ( sv )
approximation , which is discussed below , we need in addition to the virtual term also the contributions from soft gluon emission .
since the soft contributions are proportional to the lo born cross section eq .
( [ eq::nlo_fact2 ] ) can immediately be applied to the sv cross section . in the following
we discuss the construction of the sv approximation for the cross section ( see also ref .
@xcite ) . for simplicity
we consider for the following schematic reasoning the total cross section @xmath28 .
note , however , that the arguments also hold for @xmath29 . in a first step we split @xmath28 according to @xmath30 where the two terms on the right - hand side correspond to the virtual corrections ( including ultra - violet counterterms ) and the real corrections ( including the contributions from the factorization of initial - state singularities ) .
they are individually divergent and only the sum is finite . in the next step we re -
organize the terms on the right - hand side such that @xmath28 can be written as @xmath31 where `` sv '' refers to `` soft - virtual '' and `` h '' to `` hard '' .
this is achieved by splitting @xmath32 into a divergent and a finite term and separating @xmath33 into a ( divergent ) soft and ( finite ) hard contribution .
the soft contribution is combined with @xmath32 to obtain @xmath34 such that @xmath35 @xmath34 and @xmath36 are separately finite .
@xmath37 is constructed following ref .
@xcite ; explicit expressions can be found in refs .
note that the finite part is constructed solving the equation @xmath38 for @xmath39 . in this paper @xmath32
is computed including top quark mass effects .
mass effects are automatically taken into account in @xmath37 and @xmath40 since these contributions are proportional to the exact lo cross section @xcite .
note that eqs .
( [ eq::sig_split ] ) , ( [ eq::sig_sv_h ] ) , ( [ eq::sig_sv_h_2 ] ) and ( [ eq::sig_fin ] ) also hold for the differential cross section @xmath41 and thus a factorization as suggested in eq .
( [ eq::nlo_fact2 ] ) can be performed for the soft - virtual contribution .
to be more specific we present the nlo and nnlo differential version of eq .
( [ eq::sig_fin ] ) which reads @xcite @xmath42 \frac{{\rm d}\sigma^{(0)}}{{\rm d}t } \ , , \nonumber\\ & \frac{{\rm d}\sigma_v^{(2)}}{{\rm d}t } = \frac{{\rm d}\sigma_{v,\rm fin}^{(2)}}{{\rm d}t } + 2 \mbox{re}\left[i_g^{(1)}\right ] \frac{{\rm d}\sigma_{v,\rm fin}^{(1)}}{{\rm d}t } + \left\ { \left|i_g^{(1)}\right|^2 + 2 \mbox{re}\left[\left(i_g^{(1)}\right)^2\right ] + 2 \mbox{re}\left[i_g^{(2)}\right ] \right\ } \frac{{\rm d}\sigma^{(0)}}{{\rm d}t } \ , .
\label{eq::sig_fin_2}\end{aligned}\ ] ] where @xmath43 with @xmath44 ( @xmath45 ) being the incoming ( outgoing ) momentum of a gluon ( higgs boson ) , @xmath46 and explicit expressions for the operators @xmath47 can be found in ref .
@xcite .
we adopt the notation from ref .
@xcite and parametrize radiative corrections to the partonic cross section via @xmath48 with @xmath49 and @xmath50 where the renormalization scale dependence in the strong coupling constant @xmath51 is suppressed . from eq .
( [ eq::g ] ) one obtains for the total cross section @xmath52 with @xmath53 in the second line of eq .
( [ eq::g_2 ] ) we split @xmath54 into soft - virtual and hard contribution . note that in our approach we do not have access to @xmath55 .
actually , at nnlo we only have @xmath56 at hand and at nlo only the heavy top expansion of @xmath57 is available to us
. explicit results for @xmath56 can be found in ref .
@xcite including higher order terms in @xmath58 specifying , however , the renormalization and factorization scale to @xmath59 . for completeness
we provide the nlo and nnlo results for @xmath56 for generic @xmath60 in the limit @xmath61 .
the results read @xmath62 - 4c_ald_0 + 8c_ad_1 \ , , \nonumber\\ g_\textrm{sv}^{(2 ) } & = d_{-1 } \bigg\{-\frac{4\pi^2}{3}c_a^2l^2 + \frac{11\pi^2}{108}n_l^2 + { \frac{\sigma^{(2)}_\textrm{fin}}{\sigma^{\rm lo}}}+ l\bigg[-\frac{1}{3}{\frac{\sigma^{(1)}_\textrm{fin}}{\sigma^{\rm lo}}}n_l \nonumber\\ & \quad + c_a\left(\left(-\frac{1}{3 } - \frac{2\pi^2}{9}\right ) n_l + \frac{11}{6 } { \frac{\sigma^{(1)}_\textrm{fin}}{\sigma^{\rm lo}}}\right)+ c_a^2\left(\frac13 + \frac{11\pi^2}{9 } - 38\zeta_3\right)\bigg ] \nonumber\\ & \quad + c_a^2\left(\frac{607}{81 } + \frac{517\pi^2}{108 } - \frac{\pi^4}{80 } - \frac{407\zeta_3}{36}\right ) + c_a\left[\frac{2\pi^2}{3}{\frac{\sigma^{(1)}_\textrm{fin}}{\sigma^{\rm lo}}}+ n_l\left(-\frac{82}{81 } - \frac{11\pi^2}{8 } + \frac{37}{18}\zeta_3\right)\right]\bigg\}\nonumber\\ & \quad + d_0 \bigg\{l^2\left(-\frac{11}{3}c_a^2 + \frac{2}{3}c_an_l\right ) + c_an_l \left(\frac{56}{27 } - \frac{4\pi^2}{9}\right ) + l\bigg[c_a^2 \left(-\frac{134}{9 } + \frac{10\pi^2}{3}\right ) \nonumber\\ & \quad + c_a \left(\frac{20}{9}n_l - 4{\frac{\sigma^{(1)}_\textrm{fin}}{\sigma^{\rm lo}}}\right)\bigg ] + c_a^2\left(-\frac{404}{27 } + \frac{22\pi^2}{9 } + 78\zeta_3\right ) \bigg\ } + d_1 \bigg[16c_a^2l^2\nonumber\\ & \quad + l\left(\frac{44}{3}c_a^2 - \frac{8}{3 } c_an_l\right ) + c_a^2\left(\frac{268}{9 } - \frac{20\pi^2}{3}\right ) + c_a\left(-\frac{40}{9}n_l + 8{\frac{\sigma^{(1)}_\textrm{fin}}{\sigma^{\rm lo}}}\right)\bigg ] \nonumber\\ & \quad + d_2 \bigg[-\frac{44}{3}c_a^2 - 48c_a^2l + \frac{8}{3}c_an_l\bigg ] + 32c_a^2d_3 \ , , \label{eq::g_sv}\end{aligned}\ ] ] where @xmath63 are qcd colour factors , @xmath64 is the number of massless quarks , @xmath65 and @xmath66_+ \,.\end{aligned}\ ] ] the quantities @xmath67 and @xmath68 are evaluated for @xmath69 . they are obtained from @xmath70 and @xmath71 in eq .
( [ eq::sig_fin_2 ] ) after integration over @xmath72 . in the next section
we apply the formalism described in this section at nlo and compare to the results obtained in ref .
the numerical effects at nnlo are presented in section [ sec::impr_nnlo ] using the mass corrections computed in section [ sec::nnlo ] .
in this section we restrict ourselves to nlo and compare the results of ref .
@xcite with the alternative factorization discussed in the previous section . to obtain numerical results
we employ mstw2008 parton distribution functions ( pdfs ) @xcite and consistently use n@xmath73lo pdfs to compute n@xmath73lo ( @xmath74 ) cross sections .
we assume the energy of the lhc to be @xmath75 and adopt the values of the strong coupling constants @xmath76 that we use in our computation from the mstw pdf fit : @xmath77 we renormalize the top quark in the on - shell scheme and use @xmath78 gev @xcite .
for the higgs boson we use @xmath79 gev @xcite . for numerical results shown in this section the renormalization and factorization scales
have been set to @xmath80 .
partonic cross section as a function the center - of - mass energy including various orders in the inverse top quark mass .
the dotted and dashed curves show the breakup of the complete result ( solid lines ) into soft - virtual and hard contribution.,title="fig : " ] + in fig .
[ fig::sig_part ] we show the results for the partonic cross section as a function of @xmath7 from fig .
6 of ref .
@xcite , see solid lines .
the splitting of these results into soft - virtual and hard contributions is shown as dotted and dashed curves , respectively .
for the complete result and soft - virtual approximation the infinite - top quark result corresponds to the ( red ) second line from below .
the lowest line includes in addition the @xmath81 corrections .
the curves including higher order mass corrections are above the infinite - top quark result .
the topmost curves include @xmath16 terms .
the hard contribution shows a quite flat behaviour above @xmath82 gev and exhibits smaller shifts when including top mass corrections .
furthermore , the infinite top mass result corresponds to the topmost curve and the lowest curve includes @xmath16 terms . from fig .
[ fig::sig_part ] it is evident that the hard contribution is numerically much smaller than the soft - virtual one .
partonic cross section as a function the center - of - mass energy including various orders in the inverse top quark mass .
the dashed and solid curves correspond to the factorization for the total and differential cross section , respectively .
the colour coding is taken over from fig .
[ fig::sig_part].,title="fig : " ] + in fig .
[ fig::sig_part_2 ] we compare the solid curves from fig . [ fig::sig_part ] ( which are dotted in this plot ) with the results obtained with the help of df applied to the sv approximation and adding the hard contribution as given by the dashed lines in fig .
[ fig::sig_part ] .
the relative position of the lines and the colour coding is as in fig .
[ fig::sig_part ] . for lower values of @xmath7 the two approaches
lead to comparable results .
however , the df curves have their maximum at lower values of @xmath7 and lead to a smaller cross section for larger values of @xmath7 .
furthermore , one observes a drastic improvement in the convergence behaviour when including higher order mass corrections .
in particular , for @xmath83 gev the difference between the infinite top mass result and the one including @xmath16 terms amounts to only @xmath84 fb to be compared with @xmath85 fb for the dashed curves .
partonic nlo @xmath86 factor for the factorization performed at the level of the total ( dashed ) and differential ( solid ) cross section .
, title="fig : " ] + at this point we want to stress that the splitting between the soft - virtual and hard contribution in eq .
( [ eq::sig_sv_h ] ) is arbitrary .
in fact , the soft - virtual contribution of @xmath54 , @xmath56 , is constructed for the limit @xmath87 and thus it is possible to replace @xmath56 by @xmath88 with @xmath89 ( see , e.g. , discussions in refs .
the hard contribution is modified accordingly such that the sum of @xmath90 does not change .
one observes that for @xmath91 the soft - virtual contribution approximates the partonic nlo contribution to the cross section very well .
] such that at the hadronic level the deviations are below @xmath92 .
based on this observation we use @xmath91 for the nnlo cross section where we only have the soft - virtual approximation at hand .
furthermore , better results are obtained by replacing @xmath93 in eq .
( [ eq::g_sv ] ) by @xmath94 which is justified since @xmath95 in the soft limit .
note that in ref . @xcite it has been observed that the soft - virtual approximation constructed in mellin space approximates the full ( effective - theory ) result with an accuracy of 2% .
it is interesting to look at the partonic @xmath86 factor which is defined via @xmath96 results for the two methods to factorize the exact lo term are plotted in fig .
[ fig::sig_k_part ] as a function of @xmath7 where the dashed curves are already shown in ref .
one observes that df leads to a lower @xmath86 factor and that the spread among the various @xmath21 orders is smaller .
furthermore , it is interesting to note that for df the top quark pair threshold behaviour of the lo term is not washed out in contrast to the dashed curves .
it is common to both factorization methods that there is a strong raise when approaching the threshold for higgs boson pair production ( see also discussion in ref .
@xcite ) .
nlo hadronic cross section and @xmath86 factor as a function of @xmath97.,title="fig : " ] + fig .
[ fig::sig_hadr ] shows the hadronic cross section @xmath98 for higgs boson pair production including nlo corrections as a function of @xmath99 which is a technical upper cut on the partonic center - of - mass collision energy .
it is introduced via @xmath100 where the luminosity function is given by @xmath101 @xmath102 are the gluon distribution functions in the @xmath103 scheme .
note that in the soft limit @xmath97 is a good approximation to @xmath23 .
the various lines in fig .
[ fig::sig_hadr ] correspond to the inclusion of different orders in @xmath21 at nlo . for convenience
we show on the right end of fig .
[ fig::sig_hadr ] the total cross section for @xmath104 tev .
note that the approximation used for the computation of the @xmath105 terms is not valid for large values of @xmath106 ( neither is the effective - theory result ) .
however , it can be used as an estimate of the mass correction terms . using the spread as an estimate for the uncertainty
we conclude that a finite top mass induces a @xmath107% uncertainty on top of the infinite top quark mass result . the lower panel of fig .
[ fig::sig_hadr ] shows the hadronic @xmath86 factor which is obtaind from eq .
( [ eq::k_nlo ] ) by replacing @xmath28 by @xmath98 and using nlo pdfs in the numerator and lo pdfs in the denominator .
@xmath108 raises close to threshold , however , for @xmath109 gev one observes a flat behaviour of @xmath110 ( for @xmath111 ) .
splitting of partonic cross section ( solid lines ) into real ( uppser dashed lines ) and virtual ( lower dotted lines ) contributions ( see text for details).,title="fig : " ] + top quark mass effects to double higgs boson production have also been considered in ref .
@xcite . in the approximation used in that reference the real corrections
are treated exactly , however , the infinite top quark mass approximation is used for the virtual corrections .
a decrease of the cross section by about 10% due to finite top quark mass is reported . in fig .
[ fig::sig_part_real_virt ] we split poles which cancel in the sum . in fig .
[ fig::sig_part_real_virt ] the finite contributions are plotted . ]
the partonic results for the solid lines of fig .
[ fig::sig_part ] ( which corresponds to the factorization according to eq .
( [ eq::nlo_fact1 ] ) ) into virtual corrections ( including the ultra - violet counterterms ; dotted lines ) and the real - radiation part which include the contributions from the splitting functions ( dashed lines ) .
we observe that for @xmath112 gev , the region where our approximation is valid , the top quark mass corrections to the real radiation part ( upper dashed curves ) reduce the infinite top result by about 10% , in agreement with the observations of ref . @xcite . on the other hand , the top mass effects to the virtual contribution
leads to a positive shift as compared to the effective theory result .
summing real and virtual corrections leads to an overall positive effect from top mass corrections as can be seen by the solid curves , see also fig .
[ fig::sig_part_2 ] .
note that up to @xmath113 gev top mass corrections are dominated by the @xmath81 terms .
in this section we compute the virtual corrections to the nnlo cross section including top quark mass corrections .
afterwards we construct @xmath39 as described in section [ sec::fac ] and use eq .
( [ eq::g_sv ] ) to evaluate the partonic and hadronic cross sections including @xmath114 correction terms .
we have applied two methods to compute the virtual corrections . in the first one we consider the amplitudes for @xmath15 up to three - loop order and in the second one the forward scattering amplitude is considered , which , after taking the imaginary part , directly leads to the total cross section . in both cases
we perform an asymptotic expansion for large top quark mass .
the first approach has the advantage that it is straightforward to introduce higgs boson decays whereas the second approach can immediately be applied to real corrections .
+ + + nnlo calculations require corrections up to three - loop order to the process @xmath15 .
typical contributing feynman diagrams at lo , nlo and nnlo are shown in fig .
[ fig::gghh ] .
they are generated with the help of qgraf @xcite .
note that in this approach no contributions with external ghosts have to be considered since we project on physical states .
the transformation to form @xcite code is done with the program q2e
@xcite and the asymptotic expansion for large top quark masses is realized with the help of exp @xcite . after expanding
the identified hard subgraphs in the small quantities one arrives at one - scale vacuum integrals up to three loops and massless one- and two - loop four - point diagrams with two massless and two massive external momenta .
the vacuum integrals are computed using matad @xcite . in the case of the four - point integrals we apply fire @xcite to express them as linear combinations of master integrals .
the latter are shown in fig . [ fig::gghh_ml ] where we have @xmath115 , and @xmath116 .
analytic results for all master integrals can be found in the literature ( see , e.g. , refs .
@xcite ) . in this paper
we only show results for the triangle graph in the second line of fig .
[ fig::gghh_ml ] since for our purpose the representations given in refs .
@xcite are less suited .
we use instead @xmath117 + \mathcal{o}(\epsilon^2 ) \bigg\}\ , , { \addtocounter{equation}{1}\tag{\theequation}}\label{eq::i14}\end{aligned}\ ] ] with @xmath118 @xmath119 are goncharov polylogarithms @xcite with weight @xmath120 and argument @xmath121 defined through @xmath122 with @xmath123 and @xmath124 the functions of the @xmath125 term in eq .
( [ eq::i14 ] ) can be expressed in terms of logarithms and dilogarithms via @xmath126 we have cross checked the numerical result for the @xmath125 and @xmath127 terms of @xmath128 in eq .
( [ eq::i14 ] ) against fiesta @xcite . using this method
we have computed terms up to order @xmath16 at nlo @xcite and terms up to order @xmath114 at nnlo . as an important check
we have computed the @xmath81 corrections for general qcd gauge parameter which drops out in the final expression .
one- and two - loop master integrals needed after applying asymptotic expansion to the amplitude @xmath15 .
all internal lines are massless , @xmath115 , and @xmath116.,title="fig : " ] one- and two - loop master integrals needed after applying asymptotic expansion to the amplitude @xmath15 .
all internal lines are massless , @xmath115 , and @xmath116.,title="fig : " ] one- and two - loop master integrals needed after applying asymptotic expansion to the amplitude @xmath15 .
all internal lines are massless , @xmath115 , and @xmath116.,title="fig : " ] + one- and two - loop master integrals needed after applying asymptotic expansion to the amplitude @xmath15 .
all internal lines are massless , @xmath115 , and @xmath116.,title="fig : " ] one- and two - loop master integrals needed after applying asymptotic expansion to the amplitude @xmath15 .
all internal lines are massless , @xmath115 , and @xmath116.,title="fig : " ] one- and two - loop master integrals needed after applying asymptotic expansion to the amplitude @xmath15 .
all internal lines are massless , @xmath115 , and @xmath116.,title="fig : " ] + one- and two - loop master integrals needed after applying asymptotic expansion to the amplitude @xmath15 .
all internal lines are massless , @xmath115 , and @xmath116.,title="fig : " ] one- and two - loop master integrals needed after applying asymptotic expansion to the amplitude @xmath15 .
all internal lines are massless , @xmath115 , and @xmath116.,title="fig : " ] one- and two - loop master integrals needed after applying asymptotic expansion to the amplitude @xmath15 .
all internal lines are massless , @xmath115 , and @xmath116.,title="fig : " ] + one- and two - loop master integrals needed after applying asymptotic expansion to the amplitude @xmath15 . all internal lines are massless , @xmath115 , and @xmath116.,title="fig : " ] + from the calculation of @xmath15 one obtains in a first step results for @xmath129 .
integration over phase space then leads to @xmath24 . for the results in section [ sec::impr_nnlo ]
this integration is performed numerically .
the second method is based on the use of the optical theorem in analogy to the nlo calculation performed in ref .
this method serves as an important cross check . in the following
we provide some of the technical details * the amplitudes for @xmath130 are generated with the help of qgraf @xcite . * in a first step about 17 million diagrams are generated .
however , most of them do not contain a cut through exactly two higgs bosons .
for this reason we post - process the qgraf output and filter @xcite the amplitudes describing the virtual corrections to @xmath15 .
typical feynman diagrams are shown in fig .
[ fig::gg2gg ] . * form @xcite code is then generated by passing the output via q2e @xcite , which transforms feynman diagrams into feynman amplitudes , to exp @xcite .
* our in - house form code applies projectors @xmath131 for each pair of external gluons which includes also non - physical degrees of freedom in the sum .
thus also contributions with ghosts in the initial state have to be considered .
note that this is in contrast to single higgs boson production which has no virtual contributions with ghosts in the initial state .
lo , nlo and nnlo feynman diagrams needed for the forward scattering amplitude @xmath130 .
solid lines refer to top quarks , curly lines to gluons and dashed lines to higgs bosons . at nnlo
only virtual contributions are shown .
wavy lines denote the cuts.,title="fig : " ] lo , nlo and nnlo feynman diagrams needed for the forward scattering amplitude @xmath130 .
solid lines refer to top quarks , curly lines to gluons and dashed lines to higgs bosons . at nnlo
only virtual contributions are shown .
wavy lines denote the cuts.,title="fig : " ] + lo , nlo and nnlo feynman diagrams needed for the forward scattering amplitude @xmath130 .
solid lines refer to top quarks , curly lines to gluons and dashed lines to higgs bosons . at nnlo
only virtual contributions are shown .
wavy lines denote the cuts.,title="fig : " ] + lo , nlo and nnlo feynman diagrams needed for the forward scattering amplitude @xmath130 .
solid lines refer to top quarks , curly lines to gluons and dashed lines to higgs bosons . at nnlo
only virtual contributions are shown .
wavy lines denote the cuts.,title="fig : " ] lo , nlo and nnlo feynman diagrams needed for the forward scattering amplitude @xmath130 .
solid lines refer to top quarks , curly lines to gluons and dashed lines to higgs bosons . at nnlo
only virtual contributions are shown .
wavy lines denote the cuts.,title="fig : " ] + lo , nlo and nnlo feynman diagrams needed for the forward scattering amplitude @xmath130 .
solid lines refer to top quarks , curly lines to gluons and dashed lines to higgs bosons . at nnlo
only virtual contributions are shown .
wavy lines denote the cuts.,title="fig : " ] lo , nlo and nnlo feynman diagrams needed for the forward scattering amplitude @xmath130 .
solid lines refer to top quarks , curly lines to gluons and dashed lines to higgs bosons . at nnlo
only virtual contributions are shown .
wavy lines denote the cuts.,title="fig : " ] + lo , nlo and nnlo feynman diagrams needed for the forward scattering amplitude @xmath130 .
solid lines refer to top quarks , curly lines to gluons and dashed lines to higgs bosons . at nnlo
only virtual contributions are shown .
wavy lines denote the cuts.,title="fig : " ] lo , nlo and nnlo feynman diagrams needed for the forward scattering amplitude @xmath130 .
solid lines refer to top quarks , curly lines to gluons and dashed lines to higgs bosons . at nnlo
only virtual contributions are shown .
wavy lines denote the cuts.,title="fig : " ] + resulting feynman diagrams after shrinking the top quark loops to a point according to the rule of asymptotic expansion .
the blob represents one - loop vacuum integrals.,title="fig : " ] + ( a ) ( b ) ( c ) the application of the asymptotic expansion for large top quark mass leads to a factorization of the five - loop integrals into the following contributions : 1 .
four one - loop vacuum integrals times one - loop integrals which are already present at lo , see fig .
[ fig::gg2gg_ae](a ) ; 2 .
three one - loop vacuum integrals times two - loop integrals , see fig .
[ fig::gg2gg_ae](b ) ; 3 .
two one - loop vacuum integrals times three - loop integrals , see fig .
[ fig::gg2gg_ae](c ) ; 4 .
two - loop times one - loop vacuum integrals times two - loop integrals ; 5 .
three - loop times one - loop vacuum integrals times one - loop integrals .
the mass scale in the vacuum integrals is given by the top quark .
they are again computed with the help of matad @xcite . for the remaining integrals ,
which depend on @xmath132 , we use the in - house programs rows @xcite and topoid @xcite to perform the reduction to master integrals .
the latter are depicted in fig .
[ fig::gg2gg_mi ] .
phase space master integrals occurring in the amplitude for @xmath130 .
the first line contains lo and nlo integrals .
the integrals in line two and three are needed for the nnlo virtual corrections .
single and double lines represent massless and higgs propagators , respectively .
double lines with gray - shaded interspace ( last two diagrams in first row ) correspond to higgs boson propagators which shall not be cut .
a cross marks an inverse propagator.,title="fig : " ] phase space master integrals occurring in the amplitude for @xmath130 .
the first line contains lo and nlo integrals .
the integrals in line two and three are needed for the nnlo virtual corrections .
single and double lines represent massless and higgs propagators , respectively .
double lines with gray - shaded interspace ( last two diagrams in first row ) correspond to higgs boson propagators which shall not be cut .
a cross marks an inverse propagator.,title="fig : " ] phase space master integrals occurring in the amplitude for @xmath130 .
the first line contains lo and nlo integrals .
the integrals in line two and three are needed for the nnlo virtual corrections .
single and double lines represent massless and higgs propagators , respectively .
double lines with gray - shaded interspace ( last two diagrams in first row ) correspond to higgs boson propagators which shall not be cut .
a cross marks an inverse propagator.,title="fig : " ] phase space master integrals occurring in the amplitude for @xmath130 .
the first line contains lo and nlo integrals .
the integrals in line two and three are needed for the nnlo virtual corrections .
single and double lines represent massless and higgs propagators , respectively .
double lines with gray - shaded interspace ( last two diagrams in first row ) correspond to higgs boson propagators which shall not be cut .
a cross marks an inverse propagator.,title="fig : " ] phase space master integrals occurring in the amplitude for @xmath130 .
the first line contains lo and nlo integrals .
the integrals in line two and three are needed for the nnlo virtual corrections .
single and double lines represent massless and higgs propagators , respectively .
double lines with gray - shaded interspace ( last two diagrams in first row ) correspond to higgs boson propagators which shall not be cut .
a cross marks an inverse propagator.,title="fig : " ] phase space master integrals occurring in the amplitude for @xmath130 .
the first line contains lo and nlo integrals .
the integrals in line two and three are needed for the nnlo virtual corrections .
single and double lines represent massless and higgs propagators , respectively .
double lines with gray - shaded interspace ( last two diagrams in first row ) correspond to higgs boson propagators which shall not be cut .
a cross marks an inverse propagator.,title="fig : " ] + phase space master integrals occurring in the amplitude for @xmath130 .
the first line contains lo and nlo integrals .
the integrals in line two and three are needed for the nnlo virtual corrections .
single and double lines represent massless and higgs propagators , respectively .
double lines with gray - shaded interspace ( last two diagrams in first row ) correspond to higgs boson propagators which shall not be cut .
a cross marks an inverse propagator.,title="fig : " ] phase space master integrals occurring in the amplitude for @xmath130 .
the first line contains lo and nlo integrals .
the integrals in line two and three are needed for the nnlo virtual corrections .
single and double lines represent massless and higgs propagators , respectively .
double lines with gray - shaded interspace ( last two diagrams in first row ) correspond to higgs boson propagators which shall not be cut .
a cross marks an inverse propagator.,title="fig : " ] phase space master integrals occurring in the amplitude for @xmath130 .
the first line contains lo and nlo integrals .
the integrals in line two and three are needed for the nnlo virtual corrections .
single and double lines represent massless and higgs propagators , respectively .
double lines with gray - shaded interspace ( last two diagrams in first row ) correspond to higgs boson propagators which shall not be cut .
a cross marks an inverse propagator.,title="fig : " ] phase space master integrals occurring in the amplitude for @xmath130 .
the first line contains lo and nlo integrals .
the integrals in line two and three are needed for the nnlo virtual corrections .
single and double lines represent massless and higgs propagators , respectively .
double lines with gray - shaded interspace ( last two diagrams in first row ) correspond to higgs boson propagators which shall not be cut .
a cross marks an inverse propagator.,title="fig : " ] phase space master integrals occurring in the amplitude for @xmath130 .
the first line contains lo and nlo integrals .
the integrals in line two and three are needed for the nnlo virtual corrections .
single and double lines represent massless and higgs propagators , respectively .
double lines with gray - shaded interspace ( last two diagrams in first row ) correspond to higgs boson propagators which shall not be cut .
a cross marks an inverse propagator.,title="fig : " ] phase space master integrals occurring in the amplitude for @xmath130 .
the first line contains lo and nlo integrals .
the integrals in line two and three are needed for the nnlo virtual corrections .
single and double lines represent massless and higgs propagators , respectively .
double lines with gray - shaded interspace ( last two diagrams in first row ) correspond to higgs boson propagators which shall not be cut .
a cross marks an inverse propagator.,title="fig : " ] + phase space master integrals occurring in the amplitude for @xmath130 .
the first line contains lo and nlo integrals .
the integrals in line two and three are needed for the nnlo virtual corrections .
single and double lines represent massless and higgs propagators , respectively .
double lines with gray - shaded interspace ( last two diagrams in first row ) correspond to higgs boson propagators which shall not be cut .
a cross marks an inverse propagator.,title="fig : " ] phase space master integrals occurring in the amplitude for @xmath130 .
the first line contains lo and nlo integrals .
the integrals in line two and three are needed for the nnlo virtual corrections .
single and double lines represent massless and higgs propagators , respectively .
double lines with gray - shaded interspace ( last two diagrams in first row ) correspond to higgs boson propagators which shall not be cut .
a cross marks an inverse propagator.,title="fig : " ] phase space master integrals occurring in the amplitude for @xmath130 .
the first line contains lo and nlo integrals .
the integrals in line two and three are needed for the nnlo virtual corrections .
single and double lines represent massless and higgs propagators , respectively .
double lines with gray - shaded interspace ( last two diagrams in first row ) correspond to higgs boson propagators which shall not be cut .
a cross marks an inverse propagator.,title="fig : " ] phase space master integrals occurring in the amplitude for @xmath130 .
the first line contains lo and nlo integrals .
the integrals in line two and three are needed for the nnlo virtual corrections .
single and double lines represent massless and higgs propagators , respectively .
double lines with gray - shaded interspace ( last two diagrams in first row ) correspond to higgs boson propagators which shall not be cut .
a cross marks an inverse propagator.,title="fig : " ] phase space master integrals occurring in the amplitude for @xmath130 .
the first line contains lo and nlo integrals .
the integrals in line two and three are needed for the nnlo virtual corrections .
single and double lines represent massless and higgs propagators , respectively .
double lines with gray - shaded interspace ( last two diagrams in first row ) correspond to higgs boson propagators which shall not be cut .
a cross marks an inverse propagator.,title="fig : " ] phase space master integrals occurring in the amplitude for @xmath130 .
the first line contains lo and nlo integrals .
the integrals in line two and three are needed for the nnlo virtual corrections .
single and double lines represent massless and higgs propagators , respectively .
double lines with gray - shaded interspace ( last two diagrams in first row ) correspond to higgs boson propagators which shall not be cut .
a cross marks an inverse propagator.,title="fig : " ] all three - loop master integrals factorize into a two - loop form factor contribution and the one - loop master integral of the lo calculation . from the latter only the imaginary part
is needed which is well known .
the results for the two - loop form factor integrals can , e.g. , be found in refs .
@xcite .
the two - loop master integrals are more involved .
a numerical calculation would probably be possible , however , we follow the approach outlined in ref .
@xcite for the nlo master integrals and perform an expansion in @xmath133 [ cf .
( [ eq::delta ] ) ] .
all integrals contain a massless sub - loop for which analytic expressions are known .
the massless two - point function can be expressed in terms of @xmath134 functions and results for the triangle with two massless external legs and the ( crossed ) box can be found in ref .
analytic expressions for the triangle diagram with squared external momenta @xmath135 are given in eq .
( [ eq::i14 ] ) . after expanding in @xmath132 the remaining phase - space integration can be performed analytically .
we have computed expansion terms up to order @xmath136 and found agreement with the results obtained in the previous subsection which for this purpose have also been integrated analytically after performing the expansion in @xmath132 .
the approach based on the optical theorem requires special care in the treatment of the imaginary parts originating from the two - loop form factor diagrams .
such contributions either correspond to @xmath137 or @xmath138 . in the former case
the two - loop integrals originate from the product of two one - loop contributions containing factors @xmath139 and @xmath140 , respectively , which finally leads to @xmath141 . in the other case one has @xmath142 .
the corresponding discussion for single higgs production can be found in ref .
@xcite .
note that the approach based on the @xmath15 amplitudes leads to simpler intermediate expressions .
thus , it is possible to allow for a general qcd gauge parameter @xmath143 when computing the @xmath81 terms . furthermore , also the @xmath114 corrections could be evaluated ( for @xmath144 ) whereas in the optical theorem approach only @xmath81 terms could be computed in feynman gauge .
however , let us mention that this approach can be used in a straightforward way to compute nnlo top quark mass effects to the hard contribution of the total cross section whereas in the approach of the previous subsection this is less obvious . to conclude this section
let us summarize our procedure to obtain the sv corrections at nnlo .
we compute virtual corrections to @xmath15 including @xmath114 corrections .
note that we have @xmath145 .
using eq .
( [ eq::sig_fin ] ) we construct @xmath146 which enters @xmath147 in eq .
( [ eq::g_sv ] ) . the differential and total cross section
is then obtained with the help of eqs .
( [ eq::g ] ) and ( [ eq::sigma_tot ] ) .
in this section we discuss the effect of the @xmath81 and @xmath114 terms on the nnlo cross section for the production of higgs boson pairs .
the cross section in the infinite top quark mass limit has been computed in ref .
@xcite . in ref .
@xcite the three - loop matching coefficient has been added , completing the nnlo prediction , and the virtual corrections from ref .
@xcite have been cross checked .
comparison of the lo , nlo and nnlo contributions to the partonic cross section . at
lo the exact result is shown and at nlo and nnlo the first three terms in the large-@xmath148 expansion are shown .
for all curves the nnlo - value for @xmath51 is used . for the renormalization and factorization scale
we use @xmath111 .
, title="fig : " ] + the results for the virtual corrections computed in section [ sec::nnlo ] are inserted in the formalism described in section [ sec::fac ] to construct the quantity @xmath146 which enters eq .
( [ eq::g_sv ] ) .
the result for the partonic cross section is shown in fig .
[ fig::sig_part_lo_nlo_nnlo ] as a function of the partonic center - of - mass energy where the exact lo result is compared with nlo and nnlo . at nlo and nnlo three terms in the mass expansion are shown . at nlo to have a direct comparison with nnlo . ]
furthermore , at nnlo the sv approximation is shown for @xmath91 ( cf .
discussion in section [ sec::nlo ] ) .
note that the nnlo curves peak for smaller values of @xmath7 than at nlo and lo .
as far as the top quark mass corrections are concerned the same pattern is observed as at nlo : the correction term of order @xmath21 decreases the infinite top quark mass result which is overcompensated by the @xmath149 term resulting in a small positive correction .
nnlo partonic cross section for different scales and for @xmath150 ( dotted ) and @xmath91 ( solid ) .
only the @xmath151 result is shown .
, title="fig : " ] + in fig .
[ fig::sig_part_nnlo ] we compare the nnlo - sv contribution for two different scales , @xmath111 and @xmath152 .
this plot furthermore shows the effect of @xmath150 and @xmath91 .
note that the choice @xmath91 , which we expect to better approximate the complete result , leads to an increase of the cross section .
hadronic lo , nlo and nnlo - sv ( with @xmath91 ) cross sections as a function of @xmath97 . for their evaluation the respective value of @xmath51
is used . at
lo the exact result and at nlo only the @xmath151 term is shown . at nnlo
the @xmath151 , @xmath153 and @xmath149 results are plotted .
the results in the right panel with `` @xmath154 '' at the bottom correspond to the prediction of the total cross section . for this plot
@xmath111 has been used .
, title="fig : " ] + the hadronic cross section as a function of @xmath97 is shown in fig .
[ fig::sig_hadr_lo_nlo_nnlo ] for @xmath111 . at
lo the exact result is used and the nlo curve is based on the @xmath151 results . using instead the @xmath155 terms leads to an upwards shift of about 5% as can be seen in fig .
[ fig::sig_hadr ] . at nnlo
three curves are shown which include terms up to @xmath151 , @xmath153 and @xmath149 .
as at nlo one observes good convergence up to @xmath156 gev . for higher values of @xmath99 the @xmath153 curve is below and the @xmath149 curve above the infinite top quark mass result leading to a @xmath157 effect for the total cross section on the rightmost part of the plot . to be conservative , we thus estimate that the nnlo top quark mass effects lead to an uncertainty of @xmath158 .
note that the nnlo corrections amount to about 20% of the lo result .
hadronic nlo ( dotted ) and nnlo ( solid ) @xmath86 factor as a function of @xmath97.,title="fig : " ] + fig .
[ fig::sig_k_hadr_nnlo ] shows the hadronic @xmath86 factor at nnlo which is defined by @xmath159 as a function of @xmath106 .
for comparison also the nlo result from fig .
[ fig::sig_hadr ] is shown as dotted curve using the @xmath0 result .
one observes that the various @xmath21 orders lead to similar results for @xmath160 .
furthermore , there is a strong raise close to threshold which is due to the steeper behaviour of the nnlo cross section as can be seen in the inlay of fig .
[ fig::sig_part_lo_nlo_nnlo ] . for higher values of @xmath97 , in particular for the total cross section , @xmath160 approaches @xmath161 .
.[tab : nnlo]total hadronic cross section at lo , nlo and nnlo - sv including top quark mass effects using @xmath162 and @xmath91 . [ cols="<,<,<,^",options="header " , ] results for the total cross section at lo , nlo and nnlo are summarized in table [ tab : nnlo ] for @xmath111 . at nlo
only @xmath151 terms are included in the analysis whereas at nnlo @xmath151 , @xmath153 and @xmath149 terms are considered . in this way we can estimate the top mass effects of the nnlo term . besides the cross section also the @xmath86 factor is shown . at nnlo
we use @xmath91 and we furthermore show the relative deviation due to @xmath163 terms defined through @xmath164 the two known mass correction terms lead to a change of the cross section by about @xmath165 . assuming a similar pattern as at nlo we thus estimate that nnlo top quark mass corrections change the effective - theory result by at most @xmath166 .
we compute nlo and nnlo corrections to double higgs boson production in gluon fusion beyond the effective - theory approach .
the starting point of the calculation are full - theory feynman diagrams .
we perform an asymptotic expansion in the limit where the top quark mass is large and compute at nnlo three terms in the @xmath1 expansion for the virtual corrections .
they are used to construct a soft - virtual approximation for the production cross section . in the limit @xmath0 the effective - theory result of ref .
@xcite is confirmed and @xmath81 and @xmath114 terms are added . the main result of this paper is illustrated in fig . [ fig::sig_hadr_lo_nlo_nnlo ] where the hadronic cross section is shown as a function of @xmath97 ( a technical cut on the partonic center - of - mass energy ) .
the curves including @xmath1 corrections deviate from the infinite mass result only by a few per cent which leads us to the estimate that the effective - theory result is accurate to @xmath166 .
analog results for the mass corrections at nlo are shown in fig .
[ fig::sig_hadr ] which constitutes an update of ref .
@xcite . here
we estimate the uncertainty to @xmath167 .
we want to stress that the results obtained in this paper provide excellent approximations for small values of the partonic center - of - mass energy , say below @xmath168 gev .
although in this region the cross section is small it is of interest since there the cross section has a characteristic behaviour .
furthermore , it is possible to use our result in this region as a benchmark for future calculations taking into account the exact dependence on @xmath148 .
the methods described in section [ sec::nnlo ] can also be used to compute top mass corrections to the real radiation part .
however , the simplifications used in ref .
@xcite where results have been obtained for @xmath169 do not apply once finite mass effects are considered .
the calculation is much more challenging since significantly more feynman diagrams contribute and more complicated master integrals have to be computed . in this paper for the first time the effect of a finite top quark mass has been examined for the nnlo cross section for double higgs boson production . whereas at nlo an exact calculation is within reach this is certainly not the case at nnlo .
thus our results become particular important once our nlo approximations are compared to an exact calculation which increases the confidence in the uncertainty estimate . furthermore ,
one probably can obtain a prescription to tune the approximation procedure and hence reduce the uncertainty at nnlo .
we would like to thank kirill melnikov for providing to us the analytic result for @xmath128 in eq .
( [ eq::i14 ] ) , for many useful discussions and for carefully reading the manuscript .
this work is supported by the deutsche forschungsgemeinschaft through grant ste 945/2 - 1 and by kit through its distinguished researcher fellowship program .
parts of this work were supported by the european commission through contract pitn - ga-2012 - 316704 ( higgstools ) .
j.g . would like to express a special thanks to the mainz institute for theoretical physics ( mitp ) for its hospitality and support .
a. djouadi , w. kilian , m. mhlleitner and p. m. zerwas , eur .
j. c * 10 * ( 1999 ) 45 [ hep - ph/9904287 ] .
u. baur , t. plehn and d. l. rainwater , phys .
d * 67 * ( 2003 ) 033003 [ hep - ph/0211224 ] .
u. baur , t. plehn and d. l. rainwater , phys .
d * 69 * ( 2004 ) 053004 [ hep - ph/0310056 ] .
m. j. dolan , c. englert and m. spannowsky , jhep * 1210 * ( 2012 ) 112 [ arxiv:1206.5001 [ hep - ph ] ] .
a. papaefstathiou , l. l. yang and j. zurita , phys .
d * 87 * ( 2013 ) 011301 [ arxiv:1209.1489 [ hep - ph ] ] .
j. baglio , a. djouadi , r. grber , m. m. mhlleitner , j. quevillon and m. spira , jhep * 1304 * ( 2013 ) 151 [ arxiv:1212.5581 [ hep - ph ] ] .
f. goertz , a. papaefstathiou , l. l. yang and j. zurita , jhep * 1306 * ( 2013 ) 016 [ arxiv:1301.3492 [ hep - ph ] ] .
m. gouzevitch , a. oliveira , j. rojo , r. rosenfeld , g. p. salam and v. sanz , jhep * 1307 * ( 2013 ) 148 [ arxiv:1303.6636 [ hep - ph ] ] .
e. w. n. glover and j. j. van der bij , nucl .
b * 309 * ( 1988 ) 282 .
t. plehn , m. spira and p. m. zerwas , nucl .
b * 479 * ( 1996 ) 46 [ erratum - ibid .
b * 531 * ( 1998 ) 655 ] [ hep - ph/9603205 ] .
s. dawson , s. dittmaier and m. spira , phys .
d * 58 * , 115012 ( 1998 ) [ hep - ph/9805244 ] .
j. grigo , j. hoff , k. melnikov and m. steinhauser , nucl .
b * 875 * ( 2013 ) 1 [ arxiv:1305.7340 [ hep - ph ] ] .
f. maltoni , e. vryonidou and m. zaro , jhep * 1411 * ( 2014 ) 079 [ arxiv:1408.6542 [ hep - ph ] ] .
d. de florian and j. mazzitelli , phys .
b * 724 * ( 2013 ) 306 [ arxiv:1305.5206 [ hep - ph ] ] .
d. de florian and j. mazzitelli , phys . rev .
111 , * 201801 * ( 2013 ) [ arxiv:1309.6594 [ hep - ph ] ] .
j. grigo , k. melnikov and m. steinhauser , nucl .
b * 888 * ( 2014 ) 17 [ arxiv:1408.2422 [ hep - ph ] ] .
d. y. shao , c. s. li , h. t. li and j. wang , arxiv:1301.1245 [ hep - ph ] .
d. de florian and j. mazzitelli , arxiv:1505.07122 [ hep - ph ] . c. anastasiou and k. melnikov , nucl
. phys .
b * 646 * , 220 ( 2002 ) [ hep - ph/0207004 ] .
j. grigo , j. hoff , k. melnikov and m. steinhauser , pos radcor * 2013 * ( 2013 ) 006 [ arxiv:1311.7425 [ hep - ph ] ] .
j. grigo and j. hoff , arxiv:1407.1617 [ hep - ph ] .
d. de florian and j. mazzitelli , jhep * 1212 * ( 2012 ) 088 [ arxiv:1209.0673 [ hep - ph ] ] .
s. catani , phys .
b * 427 * ( 1998 ) 161 [ hep - ph/9802439 ] .
a. d. martin , w. j. stirling , r. s. thorne and g. watt , eur .
j. c * 63 * ( 2009 ) 189 [ arxiv:0901.0002 [ hep - ph ] ] . k. a. olive _ et al . _
[ particle data group collaboration ] , chin .
c * 38 * ( 2014 ) 090001 .
g. aad _ et al . _ [ atlas and cms collaborations ] , phys .
* 114 * ( 2015 ) 191803 [ arxiv:1503.07589 [ hep - ex ] ] . c. anastasiou , c. duhr , f. dulat , e. furlan , t. gehrmann , f. herzog and b. mistlberger , phys .
b * 737 * ( 2014 ) 325 [ arxiv:1403.4616 [ hep - ph ] ] .
f. herzog and b. mistlberger , arxiv:1405.5685 [ hep - ph ] .
p. nogueira , j. comput .
* 105 * ( 1993 ) 279 .
j. kuipers , t. ueda , j. a. m. vermaseren and j. vollinga , comput .
commun .
* 184 * ( 2013 ) 1453 [ arxiv:1203.6543 [ cs.sc ] ] .
r. harlander , t. seidensticker and m. steinhauser , phys .
b * 426 * ( 1998 ) 125 [ hep - ph/9712228 ] .
t. seidensticker , hep - ph/9905298 .
m. steinhauser , comput .
commun .
* 134 * ( 2001 ) 335 [ hep - ph/0009029 ] .
a. v. smirnov and v. a. smirnov , arxiv:1302.5885 [ hep - ph ] .
a. v. smirnov , comput .
commun .
* 189 * ( 2014 ) 182 [ arxiv:1408.2372 [ hep - ph ] ] .
t. g. birthwright , e. w. n. glover and p. marquard , jhep * 0409 * ( 2004 ) 042 [ hep - ph/0407343 ] .
r. k. ellis and g. zanderighi , jhep * 0802 * ( 2008 ) 002 [ arxiv:0712.1851 [ hep - ph ] ] .
f. chavez and c. duhr , jhep * 1211 * ( 2012 ) 114 [ arxiv:1209.2722 [ hep - ph ] ] .
a. b. goncharov , math .
lett . * 5 * ( 1998 ) 497 [ arxiv:1105.2076 [ math.ag ] ] .
a. v. smirnov , comput .
commun .
* 185 * ( 2014 ) 2090 [ arxiv:1312.3186 [ hep - ph ] ] .
j. hoff , `` methods for multiloop calculations and higgs boson production at the lhc '' , dissertation , kit , 2015 .
j. a. m. vermaseren , math - ph/0010025 .
t. gehrmann , t. huber and d. maitre , phys .
b * 622 * ( 2005 ) 295 [ hep - ph/0507061 ] .
a. pak , m. rogal and m. steinhauser , jhep * 1109 * ( 2011 ) 088 [ arxiv:1107.3391 [ hep - ph ] ] . | we compute next - to - next - to - leading order qcd corrections to the gluon - induced production cross section of higgs boson pairs in the large top quark mass limit using the soft - virtual approximation . in the limit of infinitely - heavy top quark
we confirm the results in the literature .
we add two more expansion terms in the inverse top quark mass to the @xmath0 result .
since the @xmath1 expansion converges poorly , we try to improve on it by factorizing the exact leading order cross section .
we discuss two ways of doing that and conclude that the finite top quark mass effects shift the cross section at most by about 10% at next - to - leading order and by about 5% at next - to - next - to - leading order .
pacs numbers : 14.80.bn , 12.38.bx , 14.65.ha |
the basic idea in the feynman - schwinger representation is to replace path integrals over quantum fields with path integrals over particle trajectories .
the step of going from field configurations to particle trajectories dramatically reduces the number of degrees of freedom . in this talk we present analytic and numerical applications of feynman - schwinger representation @xcite to nonperturbative problems
the first applictaion is the calculation of 1-body propagator in massive scalar qed .
the second application involves the calculation of 2-body bound state masses in scalar @xmath0 interaction .
massive scalar qed in 0 + 1 dimension is a simple interaction that enables one to obtain a fully analytical result for the dressed and bound state masses within the fsr approach . in 0 + 1 dimension
_ contribution of the matter loops identically vanish and therefore quenched fsr calculations give the exact result _ @xcite . in this section
we compare the self energy result obtained by three different approaches ; namely the simple bubble sum , the dyson - schwinger equation , and the feynman - schwinger representation . the minkowski metric expression for the scalar qed lagrangian in feynman gauge is given by @xmath1 where @xmath2 represents the gauge field of mass @xmath3 , and @xmath4 is the charged field of mass @xmath5 .
the field tensor @xmath6 is zero in 0 + 1 dimensions , and the dynamics is described by the gauge fixing term @xmath7 .
the kind of diagrams included in each method are displayed in fig . [ allthree.sqed ] .
the main difference between the rainbow dyson - schwinger and the feynman - schwinger diagrams is the crossed diagrams .
these diagrams involve photon lines that cross each other .
the fsr approach also includes all possible four - point interaction contributions while the rainbow dse only includes the tadpole type four - point interactions . in principle
all four - point interactions can also be incorporated into the simple bubble sum and the rainbow dse . in fig .
[ mvsg2.sqed ] we display all dressed mass results .
the bubble summation develops a complex mass pole beyond a critical coupling @xmath8 ( gev)@xmath9 . at the critical point
a ` collision ' takes place with another real solution , leading to two complex conjugated solutions with increasing @xmath10 .
the result obtained from the dyson - schwinger equation displays a similar characteristic . at low coupling strengths the rainbow ds and
the bubble results are very close and they converge to the exact result given by the feynman - schwinger approach . similar to the bubble result the ds result develops a complex mass pole at a critical coupling of @xmath11 ( gev)@xmath9 .
these results clearly show that the solution of _ the dyson - schwinger equation in rainbow approximation is not a good approximation to the exact result_. the second application of the fsr involves bound states in scalar @xmath0 interaction .
the application of the fsr to scalar particles have been considered in references @xcite .
the euclidean lagrangian for this theory is given by @xmath12\chi+\frac{1}{2}\,\phi(\mu^2-\partial^2)\phi .
\label{lagr0}\ ] ] here we present the results for the 2-body bound states .
the final result for the two - body propagator involves a quantum mechanical path integral that sums up contributions coming from all possible _ trajectories _ of _ particles _ @xmath13 } , \label{g1.phi3}\ ] ] where @xmath14 $ ] term involves kinetic and interaction energy contributions .
further details on the treatment of scalar interactions can be found in refs .
the interaction kernel @xmath15 is defined by @xmath16 where @xmath3 is chosen to be @xmath17 .
the ultraviolet singularity in the kernel @xmath18 eq .
( [ kernel.phi3 ] ) can be regularized using a pauli - villars regularization prescription .
the pauli - villars mass we choose is @xmath19 . in figure [ mvsg2.2b ]
we present the comparison of the 2-body bound state masses obtained by the fsr to various bound state equations .
the fsr calculation involves summation of all ladder and crossed ladder diagrams , and excludes the self energy contributions .
according to figure [ mvsg2.2b ] all bound state equations underbind . among the manifestly covariant equations the gross equation gives the closest result to the exact calculation obtained by the fsr method .
this is due to the fact that in the limit of infinitely heavy - light systems the gross equation effectively sums all ladder and crossed ladder diagrams
. equal - time equation also produces a strong binding but the inclusion of retardation effects pushes the equal - time results away from the exact results ( mandelzweig - wallace equation ) .
in particular the bethe - salpeter equation in the ladder approximation ( bse in figure [ mvsg2.2b ] ) gives the lowest binding .
a comparison of the ladder bethe - salpeter , gross , and the fsr results shows that _ the exchange of crossed ladder diagrams plays a very significant role . _
results presented in this talk clearly shows that _ approximate calculations in field theory may lead to serious deviations from the exact results . _
therefore it is important to develop rigorous nonperturbative methods .
the fsr is a promising candidate for doing nonperturbative calculations in field theory
. 9 r.p .
feynman , phys .
rev 80 ( 1950 ) , 440 ; j. schwinger , phys . rev . * 82 * ( 1951 ) , 664 yu . a. simonov , nucl . phys .
b * 307 * ( 1988 ) , yu .
a. simonov and j. a. tjon , ann
. phys . * 228 * , 1 ( 1993 ) .
t. nieuwenhuis , j. a. tjon , phys .
lett . * 77 * , 814 ( 1996 ) n. brambilla , a. vairo , phys .
rev d. * 56 * , 1445 ( 1997 ) .
avkli , f. gross , j. a. tjon , hep - ph/9907445 , accepted for publication in phys .
d. avkli , hep - ph/9910502 , accepted for publication in comp .
avkli , j. a. tjon , f. gross , phys .
c * 60 * : 055210 , 1999 , erratum - ibid . * c * 61 : 069901 , 2000 . | in nuclear and particle physics one is often faced with problems where perturbation theory is not applicable .
an example of this is the description of bound states .
therefore , an exact solution of field theory to all orders is an unavoidable and interesting problem .
path integrals provide a framework for exact solutions in field theory . in this talk
i will present an economical method of evaluating path integrals using the feynman - schwinger representation ( fsr ) . |
these can be most transparently studied by considering the hubbard model , which describes electrons hopping on a lattice which interact with each other via an on - site repulsion .
the hubbard hamiltonian reads @xmath76 in standard notations .
let us consider the tasaki lattice @xcite ( fig . [ fig01s ] ) as a representative for a 2d system with a dispersionless ( flat ) band .
tasaki s lattice decoration @xcite can be performed in arbitrary dimension ( and also for other lattices ) .
this allows a direct comparison of 1d and 2d . for the lowest - energy one - electron band to be completely dispersionless , the two relevant hopping integrals obey the relation @xmath77 , where @xmath71 is the coordination number of the underlying lattice .
the one - electron states in the flat band can be taken as localized on trapping cells . in 2d
each trapping cell consists of one site of the underlying square lattice and four neighboring decorating sites ( fig .
[ fig01s ] ) . a localized eigenstate of energy @xmath78 of an electron with spin @xmath79 is given by @xmath80 , where @xmath81 runs over the sites of the underlying square lattice , @xmath82 , and @xmath83 denotes the vacuum state .
localized many - particle ground states for @xmath84 electrons and @xmath2 require ( i ) that each cell be empty or singly occupied and ( ii ) that electrons in adjacent cells be in a symmetric collective spin state . as a result
, the spin degeneracy for a contiguous cluster of @xmath5 electrons is reduced from @xmath85 to that of a spin @xmath86 multiplet , @xmath87 .
this is all that s needed for the mapping to pcp : all ground states can be labelled by the possible geometric configurations of @xmath8 electrons distributed over @xmath9 cells ( they are labelled by @xmath89 ) , and the non - trivial weight of each state is given by eq .
( [ degeneracy ] ) , where @xmath13 denotes the number of separated clusters , and @xmath12 denotes the number of electrons in cluster @xmath90 .
we have verified by exact diagonalization that this set of localized many - body states spans the ground - state space for tasaki lattices up to @xmath91 sites .
numerical results for degeneracies are given in table [ tab1 ] and agree with the corresponding number of geometric configurations .
note that even for the small lattices considered in this context , not all sectors up to @xmath92 electrons are numerically accessible .
in addition , the hubbard model allows for higher electron fillings which are beyond the the geometric picture of the present paper .
one - dimensional pcp can be analyzed using a transfer - matrix method .
the crucial step is to choose suitable representative configurations for the quantum states with different values of @xmath93 , for example by putting all the down spins right from the up spins in a cluster @xcite .
we set @xmath94 and introduce the grand partition function @xmath95 of a percolating system as the sum of probabilities of all possible random realizations .
@xmath95 will be written in terms of the transfer matrix @xmath96 as @xmath97 .
the transfer matrix for the pauli - correlated percolation in one dimension then reads @xcite @xmath98 the matrix elements @xmath99 correspond to the pair of neighboring sites @xmath90 and @xmath100 and acquire the value 1 ( @xmath101 ) if the site @xmath90 is empty ( occupied ) . in order to determine the results in terms of @xmath31 rather than @xmath101
, we first calculate the average occupation number of the site @xmath102 which should be equal to @xmath31 .
thus we have @xmath103 in what follows we consider the thermodynamic limit @xmath104 . using the left and
right eigenvectors of @xmath96 , one obtains @xmath105 $ ] , and inversion of this function leads to @xmath106 $ ] .
we turn to the calculation of the average number of clusters of size @xmath21 ( normalized by the lattice size @xmath107 ) @xmath62 @xcite . to fix the cluster of length @xmath21 we start with an empty site
, then we have a string ( cluster ) of @xmath21 occupied sites , and the last site of this string is followed by an empty one . to calculate @xmath62 we have to replace the product of a sequence of @xmath108 @xmath96-matrices by the product @xmath109 , where @xmath110 that yields @xmath111 after straightforward calculations we arrive at @xmath112 see eq .
( [ nl ] ) . in these calculations
we have used the relation @xmath113 next , we use the transfer - matrix approach to calculate the ( pair ) site - occupation correlation function @xmath114 .
using the matrix @xmath115 defined in eq .
( [ av_oc - nu ] ) and calculating @xmath116 we get @xmath117 finally , we calculate the pair connectivity @xmath118 ( the probability that two sites @xmath8 and @xmath119 are both occupied and belong to the same cluster ) . for this purpose
we have to consider the quantity @xmath120 for completeness we mention that the transfer - matrix approach can also be applied to the standard percolation in one dimension . in this case we have @xmath121 and formulas ( [ av_oc - nu ] ) , ( [ n_l ] ) , and ( [ connect ] ) yield @xmath122 , @xmath123 , @xmath124 , and @xmath125 , respectively .
these results can of course also be obtained by more elementary means and are well - known in standard percolation theory @xcite . | we investigate the location and nature of the para - ferro transition of interacting electrons in dispersionless bands using the example of the hubbard model on the tasaki lattice .
this case can be analyzed as a geometric site - percolation problem where different configurations appear with nontrivial weights .
we provide a complete exact solution for the 1d case and develop a numerical algorithm for the 2d case . in two dimensions
the paramagnetic phase persists beyond the uncorrelated percolation point , and the grand - canonical transition is via a first - order jump to an _ unsaturated _ ferromagnetic phase . _
introduction._the interplay of the coulomb interaction with the pauli principle was already recognized by heisenberg @xcite to give rise to a ferromagnetic exchange interaction , also encoded in hund s rule about aligned spins in a partially filled shell . for a many - body system of correlated electrons with a flat band ,
when the interaction energy completely dominates over the kinetic energy , the ferromagnetic instability is one of the few problems for which exact results are available , albeit for a restricted range of fillings @xcite .
flat band systems are receiving a great deal of attention right now , in particular with the view of realizing new many - body phases there ( see @xcite and references therein ) ; in this context , the possibility of ferromagnetism as a many - body instability is also being considered @xcite .
it is therefore timely to provide a detailed study of the phase diagram and the critical properties of this form of magnetism : we analyze a flat - band ferromagnet with an on - site hubbard interaction of strength @xmath0 . for @xmath1 ,
any state involving electrons occupying the flat band only is trivially a ground state .
crucially , this degeneracy is only partially lifted when a repulsive @xmath2 is switched on .
first , since the flat band permits well - localized real - space electronic wave functions , at low density electrons can be placed on the lattice so that they do not overlap .
second , even if they do overlap , they can still avoid paying an energy penalty @xmath3 : the basic reason is that the pauli principle , by demanding an antisymmetric pair wave function , makes the overlap between two electrons on the same site vanish provided they are in a symmetric spin state .
this is the origin of flat - band ferromagnetism . as the density of electrons increases ,
ferromagnetic clusters of increasing size appear .
the degeneracy , @xmath4 , of a ferromagnetic cluster containing @xmath5 electrons , gives differing weights to different clustering of electrons .
the ferromagnetic transition corresponds to the emergence of a cluster containing a nonzero fraction of the electrons . an early remark by mielke @xcite likened this problem to one of percolation .
mielke and tasaki @xcite noted that , for a class of flat - band ferromagnets on particular decorated lattices , the percolation problem in question is not a standard one @xcite but rather one including nontrivial weights . here , we develop this analogy in detail .
first of all , we point out that the interaction between the clusters , on account of its `` statistical origin '' in the pauli principle , is unusual in that it is range - free and purely geometric two particles interact only if they form part of the same cluster .
the interaction is genuinely many - body in that it can not be decomposed into a sum of pairwise terms .
it is effectively repulsive and only depends on the size of the cluster , irrespective of its shape . despite its long range ,
the statistical interaction does saturate .
this motivates the study of the resulting unusual percolation problem , which we call pauli - correlated percolation ( pcp ) .
we find that it has a number of interesting features in its own right .
it provides an instance of a problem in the quantum physics of strongly correlated electrons which can be `` reduced '' to a highly nontrivial problem in _ classical _ statistical mechanics , on which an entirely different set of tools can be brought to bear .
we first demonstrate some special features of this problem by providing a complete exact solution of the one - dimensional ( 1d ) version of this model , which corresponds to a sawtooth lattice potentially realized in strongly correlated sawtoothlike compounds such as @xmath6 @xcite .
unlike standard percolation , this exhibits a tendency to break up large clusters as well as a development of spatial ( anti)correlations .
its percolation transition at full filling is continuous .
next , we carry out an analysis of the phase diagram for the two - dimensional ( 2d ) tasaki lattice , a decorated square lattice ( see left panel of fig .
1 ) . using a numerical algorithm custom - tailored to the problem at hand by extending the hoshen - kopelman and newman - ziff algorithms @xcite for standard percolation , we establish that the ferromagnetic transition does indeed take place at a filling comfortably in excess of the corresponding well - known percolation transition on the square lattice at @xmath7 @xcite . in the grand - canonical ensemble ,
this transition is of first order ; in the canonical ensemble we find concomitant phase - separated states ( see fig . [ fig01 ] for some examples ) .
_ pauli - correlated percolation and flat - band ferromagnetism._as a representative system with a dispersionless ( flat ) band , let us consider the tasaki model @xcite , although our approach in principle can be adapted to other flat - band lattices .
the enumeration of all ground states of the repulsive hubbard model on the tasaki lattice maps to a percolation problem where each occupied site on a hypercubic lattice corresponds to an electron localized in a trapping cell ( whose wave function only overlaps with that of electrons in adjacent cells . )
( the details of this mapping , which are unimportant for the following , are relegated to section [ sec : supplpcp ] ) .
all ground states can be labeled by the possible geometric configurations of @xmath8 electrons distributed over @xmath9 traps , labeled by @xmath10 , and a nontrivial weight of each state @xcite @xmath11 which arises because of the spin degeneracy of the ferromagnetic cluster of size @xmath12 in configuration @xmath10 ( @xmath13 denotes the number of clusters in the system ) . here
@xmath14 is a fugacity which can be used to tune the number of electrons in a grand - canonical ensemble .
the expectation value of an operator @xmath15 is given by the usual expression @xmath16 for the grand - canonical ensemble , the sum over @xmath10 runs over all configurations of @xmath17 electrons while for the canonical ensemble , it is restricted to configurations with a given number of electrons @xmath8 . from the point of view of magnetism , a particularly important observable is the square of the total spin @xmath18 which can be written for a particular geometric configuration @xmath10 in two equivalent ways @xmath19 in the first form , the contribution from each cluster is manifest while the second form relates it to @xmath20 , the normalized number of clusters of size @xmath21 , _
i.e. _ , a quantity which plays a central role in percolation theory @xcite .
_ quantum - statistical interaction._important differences arise between our pauli - correlated percolation and the standard one [ with trivial weight factor @xmath22 .
the weight factor can be cast as a pseudo - boltzmann weight of statistical origin , @xmath23 $ ] .
the resulting effective entropic interaction , @xmath24 , has the following properties .
first , it is repulsive a group of @xmath5 electrons has maximal weight @xmath25 if they form isolated one - electron `` clusters , '' and minimal weight @xmath4 if they form a single cluster .
these extreme cases show that @xmath26 saturates , _
i.e. _ , is never superextensive unlike other long - range interactions . befitting its quantum statistical origin
, the interaction is range - free the shape of the cluster is unimportant , only its number of electrons matters .
note also that the interaction is a genuinely many - body one : due to the form @xmath27 it can not be written as a sum of two - particle terms .
taking all of this together demonstrates that this interaction gives rise to an entirely novel `` pauli - correlated '' percolation problem , of interest in its relevance to flat - band ferromagnetism and as a physically motivated example of a nonstandard percolation problem with an unusual weight .
( top ) and @xmath28 ( bottom ) at @xmath29 for pcp ( solid line ) and standard percolation ( dotted line ) in 1d .
right : deviation of @xmath30 from saturation for large @xmath31 for pcp and standard percolation in 2d . ]
_ exact solution in one dimension._we first provide a complete solution of the 1d tasaki model ( sawtooth chain ) @xcite .
a solution of the problem can be obtained with the help of a transfer matrix @xcite despite the long - range nature of the statistical interaction ( for technical details see [ sec : supplpcp ] ) . for a given electron density @xmath32
we find @xmath33 this cluster - size distribution ( fig .
[ fig01a ] , left panel ) has a maximum at @xmath34 for @xmath35 moving along @xmath36 for @xmath37 .
this is unlike the standard percolation result @xmath38 @xcite , which drops monotonically with @xmath21 and thus has a maximum at @xmath39 .
the macroscopic magnetic moment vanishes for @xmath40 , with a continuous onset at percolation , @xmath41 , as @xmath42 : @xmath43 the connected pair correlation function @xmath44 yields a correlation length @xmath45 that diverges as @xmath46 when @xmath47 .
by contrast , for standard percolation there are no nontrivial pair correlations : @xmath48 . the negative sign in eq .
( [ eq : g ] ) shows that the interaction is repulsive electron positions anticorrelate .
the long - range and many - body nature of the interactions leads to a nontrivial cluster size distribution favoring an approximately uniform spacing of vacant cells . _
the phase diagram in 2d._the 2d case is not amenable to exact solution .
here we examine the 2d pcp numerically . due to the nontrivial weights ( [ degeneracy ] ) , simple random sampling used for the conventional percolation is insufficient .
going beyond standard numerical schemes @xcite we have implemented efficient importance sampling on @xmath49 square lattices with periodic boundary conditions as follows . in the grand - canonical ensemble ,
we simply choose a site and if it is empty ( occupied ) , propose to insert ( remove ) an electron . in the canonical ensemble
we generate a new configuration @xmath50 from the given one @xmath51 by random permutation of two sites in order to ensure a fixed number of electrons .
the new configuration is accepted with the metropolis probability @xmath52 $ ] . in addition , we have employed exchange monte carlo steps @xcite for the grand - canonical simulations .
clusters are identified in two different ways : 1 ) using a modified newman - ziff algorithm @xcite which locally updates cluster labeling for fixed number of occupied sites and 2 ) using the hoshen - kopelman algorithm @xcite which makes a global update .
our central results are the following . _
the percolation transition _ is of first order as already suggested by visual inspection of individual configurations ( fig .
[ fig01 ] ) . vs. chemical potential @xmath53 , controlling the filling of the flat band in 2d .
inset : histograms of density for a `` finite - size '' critical value @xmath54 . ] grand - canonical simulations exhibit a jump at a chemical potential @xmath55 between densities @xmath56 and @xmath57 , fixed by equal - sized peaks in the histograms as shown in fig .
[ fig05 ] .
we estimate the jump to occur between densities @xmath56 around @xmath58 and @xmath59 . in between , in our _ canonical _
simulations for finite systems , ferromagnetism appears to set on smoothly .
figure [ fig03 ] shows @xmath30 [ where @xmath60 for systems up to @xmath61 sites .
additionally , the cluster - size distribution @xmath62 indicates the emergence of a large component without passing through a scale - free critical distribution . for the 2d tasaki lattice ( up to @xmath63 ) .
right : normalized number of clusters @xmath62 for @xmath64 .
] _ extent of nonpercolating ( paramagnetic ) phase._the critical density for pcp exceeds that of the standard case ( @xmath65 @xcite ) , see fig .
[ fig03 ] , where the macroscopic moment at @xmath66 is seen to scale to zero with system size .
this reflects the breakup of the large clusters due to the repulsive effective interactions .
_ percolating phase with unsaturated ferromagnetism._for higher densities , there appears a regime of _ unsaturated _ ferromagnetism , where @xmath67 , illustrated by @xmath68 in figs .
[ fig01 ] and [ fig03 ] .
the existence of this regime is transparent from the percolation viewpoint : in standard percolation the largest cluster excludes a nonzero density , @xmath69 for @xmath70 , @xmath71 being the coordination number of the lattice . for pcp , with
its _ repulsive _ interactions , this is amplified ( see right panel in fig . [ fig01a ] ) .
however the power law is identical to that of standard percolation , showing that the repulsive interactions do not immediately lead to a breakup of the largest cluster , presumably on account of the high entropic cost of arranging voids into continuous lines separating two clusters .
we note some features of the canonical ensemble arising in the phase - coexistence regime . the high - density phase appears to form first as a compact nonpercolating object with a macroscopic magnetic moment [ the configuration at @xmath72 shown in fig .
[ fig01](d ) is in this region ] . for higher densities , including the case @xmath73 , the ferromagnetic phase spans across the system [ compare fig .
[ fig01](e ) ] .
the details of the phase - separated regime therefore contain much which is different from standard percolation including the bootstrap and correlated variants @xcite , in particular with regard to properties which are of interest to flat - band ferromagnetism .
these topics are the subject of ongoing studies @xcite .
_ conclusions and perspectives._we have considered pauli - correlated percolation , an unusual percolation problem arising in a strongly correlated flat - band system , where the weights of the geometrical configurations take nontrivial values due to the spin degeneracy and the pauli principle .
the pauli - correlated percolation problem can be examined exactly in 1d and simulated efficiently in 2d .
we found that the effectively repulsive interaction leads to a breaking up of the clusters , and thus to a first - order grand - canonical transition in 2d , at a density which is higher than that of standard site percolation .
for the underlying 2d tasaki - hubbard model our results imply ground - state ferromagnetism in a range of electron fillings from @xmath74 to @xmath75 . besides the 1d realization mentioned above @xcite and the hope of discovering corresponding quasi-2d materials , in @xcite the possible realization of flat - band ferromagnetism in organic polymers was discussed . on the other hand
the 2d version of the tasaki lattice is so simple that it seems to be a reasonable candidate for realization as a system of cold atoms in optical lattices @xcite . _
acknowledgments._the authors would like to thank j. chalker and a. nahum for valuable discussions .
m.m .
thanks daad ( a/10/84322 ) and dfg ( sfb 602 ) for support of his stays at the university of gttingen and university of magdeburg in 2010 - 2011 .
a.h . acknowledges support by the dfg through a heisenberg fellowship ( project ho 2325/4 - 2 ) .
o.d . acknowledges the kind hospitality of university of magdeburg in october - december of 2011 .
j.r . and o.d .
thank the dfg for support ( ri 615/21 - 1 ) .
99 w. heisenberg , z. phys . * 38 * , 411 ( 1926 ) ; z. phys . * 49 * , 619 ( 1928 ) .
a. mielke , j. phys .
a * 24 * , l73 ( 1991 ) ; * 24 * , 3311 ( 1991 ) ; * 25 * , 4335 ( 1992 ) ; phys .
lett .
a * 174 * , 443 ( 1993 ) .
h. tasaki , phys .
rev .
lett . * 69 * , 1608 ( 1992 ) .
a. mielke and h. tasaki , commun .
math .
phys . * 158 * , 341 ( 1993 ) .
t. okubo , m. yamada , a. thamizhavel , s. kirita , y. inada , r. settai , h. harima , k. takegahara , a. galatanu , e. yamamoto and y. onuki , j. phys . :
condensed matter * 15 * , l721 ( 2003 ) ; z. gulcsi , a. kampf , and d. vollhardt , prog .
theor .
. suppl . * 176 * , 1 ( 2008 ) .
o. derzhko , a. honecker , and j. richter , phys .
rev .
b * 76 * , 220402(r ) ( 2007 ) .
o. derzhko , j. richter , a. honecker , m. maksymenko , and r. moessner , phys .
rev .
b * 81 * , 014421 ( 2010 )
. d. n. sheng , z .- c .
gu , k. sun , and l. sheng , nat .
commun .
* 2 * , 389 ( 2011 ) ; k. sun , z. gu , h. katsura , and s. das sarma , phys .
rev .
lett . * 106 * , 236803 ( 2011 ) .
h. katsura , i. maruyama , a. tanaka , and h. tasaki , europhys . lett . * 91 * , 57007 ( 2010 ) . c. wu , d. bergman , l. balents , and s. das sarma , phys .
rev .
lett . * 99 * , 070401 ( 2007 ) ; y .- f .
wang , z .- c .
gu , c .- d . gong , and d. n. sheng , phys .
rev .
lett .
* 107 * , 146803 ( 2011 ) .
e. tang , j .- w .
mei , and x .- g .
wen , phys .
rev .
lett . * 106 * , 236802 ( 2011 ) ; j. schulenburg , a. honecker , j. schnack , j. richter , and h .- j .
schmidt , phys .
rev .
lett .
* 88 * , 167207 ( 2002 ) ; j. richter , o. derzhko , and j. schulenburg , phys .
rev .
lett . * 93 * , 107206 ( 2004 ) .
z. gulcsi , a. kampf , and d. vollhardt , phys .
rev .
lett . * 105 * , 266403 ( 2010 ) .
t. neupert , l. santos , c. chamon , and c. mudry , phys .
rev .
lett .
* 106 * , 236804 ( 2011 ) ; t. neupert , l. santos , s. ryu , c. chamon , and c. mudry , phys . rev .
lett . * 108 * , 046806 ( 2012 ) .
d. stauffer , phys .
rep . * 54 * , 1 ( 1979 ) .
m. b. isichenko , rev .
mod .
phys . * 64 * , 961 ( 1992 ) .
j. hoshen and r. kopelman , phys .
rev .
b * 14 * , 3438 ( 1976 ) . m. e. j. newman and r. m. ziff , phys .
rev .
lett . * 85 * , 4104 ( 2000 ) ; phys .
rev .
e * 64 * , 016706 ( 2001 ) .
k. hukushima and k. nemoto , j. phys .
soc .
jpn . * 65 * , 1604 ( 1996 ) .
j. chalupa , p. l. leath , and g. r. reich , j. phys .
c : solid state phys .
* 12 * , l31 ( 1979 ) .
a. weinrib , phys .
rev .
b * 29 * , 387 ( 1984 ) .
m. maksymenko , in progress ( 2012 ) .
v. apaja , m. hyrks , and m. manninen , phys .
rev .
a * 82 * , 041402(r ) ( 2010 ) ; m. hyrks , v. apaja , and m. manninen , arxiv:1201.0468 . |
entanglement is the quantum correlation exhibiting nonlocal ( nonseparable ) properties .
it is supposed to be never strengthened , on average , by local operations and classical communication ( locc ) .
in particular , entanglement in _
multi_-parties is of fundamental interest in quantum many - body theory @xcite , and makes quantum information processing ( qip ) , e.g. , distillation protocol , more efficient than that relying on entanglement only in _
two_-parties @xcite . here , we classify and characterize the multipartite entanglement which has yet to be understood , compared with the bipartite one . for the single copy of bipartite pure states on @xmath3 ,
we are interested in whether a state @xmath4 can convert to another state @xmath5 by locc .
it is convenient to consider the schmidt decomposition , @xmath6 where the computational basis @xmath7 is transformed to a local biorthonormal basis @xmath8 , and the schmidt coefficients @xmath9 can be taken as @xmath10 .
we call the number of nonzero @xmath11 the ( schmidt ) local rank @xmath12
. then the locc convertibility is given by a majorization rule between the coefficients @xmath9 of @xmath4 and those of @xmath5 @xcite .
this suggests that the structure of entangled states consists of partially ordered , continuous classes labeled by a set of @xmath9 . in particular , @xmath4 and @xmath5 belong to the same class under the locc classification if and only if all continuous @xmath9 coincide .
suppose we are concerned with a coarse grained classification by the so - called stochastic locc ( slocc ) @xcite , where we identify @xmath4 and @xmath5 that are interconvertible back and forth with ( maybe different ) nonvanishing probabilities .
this is because @xmath4 and @xmath5 are supposed to perform the same tasks in qip although their probabilities of success differ .
later , we find that this slocc classification is still fine grained to classify the multipartite entanglement .
mathematically , two states belong to the same class under slocc if and only if they are converted by an _
invertible _ local operation having a nonzero determinant @xcite .
thus the slocc classification is equivalent to the classification of orbits of the natural action : direct product of general linear groups @xmath13 @xcite .
the local rank @xmath12 in eq.([eq : schmidt ] ) @xcite , equivalently the rank of @xmath14 , is found to be preserved under slocc .
a set @xmath15 of states of the local rank @xmath16 is a _ closed _ subvariety under slocc and @xmath17 is the singular locus of @xmath15 .
this is how the local rank leads to an `` onion '' structure ( mathematically the stratification ) : @xmath18 and @xmath19 give @xmath20 classes of entangled states .
since the local rank can decrease by _ noninvertible _ local operations , i.e. , general locc @xcite , these classes are totally ordered such that , in particular , the outermost generic set @xmath21 is the class of maximally entangled states and the innermost set @xmath22 is that of separable states . for the single copy of multipartite pure states ,
@xmath23 there are difficulties in extending the schmidt decomposition for a multiorthonormal basis @xcite . moreover , an attempt to use the tensor rank of @xmath24 @xcite falls down since @xmath15 , defined by it , is not always closed @xcite . in the 3 qubits , dr _ et al .
_ showed that slocc classifies the whole states @xmath25 into _ finite _ classes and in particular there exist two inequivalent , greenberger - horne - zeilinger ( ghz ) and w , classes of the tripartite entanglement @xcite .
they also pointed out that this case is exceptional since the action @xmath26 has _ infinitely many _ orbits in general ( e.g. , for @xmath27 ) . in this paper
, we classify multipartite entanglement in a unified manner based on the hyperdeterminant .
the advantages are three - fold .
( i ) this classification is equivalent to the slocc classification when slocc has finitely many orbits .
so it naturally includes the widely known bipartite and @xmath1-qubit cases .
( ii ) in the multipartite case , we need further slocc invariants in addition to the local ranks . for example , in the @xmath1-qubit case @xcite , the @xmath1-tangle @xmath28 , just the absolute value of the hyperdeterminant ( see eq .
( [ eq : tau ] ) ) , is utilized to distinguish ghz and w classes .
this work clarifies why the @xmath1-tangle @xmath28 appears and how these slocc invariants are related to the hyperdeterminant in general .
( iii ) our classification is also useful to multipartite mixed states .
a mixed state @xmath29 can be decomposed as a convex combination of projectors onto pure states .
considering how @xmath29 needs at least the outer class in the onion structure of pure states , we can also classify multipartite mixed states into the totally ordered classes ( for details , see appendix [ sec : mixed ] ) .
we concentrate on the pure states here .
the rest of the paper is organized as follows . in sec .
[ sec:2 ] , a duality between separable states and entangled states is introduced .
we find that the hyperdeterminant , associated to this duality , and its singularities lead to the slocc - invariant onion - like structure of multipartite entanglement .
the characteristics of the hyperdeterminant and its singularities are explained in sec .
[ sec:3 ] .
classifications of multipartite entangled states are exemplified in sec .
so as to reveal how they are ordered under slocc .
finally , the conclusion is given in sec .
[ sec : conclude ] .
in this section , we find that there is a duality between the set of separable states and that of entangled states . this duality derives the hyperdeterminant our classification is based on .
to introduce our idea , we first recall the geometry of pure states . in a complex ( finite ) @xmath20-dimensional hilbert space @xmath30 ,
let @xmath4 be a ( not necessarily normalized ) vector given by @xmath20-tuple of complex amplitudes @xmath31 in a computational basis .
the physical state in @xmath30 is a ray , an equivalence class of vectors up to an overall nonzero complex number .
then the set of rays constitutes the complex projective space @xmath32 ( the projectivization of @xmath30 ) , and @xmath33 , considered up to a complex scalar multiple , gives homogeneous coordinates in @xmath32 . for a composite system which consists of @xmath34 and @xmath35 , the whole hilbert space is the tensor product @xmath36 and
the associated projective space is @xmath37 .
a set @xmath38 of the separable states is the mere cartesian product @xmath39 , whose dimension @xmath40 is much smaller than that of the whole space @xmath25 , @xmath41 .
this @xmath38 is a closed , smooth algebraic subvariety ( segre variety ) defined by the segre embedding into @xmath42 @xcite , @xmath43 denoting homogeneous coordinates in @xmath42 by @xmath44 , we find that the segre variety @xmath38 is given by the common zero locus of @xmath45 homogeneous polynomials of degree @xmath46 : @xmath47 where @xmath48 .
note that this condition implies that all @xmath49 minors of `` matrix '' @xmath50 equal 0 ; i.e. , the rank of @xmath50 is 1 .
thus we have @xmath51 , which agrees with the slocc classification by the local rank in the bipartite case .
now consider the multipartite cartesian product @xmath52 in the segre embedding into @xmath53 . because this segre variety @xmath38 is the projectivization of a variety of the matrices @xmath54
, it gives a set of the completely separable states in @xmath55 . by another segre embedding ,
say @xmath56 , we also distinguish a set of separable states where only 1st and 2nd parties can be entangled , i.e. , when we regard 1st and 2nd parties as one party , an element of this set is completely separable for `` @xmath57 '' parties .
this is how , also in the multipartite case , we can classify all kinds of _ separable _ states , typically lower dimensional sets . note that , in the multipartite case , this check for the separability is more strict than the check by local ranks @xcite . we rather want to classify _ entangled _ states , typically higher dimensional complementary sets of separable states .
our strategy is based on the duality in algebraic geometry ; a hyperplane in @xmath58 forms the point of a dual projective space @xmath59 , and conversely every point @xmath60 of @xmath58 is tied to a hyperplane @xmath61 in @xmath59 as the set of all hyperplanes in @xmath58 passing through @xmath60 .
remarkably , the projective duality between projective subspaces , like the above example , can be extended to an involutive correspondence between irreducible algebraic subvarieties in @xmath58 and @xmath59 .
so we define a projectively dual ( irreducible ) variety @xmath62 as the closure of the set of all hyperplanes tangent to the segre variety @xmath38 .
let us observe ( and see the reason later ) that , in the bipartite case seen in sec .
[ sec : intro ] , the variety @xmath63 of the _ degenerate _
@xmath64 matrices @xmath65 is projectively dual to the variety @xmath66 of the matrices @xmath67 . that is , @xmath63 is the dual variety @xmath68 . following an analogy with a 2-dimensional ( bipartite ) case ,
an @xmath2-dimensional matrix @xmath69 is called _ degenerate _ if and only if it ( precisely , its projectivization ) lies in the projectively dual variety @xmath68 of the segre variety @xmath38 .
in other words , identifying the space of @xmath2-dimensional matrices with its dual by means of the pairing , @xmath70 we see that @xmath71 is degenerate if and only if its orthogonal hyperplane @xmath72 is tangent to @xmath38 at some nonzero point @xmath73 .
analytically , a set of equations , @xmath74 ( @xmath75 and @xmath76 ) , has at least a nontrivial solution @xmath77 of every @xmath78 , and then @xmath79 is called a critical point .
the above condition is also equivalent to saying that the kernel @xmath80 of @xmath81 is not empty , where @xmath80 is the set of points @xmath82 such that , in every @xmath83 , @xmath84 for the _ arbitrary _ @xmath85 . in the case of @xmath86 , the condition for eqs.([eq : critical ] )
coincides with the usual notion of degeneracy and means that @xmath71 does not have the full rank .
it shows that @xmath68 is nothing but @xmath63 .
in particular , @xmath68 , defined by this condition , is of codimension 1 and is given by the ordinary determinant @xmath87 , if and only if @xmath71 is a square @xmath88 matrix .
in the @xmath2-dimensional case , if @xmath68 is a hypersurface ( of codimension 1 ) , it is given by the zero locus of a unique ( up to sign ) irreducible homogeneous polynomial over @xmath89 of @xmath90 .
this polynomial is the hyperdeterminant introduced by cayley and is denoted by @xmath91 . as usual , if @xmath68 is not a hypersurface , we set @xmath91 to be 1
. remember that , in the _ bipartite _ case , we classify the states @xmath92 as the generic entangled states , the states @xmath93 as the next generic entangled states , and so on .
likewise , we aim to classify the _ multipartite _ entangled states into the onion structure by the dual variety @xmath68 ( @xmath94 ) , its singular locus @xmath95 and so on ( i.e. , by every closed subvariety ) , instead of the tensor rank @xcite .
in order to classify multipartite entanglement into the slocc - invariant onion structure , we explore the dual variety @xmath68 ( zero hyperdeterminant ) and its singular locus in this section .
we utilize the hyperdeterminant , the generalized determinant for higher dimensional matrices by gelfand _
its absolute value is also known as an entanglement measure , the concurrence @xmath96 @xcite or 3-tangle @xmath28 @xcite for the 2 or 3-qubit pure case , respectively .
@xmath97 the following useful facts are found in @xcite . without loss of generality , we assume that @xmath98 .
the @xmath2-dimensional hyperdeterminant @xmath91 of format @xmath99 exists , i.e. , @xmath68 is a hypersurface , if and only if a `` polygon inequality '' @xmath100 is satisfied . for @xmath86 , this condition
is reduced to @xmath101 as desired , and @xmath91 coincides with @xmath102 .
the matrix format is called boundary if @xmath103 and interior if @xmath104 .
note that ( i ) the boundary format includes the `` bipartite cut '' between the 1st party and the others so that it is mathematically tractable .
( ii ) the interior format includes the @xmath105-qubit case .
we treat hereafter the format where the polygon inequality holds and @xmath68 is the largest closed subvariety , defined by the hypersurface @xmath106 .
@xmath91 is relatively invariant ( invariant up to constant ) under the action of @xmath107 . in particular
, interchanging two parallel slices ( submatrices with some fixed directions ) leaves @xmath91 invariant up to sign , and @xmath91 is a homogeneous polynomial in the entries of each slice . since it is ensured that @xmath68 , @xmath95 , and further singularities are invariant under slocc , our classification is equivalent to or coarser than the slocc classification .
later , we see that the former and the latter correspond to the case where slocc gives finitely and infinitely many classes , respectively
. it would not be easy to calculate @xmath108 directly by its definition that eqs.([eq : critical ] ) have at least one solution .
still , the schlfli s method enables us to construct @xmath109 of format @xmath110 ( @xmath2 qubits ) by induction on @xmath2 @xcite . for @xmath86 , by definition @xmath111 .
suppose @xmath112 , whose degree of homogeneity is @xmath113 , is given .
associating an @xmath114-dimensional matrix @xmath115 ( @xmath116 ) to a family of @xmath2-dimensional matrices @xmath117 linearly depending on the auxiliary variable @xmath118 , we have @xmath119 . due to theorem 4.1 and 4.2 of @xcite , the discriminant @xmath120 of @xmath119
gives @xmath121 with an extra factor @xmath122 .
the sylvester formula of the discriminant @xmath120 for binary forms enables us to write @xmath123 in terms of the determinant of order @xmath124 ; @xmath125 where each @xmath126 is the coefficient of @xmath127 in @xmath119 , i.e. , @xmath128 .
note that because for @xmath129 or @xmath1 , the extra factor @xmath122 is just a nonzero constant , @xmath130 for the 3 or 4 qubits is readily calculated , respectively .
it would be instructive to check that @xmath131 in eq.([eq : tau ] ) is obtained in this way .
on the other hand , for @xmath132 , @xmath122 is the chow form ( related resultant ) of irreducible components of the singular locus @xmath95 .
these are due to the fact that @xmath95 has codimension @xmath46 in @xmath25 for any format of the dimension @xmath105 except for the format @xmath133 ( @xmath1-qubit case ) , which was conjectured in @xcite and was proved in @xcite .
so we have to explore @xmath95 not only to classify entangled states in the @xmath2 qubits , but to calculate @xmath123 inductively .
although @xmath134 has yet to be written explicitly , only its degree @xmath113 of homogeneity is known ( in corollary 2.10 of @xcite ) to grow very fast as @xmath135 for @xmath136 .
we describe the singular locus of the dual variety @xmath68 .
the technical details are given in @xcite .
it is known that , for the boundary format , the next largest closed subvariety @xmath95 is always an irreducible hypersurface in @xmath68 ; in contrast , for the interior one , @xmath95 has generally two closed irreducible components of codimension 1 in @xmath68 , _ node_-type @xmath137 and _
cusp_-type @xmath138 singularities .
the rest of this subsection can be skipped for the first reading .
it is also illustrated for the @xmath1-qubit case in appendix [ sec : rep_3qubit ] . .
@xmath137 corresponds to the bitangent of @xmath38 , where both tangencies are of the first order .
@xmath138 corresponds to the tangent at an inflection point of @xmath38 , where its tangency is of the second order . ]
first , @xmath137 is the closure of the set of hyperplanes tangent to the segre variety @xmath38 at more than one point ( cf .
[ fig : sing_xv ] ) .
@xmath137 can be composed of closed irreducible subvarieties @xmath139 labeled by the subset @xmath140 including @xmath141 .
indicating that two solutions @xmath142 of eq.([eq : critical ] ) coincide for @xmath143 , the label @xmath144 distinguishes the pattern in these solutions . in order to rewrite @xmath139 ,
let us pick up a point @xmath145 such that its homogeneous coordinates @xmath146 for @xmath143 and @xmath147 for @xmath148 .
it is convenient to label the positions of @xmath149 in each @xmath150 by a multi - index @xmath151 $ ] .
for example , @xmath152 is labeled by @xmath153 $ ] and @xmath154 is just written by @xmath155 . according to eqs.([eq : critical ] ) , @xmath156 , tangent to @xmath38 at @xmath145 , consists of the matrices @xmath71 of all @xmath157 such that @xmath158 $ ] differs from @xmath151 $ ] of @xmath145 in at most one index .
then we can define @xmath139 as @xmath159 where @xmath160 acts on @xmath25 from the right and the bar stands for the closure .
second , @xmath138 is the set of hyperplanes having a critical point which is not a simple quadratic singularity ( cf .
[ fig : sing_xv ] ) .
precisely , the quadric part of @xmath81 at @xmath155 is a matrix @xmath161 , where the pairs @xmath162 are the row and the column index , respectively . denoting by @xmath163 the variety of the hessian @xmath164 in @xmath165
, we can define @xmath138 as @xmath166 this @xmath138 is already closed without taking the closure .
according to sec . [ sec:2 ] and [ sec:3 ]
, we illustrate the classification of multipartite pure entangled states for typical cases .
the classification of the 3 qubits under slocc has been already done in @xcite .
surprisingly , gelfand _ et al . _ considered the same mathematical problem by @xmath131 in example 4.5 of @xcite .
our idea is inspired by this example .
we complement the gelfand _
s result , analyzing additionally the singularities of @xmath68 in detail .
the dimensions , representatives , names , and varieties of the orbits are summarized as follows . the basis vector @xmath167 is abbreviated to @xmath168 . + dim 7 : @xmath169 ghz @xmath170 . +
dim 6 : @xmath171 w @xmath172 . + dim 4 : @xmath173 + biseparable @xmath174 @xmath175 for @xmath176 .
+ @xmath177 are three closed irreducible + components of @xmath178 . + dim 3 : @xmath179 completely separable @xmath180 + @xmath181 .
@xmath182 has the onion structure of six orbits on @xmath25 ( see fig . [
fig : onion_3bit ] ) , by excluding the orbit @xmath183 .
the dual variety @xmath68 is given by @xmath184 ( cf .
( [ eq : tau ] ) ) .
its dimension is @xmath185 .
the outside of @xmath68 is generic tripartite entangled class of the maximal dimension , whose representative is ghz .
this suggests that almost any state in the @xmath1 qubits can be locally transformed into ghz with a finite probability , and vice versa .
next , we can identify @xmath95 as @xmath138 , which is the union of three closed irreducible subvarieties @xmath186 for @xmath187 @xcite ( also see appendix [ sec : rep_3qubit ] ) .
for example , @xmath188 means by definition that , in addition to the condition of @xmath68 in sec .
[ sec : dual ] , there exists some nonzero @xmath189 such that @xmath190 for any @xmath191 ; i.e. , a set of linear equations @xmath192 for @xmath193 has a nontrivial solution @xmath189 .
this indicates that the `` bipartite '' matrix @xmath194 never has the full rank ( i.e. , six @xmath195 minors in eq.([eq : b1 ] ) are zero ) .
we can identify @xmath188 as the set @xmath196 , seen in sec .
[ sec : segre ] , of biseparable states between the 1st party and the rest of the parties .
its dimension is @xmath197 .
likewise , @xmath186 for @xmath198 gives the biseparable class for the 2nd or 3rd party , respectively .
so , the class of @xmath199 is found to be tripartite entangled states , whose representative is w. we can intuitively see that , among genuine tripartite entangled states , w is rare , compared to ghz @xcite . finally , the intersection of @xmath186 is the completely separable class @xmath180 , given by the segre variety @xmath38 of dimension @xmath1 .
another intuitive explanation about this procedure is seen in appendix [ sec : rep_3qubit ] .
-qubit case .
we utilize a duality between the smallest closed subvariety @xmath38 and the largest closed subvariety @xmath68 .
the dual variety @xmath68 ( zero hyperdeterminant ) and its singularities constitute slocc - invariant closed subvarieties , so that they classify the multipartite entangled states ( slocc orbits ) . ]
now we clarify the relationship of six classes by _ noninvertible _ local operations . because noninvertible local operations cause the decrease in local ranks @xcite , the partially ordered structure of entangled states in the @xmath1 qubits , included in fig .
[ fig : part_order ] , appears .
two inequivalent tripartite entangled classes , ghz and w , have the same local ranks @xmath200 for each party so that they are not interconvertible by the noninvertible local operations ( i.e. , general locc ) .
two classes hold different physical properties @xcite ; the ghz representative state has the maximal amount of generic tripartite entanglement measured by the @xmath1-tangle @xmath201 , while the w representative state has the maximal amount of ( average ) @xmath46-partite entanglement distributed over @xmath1 parties ( also @xcite ) . under locc ,
a state in these two classes can be transformed into any state in one of the three biseparable classes @xmath202 , where the @xmath203-th local rank is @xmath149 and the others are @xmath46 .
three classes @xmath204 never convert into each other .
likewise , a state in @xmath204 can be locally transformed into any state in the completely separable class @xmath180 of local ranks @xmath205 .
this is how the onion - like classification of slocc orbits reveals that multipartite entangled classes constitute the partially ordered structure .
it indicates significant differences from the totally ordered one in the bipartite case .
( i ) in the @xmath1-qubit case , all slocc invariants we need to classify is the hyperdeterminant @xmath131 in addition to local ranks .
( ii ) although noninvertible local operations generally mean the transformation further inside the onion structure , an outer class can not necessarily be transformed into the _ neighboring _ inner class .
a good example is given by ghz and w , as we have just seen . before proceeding to the @xmath132-qubit case , we drop in the format @xmath207 , which would give an insight into the structure of multipartite entangled states when each party has a system consisting of more than two levels .
this case is interesting since on the one hand ( contrary to the @xmath1-qubit case ) , it is typical that ghz and w are included in @xmath95 ; on the other hand ( similarly to the bipartite or @xmath1-qubit cases ) , slocc has still finite classes so that it becomes another good test for the equivalence to the slocc classification . besides
, it is a boundary format so that several subvarieties can be explicitly calculated , and enables us to analyze entanglement in the qubits system using an auxiliary level , like ion traps . + dim 11 : @xmath208 @xmath209 . +
dim 10 : @xmath210 @xmath211 .
+ dim 9 : @xmath212 , ghz @xmath213 . + dim 8 : @xmath214 , w @xmath215 .
+ dim 6 : @xmath216 biseparable @xmath217 + @xmath218 . + dim 5 : @xmath219 , biseparable @xmath220 @xmath221 . + dim 4 : @xmath222 , completely separable @xmath180 + @xmath223 .
the onion structure consists of eight orbits on @xmath25 under slocc ( see fig .
[ fig : onion_3x2x2 ] ) .
generic entangled states of the outermost class are given by nonzero @xmath108 , which can be calculated in the _ boundary _ format as the determinant associated with the cayley - koszul complex .
although this is one of the gelfand _
_ s recent successes for generalized discriminants , we avoid its detailed explanation here . according to theorem 3.3 of @xcite
, we have @xmath224 of degree @xmath225 , where @xmath226 is the @xmath227 minor of @xmath228 without the @xmath203-th column , respectively .
next , it is characteristic that @xmath95 is @xmath188 @xcite .
similarly to the @xmath1-qubit case in sec .
[ sec:3qubit ] , @xmath188 means that the `` bipartite '' matrix in eq .
( [ eq : a_3x2x2 ] ) does not have the full rank , i.e. , all four @xmath229 minors @xmath230 in eq .
( [ eq : a_3x2x2 ] ) are zero .
the slocc orbits which appear inside @xmath95 are essentially the same as the 3-qubit case .
format , including the @xmath1-qubit case . each class , corresponding to the slocc orbit ,
is labeled by the representative , local ranks , and the name .
noninvertible local operations , indicated by dashed arrows , degrade `` higher '' entangled classes into `` lower '' entangled ones . ] format , including the @xmath1-qubit case .
each class , corresponding to the slocc orbit , is labeled by the representative , local ranks , and the name .
noninvertible local operations , indicated by dashed arrows , degrade `` higher '' entangled classes into `` lower '' entangled ones . ] thus we obtain the partially ordered structure of multipartite entangled states as fig . [
fig : part_order ] .
the tripartite entanglement consists of four classes .
because the class of @xmath231 , whose representative is @xmath232 , and that of @xmath199 , whose representative is @xmath233 , have the same local ranks @xmath234 , they do not convert each other in the same reason as ghz and w do not .
however , the former two classes of the local ranks @xmath234 can convert to the latter two classes of @xmath200 by noninvertible local operations ( i.e. , locc ) . and
we can `` degrade '' these tripartite entangled classes into the biseparable or completely separable classes by locc in a similar fashion to the @xmath1 qubits .
we notice that 3 grades in the @xmath1-qubit case changed to 4 grades in the @xmath207 ( @xmath149-qutrit and @xmath46-qubit ) case .
in general , the partially ordered structure becomes `` higher '' , as the system of each party becomes the higher dimensional one .
we also see how the tensor rank @xcite is inadequate for the onion - like classification of slocc orbits .
further in the @xmath0-qubit case , our classification works .
the outermost class @xmath235 of generic @xmath2-partite entangled states is given by @xmath236 .
in @xmath237 , @xmath238 of degree 24 is explicitly calculated by the schlfli s construction in sec .
[ sec : schlafli ]
. it would be suggestive to transform any generic @xmath239-partite state ( @xmath240 ) to the `` representative '' of the outermost class by invertible local operations , @xmath241 where the continuous complex coefficients @xmath242 should satisfy @xmath243 thus three complex parameters remain in the outermost class ( since we consider rays rather than normalized state vectors ) .
this means that there are infinitely many same dimensional slocc orbits in the @xmath239 qubits , and the slocc orbits never locally convert to each other when their sets of the parameters are distinct .
it is also the case for the @xmath244 qubits .
note that , in @xmath245 , this outermost class @xmath246 corresponds to the family of generic states in verstraete _ et al .
_ s classification of the @xmath239 qubits by a different approach ( generalizing the singular value decomposition in matrix analysis to complex orthogonal equivalence classes ) , and @xmath68 contains their other special families @xcite .
the next outermost class is @xmath247 . in the @xmath239 qubits , @xmath95 is shown to consist of eight closed irreducible components of codimension 1 in @xmath68 ; @xmath248 , and six @xmath249 for @xmath250 @xcite .
they neither contain nor are contained by each other .
their intersections also give ( finitely ) many lower dimensional genuine @xmath239-partite entangled classes .
since the @xmath239-partite entangled classes necessarily have the same local ranks @xmath251 , these classes are not interconvertible by noninvertible local operations ( i.e. , any locc ) . as typical examples , ghz ( the maximally entangled state in bell s inequalities @xcite ) , @xmath252 ( i.e. , @xmath253 and the others are @xmath254 ) is included in the intersection of @xmath255 and six @xmath249 , but is excluded from @xmath138 .
in contrast , w , @xmath256 ( i.e. , @xmath257 and the others are @xmath254 ) is included in the intersection of @xmath138 and six @xmath249 but is excluded from @xmath255 . in the @xmath258 qubits , @xmath95
is shown to consist of just two closed irreducible components @xmath138 and @xmath255 @xcite .
we find that ghz and w are contained not only in @xmath259 but in @xmath95 ; i.e. , they have nontrivial solutions in eqs.([eq : critical ] ) , satisfying the singular conditions .
they correspond to different intersections of further singularities , similarly to the @xmath239 qubits .
in other words , they are peculiar , living in the border dimensions between entangled states and separable ones , in brief , the dual variety @xmath68 and its singularities lead to the _ coarse _ onion - like classification of slocc orbits , when slocc gives infinitely many orbits .
the partially ordered structure of multipartite pure entangled states becomes `` wider '' , as the number @xmath2 of parties increases .
although many inequivalent @xmath2-partite entangled classes appear in the @xmath2 qubits , they never locally convert to each other , as observed in @xcite . in particular ,
the majority of the @xmath2-partite entangled states never convert to ghz ( or w ) by locc , and the opposite conversion is also not possible .
this is a significant difference from the bipartite or @xmath1-qubit case , where almost any entangled state and the maximally entangled state ( ghz ) can convert to each other by locc with nonvanishing probabilities .
we have presented the onion - like classification of multipartite entanglement ( slocc orbits ) by the dual variety @xmath68 , i.e. , the hyperdeterminant @xmath108 .
it leads to the partially ordered structure , such as fig .
[ fig : part_order ] , of inequivalent multipartite entangled classes of pure states , which is significantly different from the totally ordered one in the bipartite case .
local ranks are not enough to distinguish these classes , and we need to calculate slocc invariants associated with @xmath108 . in other words , the generic entangled class of the maximal dimension ( the outermost class ) is given by the outside of @xmath68 ( @xmath260 ) , and other multipartite entangled classes appear as @xmath68 or its different singularities . analytically , the classification of multipartite entanglement corresponds to that of the number and pattern of the solutions in eqs .
( [ eq : critical ] ) .
this work reveals that the situation of the widely known bipartite or @xmath1-qubit cases , where the maximally entangled states in bell s inequalities belong to the generic class , is exceptional .
lying far inside the onion structure , the maximally entangled states ( ghz ) are included in the lower dimensional peculiar class in general , e.g. , for the @xmath261 qubits .
it suggests two points .
the majority of multipartite entangled states can not convert to ghz by locc , and vice versa .
so , we have given an alternative explanation to this observation , first made in @xcite by comparing the number of local parameters accessible in slocc with the dimension of the whole hilbert space .
moreover , there seems no a priori reason why we choose ghz states as the _ canonical _ @xmath2-partite entangled states , which , for example , constitute a minimal reversible entanglement generating set ( mregs ) in asymptotically reversible locc @xcite .
since the onion - like classification is given by every closed subset , not only our work enables us to see intuitively why , say in the @xmath1 qubits , the w class is rare compared to the ghz class , but it can be also extended to the classification of multipartite mixed states ( see appendix [ sec : mixed ] ) .
the onion - like classification seems to be reasonable in the sense that it coincides with the slocc classification when slocc gives finitely many orbits , such as the bipartite or @xmath1-qubit cases .
so two states belonging to the same class can convert each other by invertible local operations with nonzero probabilities . on the other hand , when slocc gives infinitely many orbits , this classification is still slocc - invariant , but may contain in one class infinitely many _ same _ dimensional slocc orbits which can not locally convert to each other even probabilistically .
for example , in the @xmath239-qubit case , the generic entangled class in eq.([eq : generic_4bit ] ) has three nonlocal continuous parameters .
note that it can be possible to make the onion - like classification finer , by characterizing the nonlocal continuous parameters in each class .
then , we may ask , what is the physical interpretation of the onion - like classification in the case of infinitely many slocc orbits ? although a simple answer has yet to be found , we discuss two points .
( i ) let us consider _ global _ unitary operations which create the multipartite entanglement .
on the one hand , states in distinct classes would have the different complexity of the global operations , since they have the distinct number and pattern of nonlocal parameters . on the other hand ,
states in one class are supposed to have the equivalent complexity , since they just correspond to different `` angles '' of the global unitary operations .
( ii ) we can consider the case where _ more than one _ state are shared , including the asymptotic case . even in two shared states
, there can exist a local conversion which is impossible if they are operated separately , such as the catalysis effect @xcite .
so we can expect that we do it more efficiently in this situation , and the coarse classification may have some physical significance .
this problem remains unsettled even in the bipartite case .
finally , two related topics are discussed .
( i ) the absolute value @xmath262 of the hyperdeterminant , representing the amount of generic entanglement , is an entanglement monotone by vidal @xcite .
this never conflicts with the property that the maximally entangled states in bell s inequalities ( ghz ) generally have a zero @xmath109 .
a single entanglement monotone is insufficient to judge the locc convertibility , and generic entangled states of the nonzero @xmath109 can not convert to ghz in spite of decreasing @xmath262 .
( ii ) the @xmath1-tangle @xmath263 first appeared in the context of so - called entanglement sharing @xcite ;
i.e. , in the @xmath1 qubits , there is a constraint ( trade - off ) between the amount of @xmath46-partite entanglement and that of @xmath1-partite entanglement . by using the entanglement measure ( concurrence @xmath96 ) for the @xmath46-qubit _ mixed _ entangled states ,
this is written as @xmath264 , and @xmath28 is defined by @xmath265 for the @xmath1-qubit _ pure _ entangled states .
we expect that , in turn , the hyperdeterminant @xmath109 gives a clue to find the entanglement measure of more than @xmath46-qubit _ mixed _ states .
the author would like to thank m. wadati , m. murao , g. kato , the members in the erato project of quantum computation and information , and the participants in the sixth quantum information technology symposium in japan on may ( 2002 ) for the most helpful discussions .
the work is partially supported by the erato project .
in sec . [ sec:4 ] , we have classified entangled classes ( slocc orbits ) , utilizing slocc - invariant closed subvarieties such as the dual variety @xmath68 and its singularities . in this appendix
, we give an intuitive explanation about our technique .
we obtain entangled classes _ by their representatives_. the @xmath1-qubit case is exemplified , and the notation and terminology of sec .
[ sec : sing ] is followed .
as the representative of the outermost generic entangled class , almost any state ( indeed , satisfying @xmath266 ) in the whole space @xmath267 is qualified .
the ghz state @xmath212 is chosen among them , since it can be seen as the multidimensional analog of the identity matrix .
we look for the representative of the dual variety @xmath68 , which is qualified as that of the next outermost entangled class . when @xmath68 is the hyperplane tangent to the segre variety @xmath38 at @xmath155 such that @xmath268 ( @xmath176 ) , the `` @xmath155-section '' of @xmath68 is given as @xmath269 in order that eqs.([eq : critical ] ) have the nontrivial solution @xmath155 .
this suggests that the representative of @xmath68 is the w state @xmath214 , since the states given by eq.([eq : xv_o ] ) and w convert to each other under some invertible local operations @xmath270 as @xmath271 the candidates for the next outer entangled class are two , node - type @xmath137 and cusp - type @xmath138 , singularities of @xmath68 .
we first consider the representative of @xmath188 . according to eq.([eq : node ] ) , the @xmath155-section of @xmath188 is given as @xmath272 we find that the representative of @xmath188 is the biseparable state @xmath219 in @xmath220 , checking that @xmath273 in the same manner , @xmath274 or @xmath275 represents the biseparable class @xmath276 or @xmath277 , respectively .
let us second analyze the representative of @xmath138 .
in terms of the quadric part @xmath278 of @xmath81 at @xmath155 : @xmath279 the @xmath155-section of @xmath138 is given as @xmath280 we have three possibilities for @xmath281 . in the case of @xmath282 , this component of @xmath138 represents the biseparable class @xmath220 , since @xmath283 likewise , in the case of @xmath284 or @xmath285 , each component of @xmath138 corresponds to the biseparable class @xmath276 or @xmath277 , respectively . remembering that each @xmath204 is characterized by @xmath186 for @xmath176
, we have shown that @xmath138 has three irreducible components @xmath186 .
thus , the next outer entangled classes are three biseparable classes @xmath204 , which never contain nor are contained by each other . in general , remaining entangled classes are given by further singularities of @xmath68 such as combinations of the above @xmath137 and @xmath138 , or genuine higher singularities .
in the @xmath1-qubit case , since we see that @xmath186 , representing the biseparable class @xmath204 , is just characterized as @xmath286 , there remains just one smaller closed irreducible subvariety @xmath287 as their intersection @xmath288 .
this segre variety @xmath38 represents the completely separable class @xmath180 , whose representative is @xmath222 . in the text ,
we have carried out the above procedure in the `` @xmath155-free '' manner ( @xmath155 should be taken as _
any _ state on @xmath38 ) , and have obtained entangled classes as ( difference ) subsets .
it enables us to decide readily which entangled class a given state @xmath4 belongs to .
after the classification of entangled classes , we can clarify their partially ordered structure under noninvertible local operations in the same manner as in the text .
the onion structure is also useful for the slocc - invariant classification of mixed entangled states .
a mixed state @xmath29 can be written as a convex combination of projectors onto pure states ( extremal points ) , @xmath289 where @xmath290 is the pure state belonging to the slocc orbit @xmath291 of an index @xmath292 .
@xmath292 is labeled by the closed subvariety @xmath293 ( i.e. , the closure of @xmath291 ) such as @xmath68 , @xmath139 , @xmath138 , and @xmath38 .
note that , in the multipartite case , there can be many closed subvarieties @xmath293 which never contain nor are contained by each other ; for example , @xmath188 , @xmath274 , and @xmath275 in fig .
[ fig : onion_3bit ] .
so , by taking the union of these `` competitive '' closed subvarieties @xmath293 ( it will form their convex hull in the space of @xmath29 ) , we pick up only _ totally ordered _
ones , e.g. , @xmath25 , @xmath68 , @xmath294 , and @xmath38 in fig .
[ fig : onion_3bit ] , for convenience later . now , we are concerned with at most how various classes of pure entangled states the mixed state @xmath29 consists of .
we take the maximal closure of @xmath293 appeared in eq.([eq : rho ] ) and denote it by @xmath295 . however , since @xmath29 can be decomposed into the form of eq.([eq : rho ] ) in infinitely many ways , we should take the minimal closure of @xmath295 over all possible decompositions , and write it as @xmath296 .
every convex subset @xmath297 of @xmath298 is closed such that @xmath297 of the smaller @xmath292 is contained by that of the larger one .
in other words , @xmath297 of the larger @xmath292 consists of more classes ( slocc orbits ) of pure entangled states .
that is how the mixed state @xmath29 is classified into the _ closed _ convex subsets @xmath297 under slocc . in the bipartite case
, @xmath296 is called the schmidt number @xcite ( since @xmath293 is just labeled by the schmidt local rank , as seen in sec .
[ sec : intro ] ) .
also in the @xmath1-qubit case , this kind of the classification has been done in @xcite , and four classes appear , following the above recipe : + ( i ) ghz class @xmath299 ( consisting of all pure states ) ; + ( ii ) w class @xmath300 ( consisting of the pure w , biseparable , or separable states ) ; + ( iii ) biseparable class @xmath301 ( consisting of the pure biseparable or separable states ) ; + ( iv ) separable class @xmath302 ( consisting of only the pure separable states ) . needless to say , the trouble is considering all possible decompositions in eq .
( [ eq : rho ] ) .
so , it is very difficult to give the criterion to distinguish each closed convex subset @xmath297 , even to distinguish the separable subset @xmath302 .
still , it would be interesting to observe that a witness operator @xmath303 , which forms the tangent hyperplane @xmath304 detecting @xmath297 a given @xmath29 belongs to , shares the same idea as our dual variety .
for example , let us consider two einstein - podolsky - rosen ( epr ) pairs @xmath307 in the @xmath239 qubits .
since their local ranks are @xmath251 , we can not distinguish this state from genuine @xmath239-partite entangled states ( cf .
[ sec : nqubit ] ) .
in contrast , we readily find that the state is included in @xmath308 , so that it is separable ( not genuine @xmath239-partite entangled ) . if there exists a noninvertible local operation between slocc orbits , some of the local ranks decrease
. note that the converse is generally not the case .
however , in our examples in sec .
[ sec:4 ] , we do find some noninvertible local operations for the decrease of local ranks .
g. vidal , j. mod . opt . * 47 * , 355 ( 2000 ) .
the proof is given in the same way that the 3-tangle @xmath309 is proved an entanglement monotone in appendix b of @xcite , by generalizing the degree 4 of homogeneity of @xmath309 to @xmath113 of @xmath262 .
it follows the arithmetic mean - geometric mean inequality that @xmath262 is greater than or equal to the average of @xmath262 resulted from any local povm .
the absolute value should be taken because @xmath109 is invariant up to sign under permutations of the parties . | we find that multidimensional determinants `` hyperdeterminants '' , related to entanglement measures ( the so - called concurrence or 3-tangle for the 2 or 3 qubits , respectively ) , are derived from a duality between entangled states and separable states . by means of the hyperdeterminant and its singularities ,
the single copy of multipartite pure entangled states is classified into an onion structure of every closed subset , similar to that by the local rank in the bipartite case .
this reveals how inequivalent multipartite entangled classes are partially ordered under local actions . in particular , the generic entangled class of the maximal dimension , distinguished as the nonzero hyperdeterminant , does not include the maximally entangled states in bell s inequalities in general ( e.g. , in the @xmath0 qubits ) , contrary to the widely known bipartite or @xmath1-qubit cases .
it suggests that not only are they never locally interconvertible with the majority of multipartite entangled states , but they would have no grounds for the canonical @xmath2-partite entangled states .
our classification is also useful for the mixed states . |
the universality of search behaviors is reflected in multitudes of studies in different fields including control systems , distributed computing and biology .
we use tools from distributed computing to study a biologically inspired scenario in which a group of agents , initially located at one central location , cooperatively search for treasures in the plane .
the goal of the search is to locate nearby treasures as fast as possible and at a rate that scales well with the number of participating agents . a variety of animal species search for food around a central location that serves the search s initial point , final destination or both @xcite .
this central location could be a food storage area , a nest where offspring are reared or simply a sheltered or familiar environment .
central place foraging holds a strong preference to locating nearby food sources before those that are further away .
possible reasons for that are , for example : ( 1 ) decreasing predation risk @xcite , ( 2 ) increasing the rate of food collection once a large quantity of food is found @xcite , ( 3 ) holding a territory without the need to reclaim it @xcite , and ( 4 ) the ease of navigating back after collecting the food using familiar landmarks @xcite . searching in groups
can increase foraging efficiency @xcite . in some extreme cases , food is so scarce that group searching is believed to be required for survival @xcite .
proximity of the food source to the central location is again important in this case .
for example , in the case of recruitment , a nearby food source would be beneficial not only to the individual that located the source but also increase the subsequent retrieval rate for many other collaborators @xcite .
foraging in groups can also facilitate the defense of larger territories @xcite .
eusocial insects ( e.g. , bees and ants ) engage in highly cooperative foraging , which can be expected as these insects reduce competition between individuals to a minimum and share any food that is found .
social insects often live in a single nest or hive , which naturally makes their foraging patterns central .
little is known about the communication between foragers , but it is believed that in some scenarios communication may become impractical @xcite .
this holds , for example , if the foragers start the search at different times and remain far apart ( which may be necessary to avoid unnecessary overlaps ) .
hence , the question of how efficient the search can be if the communication is limited , or altogether absent , is of great importance . in this paper
, we theoretically address general questions of collective searches in the particular natural setting described above .
more precisely , we introduce the _ ants nearby treasure search ( ants ) _ problem , in which @xmath0 identical ( probabilistic ) agents , initially placed at some central location , collectively search for a treasure in the two - dimensional plane .
the treasure is placed by an adversary at some target location at distance @xmath1 from the central location , where @xmath1 is unknown to the agents .
the goal of the agents is to find the treasure as fast as possible , where the time complexity is evaluated as a function of both @xmath0 and @xmath1 . in the context of search algorithms , evaluating the time as a function of @xmath1
was first introduced in a classical paper by baeza - yates _ et al . _
@xcite , who studied the _ cow - path _ problem ( studied also in @xcite ) .
the ants problem generalizes the cow - path problem , as it considers multiple identical agents instead of a single agent ( a cow in their terminology ) .
indeed , in this distributed setting , we are concerned with the _ speed - up _ measure ( see also , @xcite ) , which aims to capture the impact of using @xmath0 searchers in comparison to using a single one . note that the objectives of quickly finding nearby treasures and having significant speed - up may be at conflict .
that is , in order to ensure that nearby treasures are quickly found , a large enough fraction of the search force must be deployed near the central location . in turn , this crowding can potentially lead to overlapping searches that decrease individual efficiency .
it is a rather straightforward observation that the time required for finding the treasure is @xmath2 .
our focus is on the question of how agents can approach this bound if their communication is limited or even completely absent . in particular , as information of the size of the foraging group may not be available to the individual searchers , we concentrate our attention on the question of how important it is for agents to know ( or estimate ) their total number . as we later show , the lack of such knowledge may have a non - negligible impact on the performance .
we introduce and investigate the ants problem , a generalization of the cow - path problem .
first , we show that if the agents have a constant approximation of their total number @xmath0 , then there exists a rather simple search algorithm whose expected running time is @xmath3 , making it @xmath4-competitive .
we then turn our attention to _ uniform _ searching algorithms , in which the agents are not assumed to have any information regarding @xmath0 .
we completely characterize the speed - up penalty that must be paid when using uniform algorithms .
specifically , we show that for a function @xmath5 such that @xmath6 converges , there exists a uniform search algorithm that is @xmath7-competitive . on the other hand
, we show that if @xmath6 diverges , then there is no uniform search algorithm that is @xmath7-competitive . in particular
, this implies that , for every constant @xmath8 , there exists a uniform search algorithm that is @xmath9-competitive , but there is no uniform search algorithm that is @xmath10-competitive .
hence , the penalty for using uniform algorithms is slightly more than logarithmic in the number of agents .
in addition , we give a lower bound for the intermediate setting in which the agents are given some estimation of @xmath0 .
as a special case , this lower bound implies that for any constant @xmath11 , if each agent is given a ( one - sided ) @xmath12-approximation of @xmath0 , then the competitiveness is @xmath13 .
informally , our results imply that the agents can potentially perform well without any knowledge of their total number @xmath0 , however , to further improve they must use non trivial information regarding @xmath0 .
finally , we propose a uniform search algorithm that is concurrently efficient and extremely simple and , as such , it may imply some relevance for actual biological scenarios .
our work falls within the scope of natural algorithms , a recent attempt to investigate biological phenomena from an algorithmic perspective @xcite .
collective search is a classical problem that has been extensively studied in different fields of science .
group living and food sources that have to be actively sought after make collective foraging a widespread biological phenomenon .
social foraging theory @xcite makes use of economic and game theory to optimize food exploitation as a function of the group size and the degree of cooperativity between agents in different environmental settings .
social foraging theory has been extensively compared to experimental data ( see , e.g. , @xcite ) but does not typically account for the spatial characteristics of resource abundance .
central place foraging theory @xcite assumes a situation in which food is collected from a patchy resource and is returned to a particular location , such as a nest .
this theory is used to calculate optimal durations for exploiting food patches at different distances from the central location and has also been tested against experimental observations @xcite .
collective foraging around a central location is particularly interesting in the case of social insects where large groups forage cooperatively with , practically , no competition between individuals .
harkness and mardouras @xcite have conducted a joint experimental and modeling research into the collective search behavior of non - communicating desert ants .
modeling the ants trajectories using biased random walks , they reproduce some of the experimental findings and demonstrate significant speed - up with group size
. in bold contrast to these random walks , reynolds @xcite argues that lvy flights with a power law that approaches unity is the optimal search strategy for cooperative foragers as traveling in straight lines tends to decrease overlaps between searchers . from an engineering perspective , the distributed cooperation of a team of autonomous agents ( often referred to as robots or uavs : unmanned aerial vehicles ) is a problem that has been extensively studied .
these models extend single agent searches in which an agent with limited sensing abilities attempts to locate one or several mobile or immobile targets @xcite .
the memory and computational capacities of the agent are typically large and many algorithms rely on the construction of cognitive maps of the search area that includes current estimates that the target resides in each point @xcite .
the agent then plans an optimal path within this map with the intent , for example , of optimizing the rate of uncertainty decrease @xcite .
cooperative searches typically include communication between the agents that can be transmitted up to a given distance , or even without any restriction .
models have been suggested where agents can communicate by altering the environment to which other agent then react @xcite .
cooperation without communication has also been explored to some extent @xcite but the analysis puts no emphasis on the speed - up of the search process .
in addition , to the best of our knowledge , no works exist in this context that put emphasis on finding nearby targets faster than faraway one .
similar problems studied in this context are pattern formation @xcite , rendezvous @xcite , and flocking @xcite .
it is important to stress that in all those engineering works , the issue of whether the robots know their total number is typically not addressed , as obtaining such information does not seem to be problematic .
furthermore , in many works , robots are not identical and have unique identities . in the theory of computer science ,
the exploration of graphs using mobile agents is a central question .
most of the research for graph exploration is concerned with the case of a single deterministic agent exploring a finite graph ( typically , with some restrictions on the resources of the agent and/or on the graph structure ) .
for example , in @xcite the agent explores strongly connected directed finite graphs , and in @xcite the agent explores undirected finite graphs .
when it comes to probabilistic searching , the random walk is a natural candidate , as it is extremely simple , uses no memory , and trivially self - stabilizes .
unfortunately , however , the random walk turns out to be inefficient in a two - dimensional infinite grid .
specifically , in this case , the expected hitting time is infinite , even if the treasure is nearby .
evaluating the time to find the treasure as a function of @xmath1 , the initial distance to the treasure , was studied in the context of the cow - path problem .
one of the first papers that studied the cow - path problem is a paper by baeza - yates _ et al . _
@xcite , in which the competitive ratio for deterministically finding a point on the real line is shown to be nine . considering the two - dimensional case , baeza - yates _ et al .
_ proved that the spiral search algorithm is optimal up to lower order terms .
randomized algorithms for the problem were studied by kao _
et al . _
@xcite , for the infinite star topology .
et al . _
@xcite studied an early variant of the cow - path problem on a binary tree .
recently , demaine _ et al . _
@xcite considered the cow - path problem with a double component price : the first is distance and the second is turn cost .
lpez - ortiz and sweet @xcite extended the cow - path problem by considering @xmath0 agents .
however , in contrast to the ants problem , the agents they consider are not identical , and the goal is achieved by ( centrally ) designing a different specific path for each of the @xmath0 agents .
in general , the more complex setting of using multiple identical agents has received much less attention .
exploration by deterministic multiple agents was studied in , e.g. , @xcite .
to obtain better results when using several identical deterministic agents , one must assume that the agents are either centrally coordinated or that they have some means of communication ( either explicitly , or implicitly , by being able to detect the presence of nearby agents ) .
when it comes to probabilistic agents , analyzing the speed - up measure for @xmath0-random walkers has recently gained attention . in a series of papers , initiated by alon _
et al . _
@xcite , a speed - up of @xmath14 is established for various finite graph families , including , in particular , expanders and random graphs @xcite .
while some graph families enjoy linear speed - up , for many graph classes , to obtain linear speed - up , @xmath0 has to be quite small .
in particular , this is true for the two - dimensional @xmath15-node grid , where a linear speed up is obtained when @xmath16 . on the other hand , the cover time of @xmath17-dimensional @xmath15-node grid is always @xmath18 , regardless of @xmath0 .
hence , when @xmath0 is polynomial in @xmath15 , the speed up is only logarithmic in @xmath0 . the situation with infinite grids is even worse .
specifically , though the @xmath0-random walkers would find the treasure with probability one , the expected time to find the treasure becomes infinite .
the question of how important it is for individual processors to know their total number has recently been addressed in the context of locality . generally speaking , it has been observed that for several classical local computation tasks , knowing the number of processors is not essential @xcite . on the other hand , in the context of local decision , some evidence exist that such knowledge may be crucial for non - deterministic distributed decision @xcite .
we consider the ants problem , where @xmath0 mobile _ agents _ ( robots ) are searching for a _ treasure _ on the two - dimensional plane .
each agent has a field of view bounded by a small positive constant , hence , for simplicity , we can assume that the agents are actually walking on the integer two - dimensional infinite grid @xmath19 .
all @xmath0 agents starts the search from a central node @xmath20 of @xmath21 , called the _
source_. an adversary locates the treasure at some node @xmath22 of @xmath21 , referred to as the _ target _ node ; the agents have no a priori information about the location of @xmath22 . the goal of the agents it to _ find _ the treasure : this task is accomplished once at least one of the agents visits the node @xmath22 . the agents are probabilistic machines that can move on the grid , but can not communicate between themselves .
all @xmath0 agents are identical ( execute the same protocol ) .
an agent can traverse an edge of the grid in both directions .
we do not restrict the internal storage and computational power of agents , nevertheless , we note that all our upper bounds use simple procedures that can be implemented using relatively short memory . for example , with respect to navigation , our constructions only assume the ability to perform four basic procedures , namely : ( 1 ) choose a direction uniformly at random , ( 2 ) walk in a `` straight '' line to a prescribed distance , ( 3 ) perform a _ spiral search _ around a node is a particular deterministic local search algorithm ( see , e.g. , @xcite ) that starts at @xmath23 and enables the agent to visit all nodes at distance @xmath24 from @xmath23 by traversing @xmath25 edges , for every integer @xmath25 . for our purposes ,
since we are concerned only with asymptotic results , we can replace this atomic navigation procedure with any procedure that guarantees such a property . for simplicity , in the remainder of the paper , we assume that for every integer @xmath25 , the spiral search of length @xmath25 starting at a node @xmath23 visits all nodes at distance at most @xmath26 from @xmath23 ] , and ( 4 ) return to the source node . on the other hand , for our lower bounds to hold , we do not require any restriction on the navigation capabilities . regarding the time complexity , we assume that traversing an edge takes one unit of time . furthermore ,
for the ease of presentation , we assume that the agents are synchronous , that is , each edge traversal costs precisely one unit of time ( and all internal computations are performed in zero time ) .
indeed , this assumption can easily be removed if we measure the time according to the slowest edge - traversal .
we also assume that all agents start the search simultaneously at the same time , denoted by @xmath28 .
this assumption can also be easily removed by starting to count the time after the last agent initiates the search .
we measure the cost of an algorithm by its _ expected running time _ , that is , the expected time ( from time @xmath28 ) until at least one of the agents finds the treasure .
we denote the expected running time of algorithm @xmath29 by @xmath30 .
the _ distance _ @xmath31 between two nodes @xmath32 and @xmath23 of @xmath21 is simply the hop distance between them , i.e. , the number of edges on a shortest path connecting @xmath32 and @xmath23 in @xmath21 .
the distance between a node @xmath32 and the source node @xmath20 is simply written @xmath33 , i.e. @xmath34 . let @xmath1 be the distance between the source node @xmath20 and the target node @xmath22 , i.e. , @xmath35 .
for a node @xmath32 of @xmath21 and an integer @xmath36 , let @xmath37 be the ball of radius @xmath36 centered at the node @xmath32 , formally , @xmath38 .
for short , we denote the ball of radius @xmath36 around the source node @xmath20 by @xmath39 , that is , @xmath40 .
note that if an agent knows @xmath1 , then it can potentially find the treasure in time @xmath41 , by going to any node at distance @xmath1 , and then performing a circle around the source of radius @xmath1 ( assuming , of course , that its navigation abilities enable it to perform such a circle ) . on the other hand , without knowledge about @xmath1 ,
an agent can find the treasure in time @xmath42 by performing a spiral search around the source .
when considering @xmath0 agents , it is easy to see whose expected running time is @xmath43
. clearly , @xmath44 , because it takes time @xmath1 to merely reach the treasure .
assume , towards a contradiction , that @xmath45 . in any execution of @xmath29 , by time @xmath46 , the @xmath0 agents can visit a total of at most @xmath47 nodes .
hence , by time @xmath46 , more than half of the nodes in @xmath48 were not visited .
therefore , there must exist a node @xmath49 such that the probability that @xmath32 is visited by time @xmath46 ( by at least one of the agents ) is less than @xmath50 .
if the adversary locates the treasure at @xmath32 , then the expected time to find the treasure is strictly greater than @xmath43 , which contradicts the assumption . ] that the expected running time is @xmath51 , even if the total number of agents @xmath0 is known to all agents , and even if we relax the model and allow agents to freely communicate between each other .
it follows from theorem [ thm : know - k ] that if @xmath0 is known to agents then there exists a search algorithm whose expected running time is asymptotically optimal , namely , @xmath3 .
we evaluate the performance of an algorithm that does not assume the precise knowledge of @xmath0 with respect to this aforementioned optimal time .
formally , let @xmath52 be a function of @xmath0 .
an algorithm @xmath29 is _
@xmath52-competitive _ if @xmath53 for every integers @xmath0 and @xmath1 .
we shall be particularly interested in the performances of _ uniform _ algorithms : these are algorithms in which no information regarding @xmath0 is available to agents .
( the term `` uniform '' is chosen to stress that agents execute the same algorithm regardless of their number , see , e.g. , @xcite . )
our first theorem asserts that agents can achieve an asymptotically optimal running time if they know the precise value of @xmath0 . as a corollary
, it follows that , in fact , to obtain such a bound , it is sufficient to assume that agents only know a constant approximation of @xmath0 .
[ thm : know - k ] assume that the agents know their total number , @xmath0
. then , there exists a ( non - uniform ) search algorithm running in expected time @xmath3 . for @xmath54 ,
set @xmath55 .
consider the following algorithm ( see also figure [ fig : phasei ] for an illustration ) .
.,height=226 ] fix a positive integer @xmath56 and consider the time @xmath57 until each agent completed @xmath56 phases @xmath58 with @xmath59 .
each time an agent performs phase @xmath58 , the agent finds the treasure if the chosen node @xmath32 belongs to the ball @xmath60 around @xmath22 , the node holding the treasure .
note that at least some constant fraction of the ball @xmath60 is contained in @xmath61 , since @xmath62 .
the probability of choosing a node @xmath32 in that fraction is thus @xmath63 , which is at least @xmath64 for some positive constant @xmath65 .
thus , the probability that by time @xmath57 none of the @xmath0 agents finds the treasure ( while executing their respective @xmath56 phases @xmath58 ) is at most @xmath66 , which is at most @xmath67 for some constant @xmath68 greater than @xmath69 . for @xmath54 ,
let @xmath70 be the time required to execute a phase @xmath58 .
note that @xmath71 .
hence , the time until all agents complete stage @xmath72 for the first time is @xmath73 now fix @xmath74 .
it follows that for any integer @xmath56 , all agents complete their respective stages @xmath75 by time @xmath76 .
observe that by this time , all agents have completed at least @xmath77 phases @xmath58 with @xmath78 .
consequently , the probability that none of the @xmath0 agents finds the treasure by time @xmath79 is at most @xmath80 .
hence , the expected running time is at most @xmath81 this establishes the theorem .
fix a constant @xmath82 .
we say that the agents have a _
@xmath83-approximation _ of @xmath0 , if , initially , each agent @xmath84 receives as input a value @xmath85 satisfying @xmath86 . [ cor : known ] fix a constant @xmath82 .
assume that the agents have a @xmath83-approximation of their total number .
then , there exists a ( non - uniform ) search algorithm that is @xmath4-competitive .
each agent @xmath84 executes algorithm @xmath87 ( see the proof of theorem [ thm : know - k ] ) with the parameter @xmath0 equal to @xmath88 . by the definition of a @xmath83-approximation
, the only difference between this case and the case where the agents know their total number , is that for each agent , the time required to perform each spiral search is multiplied by a constant factor of at most @xmath89 .
therefore , the analysis in the proof of theorem [ thm : know - k ] remains essentially the same and the running time is increased by a multiplicative factor of at most @xmath89 .
we now turn our attention to the case of uniform algorithms .
[ thm : comp ] consider a function @xmath90 such that @xmath91 .
there exists a uniform search algorithm that is @xmath7-competitive .
consider the uniform search algorithm @xmath92 described below .
let us analyze the performances of the algorithm and show that its expected running time is @xmath93 , where @xmath94 .
we first note that it suffices to prove the statement when @xmath95 . indeed ,
if @xmath96 , then we may consider only @xmath1 agents among the @xmath0 agents and obtain an upper bound on the running time of @xmath97 , which is less than @xmath98 .
[ as:1 ] for every integer @xmath56 , the time until all agents complete big - stage @xmath56 is @xmath99 . for the assertion to hold ,
it is sufficient to prove that stage @xmath58 in big - stage @xmath56 takes time @xmath100 . to this end ,
notice that phase @xmath72 takes time @xmath101 .
therefore , stage @xmath58 takes time @xmath102 this establishes assertion [ as:1 ] .
set @xmath103 and fix an integer @xmath104
. then , there exists @xmath105 such that @xmath106 .
+ [ as:2 ] there exists a constant @xmath107 such that the probability that none of the agents finds the treasure while executing phase @xmath72 of stage @xmath58 is at most @xmath108 . to see this , first note that the treasure is inside the ball @xmath109
indeed , @xmath110 .
the total number of nodes in @xmath109 is @xmath111 , and at least half of the ball of radius @xmath112 around the treasure is contained in @xmath109 .
consequently , the probability for an agent @xmath84 to choose a node @xmath32 in a ball of radius @xmath112 around the treasure in phase @xmath72 of stage @xmath58 is @xmath113 if this event happens , then the treasure is found during the corresponding spiral search of agent @xmath84 . as a result
, there exists a positive constant @xmath114 such that the probability that none of the @xmath0 agents finds the treasure during phase @xmath72 of stage @xmath58 is at most @xmath115 .
this establishes assertion [ as:2 ] . by the time that all agents have completed their respective big - stage @xmath75
, all agents have performed @xmath116 stages @xmath58 with @xmath104 . by assertion [ as:2 ] , for each such @xmath58 , the probability that the treasure is not found during stage @xmath58 is at most @xmath108 for some constant @xmath117 .
hence , the probability that the treasure is not found during any of those @xmath116 stages is at most @xmath118 for some constant @xmath119 .
assertion [ as:1 ] ensures that all agents complete big - stage @xmath75 by time @xmath120 , so the expected running time is @xmath121 , as desired . setting @xmath122 yields the following corollary .
[ cor : comp ] for every positive constant @xmath123 , there exists a uniform search algorithm that is @xmath124-competitive .
we say that a function @xmath90 is _ sub - exponential _ if @xmath5 is non - decreasing and there exists a constant @xmath125 such that @xmath126 , for every @xmath127 .
[ th : lowerbound ] consider a sub - exponential function @xmath5 such that @xmath128 .
there is no uniform search algorithm that is @xmath7-competitive .
suppose that there exists a uniform search algorithm with running time less than @xmath129 .
hence , if @xmath130 , then @xmath131 , where @xmath132 .
suppose , towards a contradiction , that @xmath133 .
let @xmath43 be a ( sufficiently large ) integer and set @xmath134 .
that is , for the purpose of the proof , we assume that the treasure is actually placed at some faraway distance @xmath1 greater than @xmath46 .
this means , in particular , that by time @xmath46 the treasure has not been found yet . for every integer @xmath135 ,
set @xmath136 fix an integer @xmath58 in @xmath137 $ ] , and consider @xmath138 , the ball of radius @xmath139 around the source node
. we now deal with the case where the algorithm is executed by @xmath140 agents . for every subset @xmath141 of nodes in @xmath138 ,
let @xmath142 be the random variable indicating the number of nodes in @xmath141 that were visited by at least one of the @xmath140 agents by time @xmath46 .
( we write @xmath143 for @xmath144 . )
note that @xmath145 and , therefore , for each node @xmath146 the expected time to visit @xmath32 is at most @xmath147 .
thus , by markov s inequality , the probability that @xmath32 is visited by time @xmath46 is at least @xmath50 , i.e. , @xmath148 .
hence , @xmath149 .
now consider an integer @xmath58 in @xmath150 $ ] and set @xmath151 .
by linearity of expectation , @xmath152 .
consequently , by time @xmath46 , the expected number of nodes in @xmath153 that an agent visits is @xmath154 where the second equality follows from the fact that @xmath155 , and the third equality follows from the facts that @xmath156 and @xmath5 is sub - exponential .
in other words , for every integer @xmath58 in @xmath150 $ ] , the expected number of nodes in @xmath153 that each agent visits by time @xmath46 is @xmath157 . since the sets @xmath153 are pairwise disjoint
, the linearity of expectation implies that the expected number of nodes that an agent visits by time @xmath46 is @xmath158 consequently , @xmath159 must converge as @xmath43 goes to infinity , and hence so does @xmath160 .
this contradicts the assumption on @xmath5 . setting @xmath161 yields the following statement .
[ cor : lowerbound ] there is no uniform search algorithm that is @xmath10-competitive .
we now present a lower bound for the competitiveness of search algorithms assuming that agents are given approximations of their total number @xmath0 . as a special case , our lower bound implies that for any constant @xmath11 , if agents are given an estimation @xmath162 such that @xmath163 , then the competitiveness is @xmath13 .
that is , the competitiveness remains logarithmic even for relatively good approximations of @xmath0 .
formally , let @xmath164 $ ] .
we say that the agents have a _ @xmath12-approximation of @xmath0 _ if each agent @xmath84 receives as input an estimation @xmath165 for @xmath0 that satisfies : @xmath166 ( for example , if the agents have a @xmath12-approximation of @xmath0 , where @xmath123 is the constant function equal to @xmath50 , then , in particular , this means that if all agents are given the same value @xmath162 , then the real number @xmath0 of agents satisfies @xmath167 . )
[ th : lowerbound2 ] let @xmath164 $ ] .
assume that the agents have a @xmath12-approximation of @xmath0 .
if there exists a @xmath52-competitive algorithm , where @xmath168 is a non - decreasing function , then @xmath169 .
assume that there is a search algorithm for this case running in time @xmath170 , where @xmath168 is non - decreasing .
suppose that all agents receive the same value @xmath162 , which should serve as an estimate for @xmath0 .
consider an integer @xmath171 greater than @xmath172 .
set @xmath173 for the purposes of the proof , we assume that the treasure is located at distance @xmath174 , so that by time @xmath46 it is guaranteed that no agent finds the treasure . for @xmath54 , define @xmath175 fix an integer @xmath58 in @xmath176 assume for the time being , that the number of agents is @xmath177 .
note that @xmath178 hence , @xmath140 is a possible candidate for being the real number of agents . observe that all nodes in @xmath153 are at distance at most @xmath179 from the source , and that @xmath180 . by the definition , @xmath181 .
hence , @xmath182 , and therefore using the expected running time of the algorithm , it follows that for each node @xmath183 , the expected time until at least one of the @xmath140 agents covers @xmath32 is at most @xmath184 recall that we now consider the case where the algorithm is executed with @xmath140 agents . for every subset @xmath141 of nodes of @xmath21 ,
let @xmath142 be the random variable indicating the number of nodes in @xmath141 that were visited by at least one of the @xmath140 agents by time @xmath46 .
( as before , we write @xmath143 for @xmath144 . ) by markov s inequality , the probability that @xmath32 is visited by at least one of the @xmath140 agents by time @xmath46 is at least @xmath50 , i.e. , @xmath148 . hence , @xmath149 . by linearity of expectation ,
consequently , by time @xmath46 , the expected number of nodes in @xmath153 that a single agent visits is @xmath185 .
since this holds for any @xmath58 in @xmath186 , and since the sets @xmath153 are pairwise disjoint , the linearity of expectation implies that the expected number of nodes that a single agent visits by time @xmath46 is @xmath187 . since @xmath188 , this implies that @xmath189 , as desired .
this concludes the proof of the theorem .
the algorithms described in the section [ sec : upper ] are relatively simple but still require the use of non trivial iterations , which may be complex for simple and tiny agents , such as ants .
if we relax the requirement of bounding the expected running time and demand only that the treasure be found with some low constant probability , then it is possible to avoid one of the loops of the algorithms .
however , a sequence of iterations still needs to be performed . in this section
, we propose an extremely simple algorithm , coined the _ _ harmonic search algorithm__-server problem see , e.g. , @xcite . ] , which does not perform in iterations and
is essentially composed of three components : ( 1 ) choose a random direction and walk in this direction for a distance of @xmath190 , chosen randomly according to a distribution in which the probability of choosing @xmath190 is roughly inverse proportional to @xmath190 , ( 2 ) perform a local search ( e.g. , a spiral search ) for time roughly @xmath191 , and ( 3 ) return to the source .
it turns out that this extremely simple algorithm has a good probability of quickly finding the treasure , if the number of agents is sufficiently large .
more specifically , the algorithm depends on a positive constant parameter @xmath192 that is fixed in advance and governs the performance of the algorithm . for a node @xmath32 ,
let @xmath193 , where @xmath108 is the normalizing factor , defined so that @xmath194 .
( note that @xmath108 depends on @xmath192 . ) using arguments similar to those introduced earlier , e.g. , in the proofs of theorems [ thm : know - k ] and [ thm : comp ] , one can prove the following result .
let @xmath195 $ ] .
for every @xmath11 , there exists a positive real number @xmath196 such that if the number @xmath0 of agents is greater than @xmath197 , then with probability at least @xmath198 , the expected running time of the harmonic algorithm is @xmath199
we first presented an algorithm that assumes that agents have a constant approximation of @xmath0 and runs in optimal expected time , that is , in time @xmath3 .
we then showed that there exists a uniform search algorithm whose competitiveness is slightly more than logarithmic , specifically , @xmath9 , for an arbitrarily small positive constant @xmath123 .
we also presented a relatively efficient uniform algorithm , namely , the harmonic algorithm , that has and extremely simple structure .
our constructions imply that , in the absence of any communication , multiple searchers can still potentially perform rather well . on the other hand ,
our lower bounds imply that to achieve a better running time , the searchers must either communicate or utilize some information regarding @xmath0 . in particular , even providing each agent with a @xmath12-approximation to @xmath0 ( for constant @xmath11 ) does not suffice to bring the competitiveness strictly below @xmath10 .
although the issue of memory is beyond the scope of this paper , our constructions are simple and can be implemented using relatively low memory .
for example , going in a straight line for a distance of @xmath200 can be implemented using @xmath201 memory bits , by employing a randomized counting technique .
in addition , our lower bounds provide evidence that in order to achieve a near - optimal running time , agents must use non - trivial memory size , required merely to store the necessary approximation of @xmath0
. this may be useful to obtain a tradeoff between the running time and the memory size of agents . from another perspective , it is of course interesting to experimentally verify whether social insects engage in search patterns in the plane which resemble the simple uniform algorithms specified above , and , in particular , the harmonic algorithm .
two natural candidates are desert ants _ cataglyphys _ and honeybees _ apis mellifera_. first , these species seem to face settings that are similar to the one we use .
indeed , they can not rely on communication during the search due to the dispersedness of individuals @xcite and their inability to leave chemical trails ( this is due to increased pheromone evaporation in the case of the desert ant ) . additionally , the task of finding the treasure is relevant , as food sources in many cases are indeed relatively rare or patchy .
moreover , due to the reasons mentioned in section [ sec : introduction ] , finding nearby sources of food is of great importance .
second , insects of these species have the behavioral and computational capacity to maintain a compass - directed vector flight @xcite , measure distance using an internal odometer @xcite , travel to distances taken from a random power law distribution @xcite , and engage in spiral or quasi - spiral movement patterns @xcite .
these are the main ingredients that are needed to perform the algorithms described in this paper .
finally , the search trajectories of desert ants have been shown to include two distinguishable sections : a long straight path in a given direction emanating from the nest and a second more tortuous path within a small confined area @xcite .
the authors are grateful to the anonymous referees for their useful suggestions .
e. a. capaldi , a. d. smith , j. l. osborne , s. e. fahrbach , s. m. farris , d. r. reynolds , a. s. edwards , a. martin , g. e. robinson , g. m. poppy , and j. r. riley .
ontogeny of orientation flight in the honeybee revealed by harmonic radar .
, 403:537540 , 2000 .
s. das , p. flocchini , n. santoro , and m. yamashita . on the computational power of oblivious robots : forming a series of geometric patterns . in _ proc .
principles dist .
_ , ( podc10 ) , pages 267276 .
acm , 2010 .
a. korman , j .- s .
sereni , and l. viennot . toward more localized local algorithms : removing assumptions concerning global knowledge . in _ proc .
principles dist .
_ , ( podc11 ) , pages 4958 .
acm , 2011 .
i. suzuki and m. yamashita .
erratum : `` distributed anonymous mobile robots : formation of geometric patterns '' [ siam j. comput .
* 28 * ( 1999 ) , no .
4 , 13471363 ( electronic ) ; mr1681010 ] . , 36(1):279280 ( electronic ) , 2006 .
y. yang , a.a .
minai , and m.m .
decentralized cooperative search by networked uavs in an uncertain environment . in _ proc .
control conf . , ( acc04 ) _ , volume 6 , pages 55585563 .
ieee , 2004 . | we use distributed computing tools to provide a new perspective on the behavior of cooperative biological ensembles .
we introduce the _ ants nearby treasure search ( ants ) _ problem , a generalization of the classical cow - path problem @xcite , which is relevant for collective foraging in animal groups . in the ants problem ,
@xmath0 identical ( probabilistic ) agents , initially placed at some central location , collectively search for a treasure in the two - dimensional plane .
the treasure is placed at a target location by an adversary and the goal is to find it as fast as possible as a function of both @xmath0 and @xmath1 , where @xmath1 is the distance between the central location and the target .
this is biologically motivated by cooperative , central place foraging , such as performed by ants around their nest . in this type of search
there is a strong preference to locate nearby food sources before those that are further away .
we focus on trying to find what can be achieved if communication is limited or altogether absent . indeed , to avoid overlaps agents must be highly dispersed making communication difficult .
furthermore , if the agents do not commence the search in synchrony , then even initial communication is problematic .
this holds , in particular , with respect to the question of whether the agents can communicate and conclude their total number , @xmath0 .
it turns out that the knowledge of @xmath0 by the individual agents is crucial for performance .
indeed , it is a straightforward observation that the time required for finding the treasure is @xmath2 , and we show in this paper that this bound can be matched if the agents have knowledge of @xmath0 up to some constant approximation .
we present a tight bound for the competitive penalty that must be paid , in the running time , if the agents have no information about @xmath0 .
specifically , this bound is slightly more than logarithmic in the number of agents .
in addition , we give a lower bound for the setting in which the agents are given some estimation of @xmath0 .
informally , our results imply that the agents can potentially perform well without any knowledge of their total number @xmath0 , however , to further improve , they must use non trivial information regarding @xmath0 .
finally , we propose a uniform algorithm that is both efficient and extremely simple , suggesting its relevance for actual biological scenarios . |
low dimensional interacting spin systems have long been a laboratory for the discovery of novel quantum states of matter . enhanced quantum fluctuations due to reduced dimensionality
enable the appearance of multiple quantum phases with unique characteristics driven by the interplay between strong interactions , external ( e.g. , pressure and magnetic field ) and internal ( e.g. , crystal field effects ) potentials and lattice geometry that are suppressed in higher dimensions . the relative simplicity of the microscopic models facilitates the development of exact analytical solutions in many cases and powerful field theoretic and computational approaches in others providing greater insight into the emergence of these complex phases and their physical properties .
concurrent rapid advances in the synthesis and characterization of low - dimensional quantum magnets have kept the study of these systems one of the most active frontiers in condensed matter physics ( see landee and turnbull@xcite for a recent pedagogical review ) .
one of the most remarkable results in the study of quantum spin models one that substantially enhanced our understanding of long range order in quantum many body systems is the pioneering work by haldane.@xcite by studying the non - linear sigma model in ( 1 + 1 ) dimensions , haldane conjectured that the ground state of the one - dimensional ( 1d ) heisenberg antiferromagnet ( hafm ) has gapless excitations for half - odd integer spins , whereas that for integer spins is separated from all excited states by a finite spin gap ( haldane gap ) .
haldane s conjecture has inspired numerous theoretical studies of integer spins in low dimensions , primarily @xmath0 spins where the haldane phase is most robust .
these include chain mean field theory ( cmft),@xcite exact diagonalization,@xcite density matrix renormalization group ( dmrg),@xcite and quantum monte carlo ( qmc ) simulations.@xcite the theoretical studies have been complemented by the discovery of several quasi - one - dimensional ( q1d ) spin @xmath0 quantum magnets , such as agvp@xmath1s@xmath2@xcite ndmap,@xcite nenb,@xcite nenp,@xcite nino,@xcite pbni@xmath1v@xmath1o@xmath3,@xcite srni@xmath1v@xmath1o@xmath3,@xcite tmnin,@xcite and y@xmath1banio@xmath4.@xcite in these materials , the magnetic ions are arranged in chains , with weak but finite inter - chain couplings , which affect the ground state phases .
additionally , in most known @xmath0 magnets , the ubiquitous heisenberg exchange is complemented by a single - ion anisotropy .
the expanded hilbert space of the @xmath0 spins , and the interplay between multiple competing interactions , external magnetic field and different lattice geometries result in a rich variety of ground state phases .
in addition to the gapped haldane phase , examples of exotic quantum states realized in low dimensional interacting spin systems include experimentally realized bose einstein condensation ( bec ) of magnons,@xcite quantum paramagnet,@xcite and the recently proposed spin supersolid@xcite and ferronematic@xcite phases .
recent advances in synthesis techniques have made it possible to engineer quasi - low - dimensional materials where the `` effective dimensionality '' ( that is , inter - chain or inter - layer couplings ) and hamiltonian parameters ( such as the ratio of exchange interaction and single - ion anisotropy ) can be controlled.@xcite this raises the possibility of preparing materials with desired predetermined properties .
the search for such tailor - made materials has grown in recent years , as these are believed to drive the next generation of electronics .
in addition to condensed matter systems , rapid advances in the field of ultracold atoms in optical lattices have opened up a new frontier in the study of interacting many - body systems in arbitrary dimensions . the unprecedented control over number of atoms , interactions , and lattice geometry makes it an ideal testbed for preparing and studying novel quantum states .
the most remarkable property of the haldane phase is that even though there is no long - range magnetic order all spin - spin correlations decay exponentially there exists a hidden or nonlocal order measured by the string order parameter introduced by den nijs and rommelse@xcite .
the introduction of a nonlocal order parameter was a marked departure from the conventional practice of characterizing quantum many body states in terms of local order parameters and ushered in the study of topological phases an area of intense current research .
much insight into the nature of the haldane phase can be gained from the aklt ( affleck - kennedy - lieb - tasaki ) state the exact ground state of a 1d @xmath0 spin chain where the near - neighbor heisenberg interaction is supplemented by an additional interaction between neighboring spins.@xcite the aklt state may be understood by noting that the spin at each site can be thought of as a symmetric combination of two @xmath5 spins .
pairs of these spins on neighboring sites form a singlet on each bond . for a periodic chain
, this forms a unique valence bond solid ground state a singlet on each bond with a gap to lowest excitations .
but for open chains , there remains an unpaired @xmath5 moment at each end which is doubly degenerate ( see fig .
[ aklt ] for an illustration of this state ) .
consequently , the ground state is gapped in the `` bulk '' and has degenerate gapless edge states .
the state is protected by this @xmath6 symmetry .
using the current terminology of topological states , this state is a symmetry protected topological ( spt ) phase .
the haldane phase is adiabatically connected to the aklt state in other words , it has the same qualitative character .
in fact , the haldane phase is widely recognized as the earliest and best understood spt phase in interacting many body systems.@xcite degrees of freedom composed of the symmetric combination of two spin @xmath5 spins ( small solid circles ) . in the aklt state ,
pairs of @xmath5 spins on neighboring sites form a singlet state ( blue lines ) . for open boundaries
, this leaves an unpaired @xmath5 degree of freedom at each boundary . in the thermodynamic limit , these spins are completely decoupled from the rest of the lattice , leading to a fourfold degenerate ground state . ]
what happens to the haldane phase in the presence of additional interactions , such as , inter - chain coupling and single - ion anisotropy ?
the question is not simply of academic interest all real materials have nonzero interchain coupling and in many quantum magnets , the crystal electric field lifts the degeneracy of the local hilbert space in the form of a single - ion anisotropy .
theoretical studies have shown that both these interactions destroy the haldane phase at sufficiently strong interaction strengths the single - ion anisotropy drives a transition to a quantum paramagnetic phase whereas interchain coupling favors long - range afm ordering but the ground state remains gapped up to finite values of the couplings .
what is the nature of the gapped ground state away from the isotropic heisenberg point ?
does it still retain its spt character ?
the string order parameter the only definitive probe for the haldane phase is strictly defined only in one dimension and its extension to coupled chains can not be trusted without corroboration from additional measurements .
fortunately , recent advancements in the study of spt phases have yielded new probes that provide better insight into the nature of the putative haldane phase away from the isotropic heisenberg point . in this brief review ,
we summarize some of our recent progress towards fully characterizing the haldane phase in quasi - one - dimensional geometries .
we consider a model of low dimensional quantum magnets consisting of an array of weakly coupled spin-1 hafm chains .
this model can be described by a spatially anisotropic hamiltonian of @xmath0 spin operators with nearest - neighbor spin exchange interactions and on - site single - ion anisotropy : @xmath7}{\vec s}_{i}\cdot{\vec s}_{j}+d\sum_{i}\left(s_{i}^{z}\right)^{2}.\ ] ] the first two sums are over all pairs of interacting spins @xmath8 , with @xmath9 and @xmath10 $ ] referring to nearest neighbor spin pairs along and perpendicular to the chain direction , respectively , while the third sum is over all lattice sites @xmath11 . here ,
we limit ourselves to nearest neighbor interactions on bipartite lattices to avoid the sign problem in qmc , and tune the interactions @xmath12 and @xmath13 to generate systems of weakly interacting spin chains .
this is illustrated in fig .
[ lattice ] for a spatially anisotropic square lattice composed of weakly interacting spin chains . in this paper
, we consider several bipartite lattices in the quasi - one - dimensional limit . for each lattice
, we define a chain length @xmath14 as the system size along the chain direction , and a perpendicular length @xmath15 such that for 2d lattices the total number of spins @xmath16 while for 3d lattices @xmath17 .
the ratio of these two length scales forms the aspect ratio @xmath18 , which we set at @xmath19 unless otherwise stated .
we henceforth set the intrachain interaction strength to unity ( @xmath20 ) , and study the ground state properties of the above model for various single - ion anisotropy @xmath21 and interchain coupling @xmath13 .
( blue lines ) and weak interchain coupling @xmath13 ( red lines ) . ]
quantum monte carlo methods map quantum systems onto equivalent classical representations in @xmath22 dimensions , where the configuration space can be stochastically sampled by markov chain monte carlo methods.@xcite the numerical results presented here were obtained using the stochastic series expansion ( sse ) formulation of qmc.@xcite briefly , the density matrix can be expanded as a taylor series , @xmath23 . for a spin hamiltonian consisting of onsite fields and pair interactions ,
we can write @xmath24 , where the sum is over all pair interactions , or _ bonds_. the powers of @xmath25 appearing in the taylor expansion of @xmath26 can then be written as a sum over products of bond operators , @xmath27 , where @xmath28 represents a sequence of bond operators called the _ operator string_. note that the overall sign of a particular operator string depends on the number of bond operators with negative weight . to avoid the sign problem in qmc
, all the bond operators must therefore be positive definite , which in practice limits the study of heisenberg antiferromagnets to bipartite lattices , where a suitable sublattice rotation is available .
the configuration space consists of all possible rearrangements of the operator string , the length of which takes into account the expansion variable @xmath29 , while its sequence dictates the states @xmath30 appearing in the partition function , @xmath31 .
loop algorithms exist to efficiently sample this configuration space.@xcite in addition to the sse method , we implement a projective qmc method that allows us to access both ground state expectation values as well as wave function overlaps between the ground state and a trivial product state .
this method is quite similar to sse , and details are given elsewhere.@xcite
here we present recent qmc results on the ground state properties of the spin @xmath0 hafm described by the hamiltonian in eq .
( [ model ] ) .
first , we show how finite size scaling can be used to accurately determine the ground state phase boundaries between the haldane phase and neighboring magnetically ordered phases .
next , we examine the effect of lattice geometry on the critical coupling of bipartite lattices .
the phase diagram is then presented in the quasi - one - dimensional regime using a spatially anisotropic rectilinear lattice .
this phase diagram is directly compared to the experimentally determined parameters of haldane gap materials .
we also show the magnetization curves of the q1d model for the various zero - field ground state phases , and calculate the low - lying excitation spectrum of the haldane phase near the haldane gap minimum .
finally , the haldane phase in q1d geometries is shown to belong to the new class of symmetry protected topological states .
even with a simulation method such as sign - free quantum monte carlo that scales polynomially in system size , it is impossible to directly access results for macroscopic system sizes . to overcome this difficulty
, we employ finite - size scaling ( fss ) methods to extrapolate our results to the ground state thermodynamic limit .
since quantum systems map onto classical systems in @xmath32 dimensions , where @xmath33 is the dynamic critical exponent , similar procedures can be used to access the @xmath34 and @xmath35 limits . for ground states that break a continuous spin symmetry ,
the spin stiffness is a useful observable to determine the boundary of such a gapless phase ( as per the goldstone theorem ) with a gapped state of any order .
this is because the spin stiffness must be finite in the gapless phase and zero in the gapped phase .
the scaling function of the spin stiffness in the vicinity of a quantum critical point is known to be @xmath36 where @xmath37 is a hamiltonian parameter that can tune the ground state system to the quantum critical point at @xmath38 , @xmath39 is the critical exponent governing the growth of the ( spatial ) correlation length @xmath40 near @xmath38 , and @xmath33 relates @xmath40 to the correlation length in imaginary time by the relation @xmath41 . by taking the inverse temperature to scale as the system size @xmath14 ( with @xmath42 for the quantum phase transitions considered in this work )
, it is clear that @xmath43 is independent of system size at the critical point .
this is demonstrated in the right panels of fig .
[ scaling ] , where we plot @xmath44 for a @xmath45 dimensional quantum system and use @xmath46 to reach the ground state limit .
the critical point is clearly visible at @xmath47 where the quantity @xmath44 is independent of system size .
a similar scaling function is known for the square of the staggered magnetization along the @xmath33 axis @xmath48 that acts as an order parameter for the ising antiferromagnetic phase .
it is @xmath49 where @xmath50 is the anomalous dimensionality .
thus we can also see that @xmath51 will display a crossing behavior for varying @xmath14 at the phase boundaries of the ising afm phase .
this is shown in the left panels of fig .
[ scaling ] for the ising afm to haldane phase transition . , left panels ) and spin stiffness ( @xmath52 , right panels ) at the ising afm to haldane and haldane to @xmath53 afm phase transitions , respectively .
interchain coupling is fixed at @xmath54 .
the critical points @xmath55 and @xmath56 are determined by the crossing criterion ( upper panels ) assuming a dynamic critical exponent @xmath42 .
good finite - size scaling collapse is achieved near the critical points assuming mean field exponents ( @xmath57 and @xmath58 ) for a continuous phase transition at the upper critical dimension @xmath59 ( lower panels ) .
inverse temperature is scaled as @xmath46 to ensure convergence to the ground state limit . ] starting from a system of uncoupled chains in the haldane phase and slowly turning on the interchain coupling @xmath13 , the system will remain in the gapped haldane phase until a critical coupling @xmath60 is reached at which point the gap closes and long - range magnetic order develops . within the chain mean field approximation ( cmfa ) , the critical coupling @xmath60 depends only on the coordination number of chains @xmath61 .
thus , the quantity @xmath62 might be expected to be universal for unfrustrated lattices . in fig .
[ geometry ] we show results for chains arranged into square ( @xmath63 ) and honeycomb ( @xmath64 ) superlattices . while they give different results for the critical coupling ( @xmath65 and @xmath66 , respectively ) , the scaled values @xmath62 are almost in statistical agreement : @xmath67 vs. @xmath68 .
these can be compared to the cmfa value @xmath69 that acts as a lower bound.@xcite ) at the haldane to nel phase transition for chain coordination * ( a ) * @xmath63 ( i.e. square superlattice of chains ) , and * ( b ) * @xmath64 ( i.e. honeycomb superlattice of chains ) .
single - ion anisotropy is fixed at the isotropic point , @xmath70 .
inverse temperature is scaled as @xmath71 to ensure convergence to the ground state limit . adapted from k. wierschem and p. sengupta , jps conf .
* 3 * ( 2014 ) 012005 . ]
the ground state phase diagram of @xmath25 is shown in fig . [ phase1 ] . for small @xmath72 and @xmath13
the system is in the haldane phase . for sufficiently strong interchain couplings ,
the haldane gap is quenched and three - dimensional long - range magnetic order sets in .
this magnetic order is the nel antiferromagnetic state in the case of isotropic spins ( @xmath70 ) , while axial ( @xmath73 ) and planar ( @xmath74 ) anisotropy lead to ising afm and @xmath53 afm states , respectively .
additionally , there is a quantum paramagnetic ( qpm ) phase for large @xmath75 .
the haldane to @xmath53 afm phase boundary is determined by fss of the spin stiffness , while the haldane to ising afm phase boundary is determined by fss of the staggered magnetization . in the case of the haldane to nel phase transition at the isotropic point ( @xmath70 ) , fss of both the spin stiffness and staggered magnetization yield values in agreement up to the statistical uncertainty given .
@xmath13 plane with phase boundaries as indicated by the dotted black lines .
the borders of the haldane and qpm phases are obtained as polynomial fits to the present work and represent guides for the eye .
data points for the present work are determined by qmc simulations at constant @xmath21 and @xmath13 ( solid red circles and blue squares , respectively ) .
several haldane compounds are plotted as large crosshatched symbols using estimates for @xmath21 and @xmath13 from the indicated sources . adapted from k. wierschem and p. sengupta , phys .
* 112 * , 247203 ( 2014 ) . ]
one feature of interest in the phase diagram of fig .
[ phase1 ] is the ability to change the effective hamiltonian parameters of a given material by the application of hydrostatic pressure .
this has been demonstrated for nenp by zaliznyak _
et al._@xcite however , as can be seen in fig .
[ phase1 ] , nenp under pressure is actually closer to an ideal heisenberg chain , i.e. it moves towards the origin and _ away _ from the boundaries of the haldane phase .
thus , it is not a good candidate for the observation of pressure - induced quantum criticality , as has been found in the spin-1/2 dimer compound tlcucl@xmath76.@xcite by contrast , the isostructural compounds pbni@xmath1v@xmath1o@xmath3 and srni@xmath1v@xmath1o@xmath3 are found to already lie very near to a quantum critical point , making them prime candidates for the study of pressure - induced quantum criticality .
in addition to pressure , chemical substitution represents another mechanism by which the effective hamiltonian parameters of magnetic compounds can be modified .
this is already somewhat apparent in the isostructural compounds pbni@xmath1v@xmath1o@xmath3 and srni@xmath1v@xmath1o@xmath3 , where the differing influence of the pb@xmath77 and sr@xmath77 ions leads srni@xmath1v@xmath1o@xmath3 to be closer to the ising afm phase than pbni@xmath1v@xmath1o@xmath3 .
indeed , srni@xmath1v@xmath1o@xmath3 was initially thought to magnetically order below @xmath78 based on experiments on powder samples,@xcite whereas recent results on polycrystalline@xcite and single crystals@xcite have established its low temperature magnetic behavior to be consistent with a non - magnetic haldane ground state .
the prospect of chemical fine - tuning is even more exciting for molecule - based magnets , where the choice of bridging ligands has been shown to allow for the synthesis of a wide range of low dimensional structures .
of particular interest is the recently synthesized compound [ ni(hf@xmath79)(3-clpy)@xmath80bf@xmath81 that is believed to lie near the 1d gaussian critical point.@xcite we note here that this critical point is well - described by the tomonaga - luttinger liquid theory with luttinger parameter @xmath82.@xcite it is important to take the crystal symmetry into consideration , as it determines which symmetry breaking and conserving phases can occur . for example , nenp is known to have a rhombohedral crystal electric field component @xmath83 that explicitly breaks the onsite @xmath84 spin symmetry of @xmath25 down to @xmath85 .
additionally , application of a uniform longitudinal magnetic field is known to induce a staggered transverse field at the nickel sites due to a zigzag chain structure.@xcite also , note that the related compound nenc falls into the qpm phase since @xmath86,@xcite but because @xmath87 we do not include it here .
we also have calculated the phase boundary between the @xmath53 afm and qpm phases from the q1d limit all the way to the spatially isotropic 3d lattice .
this is shown in fig .
[ phase2 ] where we also plot the location of some quantum magnets in the phase diagram .
note that the phase boundary between the @xmath53 afm and qpm phases is a line of critical points belonging to the @xmath88 universality class in 3 + 1 dimensions . within the qpm phase , application of a magnetic field leads to a field
induced quantum phase transition into the _ canted _ @xmath53 afm phase .
this field induced transition has been studied by zhang _
et al._@xcite and shown to be a mean field transition because the effective dimension is @xmath89 ( the dynamical exponent is z=2 because these field induced transitions belong to the bec universality class@xcite ) .
regime of the @xmath21@xmath13 plane .
data points for the phase boundary are extracted by finite - size scaling analysis of qmc simulations performed at constant @xmath13 .
dotted line is obtained as a polynomial fit to the qmc data and delineates the @xmath53 afm and qpm phase boundary . ]
the theoretical effect of uniaxial hydrostatic pressure applied to dtn has been examined , with encouraging signs that dtn may be tuned to a quantum critical point ( qcp).@xcite this pressure - induced qcp would be of the @xmath88 universality class in 3 + 1 dimensions , as opposed to the field - induced qcp in dtn that has been shown to belong to the bec universality class in 3 + 2 dimensions.@xcite there also are several compounds much deeper in the qpm phase , such as sr@xmath76nipto@xmath90.@xcite the parameters for the compounds shown in figs . [ phase1 ] and [ phase2 ] are listed in table [ table1 ] . for completeness
, we also mention the haldane chain compounds ndmaz,@xcite nenb,@xcite ninaz,@xcite and nino,@xcite which could also fit into the phase diagram in fig .
[ phase1 ] near the related compounds ndmap and nenp but are not shown here in the interest of clarity . additionally , the hexagonal compounds csnicl@xmath76 and tmnin are outside the scope of the present model as they can not be well approximated by a cubic lattice .
tmnin is a haldane gap material while csnicl@xmath76 exhibits haldane behavior above its magnetic ordering temperature @xmath91.@xcite finally , the low temperature phase of tl@xmath79ru@xmath79o@xmath92 has been interpreted as a haldane chain system that spontaneously arises from a fundamentally three - dimensional crystal structure due to the orbital ordering.@xcite c c r r r r + compound & bravais lattice & @xmath93 & @xmath94 & @xmath12 & ref .
+ + agvp@xmath1s@xmath95 & monoclinic & 0.006 & @xmath96 & 780k & [ ] + ndmap & orthorhombic & 0.25 & @xmath97 & 33.1k & [ ] + nenp & orthorhombic & 0.16 & @xmath98 & 46k & [ ] + [email protected] & orthorhombic & 0.09 & @xmath99 & 48k & [ ] + [ ni(hf@xmath79)(3-clpy)@xmath80bf@xmath81 & monoclinic & 0.88 & & 4.86k & [ ] + pbni@xmath1v@xmath1o@xmath3 & tetragonal & -0.05 & @xmath100 & 104k & [ ] + srni@xmath1v@xmath1o@xmath3 & tetragonal & -0.04 & @xmath101 & 100k & [ ] + y@xmath1banio@xmath4 & tetragonal & -0.039 & @xmath102 & 240k & [ ] + + dtn & tetragonal & 4.05 & @xmath103 & 2.2k & [ ] + [ ni(hf@xmath79)(pyz)@xmath79]sbf@xmath90 & tetragonal & 1.22 & @xmath105 & 9k & [ ] + sr@xmath76nipto@xmath90 & rhombohedral & 8.8 & & 11k & [ ] + [ table1 ] the ground state magnetization curve can be a good way to characterize quantum magnets at low temperatures .
the direct application of an external magnetic field is an efficient probe of the underlying magnetic phases of a magnetic compound . for longitudinal magnetic fields , the hamiltonian in eq.([model ] )
comes with a conservation law such that the ground state at any field must occur within a single magnetization sector defined by the quantum number @xmath106 .
this can lead to several interesting features .
first , note that the fully polarized state @xmath107 is an exact eigenstate of the hamiltonian , since the off - diagonal terms come in pairs @xmath108 such that @xmath107 is always annihilated , leaving only the action of the diagonal terms to set its energy .
additionally , the saturation field @xmath109 can often be obtained by setting the energy difference between @xmath107 and the lowest energy state in the @xmath110 sector to zero . for the model considered here on an anisotropic cubic lattice
, it can be shown that @xmath111 as long as @xmath21 is not strongly easy - axis .
it has previously been demonstrated that for @xmath73 in the @xmath112 limit it is the @xmath113 sector that instead determines the saturation field , while for intermediate @xmath21 a discontinuity in @xmath114 develops at a first order transition between the fully polarized and canted @xmath53 afm phases.@xcite gapped and gapless phases respond differently to an applied magnetic field .
for the longitudinal magnetic field @xmath115 the ising afm , haldane , and qpm phases are all gapped and therefore stay in the @xmath116 magnetization sector until the field reaches some critical value @xmath117 . in the case of the ising afm phase , there is a first order phase transition the so - called _ spin flop _ transition that occurs , whereby one sublattice of spins that was previously aligned against the field `` flops '' over and becomes partially aligned with the field instead , while at the same time the other sublattice of spins reduces its relative polarization .
these spins retain some antiferromagnetism by developing a staggered magnetization in the @xmath53 plane the so - called _ canted _ @xmath53 afm phase . in the case of the haldane and qpm phases , there is instead a continuous phase transition at @xmath117 , with field - induced critical points belonging to the bec universality class . in fig .
[ magnetization ] we plot the magnetization curves for @xmath118 for zero - field ground states in the haldane , @xmath53 afm and qpm phases .
it is easy to see that the gapless @xmath53 afm phase immediately responds to the applied field , while the magnetization curves for the haldane and qpm phases show signs of a small gap , with @xmath119 .
all three curves appear similar in their approach to the fully polarized state , with the increase in slope due to a combined effect of low spin number ( @xmath0 , where we remind the reader that for the most quantum spin @xmath5 the slope diverges at @xmath109 as is known from the bethe ansatz ) and reduced dimensionality .
these features are most easily seen in the corresponding differential susceptibility , @xmath120 .
another feature to notice is the pronounced hump in the differential susceptibility around @xmath121 when @xmath70 , followed by a local minimum .
this feature has previously been pointed out by kashurnikov _
_ in the magnetization curve of the haldane phase of a 1d chain,@xcite and is characteristic of the haldane phase . , the interchain coupling is set to @xmath118 with inverse temperature @xmath46 sufficient to reach the ground state regime for @xmath122 . ]
one of the most characteristic features of the haldane phase is the haldane gap .
this arises directly in the haldane conjecture due to the difference between integer and half - odd - integer spins .
it was also one of the first features to be confirmed , with early exact diagonalization results showing a finite gap for the spin-1 heisenberg model.@xcite the accurate determination of this gap was also one of the first applications of the powerful dmrg method.@xcite the low lying states can also be estimated with an upper bound estimator @xmath123 that can be derived from well known sum rules.@xcite when the single - mode approximation holds reasonably well , this estimator can be quite accurate .
the spectrum of the haldane phase in 1d is known from dmrg studies,@xcite and in particular at @xmath124 the spectrum is sharply peaked at a single - magnon mode ( corresponding to the haldane gap ) , with a further gap to the higher multi - magnon modes .
thus , the estimator should be very accurate for @xmath124 .
we use this to calculate the low lying spectrum of coupled chains with @xmath124 along the chain direction .
our results are shown in fig .
[ dispersion ] , where the tendency for the gap to close as the interchain coupling is increased is quite clear , as is the trend for the modes about @xmath125 to become linear , indicative of the fact that the system is approaching the gapless nel afm phase . .
dashed line represents the haldane gap of uncoupled chains ( @xmath126 ) .
data shown for length @xmath122 and inverse temperature @xmath127 . adapted from k. wierschem and p. sengupta , phys .
lett . * 112 * , 247203 ( 2014 ) . ]
a nonlocal string order captures the hidden symmetry breaking in the haldane phase of the spin-1 hafm in 1d . for two sites
@xmath11 and @xmath128 in a chain , the string correlation function is given by @xmath129}s_j^z\right>. \label{eq : string}\ ] ] in the limit that the distance between sites @xmath11 and @xmath128 goes to infinity , @xmath130 becomes exactly @xmath131 in the aklt state,@xcite while in the haldane state it has been estimated to be @xmath132.@xcite although it is not possible to generalize the string order parameter to higher dimensional correlations , it is possible to calculate the string correlation function in systems of coupled spin chains for spins residing within the same chain . in fig .
[ strings ] we plot @xmath133 , which as @xmath35 should scale exponentially to zero in the absence of string order , or to a nonzero value when string order is present . thus , @xmath133 acts as a string order parameter . for comparison
, we also show the scaled spin stiffness @xmath44 , which shows that below @xmath134 the system is in the gapped haldane phase , while above this value @xmath135 is finite and we are in the gapless @xmath53 afm phase .
it is interesting to note that the string order parameter shows a qualitative change in behavior at @xmath136 .
below @xmath136 , @xmath133 shows very weak decay with system size , and may even saturate to a finite value as @xmath35 .
by contrast , in the @xmath53 afm phase , @xmath133 exhibits rapid decay with system size , and will scale to zero as @xmath35 . while this would seem to indicate that there is long - range string order in the haldane phase in q1d geometries , a note of caution must be sounded .
this is because the string correlation function has been argued to always scale exponentially to zero with distance for any nonzero value of the interchain coupling.@xcite thus , it remains a possibility that the correlation length of string order grows rapidly at @xmath136 , yet remains finite in the q1d haldane phase . and * ( b ) * the scaled spin stiffness @xmath137 across the haldane to @xmath53 afm phase boundary with aspect ratio @xmath138 , interchain coupling @xmath118 , and inverse temperature @xmath46 .
adapted from k. wierschem and p. sengupta , phys .
112 * , 247203 ( 2014 ) . ]
due to the shortcomings of the string order parameter in more than one spatial dimension , it would be nice to find alternative ways to characterize the haldane phase in q1d geometries .
since the haldane state in 1d is a nontrivial spt state , we might expect it to remain such in q1d systems .
still , direct evidence for this has been lacking .
this may have to do with the difficulty of characterizing spt states in general .
one common method is to look at the low lying entanglement spectrum,@xcite where an overall degeneracy signals a nontrivial spt state.@xcite while the entanglement spectrum can be easily accessed using powerful dmrg techniques in 1d , systems in higher dimension pose a tougher challenge . additionally , there is the possibility that the `` cut '' used to form a bipartition of the system during the construction of the reduced density matrix may break an off - site symmetry , and so misdiagnose spt phases protected by such symmetries .
fortunately , in 1d and 2d there exists an alternative way to distinguish between trivial and nontrivial spt states .
this is the so - called strange correlator of you _
et al._,@xcite which uses a mapping between the space - time correlations at the spatial boundary of a nontrivial spt state with a trivial spt state ( or vacuum ) and ( through a laplace transform ) the spatial correlations at the temporal boundary between the time - evolved quantum field theories of the respective trivial and nontrivial spt states .
since the space - time correlations at the aforementioned spatial boundary are known to possess long - range ( or , possibly quasi - long - range ) order in one and two dimensions,@xcite a long - range or quasi - long - range strange correlator is then a direct measure of spt order in 1d and 2d , while short - range behavior is a sign of a trivial product state . in 3d
, the possibility of a topologically ordered edge state allows for the construction of nontrivial spt phases with short - range strange correlators ; however , long - range or quasi - long - range behavior remains indicative of spt order .
the strange correlator is calculated as a `` mixed '' correlation function between the nontrivial spt state @xmath139 and the trivial spt state @xmath140 , @xmath141 in practice , @xmath140 is taken as a symmetric product state , which is by definition a trivial spt state .
the state @xmath139 can be our putative spt state in question , and thus a ( quasi- ) long - range strange correlator becomes a direct probe of the nontrivial spt character of the state @xmath139 .
it should also be mentioned that this only applies to states that are gapped and symmetric , i.e. they are either trivial or nontrivial spt states to begin with . for the case considered here
, there is a host of evidence that the q1d haldane phase is just such a state .
the strange correlator has recently been implemented using qmc methods.@xcite in fig .
[ strange ] we show the finite - size scaling of a finite - size order parameter @xmath142 constructed from the strange correlator . for the q1d haldane phase
, we see that @xmath143 remains finite , signaling a nontrivial spt state .
this is true for both 2d and 3d lattices .
by contrast , in the qpm phase we see evidence for exponential decay to zero , as expected for a nontrivial spt state .
thus , we have provided direct evidence that the q1d haldane phase is indeed a nontrivial spt state . , black circles ) and the quantum paramagnetic phase ( @xmath144 , red squares ) . results in the ground state limit are obtained by scaling the operator string length as @xmath145 .
solid lines and filled symbols are results in 3d with @xmath118 , while dashed lines and symbols are in 2d with @xmath146 . adapted from k. wierschem and p. sengupta , phys .
lett . * 112 * , 247203 ( 2014 ) . ]
the identification of the quasi - one - dimensional haldane phase as a nontrivial spt state in its own right naturally leads us to consider what sort of topological edge modes might be present in this system .
although these edge states are yet to be directly determined using numerical methods , there exists a solid theoretical framework from which we can speculate on their nature .
first , consider that the haldane phase in 1d supports degenerate spin-1/2 edge states .
next , if we couple haldane chains into an @xmath147-leg ladder , the edge states will form an overall singlet for @xmath147 even , but retain a degeneracy for @xmath147 odd due to kramers theorem .
this explains why even leg ladders form trivial spt states , while odd leg ladders form non - trivial spt states .
what happens to the edge states as @xmath148 ?
for infinitesimal interchain couplings , they must be none other than the spin-1/2 heisenberg antiferromagnetic chain , with a gapless linear dispersion near the antiferromagnetic ordering wave vector . similarly , the edge states of a three - dimensional q1d haldane phase correspond to the ground state of the spin-1/2 heisenberg square lattice antiferromagnet , which is likewise gapless .
it is natural to assume the edge state spectrum will not significantly alter as @xmath13 is increased while remaining in the quasi - one - dimensional haldane phase .
further , as long as the protecting symmetry of space inversion about chain bonds remains intact , these edge states are robust to any symmetry conserving perturbations .
the question then becomes : how can we measure these directly , both in simulation as well as in experiment ?
in this context , we mention that a method has recently been proposed for the experimental probe of fractional edge states in @xmath0 heisenberg chains.@xcite string order has been predicted to decay exponentially , with correlation length @xmath149 , for any finite coupling @xmath13 between chains.@xcite if this is the case , we might expect essentially 1d behavior when the length of individual chains @xmath150 . since spin-1/2 edge states have been observed in doped haldane chain compounds using non - magnetic dopant ions ( which break the spin chains into finite segments ) , we may surmise that the finite segments @xmath151 in such systems obey the relation @xmath152 , where we use the expectation value to reflect the distribution of @xmath151 values due to random placing of the dopant ions ( for a doping concentration @xmath153 , we expect @xmath154 ) . in a system such as y@xmath1banio@xmath4 where the interchain coupling is very small ( @xmath155),@xcite the resulting correlation length is so large ( @xmath156 ) that almost any finite doping @xmath153 is sufficient to guarantee that @xmath152 .
indeed , spin-1/2 edge states are observed in the system y@xmath1bani@xmath157mg@xmath158o@xmath4 with @xmath159 and @xmath160.@xcite however , it is interesting to speculate what might occur in haldane materials with relatively strong @xmath13 , such as pbni@xmath1v@xmath1o@xmath3 and srni@xmath1v@xmath1o@xmath3 : might there exists a critical doping @xmath161 , below which @xmath162 ?
if so , are spin-1/2 edge states still detectable ?
one exciting possible use of the symmetry protected edge states in haldane chains is for measurement - based quantum computation ( mbqc).@xcite in this quantum computing scheme , an entangled state is initially prepared , after which single qubit measurements are performed .
these measurements destroy the resource state , so this method is also referred to as _ one - way _ quantum computing .
the aklt state , and haldane ground states in general , have been shown to be a possible resource state for mbqc.@xcite however , in order to achieve universal quantum computation , a two dimensional resource state is required . in this fashion ,
the spin-3/2 aklt state on the honeycomb lattice has been proposed as a universal resource for mbqc.@xcite it is interesting to speculate that any vbs state with protected edge modes in two dimensions might prove to be a universal resource state for mbqc , in which case the haldane phase of spin-1 heisenberg chains weakly coupled into a two dimensional array ( as in fig . [ lattice ] ) might be a way to achieve such a state .
this would be very useful since aklt states of higher spins require additional hamiltonian terms beyond the bilinear and biquadratic ones needed for spin-1 .
spin-1 chains also have the distinction that the aklt state is adiabatically connected to the haldane phase at the heisenberg point ( i.e. as the biquadratic term is tuned to zero , the vbs state remains intact ) . in higher dimensions
, the heisenberg ground state tends towards nel order , which is the case on honeycomb and square lattices .
the remarkable properties of the haldane phase in spin-1 heisenberg antiferromagnets have spurred continued interest in haldane gap materials and their low - energy effective spin models . yet
surprisingly , no accurate determination of the phase diagram for these models in experimentally relevant geometries has been completed until recently . here , we have presented some of our recent work on these systems , including an accurate phase diagram for a square superlattice of weakly interacting heisenberg antiferromagnetic spin chains .
further , we have shown the magnetization and differential susceptibility curves for longitudinal magnetic fields , and calculated the low - lying dispersion along the @xmath163 plane in reciprocal space .
we have also presented direct evidence for the symmetry protected topological character of the haldane phase in quasi - one - dimensional geometries .
this includes extended string correlations along the weakly coupled chains , as well as a direct characterization through the long - range behavior of the strange correlator .
these calculations lead to a picture of the quasi - one - dimensional haldane phase as a symmetry protected topological phase with an effective gapless spin-1/2 heisenberg antiferromagnetic state at its surface . in this brief review ,
we have endeavored to introduce the haldane phase within the current classification of short - range entangled symmetric states of matter namely , the symmetry protected topological states .
this exciting development is stimulating renewed interest in this field . at the same time , recent developments in the growth and chemical synthesis of molecule - based magnets allow for greater control over the microscopic properties of magnetic materials .
the convergence of these two lines of inquiry will be a promising direction for future research .
in particular , the experimental identification of edge states at the surface of a quantum antiferromagnet would potentially realize the first known example of a bosonic spt state .
such a state could have potential uses in measurement - based quantum computing .
we look forward to many exciting developments in this field in the years to come !
we sincerely thank cristian batista , paul goddard , christopher landee , jamie manson , and vivian zapf for providing us with helpful comments and feedback during the preparation of this manuscript .
work of ps was partially supported by grant moe2011-t2 - 1 - 108 from the ministry of education , singapore .
this research used resources of the national energy research scientific computing center , which is supported by the office of science of the u.s .
department of energy under contract no .
de - ac02 - 05ch11231 . c. p. landee and m. m. turnbull , _ j. coord .
* 67 * ( 2014 ) 375 .
f. d. m. haldane , _ phys .
a _ * 93 * ( 1983 ) 464 . f. d. m. haldane , _ phys .
lett . _ * 50 * ( 1983 ) 1153 .
t. sakai and m. takahashi , _
* 58 * ( 1989 ) 3131 .
t. sakai and m. takahashi , _ phys .
b _ * 42 * ( 1990 ) 4537 .
r. botet and r. jullien , _ phys .
b _ * 27 * ( 1983 ) 613 .
o. golinelli , th .
jolicoeur and r. lacaze , _ phys .
b _ * 45 * ( 1992 ) 9798 .
s. r. white and d. a. huse , _ phys .
b _ * 48 * ( 1993 ) 3844 .
s. r. white and i. affleck , _ phys .
b _ * 77 * ( 2008 ) 134437 .
s. hu , b. normand , x. wang and l. yu , _ phys . rev .
b _ * 84 * ( 2011 ) 220402 .
s. moukouri and e. eidelstein , _ phys .
b _ * 84 * ( 2011 ) 193103 .
s. moukouri and e. eidelstein , _ phys .
b _ * 86 * ( 2012 ) 155112 .
a. f. albuquerque , c. j. hamer and j. oitmaa , _ phys .
b _ * 79 * ( 2009 ) 054412 .
y. j. kim and r. j. birgeneau , _ phys .
b _ * 62 * ( 2000 ) 6378 .
m. matsumoto , c. yasuda , s. todo and h. takayama , _ phys .
b _ * 65 * ( 2001 ) 014407 .
k. wierschem and p. sengupta , _ j. phys .
* 400 * ( 2012 ) 032112 .
k. wierschem and p. sengupta , _ phys .
* 112 * ( 2014 ) 247203 .
k. wierschem and p. sengupta , _ jps conf .
proc . _ * 3 * ( 2014 ) 012005 .
h. mutka , c. payen , p. molini , j. l. soubeyroux , p. colombet and a. d. taylor , _ phys .
* 67 * ( 1991 ) 497 .
t. asano , y. ajiro , m. mekata , h. yamazaki , n. hosoito , t. shinjo and h. kikuchi , _ solid state commun . _
* 90 * ( 1994 ) 125 .
m. takigawa , t. asano , y. ajiro and m. mekata , _ phys .
b _ * 52 * ( 1995 ) r13087 .
m. takigawa , t. asano , y. ajiro , m. mekata , and y. j. uemura , _ phys .
* 76 * ( 1996 ) 2173 .
z. honda , k. katsumata , y. nishiyama and i. harada , _ phys .
b _ * 63 * ( 2001 ) 064420 .
e. imr , m. ozerov , o. ignatchik , t. p. papageorgiou , j. wosnitza , s. a. zvyagin , j. krzystek , z. zhou , c. p. landee , b. r. landry , m. m. turnbull and j. l. wikaira , _
new j. phys . _
* 10 * ( 2008 ) 033008 .
j. p. renard , m. verdaguer , l. p. regnault , w. a. c. erkelens , j. rossat - mignod and w. g. stirling , _ europhys .
lett . _ * 3 * ( 1987 ) 945 .
j. p. renard , m. verdaguer , l. p. regnault , w. a. c. erkelens , j. rossat - mignod , j. ribas , w. g. stirling and c. vettier , _ j. appl
* 63 * ( 1988 ) 3538 .
l. p. regnault , i. zaliznyak , j. p. renard and c. vettier , _ phys .
b _ * 50 * ( 1994 ) 9174 .
i. a. zaliznyak , d. c. dender , c. broholm and d. h. reich , _ phys .
b _ * 57 * ( 1998 ) 5200 .
y. uchiyama , y. sasago , i. tsukada , k. uchinokura , a. zheludev , t. hayashi , n. miura and p. bni , _ phys .
* 83 * ( 1999 ) 632 .
a. zheludev , t. masuda , i. tsukada , y. uchiyama , k. uchinokura , p. bni and s .- h .
lee , _ phys .
b _ * 62 * ( 2000 ) 8921 .
b. pahari , k. ghoshray , r. sarkar , b. bandyopadhyay , and a. ghoshray , _ phys .
b _ * 73 * ( 2006 ) 012407 .
a. k. bera , b. lake , a. t. m. n. islam , b. klemke , e. faulhaber and j. m. law , _ phys .
b _ * 87 * ( 2013 ) 224423 .
v. gadet , m. verdaguer , v. briois , a. gleizes , j. p. renard , p. beauvillain , c. chappert , t. goto , k. le dang and p. veillet , _ phys .
b _ * 44 * ( 1991 ) 705 .
j. darriet and l. regnault , _ solid state commun .
_ * 86 * ( 1993 ) 409 .
g. xu , j. f. ditusa , t. ito , k. oka , h. takagi , c. broholm and g. aeppli , _ phys .
b _ * 54 * ( 1996 ) 6827 . ch .
regg , a. furrer , d. sheptyakov , th .
strssle , k. w. krmer , h .- u .
gdel and l. mlsi , _ phys .
_ * 93 * , 257201 ( 2004 ) .
v. s. zapf , m. jaime and c. d. batista , _ rev .
* 86 * ( 2014 ) 563 . v. s. zapf , d. zocco , b. r. hansen , m. jaime , n. harrison , c. d. batista , m. kenzelmann , c. niedermayer , a. lacerda and a. paduan - filho , _ phys .
* 96 * ( 2006 ) 077204 .
s. a. zvyagin , j. wosnitza , c. d. batista , m. tsukamoto , n. kawashima , j. krzystek , v. s. zapf , m. jaime , n. f. oliveira jr . and a. paduan - filho , _ phys . rev .
* 96 * ( 2006 ) 077204 .
l. yin , j. s. xia , v. s. zapf , n. s. sullivan , and a. paduan- filho , _ phys .
lett . _ * 101 * ( 2008 ) 187205 .
p. sengupta and c. d. batista , _ phys .
_ * 99 * ( 2007 ) 217205 .
p. sengupta and c. d. batista , _ phys .
lett . _ * 98 * ( 2007 ) 227201 .
k. wierschem , y. kato , y. nishida , c. d. batista , and p. sengupta , _ phys .
b _ * 86 * ( 2012 ) 201108 .
p. a. goddard , j. l. manson , j. singleton , i. franke , t. lancaster , a. j. steele , s. j. blundell , c. baines , f. l. pratt , r. d. mcdonald , o. e. ayala - valenzuela , j. f. corbey , h. i. southerland , p. sengupta and j. a. schlueter , _ phys .
* 108 * ( 2012 ) 207201 .
t. lancaster , p. a. goddard , s. j. blundell , f. r. foronda , s. ghannadzadeh , j. s. mller , p. j. baker , f. l. pratt , c. baines , l. huang , j. wosnitza , r. d. mcdonald , k. a. modic , j. singleton , c. v. topping , t. a. w. beale , f. xiao , j. a. schlueter , a. m. barton , r. d. cabrera , k. e. carreiro , h. e. tran and j. l. manson , _ phys .
* 112 * ( 2014 ) 077208 .
m. den nijs and k. rommelse , _ phys .
b _ * 40 * ( 1989 ) 4709 . i. affleck , t. kennedy , e. h. lieb and h. tasaki , _ phys . rev
* 59 * ( 1987 ) 799 .
gu and x .- g .
wen , _ phys .
b _ * 80 * ( 2009 ) 155131 . f. pollmann , a. m. turner , e. berg and m. oshikawa , _ phys . rev .
b _ * 81 * ( 2010 ) 064439 .
r. k. kaul , r. g. melko and a. w. sandvik , _ annu .
rev . condens .
matter phys .
_ * 4 * ( 2013 ) 179 .
a. w. sandvik and j. kurkijrvi , _ phys .
b _ * 43 * ( 1991 ) 5950 .
a. w. sandvik , _ j. phys .
a _ * 25 * ( 1992 ) 3667 . o. f. syljusen and a. w. sandvik , _ phys .
* 66 * ( 2002 ) 046701 .
o. f. syljusen , _ phys .
* 67 * ( 2003 ) 046701 .
f. alet , s. wessel and m. troyer , _ phys .
* 71 * ( 2005 ) 036706 .
k. wierschem and p. sengupta , _ phys .
b _ * 90 * ( 2014 ) 115157 .
j. l. manson , a. g. baldwin , b. l. scott , j. bendix , r. e. del sesto , p. a. goddard , y. kohama , h. e. tran , s. ghannadzadeh , j. singleton , t. lancaster , j. s. mller , s. j. blundell , f. l. pratt , v. s. zapf , j. kang , c. lee , m .- h .
whangbo and c. baines , _ inorg .
chem . _ * 51 * ( 2012 ) 7520 .
xia , a. ozarowski , p. m. spurgeon , a. g. baldwin , j. l. manson and m. w. meisel , arxiv:1409.5971 .
m. chiba , y. ajiro , h. kikuchi , t. kubo and t. morimoto , _ phys .
b _ * 44 * ( 1991 ) 2838 .
m. orend , s. zvyagin , a. orendov , m. sieling , b. lthi and m. w. meisel , _ phys .
b _ * 60 * ( 1999 ) 4170 .
z. zhang , k. wierschem , i. yap , y. kato , c. d. batista and p. sengupta , _ phys .
b _ * 87 * ( 2013 ) 174405 .
f. weickert , r. kchler , a. steppke , l. pedrero , m. nicklas , m. brando , f. steglich , m. jaime , v. s. zapf , a. paduan - filho , k. a. al - hassanieh , c. d. batista and p. sengupta , _ phys .
b _ * 85 * ( 2012 ) 184408 .
s. chattopadhyay , d. jain , v. ganesan , s. giri and s. majumdar , _ phys .
b _ * 82 * ( 2010 ) 094431 .
j. manson , _ private communication _ ( 2014 ) .
n. metoki , y. koike , y. matsuoka , t. kobayashi and m. yamashita , _ j. magn .
* 226 - 230 * ( 2001 ) 414 .
a. zheludev , s. e. nagler , s. m. shapiro , l. k. chou , d. r. talham and m. w. meisel , _ phys .
b _ * 53 * ( 1996 ) 15004 .
m. kenzelmann , r. a. cowley , w. j. l. buyers , z. tun , r. coldea and m. enderle , _ phys .
b _ * 66 * ( 2002 ) 024407 .
s. lee , j .-
park , d. t. adroja , d. khomskii , s. streltsov , k. a. mcewen , h. sakai , k. yoshimura , v. i. anisimov , d. mori , r. kanno and r. ibberson , _ nat .
mater . _ * 5 * ( 2006 ) 471 .
v. a. kashurnikov , n. v. prokofev , b. v. svistunov and m. troyer , _ phys . rev .
b _ * 59 * ( 1999 ) 1162 . a. w. sandvik and a. sudb , _ europhys .
_ * 36 * ( 1996 ) 443 .
l. wang and a. w. sandvik , _ phys .
* 97 * ( 2006 ) 117204 .
s. todo , m. matsumoto , c. yasuda and h. takayama , _ phys .
b _ * 64 * ( 2001 ) 224412 .
f. anfuso and a. rosch , _ phys .
b _ * 75 * ( 2007 ) 144420 .
f. anfuso and a. rosch , _ phys .
b _ * 76 * ( 2007 ) 085124 . h. li and f. d. m. haldane , _ phys
lett . _ * 101 * ( 2008 ) 010504 .
you , z. bi , a. rasmussen , k. slagle and c. xu , _ phys .
lett . _ * 112 * ( 2014 ) 247202 .
f. delgado , c. d. batista and j. fernndez - rossier , _ phys .
* 111 * ( 2013 ) 167201 . f. tedoldi , r. santachiara and m. horvati , _ phys . rev
* 83 * ( 1999 ) 412 .
r. raussendorf and h. j. briegel , _ phys .
_ * 86 * ( 2001 ) 5188 .
g. k. brennen and a. miyake , _ phys .
* 101 * ( 2008 ) 010502 .
a. miyake , _ phys .
* 105 * ( 2010 ) 040501 .
a. miyake , _ ann .
* 326 * ( 2011 ) 1656 .
wei , i. affleck and r. raussendorf , _ phys .
lett . _ * 106 * ( 2011 ) 070501 .
wei , i. affleck and r. raussendorf , _ phys .
a _ * 86 * ( 2012 ) 032328 . | we review the basic properties of the haldane phase in spin-1 heisenberg antiferromagnetic chains , including its persistence in quasi - one - dimensional geometries . using large - scale numerical simulations , we map out the phase diagram for a realistic model applicable to experimental haldane compounds .
we also investigate the effect of different chain coupling geometries and confirm a general mean field universality of the critical coupling times the coordination number of the lattice .
inspired by the recent development of characterization of symmetry protected topological states , of which the haldane phase of spin-1 heisenberg antiferromagnetic chain is a preeminent example , we provide direct evidence that the quasi - one - dimensional haldane phase is indeed a non - trivial symmetry protected topological state . |
multi - label classification is the set of classification problems where the output vector has a variable length .
the average number of labels per review varies across datasets , and in general is a function of the semantics of the text rather than the syntax .
the learning algorithm needs to estimate the number of labels and make the correct predictions .
+ previously multi - label classification problems were solved using problem transformation techniques ( converting the problem into binary classification problems per output label ) , or by adapting the algorithm to directly perform multi - label classification @xcite . this paper extracts correlation information between labels and factors the joint probabilities into the model . +
given large datasets of product reviews from amazon and twitter and manually labelled multi - label classifications for each ( ground truth ) , the algorithm aims to predict all classifications for a review @xcite .
we aim to make the algorithm portable across various datasets , i.e. , training a model on amazon reviews and using it to classify tweets from twitter .
the amazon dataset contains reviews for 400,000 products across 40,000 different categories . among these
, we focused on books and book subcategories .
the twitter dataset has 100,000 tweets .
both the datasets have the same schema , and each review and tweet has a unique i d , pre - pruned content , and a tree of all of the product labels for the review up to the book category at the root .
our approach augments the independent binary classification model .
in addition to the individual probabilities , this also looks at the probability of labels occurring together .
hence the inference algorithm models the formula + @xmath0 where @xmath1 + @xmath2 represents the probability of the independent binary model for label @xmath3 classifying as 0 or 1 given input @xmath4 .
@xmath5 represents the assignment of 0 or 1 to all @xmath6 that maximizes the joint probability .
+ the pairwise correlation probability between each pair of labels is stored in the correlation matrix .
this normalized rows of the matrix represent probability of a particular label s coupling with every other label except for itself .
this probability is computed as a prior for the entire dataset and is not dependent on the input feature vector itself .
it is represented by @xmath7 .
+ @xmath8 represents the joint probability of all pairwise combinations of predicted labels , each discounted by @xmath9 .
the probabilities obtained from the correlation matrix are given less weight than the probabilities obtained from independent models , hence the discount .
this is because co - occurrence of labels is not completely characterized by correlation , it also requires the higher order moments which significantly increase the computational overhead .
the entire algorithm is split into 3 steps after parsing preprocessing , training independent classifiers , and incorporating co - occurrence probability to make the final label set .
hyper - parameters need to be tuned for each dataset .
+ liblinear works well for larger datasets while libsvm works well for custom and smaller datasets .
liblinear only supports a linear kernel but is very fast relative to libsvm and hence used in the learning algorithms @xcite .
+ the amazon and twitter datasets are parsed to extract book reviews and their corresponding labels ( ground truth ) .
the labels are structured in a hierarchy - for books , there are 31 top level labels and for each label there are several sub - labels .
for instance a top level label is `` literature and fiction '' and few sub labels associated with it are `` folklore '' , `` mysteries '' and `` classics . ''
the reviews are labelled with these sub - level labels which we map to one of the 31 top - level labels , and use for multi - label classification .
+ training and testing of the algorithm is done on at most 3,000 reviews per dataset due to computational constraints during learning .
each book review is mapped to a bag - of - words based input feature .
on parsing a review , its content is pruned and stemmed and then a tf - idf based vector is generated which is used as the input feature @xmath4 .
the labels corresponding to this review are associated as its outputs @xmath10 .
+ the correlation matrix is built by parsing 100,000 book reviews and creating pairwise counts of different labels a 31 @xmath11 31 row normalized matrix with laplace smoothing .
after normalization the matrix is not symmetric , so the geometric mean of cells @xmath12and @xmath13 is treated as the actual correlation probability for two labels .
plots the row normalized covariance matrix for amazon dataset .
the existence of large peaks is the motivation behind the algorithm .
+ * the inputs to the learning algorithm are tf - idf based input features , their manually assigned labels , and the prior generated correlation matrix * 31 independent binary models are trained using l2-regularized svm .
this serves as the * baseline*. * for a review , the j labels with highest probabilities from the independent models are selected .
* probabilities of pairwise - combinations of these labels are computed as @xmath14 * k @xmath15 pairs with maximum probability product are chosen and distinct labels constitute the predicted set .
the algorithm does not compute the theoretical model exactly but is an approximation .
it is similar to beam search .
+ .hyperparameters for amazon dataset [ cols="<,<,<,<",options="header " , ] the correlation algorithm weighs false positives and false negatives differently , hence the @xmath16 score gives a more accurate understanding of the algorithm s performance .
@xmath16 for the correlation algorithm s test on amazon is 30.19% better than the baseline s test , 2.82% better than the baseline s test on cross - domain learning , and 0.3% worse than baseline s test on twitter .
this validates the trend in the plots .
+ the false positive counts when training and testing purely on one dataset are usually higher than the false positive counts from the baseline , as in our approach we tend to discount the error of making false positives given we meet true positives .
+ table v shows the confusion matrix for cross domain learning where the correlation algorithm s true negative count is much higher than that of the baseline . by using the correlation between output labels ,
the algorithm discards labels that independently had a high probability but were not well correlated with other high probability labels .
+ as evident , the observed training error for each dataset was much less than the observed test error , indicating high variance and low bias .
to further investigate , principal component analysis was used . since the feature vectors were based on word occurrences in text , they were significantly larger than the dataset size the observed feature size for 2500 dataset size of amazon reviews was 12000 . due to computational limitations ,
straightforward pca was infeasible .
+ sparse notation was used to represent feature vector , from python s numpy library . then sparse svd ( singular value decomposition )
found the smallest subspace that the feature matrix mapped to , keeping all singular values greater than 1 .
this significantly reduced the dataset size .
the 2500 dataset of amazon reviews , now reduced to a feature vector of size 2300 .
while the number of features was still comparable to the number of data points , it was substantially lower than the size of the raw feature vector ( 1/6 ) .
+ this experimentation exposed two facts .
the errors were still the same , but the computation time rose significantly ( about five times the previous duration ) .
it was seen that liblinear , the python library , was tuned to work with large datasets with document based feature vectors ( as compared to libsvm ) , and not for any general features .
+ since the baseline itself suffers from the issue of high variance and low bias , any augmentations over it would be unable to resolve this by themselves , and would require modifying how the tf - idf vectors are generated .
we had already stemmed the words to check this issue , but evidently this would require a more careful processing of the reviews , before they are converted to features .
moreover , since svms enforce larger margin as a metric to evaluate each point , they end with a much lower vc dimension than the size of the feature vector .
this was another reason in favor of svm over logistic regression , naive bayes , and random forest .
+ the baseline resulted in about 10% training error , while our algorithm resulted in about 15% training error .
thus there is an increased bias , although the higher variance is a more pressing issue because the test error is significantly higher than training error for both algorithms . +
using the correlation matrix to help classify reviews appeared to be a good approach , justified by peaks in correlation .
the results show , however , that when performing multi - label classification with a sufficiently sized feature set , augmentation of an independent model with second order correlation probability shows only marginal improvements .
+ the model was modified from the baseline by adding a single feature : second order correlation .
this is the greedy method of selecting features , in contrast to exploring correlation of all combinations of labels .
in such a case where we did not explore higher order correlation we could have missed , for example , that labels a , b , and c never occur together a piece of information that could have proved vital to the model s representation of the true data .
thus , if computation time and space are not an issue , higher order correlation should be considered .
+ additionally , svm , as was already known , has shown to be best for learning on large feature sets because it will attempt to reduce the set and hence generalize better .
the correlation matrix was built specific for every dataset , so to apply the learning algorithms on cross datasets a generalized correlation framework is needed .
+ currently the hyper - parameter k is optimized to choose number of predicted labels using supervised techniques .
the number of labels for each review could be predicted by using unsupervised techniques such as k - means clustering on label probabilities .
+ the datasets have different inherent structures which do not generalize well when using a common learning algorithm . for the twitter dataset ,
each feature vector is about 12% the size of a corresponding feature vector in the amazon dataset .
the features should be supplemented with options such as semantic relations in text to make classifications and context of user history .
+ the reviews have sub - level labels associated with them which are rolled up to first - level labels to make the classifications .
the algorithm should extend to make hierarchical classifications .
sincere thanks to professor ashutosh saxena and to aditya jami for advising the research team and providing the labelled datasets of product reviews from amazon and twitter . | this paper attempts multi - label classification by extending the idea of independent binary classification models for each output label , and exploring how the inherent correlation between output labels can be used to improve predictions . logistic regression , naive bayes , random forest , and svm models were constructed , with svm giving the best results : an improvement of 12.9% over binary models was achieved for hold out cross validation by augmenting with pairwise correlation probabilities of the labels . |
in a seminal series of papers @xcite baxter first introduced the concept of the so - called ` sticky hard sphere ' ( shs ) models , as the simplest - albeit crude - modelization for real fluids of spherical particles with a strong surface adhesion . in baxter s original formulation @xcite and its extension to the multi - component case @xcite ( both hereafter referred to as shs1 model )
the pair potential contains - in addition to a hard sphere ( hs ) repulsion - a infinitely deep and narrow attractive square - well , obtained according to a particular limiting procedure ( baxter s ` sticky limit ' ) that keeps the second virial coefficient finite baxter68 .
although this model appears rather pathological at first sight , it includes a number of interesting features which justify its wide popularity .
first , the ornstein - zernike ( oz ) integral equation of the statistical - mechanical theory of fluids can be analytically solved for it within the _ percus - yevick _ ( py ) approximation and the solution exhibits a gas - liquid transition @xcite .
second , baxter s model has already proven to be appropriate for describing some properties of colloidal suspensions , micelles , microemulsions and protein solutions with short - range interactions as well as some aspects of adsorption , flocculation and percolation phenomena , solvent - mediated forces , ionic mixtures , solutions with a small degree of size polydispersity and fluids of chain - like molecules ( for an illustrative , albeit not exhaustive , list of references , see @xcite ) .
all this means that , in spite of its highly idealized character and known shortcomings stell91 , the shs1 model is able to capture some important physical features of structure , thermodynamics , and phase behaviour of many real systems . however , the shs1 model has its main drawback in its problematic application to mixtures with a large number @xmath0 of components - as occurs for colloidal suspensions with large polydispersity - since this case requires the solution for a set of @xmath1 coupled quadratic equations @xcite , a task which can not be accomplished analytically .
this important fact has originated a more recent attempt of finding an alternative shs model , which could be analytically tractable even in the general multi - component case brey87,mier89,ginoza96,tutschka98,tutschka00,tutschka01,tutschka02,gazzillo00,gazzillo02a , gazzillo02b , gazzillo03,timoneda89,ginoza01 .
for pure fluids , brey _ et al . _
@xcite proposed to start from a hs - yukawa ( hsy ) potential @xmath2 with @xmath3where @xmath4 is the distance between particles , @xmath5 ( @xmath6 being boltzmann s constant , and @xmath7 the temperature ) , @xmath8 denotes the hs diameter , @xmath9 the yukawa inverse range , @xmath10 an energy , and @xmath11 a reduced temperature ( as in shs1 , the factor @xmath12 simplifies subsequent analysis ) . the definition of this second hamiltonian ( shs2 model in the following ) is completed by the definition of a ` sticky limit ' , which in this case amounts to taking @xmath13 @xcite .
it is worth remarking that , unlike its counterpart in the shs1 model , the starting potential @xmath14 is _ itself _ independent of temperature . for the shs2 model ( in both cases of pure fluids and mixtures )
, the oz equation can be solved analytically within the _ mean spherical approximation _ ( msa ) @xcite .
it turns out that the shs2-msa solution @xmath15 ( or @xmath16 ) for the baxter form of the oz equation has exactly the same @xmath4-dependence as the shs1-py solution , i.e. both are expressed in terms of a second - degree polynomial in @xmath4 .
however , the difference between the two solutions lies in the density and temperature dependence of their polynomial coefficients ( see below ) , which makes the shs2-msa solution readily usable even in the multi - component case tutschka98,tutschka00,tutschka02,gazzillo00,gazzillo02a , unlike the shs1-py one . unfortunately , the common belief that the shs1 and shs2 hamiltonians are different but equivalent representations of a _
unique _ shs potential jamnik96,jamnik01,tutschka98,tutschka00,tutschka01,tutschka02,jamnik91,hoye93 as well as a failure in appreciating the subtle distinction between _ model _ and _ solution _ @xcite have often generated a number of misunderstandings and erroneous beliefs in the literature on shs fluids .
in particular , both aforesaid py and msa solutions have sometimes been regarded as corresponding to the same shs model ( baxter s one ) jamnik96,tutschka98,tutschka00,tutschka01,tutschka02,jamnik91 .
on the contrary , in the present paper we stress that the shs1-py and shs2-msa _ solutions _ are different not only as stemming from _ different closures _ , but , more importantly , since they refer to _
different hamiltonians_. in fact , unlike shs1 , the shs2 model _ itself _ is ill - defined from a thermodynamical point of view , and thus shs1 and shs2 can not be equivalent .
we will prove this point by considering the _ exact _ second virial coefficient of the hsy potential ( [ i1 ] ) and showing that it _ diverges _ in the sticky limit ( a very short , preliminary , account of these results has been given in ref .
as we further elaborate below , this pathology is hidden in the shs2-msa compressibility equation of state ( eos ) but is mirrored by a similar singular behaviour of the corresponding msa virial and energy eos s , never investigated in previous studies .
the second and main goal of the present paper is to propose a new model ( shs3 ) which combines the advantages of both shs1 and shs2 . as shs1
, it has a finite second virial coefficient , and thus it is a well - defined model .
as shs2 however , it admits a simple analytical solution within a new closure , referred to as _ modified mean spherical approximation _ ( mmsa ) in the following .
the remarkable property of the shs3-mmsa solution for @xmath15 is that it turns out to be formally identical with the shs2-msa one , and so are all quantities which are immediately derivable from it ( such as structural properties and compressibility eos ) . as a consequence of this mapping , all shs2-msa results obtained in the past for these quantities can be recovered after an appropriate re - interpretation .
in addition , we will provide new results for other quantities , notably the virial and energy eos , for which the shs2-msa solution badly fails .
this is _ not _ the case in the shs3-mmsa solution , where the energy eos turns out to be finite , albeit mean - field - like .
finally , although the virial eos displays a singular behaviour within the shs3-mmsa as well , we will argue that , unlike the shs2 case , this divergence is a consequence of the deficiency of the mmsa closure , and _ not of the shs3 model itself_. our findings thus provide a sound theoretical basis for a critical analysis of the existing literature on shs models , and allow to clarify misunderstandings and discard incorrect results , previously reported in the literature . the plan of the remaining of the paper is the following . in section ii we will briefly recall the main features of the shs2-msa solution and describe the drawbacks of shs2 in detail . the new model along with the new closure , which constitute the central part of this work , are both discussed in section iii .
final remarks and future perspectives close the paper ( section iv ) .
for a one - component fluid of @xmath17 molecules in a volume @xmath18 with spherically symmetric interactions , the baxter form of the oz integral equation baxter71a is given by @xmath19 where @xmath20 denotes the number density , @xmath21 the direct correlation function ( dcf ) , @xmath22 , with @xmath23 being the radial distribution function ( rdf ) .
moreover , the prime denotes differentiation with respect to @xmath24 .
solving the baxter equations is tantamount to determining the factor correlation function @xmath15 , an auxiliary quantity from which @xmath25 and @xmath26 can be easily derived .
an approximate integral equation can be obtained by adding to the oz equation some approximate ` closure ' relating @xmath27 , @xmath28 and the potential @xmath29 . for potentials with a hard - core part , the _
mean spherical approximation _ ( msa ) reads @xmath30 , & & 0<r<\sigma , \\
-\beta \phi _ { { \rm tail}}\left ( r\right ) , & & r\geq \sigma , % \end{array}% \right . \label{clos2}\ ] ] where @xmath31 , @xmath32 is the potential outside the core . since @xmath33 the msa
may also be written as@xmath34 , \label{clos5}\]]where @xmath35 = \theta \left ( r-\sigma \right ) , $ ] with @xmath36 being the hs potential , and @xmath37 the heaviside step function ( @xmath38 for @xmath39 and @xmath40 for @xmath41 ) .
for the hsy potential the msa closure becomes @xmath42 where the parameters are the same as in eq .
( [ i1 ] ) .
the analytic msa solution @xmath43 for the hsy fluid @xcite yields in the sticky limit ( @xmath44 ) the shs2-msa solution @xmath45 \theta \left ( \sigma -r\right ) , \qquad \text { for \ } r\geq 0 , \label{so1}\ ] ] @xmath46 @xmath47 where @xmath48 is the packing fraction .
as remarked , the expression ( @xmath49 ) is formally identical with the shs1-py solution with the crucial difference that @xmath50 depends only on temperature ( being proportional to @xmath51 , whereas @xmath52 depends on _ both _ temperature _ and _ density ( in a more complex way ) @xcite . once @xmath15 is known , all structural and thermodynamic properties can , in principle , be calculated .
it is worth mentioning that , with a few exceptions @xcite , the thermodynamic properties of shs2 are still mostly unexplored .
however , we stress once again that the main aim of our paper is not to present new results on the shs2-msa thermodynamics ( although this task will be accomplished too ) , but to reveal a dramatic fault of the shs2 potential _ itself _ ( irrespectively of any approximate closure ) , which , surprisingly enough , has never been observed in the previous literature , but will emerge from the following simple analysis .
the hamiltonian of the shs1 model was introduced by baxter as a limiting case , through a clever definition of _ both _ a starting potential and a limit procedure in such a way that the contribution of the vanishing square - well tail to the second virial coeffient , @xmath53 , remains finite and nonzero .
it is instructive to consider the behaviour of @xmath53 for the shs2 model , by considering the result of the @xmath13 limit for the _ exact _ second virial coefficient @xmath54 of the hsy fluid .
this can be calculated from the general definition , @xmath55with @xmath56 and @xmath57 $ ] being the boltzmann factor .
one finds @xmath58where @xmath59 -1\right\ } , \]]it is now easy to show that @xmath60 diverges in the @xmath13 limit .
to this aim we note that , for @xmath61 , one has @xmath62 and hence we can use the bound @xmath63 as the right hand side of eq.([bound ] ) diverges as @xmath64 we have thus shown that @xmath65this result is exact and _ independent _ of any closure , and reflects an inconsistency of the hsy potential with the definition of sticky limit employed when setting up the shs2 hamiltonian @xcite . as a consequence ,
the shs2 hamiltonian is ill - defined from the outset , and the corresponding model ( which can not be a different representation of baxter s one ) _ must be discarded_. this is , however , a surprising result in some respect .
one may rightfully wonder why no trace of the pathological nature of the shs2 model has ever been revealed by a number of structural studies tutschka98,tutschka00,tutschka02,gazzillo00,gazzillo02a carried out on its multi - component version . as discussed in section iv , all these structural results based upon the one and multi - component shs2-msa solutions are , in fact , fully correct after their re - interpretation in terms of shs3 . before doing this , however , it is instructive to consider those shs2-msa thermodynamic properties of the one - component fluid which have _ not _ been investigated so far .
this _ will _ display the pathological nature of the model as shown next .
mier - y - teran _ et al . _
@xcite obtained , for @xmath66 , the msa _ _ compressibility _ _ @xmath67 _ _ eos @xmath68 \frac{1}{% t^{\ast } } \nonumber \\ & & \ + \left [ \frac{2-\eta } { 1-\eta } + \frac{2}{\eta } \ln ( 1-\eta ) \right ] \frac{1}{t^{\ast 2}}.\end{aligned}\]]by integrating with respect to density the _ compressibility equation _
@xcite @xmath69 ^{2}=a^{2 } , \label{eosc}\ ] ] where @xmath70 denotes the unidimensional fourier transform of @xmath15 . on the other hand , no expressions for the shs2-msa virial and energy eos were given either in ref .
@xcite or in subsequent literature on this subject , to the best of our knowledge .
we thus tackle this analysis here .
the most direct way we have followed is to consider the @xmath71 limit of the hsy - msa virial and energy pressures .
convenient expressions for these quantities were given by cummings and smith ( eqs .
( 18)-(20 ) of ref .
@xcite ) . unfortunately
, their sticky limit requires a rather elaborate analysis , which we will not report here .
however , the final result is rather simple : _ both the virial and the energy msa pressures of the hsy fluid diverge in the sticky limit .
_ for the msa energy eos , the same conclusion can be drawn by exploiting the alternative form reported by herrera _
it is worth stressing again that this important feature has never been pointed out before , to our knowledge . to convince the reader that our statement is correct , we now provide a simpler demonstration , based upon the analysis of the sticky limit of the msa second virial coefficients of the hsy fluid , as obtained from the density expansion of @xmath72 , @xmath73 and @xmath74 .
it is well known that , in any approximate theory , compressibility , virial and energy eos s may yield different virial coefficients , and the first one at which this difference begins to appear depends on the chosen closure .
we now show that , in the shs2-msa case , a non - consistency already appears at the @xmath75-level , and partly veils the singular character of the shs2 hamiltonian . nonetheless , as it will be shown below , _ both _
@xmath76 _ and _ @xmath77 are divergent , and this strongly supports the aforementioned statement about the singular character of the virial and energy eos for the shs2-msa solution .
the quantity @xmath78 can be easily computed directly from the density expansion of @xmath79 , given by eq.([eosc ] ) , upon using eqs .
( [ so2])-([so3 ] ) . to evaluate @xmath80 and @xmath81
, it proves convenient to exploit the following expressions for the msa virial and energy eos of a hsy fluid hoye77b , arrieta87 @xmath82 , \label{d6}\ ] ] @xmath83 -i\right\ } , \label{d7}\ ] ] @xmath84 using the low - density expansion of the msa approximation to @xmath85 , eq .
( @xmath86 ) , one gets @xmath87 .
\label{d9}\ ] ] @xmath88 .
\label{d10}\ ] ] @xmath89 .
\label{d11}\ ] ] where @xmath90 is the particle volume .
the difference among the three results for @xmath91 is a clear manifestation of the known thermodynamic inconsistency of the msa cummings79a , cummings83 .
it is however quite surprising that discrepancies already appear at the @xmath53-level .
note that @xmath92 .
these differences magnify with increasing @xmath9 , and become dramatic in the sticky limit which yields @xmath93 @xmath94 eq .
( [ d13 ] ) thus confirms our previous statement about the divergence of the msa virial and energy pressures of the hsy fluid in the sticky limit . on the other hand , eq .
( [ d12 ] ) gives a finite value for @xmath95 in agreement with the corresponding compressibility eos .
however , we stress that this result is due to the fact that the msa involves an approximation at the level of boltzmann factor @xcite , i.e. @xmath96 , \label{d14}\ ] ] equivalent to the linearization @xmath97
\simeq 1-\beta \phi _ { { \rm tail}}\left ( r\right ) $ ] ( as can be inferred from a comparison between the zeroth - order terms in the density expansion of @xmath98 and @xmath99 , i.e. @xmath100 $ ] and @xmath101 $ ] , respectively ) .
the replacement of the exact @xmath102 in the expression for @xmath103 with its msa counterpart ( [ d14 ] ) just leads to the @xmath104 result , in the sticky limit . summarizing
, we can conclude that the existence of the msa solution , @xmath105 , is a fortuitous consequence of the fact that the divergent character of the shs2 model is _ masked _ by the msa linearization ( @xmath106 ) . as a final point of this first part
, we wish to point out an interesting different behaviour between the internal energy in the shs2-msa and shs1-py solution .
let us now consider the _ energy _
@xmath107 _ equation , _ @xmath108 where @xmath109 is the excess internal energy . within the msa ,
one gets for our hsy fluid @xmath110 \left [ -\beta \phi _ { { \rm y}}\left ( r\right ) \right ] \\ & = & 2\pi \rho \left\ { k_{0}\int_{\sigma } ^{\infty } dr\ r\left [ 1+\gamma _ { % { \rm hsy - msa}}\left ( r\right ) \right ] \ ze^{-z(r-\sigma ) } + \frac{k_{0}^{2}}{2% } z\right\ } .\end{aligned}\ ] ] since @xmath111 is finite and sufficiently regular everywhere , it is easy to show , integrating by parts or using the property @xmath112 @xcite , that the last integral remains finite as @xmath113 .
in conclusion , @xmath114 _ diverges linearly_. this should be compared with the _ logarithmic _ divergence found in shs1-py energy @xcite .
we speculate that this difference in the internal energy behaviour parallels that in the corresponding eos .
it is then clear that , in the shs1-py case , such a weak divergence does not in itself constitute an impediment to the existence of finite eos s in the sticky limit baxter68,stell91,yuste93a . as a side comment ,
it is worth noticing that our finding contradicts ref .
@xcite , where an msa energy with a non - singular sticky contribution ( eq .
( 42 ) of that paper ) is reported .
given the above premise , one could , at this point , suspect that all previous findings based upon the shs2 model ( and notably , those referring to polydisperse colloidal fluids @xcite ) should be discarded .
this is not so , and all those results are , in fact , correct .
the second aim of our paper is to show that an analytic solution with the same functional form as @xmath115 is obtained within a new and well - defined model ( which we call shs3 ) , coupled with a simple new closure . as a consequence ,
most of the results derived from shs2 can be recovered after an appropriate re - interpretation .
we define a new shs hamiltonian , which represents the simplest correct alternative to shs2 , and is analytically solvable within a novel closure .
the basic idea hinges on baxter s trick of a logarithmic tail , combined with the infinite - ranged yukawa expression of shs2 .
the starting potential ( hereafter referred to as m3 ) is @xmath116 , & & r\geq \sigma , % \end{array}% \right .
\label{m1}\ ] ] where @xmath117 depends on @xmath7 according to eq .
( @xmath118 ) , and the sticky limit corresponds again to @xmath119 we note that at large @xmath4 values , the quantity @xmath120 , appearing in the argument of the logarithm , becomes so small that the approximation @xmath121 can be used . in other words , m3 has the same large-@xmath4 asymptotic behaviour as @xmath122 , but differs from the yukawa tail near contact .
this is depicted in figure 1 , where it can be clearly seen that this difference between the two potentials increases as @xmath9 increases .
however , for the m3 model , the boltzmann factor reads@xmath123 , \label{m2}\]]with the essential consequence that the corresponding _ exact _ second virial coefficient remains finite in the sticky limit , i.e. , @xmath124 , \ ] ] @xmath125 note that this @xmath53-expression , _ exact _ for shs3 , is formally identical with @xmath126 , which is only _ approximate _ for shs2 . as a further check that we are on the right track
, we have also computed the _ exact _ shs3 third virial coefficient @xmath127 it is remarkable that @xmath128 coincides with @xmath129 , once @xmath11 is replaced with its counterpart @xmath130 of baxter s model @xcite .
we define a _ modified mean spherical approximation _ ( mmsa ) @xmath131 , & & 0<r<\sigma , \\ f_{{\rm tail}}\left ( r\right ) = \exp \left [ -\beta \phi _ { { \rm tail}}\left ( r\right ) \right ] -1 , & & r\geq \sigma .%
\end{array}% \right .
\label{clos3}\ ] ] in terms of rdf , this closure reads @xmath132 \nonumber \\ & = & e\left ( r\right ) + e_{{\rm hs}}\left ( r\right ) \gamma \left ( r\right ) , \label{clos4}\end{aligned}\ ] ] since @xmath133 = e\left ( r\right ) $ ] .
although this closure is not completely new ( see ref.s @xcite ) , it was never formulated in the present general form .
yet it is a very natural choice , since it yields the correct zeroth - order term in the density expansion of the dcf : @xmath134 @xmath135 @xmath136 when @xmath137 the advantage of coupling the m3 potential with the mmsa is that one finds @xmath138 ( @xmath139 ) , which is identical with the usual yukawa closure .
this mapping allows the immediate identification between the solution @xmath140 with @xmath141 , and thus in the limit @xmath71 we get @xmath142 in short , the solution given by eqs .
( @xmath49)-(@xmath143 ) is recovered , but within a new _ well - defined _ model , with a _ finite _ second virial coefficient . since @xmath144 can be computed directly from @xmath145 and eq .
( [ eosc ] ) ( in fact , sticky limit and @xmath146-integration commutate ) , it results that @xmath147 expanding this quantity in powers of @xmath146 , one finds that @xmath148 @xmath149 @xmath150 , whereas @xmath151 $ ] differs from @xmath128 .
let us insert eq .
( @xmath152 ) into the _ virial _
@xmath153 _ equation , _
@xmath154 \equiv 1+\frac{2\pi } { 3}\rho j_{v}. \label{eosv}\ ] ] the integral can be written as @xmath155 , with @xmath156 = \int_{0}^{\infty } dr\ r^{3}e^{\prime } ( r)=\int_{0}^{\infty } dr\ r^{3}f^{\prime } ( r),\]]@xmath157 \\ & = & \int_{0}^{\infty } dr\ e_{{\rm hs}}\left ( r\right ) r^{3}\gamma \left ( r\right ) \left [ -\beta \phi _ { { \rm hs}}^{\prime } ( r)-\beta \phi _ { { \rm tail}% } ^{\prime } ( r)\right ] .\end{aligned}\]]integrating @xmath158 by parts , and observing that the boundary terms vanish , one finds that @xmath159 .
on the other hand , if @xmath160 is written as @xmath161 , with @xmath162 $ ] , and using @xmath163 = e_{{\rm hs}}^{\prime } \left ( r\right ) = \delta ( r-\sigma ) $ ] , one gets@xmath164 \right\ } .
\label{v1}\ ] ] in the particular case of the m3 potential , @xmath165 assumes increasingly large negative values as @xmath9 increases , whereas @xmath166 remains bounded , in such a way that the integral @xmath167 diverges in the sticky limit ( for shortness , our detailed analysis of this point is not reported here ) .
( @xmath168 ) and these results imply that the mmsa - virial eos diverges .
however , since the second virial coefficient is finite and exact , and @xmath169 diverges whereas @xmath170 is finite , the singularity of @xmath171 is not due to the hamiltonian , but is surely caused by the mmsa closure .
finally let us explore the energy ( or free energy ) route for temperature - dependent potentials .
it can be shown @xcite that , for any potential , @xmath172 = 2\pi \rho \int_{0}^{\infty } dr\ r^{2}g(r)\ \frac{% \partial \left [ \beta \phi \left ( r\right ) \right ] } { \partial \zeta } , \label{e0}\ ] ] where @xmath173 denotes an arbitrary parameter from which @xmath174 depends upon , and @xmath175 indicates the ideal gas free energy . in particular , when @xmath176 and @xmath177 , one gets@xmath178 _ { { \rm mmsa}}=2\pi \rho \left ( -\int_{0}^{\infty } dr\ r^{2}\frac{\partial e\left ( r\right ) } { \partial \beta } + \int_{\sigma } ^{\infty } dr\ r^{2}\gamma _ { { \rm mmsa}}\left ( r\right ) \frac{\partial \left [ \beta \phi _ { { \rm tail}}\left ( r\right ) \right ] } { \partial \beta } \right ) .\]]for the m3 potential , this expression becomes @xmath179 = -12\eta \ \varepsilon _ { 0}\left [ 1+\frac{1% } { z\sigma } + \frac{\gamma _ { { \rm m3-mmsa}}\left ( \sigma ; \beta , z\right ) } { % 1+z\sigma \
\beta \varepsilon _ { 0}}+{\cal o}\left ( z^{-2}\right ) \right ] , \label{e0b}\ ] ] with the boundary condition @xmath180 ( @xmath181 ) , corresponding to the hs case , this equation can be integrated with respect to @xmath182 and in the @xmath13 limit one finds@xmath183(since sticky limit and @xmath184-integration commutate ) . from @xmath185
we then get @xmath186 it is noteworthy that these results for the helmholtz free energy and the energy - eos are van der waals - like .
the effect of the surface adhesion enters these expressions only at the level of the second virial coefficient ( which is exact ) , while all higher - order virial coefficients predicted by eq .
( @xmath187 ) coincide with the pure hs ones .
the reason for this may be traced back to the fact that the mmsa closure takes into account only the zeroth - order term in the density expansion of the dcf outside the core .
three different models of ` sticky hard spheres ' have been treated in this paper .
the first two ( shs1 and shs2 ) were already present in the literature , while the last one ( shs3 ) is new and constitutes the main contribution of the present work .
all these models describe a fluid of rigid spherical particles with infinitely strong surface adhesion , defined through an appropriate ` sticky limit ' which constitutes an essential part of the model .
the choice of the starting potential strongly influences that of the approximate closure to be used to solve the oz equation analytically . in shs1
a suitably defined square - well allows the solution within the py approximation , whereas in shs2 the starting point is a hs - yukawa potential which requires the msa approximation .
however , the most crucial point is that , while shs1 is a perfectly well - defined model , shs2 is _ not , _ since its _ exact _ second virial coefficient is divergent , as we have proven in this paper .
this singularity is mirrored by a similar behaviour appearing in both the virial and energy shs2-msa eos s , but it should be emphasized that it is a weakness of the model itself and is not due to the msa approximation .
this means that the shs2 model itself is not properly defined and must be abandoned . to replace it ,
a proposal has been put forward in the second part of the paper .
our recipe is based upon the introduction of a new potential ( m3 ) , coupled with a new closure ( mmsa ) , in such a way that the baxter - oz equations turn out to be analytically solvable , with formally the same solution @xmath15 found in the shs2-msa case .
this remarkable correspondence allows a recovery of all previous results for structure factors and compressibility eos based upon the shs2-msa solution tutschka98,gazzillo00,gazzillo02a , gazzillo02b , after a re - interpretation in terms of shs3-mmsa .
the shs3 model is the simplest correct alternative to shs2 .
it is well - defined , since the sticky limit of the _ exact _ second virial coefficient , @xmath188 , remains _ finite . _ furthermore , its starting potential , m3 , is asymptotically equivalent to the hsy one at large @xmath4 values , being different from it only in the contact region .
as regards the closure , the mmsa is correct in the zero - density limit , and has , consequently , a higher thermodynamic consistency than the msa . as a matter of fact , compressibility , virial and energy routes all
generate the same , exact , @xmath53 . on the other hand
, one can not expect from the mmsa to go further in consistency , since this closure is still rather poor , taking into account only the zeroth - order term in the density expansion of the dcf outside the core .
we have shown that discrepancies are already found at the level of third virial coefficient : @xmath189 is temperature - dependent but different from @xmath190 , @xmath191 diverges , while @xmath192 is temperature - independent and equal to its hs counterpart .
clearly , these three results could coincide within a more refined ( density - dependent ) closure .
it would be very interesting to investigate whether the shs3 model admits further analytic solutions , within more sophisticated approximations ( such as , in particular , the py approximation ) .
we plan to do such analysis in future work , including also a more detailed comparison with baxter s original model .
finally , it is worth noting that the extension of the shs3 model to mixtures can be easily carried out .
we hope that the corresponding solution can provide a simple useful tool for further studies on structural and thermodynamic properties of polydisperse colloidal fluids .
this work was partially supported by the italian infm ( istituto nazionale per la fisica della materia ) .
useful discussions with lesser blum , gerhard kahl and yurii kalyuzhnyi are gratefully acknowledged .
a special thank goes to andrs santos for a suggestion concerning the virial eos . in order to avoid any possible confusion ,
we have adopted the following terminological distinction .
a shs _ model _ is univocally defined by the specification of its hamiltonian , which is characterized by _ two _ elements : the choice of a starting potential , and the definition of a ` sticky limit ' procedure . on the other hand , a particular _ solution _
is identified by _ three _ elements : the two ingredients of the model , along with a ` closure ' for the oz equation . here
, @xmath194 is the asymmetrical dirac delta function defined by : @xmath195 if @xmath196 , and @xmath197 if @xmath198 or @xmath199 ( the yukawa tail , @xmath200 gives rise to a singularity , since @xmath201@xcite ) . | a known ` sticky - hard - sphere ' model , defined starting from a hard - sphere - yukawa potential and taking the limit of infinite amplitude and vanishing range with their product remaining constant , is shown to be ill - defined .
this is because its hamiltonian ( which we call shs2 ) leads to an _ exact _ second virial coefficient which _ diverges _ , unlike that of baxter s original model ( shs1 ) .
this deficiency has never been observed so far , since the linearization implicit in the ` mean spherical approximation ' ( msa ) , within which the model is analytically solvable , partly _ masks _ such a pathology . to overcome this drawback and retain some useful features of shs2
, we propose both a new model ( shs3 ) and a new closure ( ` modified msa ' ) , whose combination yields an analytic solution formally identical with the shs2-msa one .
this mapping allows to recover many results derived from shs2 , after a re - interpretation within a correct framework .
possible developments are finally indicated . |
the scenario where our world is associated with a brane embedded in a higher dimensional space - time with non - factorizable geometry @xcite has attracted a lot of interest with the aim of solving several open questions in modern physics . most of these models were realized as time independent field configurations .
however , there have appeared several braneworld models that assumed time - dependent metrics and fields @xcite .
here we consider the braneworld scenario with non - stationary metric coefficients recently proposed in @xcite .
the braneworld is generated by 5d standing gravitational waves coupled to a phantom - like scalar field in the bulk .
a key requirement for realizing the braneworld idea is that the various bulk fields be localized on the brane . for reasons of economy and avoidance of charge universality obstruction
@xcite one would like to have a universal gravitational trapping mechanism for all fields .
however , there are difficulties to realize such mechanism with exponentially warped space - times . in the existing ( 1 + 4)-dimensional models spin @xmath0 and spin @xmath1 fields can be localized on the brane with the decreasing warp factor @xcite , spin @xmath2 field can be localized with the increasing factor @xcite , and spin @xmath3 fields are not localized at all @xcite . for the case of ( 1 + 5)-dimensions
it was found that spin @xmath0 , spin @xmath3 and spin @xmath1 fields are localized on the brane with the decreasing warp factor and spin @xmath2 fields again are localized with the increasing factor @xcite .
there exist also 6d models with non - exponential warp factors that provide gravitational localization of all kind of bulk fields on the brane @xcite , however , these models require introduction of unnatural sources
. as it was mentioned in @xcite standing wave braneworld model can provide universal gravitational trapping of zero modes of all kind of fields in case of rapid oscillations of standing gravitational waves in the bulk .
to understand the idea we want to remind that standing electromagnetic waves , so - called optical lattices , can provide trapping of various particles by scattering and dipole forces @xcite . in @xcite localization
was demonstrated through quadruple forces as well .
it is known that the motion of test particles in the field of a gravitational wave is similar to the motion of charged particles in the field of an electromagnetic wave @xcite . thus standing gravitational waves
could also provide confinement of matter via quadruple forces . as a classical example
let us consider the equations of motion of the system of spinless particles in the quadruple approximation @xcite , @xmath4 where @xmath5 is the total momentum of the matter field and @xmath6 is the quadruple moment of the stress - energy tensor for the matter field .
the oscillating metric due to gravitational waves should induce a quadruple moment in the matter fields .
if the induced quadruple moment is out of phase with the gravitational wave the system energy increases in comparison with the resonant case and the fields / particles will feel a quadruple force , @xmath7 , which ejects them out of the high curvature region , i.e. it would localize them at the nodes . in this latter
we show that standing gravitational waves induced by the brane in the bulk really can localize scalar and tensor field zero modes on the brane even for the case of exponentially increasing warp factor .
in this section we briefly recall the 5d standing wave braneworld model proposed in @xcite .
the braneworld is generated by gravity coupled to a non - self interacting scalar phantom - like field @xcite , which depends on time and propagates in the bulk .
the action of the model has the form : @xmath8,\ ] ] where @xmath9 and @xmath10 are 5d newton and cosmological constants respectively . to avoid the well - known problems with stability , which occur with ghost fields , we can associate the ghost - like field @xmath11 with the geometrical scalar field in a 5d integrable weyl model @xcite . in the weyl model a massless scalar appears through the definition of the covariant derivative of the metric tensor , @xmath12
this is a generalization of the riemannian case where the covariant derivative of the metric is zero .
the assumption ( [ d ] ) implies that the length of a vector is altered by parallel transport .
weyl s scalar field in ( [ d ] ) may imitate a massless scalar field - either an ordinary scalar or ghost - like scalar . the gravitational action for the weyl 5d integrable model can be written as @xmath13~,\ ] ] where @xmath14 is an arbitrary constant .
for example , for @xmath15 the action ( [ grav - weyl ] ) exactly coincides with ( [ action ] ) .
thus we can start with a 5d weyl model and require that we have riemann geometry on the brane by assuming that the geometrical scalar field @xmath11 is independent of 4d spatial coordinates , @xmath16 , and vanishes on the brane .
the definition of @xmath11 via ( [ d ] ) avoids the usual instability problems of ghost fields since the geometrical fields have specific couplings with matter fields and it is known that the weyl model is stable for any value of @xmath14 .
the einstein equations for the action ( [ action ] ) read : @xmath17 the metric _ ansatz _ we use is : @xmath18 where @xmath19 is a constant , which corresponds to brane width .
the peculiarity of the model ( [ metrica ] ) is that the brane , located at @xmath20 , possesses anisotropic oscillations and send a wave into the bulk ( as in @xcite ) , i.e. the brane is warped along the spatial coordinates @xmath21 through the factor @xmath22 , which depends on time @xmath23 and the extra coordinate @xmath24 . on the other hand ,
the phantom - like scalar field @xmath25 obeys the klein - gordon equation on the background space - time given by ( [ metrica ] ) , @xmath26 where @xmath27 is the sign function and overdots and primes mean derivatives with respect to @xmath23 and @xmath24 respectively .
we further rewrite the einstein equations in the form : @xmath28 where the gravitational constant has been absorbed in the definition of the scalar field : @xmath29 it turns out that the system ( [ field - eqns1 ] ) is consistent only when the fields @xmath30 and @xmath31 are related @xcite : @xmath32 the non - zero components of the ricci tensor for the metric ( [ metrica ] ) read : @xmath39~ , \nonumber \\ r_{xx}&=&r_{yy}= e^{2a|r|+u}\left [ \frac { 1}{2 } e^{-2a|r|}\ddot u - 4a^2 - 2a\epsilon(r)u ' - 2a\delta(r ) -\frac { 1}{2 } u '' \right]~ , \nonumber \\
r_{zz } & = & e^{2a|r|-2u}\left [ - e^{-2a|r|}\ddot u - 4a^2 + 4a\epsilon(r)u ' - 2a\delta(r ) + u''\right]~ , \\
r_{rr } & = & -\frac 32 u'^2 - 4a^2 - 8a\delta(r)~ , \nonumber\\ r_{rt } & = & -\frac 32 \dot uu ' ~. \nonumber\end{aligned}\ ] ] using the fine tuning : @xmath40 the system of einstein equations ( [ field - eqnsdelta ] ) reduces to a single ordinary differential equation , @xmath41 in addition , for the brane tensions @xmath34 we have @xcite : @xmath42\right)\delta \left(r\right ) ~,\\ \lambda_z & = & \frac{e^{-2u\left(t,0\right)}}{8\pi g_5}\left(6a+[u']\right)\delta \left(r\right)~ , \nonumber \\ \lambda_r & = & 0~ , \nonumber\end{aligned}\ ] ] where @xmath43 and @xmath44 $ ] denote the value of @xmath45 and the jump of its first derivative at @xmath46 , respectively . the standing wave solution to ( [ eqntou ] ) can be constructed by implementing the _ ansatz _ : @xmath47 where @xmath48 and @xmath49 are real constants . from ( [ eqntou ] )
we get the equation for the radial function : @xmath50 the general solution to this equation has the following form : @xmath51,\ ] ] where @xmath52 are arbitrary constants and @xmath53 and @xmath54 are second - order bessel functions of the first and second kind , respectively . as pointed out in @xcite , along with the solution ( [ fsol ] ) , the ghost - like field @xmath55 must be unobservable on the brane . taking into account the relations ( [ sigma = u ] ) and ( [ separation ] ) we can accomplish this requirement by setting the boundary condition : @xmath56
since @xmath53 and @xmath54 are oscillatory functions , in the case of increasing ( decreasing ) warp factor for some fixed values of the constants @xmath57 and @xmath58 the function @xmath59 can have finite ( infinite ) number of zeros .
thus the above boundary condition can be written in the form which quantizes the oscillation frequency , @xmath49 , of the standing wave in terms of the brane width , @xmath60 , i.e. @xmath61 where @xmath62 is the @xmath63 zero of the function @xmath59 .
correspondingly , the functions @xmath55 and @xmath45 vanish at the finite ( infinite ) number of points along the extra dimension .
these points - the nodes of standing wave - can be considered as 4d space - time ` islands ' , where the matter particles are assumed to be bound .
since the equations for extra components of tensor and scalar fluctuations are similar for the simplicity let us consider localization problem only for scalar fields , which are defined by the 5d action : @xmath64 corresponding klein - gordon equation has the form : @xmath65 determinant for our _ ansatz _ ( [ metrica ] ) is equal to @xmath66 and ( [ scalfieldeqn ] ) can be written as : @xmath67\phi = e^{2a|r|}\left ( e^{4a|r| } \phi ' \right)',\end{aligned}\ ] ] where primes denote derivatives with respect to @xmath24 .
let us look for the solution in the form : @xmath68 which transforms ( [ equation1 ] ) into the system of the equations , @xmath69\psi & = & e^{2a|r|}\left ( e^{4a|r|}\psi ' \right ) ' . \nonumber\end{aligned}\ ] ] in the case @xmath70 the parameters @xmath71 and @xmath72 can be regarded as momentum components along the brane .
their exact physical meaning is not clear in general case since ( [ system ] ) are equations with variable coefficients . to solve the system ( [ system ] )
we assume that the constant @xmath57 in the solution ( [ fsol ] ) is zero and the oscillation function of gravitational waves has the form : @xmath73 where @xmath58 is a constant and @xmath54 is the second - order bessel function of the second kind .
physical solutions described by bessel functions usually contain the first kind functions , @xmath74 , because of their regularity at the origin .
however , in our case we can use even bessel function of the second kind , @xmath75 , since the argument in ( [ u ] ) , @xmath76 , is always positive and becomes zero only for the case of increasing warp factor ( @xmath77 ) at @xmath78 . to show explicitly localization by standing waves we consider the extreme case with the increasing warp factor @xmath79 .
the function ( [ u ] ) is zero at the position of the brane , @xmath20 , due to fine tuning ( [ quantize ] ) , where @xmath62 now is one of the zeros of @xmath54 . for 4d massless scalar modes on the brane we have the following dispersion relation : @xmath80 when the frequency of standing gravitational waves is much larger than frequencies associated with the energies of the particles on the brane , @xmath81 we can perform time averaging of oscillating exponents in the equation for scalar field ( [ system ] ) . using the expressions : @xmath82^m dt & = & \left \ { \begin{array } { lr } 0 & ( m = 2n + 1)\\\\ 2^{-2n}\left ( 2n \right ) ! \left ( n ! \right)^{-2 } & ( m = 2n ) \end{array}\right .
e^{bu } & = & \sum \limits_{n = 0}^{+\infty } \frac{\left ( bu \right)^n}{n!}~,\end{aligned}\ ] ] where @xmath83 is an integer , for the time averages of oscillating exponents ( [ e - br ] ) we get : @xmath84^{2n } = i_0(z)~,\ ] ] where @xmath85 is the modified bessel function of the zero order with the argument : @xmath86 .\ ] ] now separating variables , @xmath87 for the extra factor @xmath88 of the wave function we obtain : @xmath89 where @xmath90 it is more convenient to put eq .
( [ rho ] ) into the form of an analogue non - relativistic quantum mechanical problem by making the change : @xmath91 for @xmath92 we find : @xmath93 where the function , @xmath94 is the analog non - relativistic potential .
1 shows the behaviour of @xmath95 in the case of first zero of the bessel function @xmath54 , i.e. when @xmath96 ) .,scaledwidth=50.0% ] to study general behaviour of extra part of the scalar zero mode wave function we explore eq .
( [ psi ] ) in two limiting regions , far from and close to the brane .
far from the brane , @xmath78 , eq .
( [ psi ] ) has the following asymptotic form : @xmath97 with the solutions @xmath98 .
taking into account the normalization condition for the wave function and @xmath79 , we choose @xmath99 .
so , far from the brane for the extra factor of the wave function ( [ ro ] ) we obtain : @xmath100 in the second limiting region , @xmath101 , i.e. near the brane , eq .
( [ psi ] ) can be approximated as : @xmath102\psi = 0~,\ ] ] where we have introduced the constant : @xmath103 ^ 2.\ ] ] the solution of ( [ zero ] ) is : @xmath104~,\end{aligned}\ ] ] where @xmath105 is a constant , @xmath106 is the gamma function and @xmath107 and @xmath108 denote whittaker functions . imposing boundary condition at the origin , @xmath109 close to the brane extra part of scalar wave function will behave as ( see fig .
2 ) : @xmath110~.\ ] ] so the extra part of scalar field zero - mode wave function @xmath111 has maximum at @xmath46 , falls off from the brane and turns into the asymptotic form ( [ rho - infinity ] ) at the infinity . ) .,scaledwidth=50.0% ] in the action of scalar fields ( [ sphi ] ) the determinant , @xmath66 , and the metric tensor with upper indices , @xmath112 , give the total exponential factor @xmath113 , which obviously increases for @xmath77 .
this is the reason why in the original brane models @xcite the scalar field zero modes , with the constant @xmath24-depended extra part of the wave function , can be localized on the brane only in the case of decreasing warp factor ( i.e. @xmath114 ) . in our model
the extra part of wave function ( [ ro ] ) is not a constant , moreover , for @xmath79 it contains the exponentially decreasing factor @xmath115 .
in this latter we have investigated localization problem of scalar and tensor fields within the 5d standing wave braneworld proposed in @xcite . since equations for the extra part of wave functions of scalar and tensor fields are similar for simplicity we have explicitly considered in details only the case of scalar fields .
we have shown existence of normalizable zero modes on the brane even for the case of increasing warp factor .
further investigations towards the localization of spinor and vector fields in this model are in progress . n. arkani - hamed , s. dimopoulos and g. dvali , phys .
* b 429 * ( 1998 ) 263 , arxiv : hep - ph/9803315 ; + i. antoniadis , n. arkani hamed , s. dimopoulos and g. dvali , phys . lett . * b 436 * ( 1998 ) 257 , arxiv : hep - ph/9804398 .
m. gogberashvili , int . j. mod
* d 11 * ( 2002 ) 1635 , arxiv : hep - ph/9812296 ; mod .
. lett . * a 14 * ( 1999 ) 2025 , arxiv : hep - ph/9904383 ; + l. randall and r. sundrum , phys .
* 83 * ( 1999 ) 3370 , arxiv : hep - ph/9905221 ; phys .
* 83 * ( 1999 ) 4690 , arxiv : hep - th/9906064 .
m. gutperle and a. strominger , jhep * 0204 * ( 2002 ) 018 , arxiv : hep - th/0202210 ; + m. kruczenski , r.c .
myers and a.w .
peet , jhep * 0205 * ( 2002 ) 039 , arxiv : hep - th/0204144 ; + v.d .
ivashchuk and d. singleton , jhep * 0410 * ( 2004 ) 061 , arxiv : hep - th/0407224 ; + c.p .
burgess , f. quevedo , r. rabadan , g. tasinato and i. zavala , jcap * 0402 * ( 2004 ) 008 , arxiv : hep - th/0310122 .
m. gogberashvili and p. midodashvili , phys . lett .
* b 515 * ( 2001 ) 447 , arxiv : hep - ph/0005298 ; europhys . lett . * 61 * ( 2003 ) 308 , arxiv : hep - th/0111132 ; + m. gogberashvili and d. singleton , phys . lett . *
b 582 * ( 2004 ) 95 , arxiv : hep - th/0310048 ; phys .
d 69 * ( 2004 ) 026004 , arxiv : hep - th/0305241 ; + m. gogberashvili , p. midodashvili and d. singleton , jhep * 0708 * ( 2007 ) 033 , arxiv : 0706.0676 [ hep - th ] .
k. bronnikov , acta .
phys . pol .
* b 4 * ( 1973 ) 251 ; + h. ellis , j. math .
* 14 * ( 1973 ) 104 ; + r.r .
caldwell , phys .
* b 545 * ( 2002 ) 23 , arxiv : astro - ph/9908168 ; + m. pospelov , int .
phys . * a 23 * ( 2008 ) 881 , arxiv : hep - ph/0412280 ; + a. das , s. kar and s. sengupta , int .
phys . * a 24 * ( 2009 ) 4457 , arxiv : 0804.1757 [ hep - th ] ; + v. dzhunushaliev , v. folomeev and m. minamitsuji , rept .
* 73 * ( 2010 ) 066901 , arxiv : 0904.1775 [ gr - qc ] .
salim and s.l .
sauti , class .
* 13 * ( 1996 ) 353 ; + o. arias , r. cardenas and i. quiros , nucl . phys . *
b 643 * ( 2002 ) 187 , arxiv : hep - th/0202130 ; + n. barbosa - cendejas and a. herrera - aguilar , jhep * 10 * ( 2005 ) 101 , arxiv : hep - th/0511050 ; phys . rev . *
d 73 * ( 2006 ) 084022 , arxiv : hep - th/0603184 ; + n. barbosa - cendejas , a. herrera - aguilar , m.a .
reyes santos and c. schubert , phys .
rev . d*77 * ( 2008 ) 126013 , arxiv : 0709.3552 [ hep - th ] ; + j.e .
madriz aguilar and c. romero , found .
* 39 * ( 2009 ) 1205 ,
arxiv : 0809.2547 [ math - ph ] ; int . j. mod .
phys . * a 24 * ( 2009 ) 1505 , arxiv : 0906.1613 [ gr - qc ] . | we investigate scalar and tensor fields in the brane model solution for the 5d space - time with standing gravitational waves in the bulk and show that even in the case of increasing warp factor there exist normalizable zero modes localized on the brane . 0.3 cm pacs numbers : 04.50.-h , 11.25.-w , 11.27.+d 0.5 cm |
comprehensive reviews on baryon flux - tubes and hybrid baryons are available .
@xcite here developments since the last review in 2002 are summarized .
_ baryon flux - tube : _ the consensus in quenched lattice qcd is that the baryon potential at sufficient distances is best described by the sum of a coulomb , y - shaped confinement , and constant term .
@xcite this statement is independent of the lattice operator , and hence physical .
@xcite the difference with abelian projected full lattice qcd is small .
@xcite the baryon flux action density @xcite is displayed in fig .
the density clearly peaks at the three quarks and the junction
. however , the detailed contour lines , as well as the y - shape , are lattice operator dependent .
@xcite abelian projected full lattice qcd shows @xcite that cromo - electic flux flows in the same directions as the flux - tubes , with solenoidal currents circulating around them .
this supports @xcite the idea that qcd is a dual superconductor . _
hybrid baryon : _ the first quenched lattice qcd calculation of the low - lying hybrid baryon potential has been performed .
@xcite the lowest gluonic excitation energy ( fig .
[ tak ] ) is consistently greater than 1 gev .
this is several hundred mev larger than the flux - tube model prediction @xcite , which already gives the highest mass prediction for the lightest @xmath1 hybrid baryon of all the models @xcite .
if the lattice calculation is correct , this hybrid baryon will likely have a mass above 2 gev , which would mean that the hybrid baryon spectrum starts at a very high mass .
in contrast to this , qcd sum rule predictions for hybrid baryon masses remain low : most recently , the @xmath2 is predicted @xcite at @xmath3 gev .
a possible reason why the lattice calculation is high is that the effect of higher excited states has not been eliminated , although it is stated @xcite that this is not the case . _
hybrid baryon production in @xmath0 decay _ , a glue - rich process , has long been thought to be promising .
@xcite particularly , bes at bepc can look for @xmath4 .
recently , several high mass peaks have been reported @xcite : a @xmath5 peak at @xmath6 mev seen in @xmath7 , and a @xmath8 mev peak seen in @xmath9 . a recent model calculation @xcite finds , in contrast to nave expectations , that hybrids are very weakly produced relative to radially excited baryons in @xmath0 decay .
this is due to a low probability of the gluon coming from the @xmath0 to fold in to the @xmath10 hybrid wavefunction , and color factors in the hybrid baryon wave function .
in fact , it is suggested that @xmath0 decay is a filter in favor of conventional baryons .
@xcite _ hybrid baryon electro - production : _ a recent analysis of new jefferson lab data finds @xcite that the amplitude for roper resonance ( @xmath11 ) production via longitudinal photons is substantial , inconsistent with a model prediction that it should be zero .
the roper hence does not fit well as a hybrid baryon .
when a baryon decays via ozi allowed quark - antiquark pair creation to a baryon and a meson , it has not hitherto been considered what happens to the flux - tubes .
there is certainly an effect , as the pair is created from gluons .
here we report on the first preliminary calculation .
such a calculation exists for meson decay to two mesons .
@xcite the structure of the decay can be investigated by fixing the initial quarks in the baryon , and evaluating the overlap of the initial flux configuration with the final baryon and mesons , for a fixed quark - antiquark pair created from the initial flux - tube : the `` flux - tube overlap '' .
this quantity enters in models of baryon decay where it is traditionally assumed to be a constant .
this is equivalent to taking the y - shaped flux structure to have no effect on decay .
in such a picture the main effect of confinement on the decay is that it specifies the decay operator for quark - antiquark pair creation , e.g. the @xmath12 decay operator @xcite .
this operator is typically taken to act locally with no knowledge of the global y - shaped structure . here
the concern is not with the local character of decay ( the decay operator ) , but with imposing constraints from y - shaped confinement ( the flux - tube overlap ) .
the baryon decay flux overlap @xmath13 measures the overlap of the flux - tube wavefunction of initial baryon a , with those of final baryon b and meson c , assuming that pair creation takes place at position @xmath14 .
the overlap is evaluated analytically in the flux - tube model of isgur and paton @xcite as was done for the meson case @xcite . in this model the flux is represented by oscillating beads , and for the baryon the flux - tubes meet at a junction .
the coordinates of the quarks and junction are displayed in fig .
[ def](b ) .
the position of the junction of the initial baryon , measured from its equilibrium position , is denoted by @xmath15 . in eq .
[ defeq ] the position of bead @xmath16 associated with quark @xmath17 is denoted by @xmath18 . for the calculation shown , the constituent quark masses we taken to be 330 mev , the string tension that of mesons ( @xmath19 gev@xmath20 gev / fm ) , @xcite and all the bead masses equal .
bead masses are fixed by the condition that their sum equals the potential energy in the flux - tubes when the system is in equilibrium .
the flux - tube overlap @xmath21 depends on the six variables @xmath14 and @xmath22 , shown in fig .
[ def](a ) .
there is also dependence on the number of beads associated with each flux - tube in the initial baryon ( @xmath23 ) . if pair creation can take place on more than one bead , @xmath21 is the sum of the possible overlaps .
( hence the sum over the number of beads on the final meson @xmath24 in eq .
[ defeq ] . ) without loss of generality pair creation associated with the flux - tube of the third quark is studied .
the components of @xmath14 perpendicular to the vector @xmath25 are @xmath26 ( in the three - quark plane ) and @xmath27 ( perpendicular to the three - quark plane ) .
the component of @xmath14 along @xmath25 is @xmath28 .
when any variables are fixed in the figures , the following choices are made , unless indicated otherwise : @xmath29 , @xmath30 gev@xmath31 fm , @xmath32 gev@xmath33 fm and @xmath34 . , the dependence of @xmath21 on the pair creation position @xmath14 is as follows .
the dependence on @xmath26 and @xmath27 is shown in fig .
[ gauss ] . a clear gaussian fall - off with flattened top with a diameter @xmath35 fm think can be seen .
this follows closely what would be expected for a thick flux - tube which falls off away from @xmath25 as predicted by the flux - tube model ( see fig . 2 of ref .
the gaussian fall - off with flattened top continues to be observed as @xmath28 is varied .
[ bump](a ) shows the dependence on @xmath28 .
the asymmetric shape is because pair creation can take place beyond the position of the third quark ( @xmath36 gev@xmath37 ) .
vanishing pair creation at @xmath38 * 0 * is an artifact of the small - oscillations approximation @xcite made in this calculation , and is _ not _ a prediction of the flux - tube model .
the dependence of @xmath21 on quark position @xmath22 variations is shown in figs .
[ bump](b ) and [ ldep ] .
the extremes of the variation are consistently ( a ) smallest @xmath21 when @xmath39 ( red line @xmath40 ) , and ( b ) largest @xmath21 when @xmath41 ( blue line @xmath42 ) .
these extremes are easily understood intuitively : when @xmath43 is small / large , there is less / more opportunity for pair creation associated with the third quark .
[ bump](b ) shows that pair creation becomes more confined to the region inside the third quark ( @xmath44 ) as @xmath43 increases .
there is also a dependence on the number of beads @xmath23 .
calculations were also performed for @xmath45 .
the number of beads determine the inter - bead spacing , a regularization parameter , which is unphysical .
the qualitative features of the dependence of @xmath21 on @xmath14 and @xmath22 discussed above are independent of @xmath23 , and hence physical .
they are also without exception what one would have expected before the calculation was performed .
the overall magnitude of decay is strongly dependent on @xmath23 , and hence unphysical .
this is as expected , since the magnitude of decay is normalized to phenomenology in pair creation models . , ,
_ baryon : _ at sufficient distances the potential is y - shaped with a clear junction .
the decay flux - tube overlap has been evaluated in the flux - tube model , and follows nave expectations .
y - shaped baryon confinement can henceforth be incorporated in models of decay .
_ hybrid baryon : _ the only known way to study them rigorously is via excited adiabatic potentials .
@xcite the low lying state in models is @xmath1 with mass 1.5 1.8 gev .
@xcite if recent lattice qcd calculations are correct , this mass is most likely above 2 gev .
a recent model calculation shows that glue - rich @xmath0 decay very weakly produces hybrid baryons .
the roper resonance does not appear to be a hybrid baryon from a recent analysis of electro - production via longitudinal photons .
collaboration with n.f .
black ( university of tennessee , knoxville ) on the baryon decay flux - tube overlap , and hospitality at menu2004 , are acknowledged .
takahashi , h. matsufuru , y. nemoto and h. suganuma , _ phys .
lett . _ * 86 * , 18 ( 2001 ) ; _ phys .
rev . _ * 65 * , 114509 ( 2002 ) ; c. alexandrou , p. de forcrand and o. jahn , _ nucl . phys .
( proc . suppl . )
_ * 119 * , 667 ( 2003 ) . | the status of conventional baryon flux - tubes and hybrid baryons is reviewed .
recent surprises are that a model prediction indicates that hybrid baryons are very weakly produced in glue - rich @xmath0 decays , and an analysis of electro - production data concludes that the roper resonance is not a hybrid baryon . the baryon decay flux - tube overlap has been calculated in the flux - tube model , and is discussed here . the behavior of the overlap follows nave expectations . |
electrokinetics(ek ) in fluid is a classical subject defined as the study on the interaction of ions with the solid interface , and the dynamics of ions in fluid under external fields @xcite .
the roots of the ek study are very broad and can be stemmed from various disciplines for decades , such as chemistry , biology , physics , and material sciences . with the development of nano technologies , the structure scale used to confine liquid can be narrowed from micro to nano .
the increase of the surface - to - volume ratio in low dimensional structures leads to new ek behaviors @xcite .
those new ek phenomena can be applied to molecular manipulation , drug mixing , and electro - osmosis , etc , which stimulates the ek study experiencing a strong revival @xcite .
+ it is well known that the ions at the liquid - solid interface attract counter ions to form electric double layers(edl ) in liquid @xcite . as an example of water at @xmath3 ,
silanol groups at the silica - water interface defined as the stern layer attract @xmath4 ions staying in the neighborhood of the interface by the electrostatic interaction .
the edl is known to play a key role in the ek behaviors . in order to study the charge distribution in the edl ,
a conventional method is to solve the poisson - boltzmann(pb ) equation @xcite . to solve the pb equation ,
two main types of boundary conditions are adopted .
one type boundary condition is called the gouy - chapmann(gc ) model , in which a surface charge density is assumed at the interface .
the other model is to define a @xmath0 potential as the boundary condition at the slip plane , where the counter ion starts its density extending into the fluid domain .
the @xmath0 potential is an important physical parameter , which reflects the surface charge density and the strength of the ek behaviors .
the @xmath0 potential can be measured indirectly by three main techniques .
one is to determine the @xmath0 potential by measuring electroosmotic mobility combined with the expression of smoluchowski velocity @xcite .
the second is by measuring the streaming current or streaming potential @xcite .
the last is by measuring electrophoresis behaviors of colloidal particles in liquid @xcite .
+ in the solving of pb equation , @xmath0 potential as the boundary condition is considered to be a constant .
similarly in the gc model , the surface charge density of the silanol groups is also considered to be pre - existing constant .
the results based on either of the two models are contradictory to the experimental observations @xcite . in aqueous electrolyte ,
the @xmath0 potential and the surface charge density as well are functional of the densities of acid , alkali and salt added into the liquid . for convenience , minus of logarithmic density of @xmath4 ions ( salt ) is denoted by ph ( pc ) throughout this paper . at a certain ph
, the @xmath0 potential can be measured to be zero where the ph takes its name of iso - electric point(iep ) . even though there still has no direct measurements on the surface charge density , the zero @xmath0 potential at iep
is understood to be attributed to the neutralization of the surface charge density by the acid .
the iep has been observed experimentally for a long time .
but it is still lack of a theoretical tool to reveal the occurrence of charge neutralization at iep .
one effective method with a name of charge regulation model(crm ) has been proposed to study the edl by fitting the experimental data @xcite .
however , the crm has not been used to account for the charge neutralization at iep .
+ recently , a surface potential trap model has been proposed to provide a new perspective on the formation of edl in liquid @xcite .
this model is established in the framework of charge conservation . the surface charge density or the @xmath0 potential
is not fixed as the boundary conditions in the model .
they can vary as the ph or pc changes in the liquid .
the model has also taken into account the dissociation of water molecules , which has been supported by the reactive molecular dynamics simulation @xcite . in this work
, we will use the potential trap model to reveal the mechanism of the charge neutralization at iep .
a cylindrical channel is considered in this work with a radius of @xmath5 and an infinite length . in this study ,
the channel surface is silica and the liquid filled in the channel is water .
the acid and alkali are used to adjust the ph and salt is added to change the pc . in such system ,
@xmath4 and @xmath6 ions can be not only from the acid and alkali , but also from the dissociation of water molecules .
the reactive molecular dynamics simulation shows that on a freshly cut silica surface there exist dangling bonds , si and si - o .
one neutral water molecule can dissociate into an @xmath7 and an @xmath4 when contacting the dangling bonds @xcite .
the @xmath7 and @xmath4 ions combine with si and si - o to form two silanol groups , which are unstable in an aqueous environment .
silanol groups at the silica surface contacting water are dissociated , leaving @xmath8 ions at the solid surface and @xmath4 in liquid .
we skip over the detail of the local chemical reaction at the interface , and propose a surface potential trap to simulate the dangling bonds at the interface physically .
the trap height is influenced by the chemical environment of the liquid , such as pc and ph .
however , this work will not be focused on the relation between the trap height and the chemical environment , but rather on the mechanism of charge neutralization at iep .
thus , we will fix the trap height as a constant at each pc .
it should be noted that the net charges of @xmath4 and @xmath6 can be nonzero at @xmath9 , which can be neutralized by the buffer ions . as an example , if @xmath10 is used to decrease the ph of water , then @xmath11 is the buffer ion .
and the @xmath12 is the buffer ion if @xmath13 is used as the alkali . for convenience ,
the ion densities of @xmath4 and @xmath14 are denoted as @xmath15 and @xmath16 respectively . in this work
, we do nt distinguish the difference between the buffer ions and the salt ions for simplicity , even though they may have different atomic structures and ionic valencies .
the ion densities of the negative and positive ions in the category including the salt and buffer ions are denoted as @xmath17 and @xmath18 respectively .
compared to the @xmath4 and @xmath7 , the salt and buffer ions are charge conserved , and can not be produced or eliminated as @xmath4 and @xmath7 can do by dissociation or combination in water .
+ the poisson equation in such system reads : @xmath19.\ ] ] with @xmath20 the dielectric constant in liquid , @xmath21 the elementary charge of electron and @xmath22 the potential in the system . @xmath23 and @xmath24 are the ionic valencies for the negative ion and positive ion of salt respectively . regarding the dissociation of water molecules , the ion densities of @xmath4 and @xmath7 follow the boltzmann distribution @xmath25 with the chemical potential @xmath26 introduced@xcite .
@xmath27 is the boltzmann constant and @xmath28 refers to the room temperature in this work . here
, the potential trap @xmath29 has been implemented in the distribution , and @xmath30 is equal to @xmath31 denoting the average ion density in the bulk neutral limit of water at @xmath3 .
the parameters of @xmath32 and @xmath33 are used in the equation to exhibit the bulk limit ion density of @xmath4 and @xmath6 renormalized by the @xmath30 in the case of @xmath9 .
the relevant mass action law for the deprotonation reaction reveals @xmath34 the potential trap takes the following function @xmath35 here , @xmath36 is the potential trap width , approximating the length of dangling bond @xmath37 .
@xmath38 is the potential trap height . in this work ,
we neglect the precise relation between the @xmath38 and pc or ph , and we use @xmath39 at @xmath40 .
since the salt and buffer ions are charge conserved , the @xmath18 and @xmath17 should be governed by @xmath41 in the cylindrical coordinator @xcite . here , @xmath42 is the salt concentration , having the value of @xmath43 .
the @xmath44 is the potential trap at the liquid - solid interface seen by the salt ions . @xmath45 and @xmath46 are introduced for the positive and negative buffer ion densities respectively .
the buffer ion densities should neutralize the net charges of @xmath4 and @xmath6 , leading to explicit expressions as @xmath47 for @xmath48 , and @xmath49 for @xmath50 . in this study
, we use monovalent salt with @xmath51 , and set @xmath52 for simplicity .
+ owing to the charge neutralization , the integration of the right hand side of the eq.([poisson ] ) over the whole computation domain should be zero , leading to the chemical potential expressed as @xmath53\ ] ] for both @xmath54 and @xmath55 cases .
here , we introduce two integral parameters for convenience @xmath56 principally , the charge densities and potential distribution can be solved from the above equations . in the solution
, the boundary condition of @xmath57 is applied at the interface . after obtaining the @xmath22 ,
the @xmath0 potential then can be defined generally by @xmath58 + it has been noted that the potential trap can cause the dissociation of water into ions @xcite .
the mount of @xmath4 ( @xmath6 ) ions contributed from the water dissociation can be obtained from the subtraction of the whole @xmath4 ( @xmath6 ) ion number in the system by the integral of @xmath59(@xmath60 ) over the whole domain . in order to show the dissociation efficiency of molecules ,
surface charge density related to the dissociation has been defined and noted as @xmath61 and @xmath62 for the @xmath4 and @xmath6 ions respectively . due to the charge conservation , @xmath61 and @xmath62
can be solved from the following equations @xmath63 to get @xmath64.\ ] ] the equality of @xmath65 is consistent to the dissociation process of water molecules in which the dissociated charges of @xmath4 and @xmath6 ions should be equal to each other . in the following
, we will use one parameter @xmath66 to replace the two denotations of @xmath61 and @xmath62 since @xmath65 .
+ the total surface charge density of @xmath6 trapped at the interface is denoted by @xmath67 , which represents the integral of @xmath16 in the potential trap @xmath68 it should be noted that the salt ions also can enter into the potential trap at large pc .
however , such mount of salt ions in the trap make little influence on the @xmath0 potential and the obtained results .
thus , we consider the stern layer comprising of only @xmath6 and exclude the salt ions out of the trap .
another important surface charge density denoted as @xmath69 is related to the @xmath6 ions density in bulk limit .
suppose all the @xmath6 ions in the bulk limit liquid are trapped at the interface , then @xmath69 is defined as @xmath70
the cylindrical channel used for the calculation is with @xmath71 to avoid the overlap of debye layer . in the calculation , the trap height is @xmath39 for @xmath72 .
results presented in fig .
1 show the absolute values of @xmath0 potentials decrease with the decrease of ph . at @xmath73 ,
the @xmath0 potentials go to zero , which is consistent to the experimental measurement of iep @xcite . with the increase of the ph , the absolute values also increase . after experiencing a nearly plateau ,
the @xmath0 potentials then start jumping to large absolute values at @xmath74 .
such @xmath0 potential behaviors in the figure have captured the main feature of the relation between the @xmath0 potentials and ph measured by experiments , except the latter shows an obvious linearity @xcite .
the deviation of the calculated results away from the linearity is attributed to our neglect of the precise relation between the trap height and the local chemical environment at various ph and pc values .
the study on the precise relation is not in the scope of this work , since we only concern about the mechanism of the charge neutralization at iep .
+ potential as a function of ph at pc=1,2,3 . in the calculation ,
@xmath75 and @xmath76 are used.,width=384,height=288 ] in order to reveal the origin of the results in fig .
1 , the surface charge densities at @xmath77 have been calculated in fig .
@xmath66 representing the mount of @xmath6 contributed from the water dissociation reaches its maximum at @xmath3 , and decreases when @xmath78 is away from 7 , no matter if the ph increases or decreases .
the decrease of @xmath66 at @xmath9 is due to the mass action law . at iep
, @xmath66 is zero within the calculation error , meaning that enough high @xmath4 density at iep suppresses the produce of @xmath4 from the water dissociation .
@xmath69 has been referred to the @xmath6 ion density in bulk limit trapped totally at the interface .
logarithmic value of @xmath69 is observed to be linearly increasing with the increase of ph due to the increase of @xmath6 ion density in bulk , and can be negligible when @xmath79 .
the physical properties of the edl basically is determined by the @xmath67 . in the figure
, it shows that the curve of @xmath67 overlaps that of @xmath66 when @xmath80 and meets the @xmath69 curve when @xmath2 .
that means there exist two sources contributing to the charges in the stern layer . at @xmath81 ,
the charges in the stern layer are mainly originated from the dissociation of water molecules .
and at @xmath2 , the charges in the stern layer are mainly from the bulk liquid .
since the @xmath66 is zero at @xmath73 due to the mass action law , no charges can be trapped in the stern layer , leading to the zero @xmath0 potential at iep .
it also indicates in the fig .
1 that the @xmath0 potential jumps to a large absolute value at @xmath74 due to the existence of enough @xmath6 ions in bulk liquid .
what is more , it should be noted that the @xmath67 and @xmath0 potential in the case of @xmath82 can not go to infinity , and should be limited by two constraints .
one constraint is due to the limited capacity of potential trap itself .
the other is due to the limited lattices at the silica surface .
the latter can also be understood as the steric effect .
if the limitations are considered , the @xmath67 and @xmath0 potentials will reach a plateau at @xmath82 .
the detail calculation of the @xmath67 and @xmath0 potential at a high ph is not the goal of this paper .
+ and @xmath76 are used .
( a ) surface charge densities of @xmath66 , @xmath69 , and @xmath67 are presented for @xmath77 .
( b ) @xmath66 is exhibited for pc=1,2,3 and the @xmath69 has also been presented for reference.,title="fig:",width=384,height=288 ] + and @xmath76 are used .
( a ) surface charge densities of @xmath66 , @xmath69 , and @xmath67 are presented for @xmath77 .
( b ) @xmath66 is exhibited for pc=1,2,3 and the @xmath69 has also been presented for reference.,title="fig:",width=384,height=288 ] in the gc model , the occurrence of iep is understood as the neutralization of the surface charges by @xmath4 ions from acid . however , in the gc model , the source for the pre - existing surface charges has not been clarified . even if the total @xmath6 ions in the bulk is used as the source to be trapped ,
the surface charge density is not large enough to give an experimentally observable @xmath0 potential , which has been confirmed by the @xmath69 curve in the fig .
2(a ) . instead
, the potential trap model can give the answer that the source for the surface charge at the ph close to iep is from the dissociation of water molecules , and at the iep the dissociation process is suppressed due to the mass action law .
+ , ( a ) by varying @xmath38 with @xmath71 , ( b ) by varying r with @xmath39.,title="fig:",width=384,height=288 ] + , ( a ) by varying @xmath38 with @xmath71 , ( b ) by varying r with @xmath39.,title="fig:",width=384,height=288 ] under the same condition of @xmath39 , the surface charge densities at various pc have been presented in fig.2(b ) . in the figure , the dissociation efficiency of water molecules reflected by @xmath66 decreases with the increase of pc .
that means salt concentration into the water can help to activate the water dissociation , but decrease the @xmath0 potential ( fig.1 ) .
such phenomena can be understood as the following .
suppose a fixed charge density at the interface , the electric field magnitude at the slip plane can be obtained by using the gaussian law .
such electric field would penetrate the debye layer to the cylinder center , resulting in a potential drop .
since the potential has been fixed to be zero at the interface as the boundary condition , the potential at the cylinder center denoted as @xmath83 should be positive with the consideration that the debye layer comprises of positive charges .
the @xmath83 is the exact value equal to the absolute value of the @xmath0 potential .
it is well known that the add of salt into water can decrease the thickness of debye layer . with the increase of the salt concentration in water ,
the decrease of the debye layer will decrease the @xmath83 if the electric field is fixed at the slip plane as a constant before entering into the debye layer .
then , according to the boltzmann distribution , the decrease of positive @xmath83 will attract more positive charges in the debye layer , which is realized by the increase of the surface negative charge density in the stern layer .
thus , the decrease of pc can increase @xmath66 , but the whole effect of the thickness decreasing of debye layer results in the decrease of @xmath0 potential .
+ it has also been found that @xmath66 is strongly dependent on the trap height @xmath38 , shown in fig.3(a ) .
a larger @xmath38 can provide more energy for the dissociation of water molecules , leading to a larger @xmath66 .
the relation between the @xmath66 and the cylinder radius @xmath5 has also been presented in fig .
3(b ) . with the increase of @xmath5 ,
more @xmath6 ions in the bulk are provided as the source for the surface charge density , shifting the @xmath69 curve to a lower ph . and
more neutral water molecules are provided for the dissociation with a larger @xmath5 , which can shift the @xmath66 curve upward to be larger .
the relation between the @xmath0 potential and @xmath38 or @xmath5 can be found elsewhere @xcite .
the potential trap model has been used to study the mechanism of the surface charge neutralization at iep .
it is found that at iep the decrease of the charge density at the interface is due to the suppress of the water dissociation , instead of the neutralization by acid .
the increase of the salt concentration in the water can increase the charge density in the stern layer , but decrease the @xmath0 potential , which is consistent to the experimental results .
9 r. j. hunter , _
@xmath0 potential in colloid science : principles and applications _ , ( academic press , london , 1981 ) .
l. journiaux and j. p. pozzi , _ laboratory measurements anomalous 0.10.5 hz streaming potential under geochemical changes : implications for electrotelluric precursors to earthquakes _ , j. geophys .
res . 102,15335 ( 1997 ) .
p. reppert , f. d. morgan , d. p. lesmes , and l. jouniaux , _ frequency dependent streaming potentials _ , j. colloid interface sci .
234 , 194 ( 2001 ) .
p. m. reppert and f. d. morgan , _ frequence dependent electroosmosis _ , j. colloid interface sci .
254 , 372(2002 ) .
reto b. schoch , jongyoon han , philippe renaud , _ transport phenomena in nanofluidics _ , rev .
mod . phys .
80 , 839(2008 ) .
s. yeh , m. seul , and b. shraiman , _ assembly of ordered colloidal aggregrates by electric - field - induced fluid flow _ , nature ( london ) 386 , 57 ( 1997 ) .
m. trau , d. a. saville , and i. a. aksay , _ assembly of colloidal crystals at electrode interfaces _ , langmuir 13,6375 ( 1997 ) .
c. faure , n. decoster , and f. argoul , _ ac field induced two - dimensional aggregation of multilamellar vesicles _
j. b 5 , 87 ( 1998 ) .
a. ramos , h. morgan , n. g. green , and a. castellanos , _ ac electrokinetics : a review of forces in microelectrode structures _
, j. phys .
d 31 , 2338 ( 1998 ) .
n. g. green , a. ramos , a. gonzlez , h. morgan , and a. castellanos , _ fluid flow induced by nonuniform ac electric fields in electrolytes on microelectrodes .
i.experimental measurements _ , phys .
e 61 , 4011(2000 ) .
a. gonzlez , a. ramos , n. g. green , a. castellanos , and h. morgan , _ fluid flow induced by nonuniform ac electric fields in electrolytes on microelectrodes .
ii . a linear double - layer analysis _ , phys . rev .
e 61 , 4019(2000 ) .
a. ajdari , _ pumping liquids using asymmetric electrode arrays _ , phys .
e 61 , r45 ( 2000 ) .
n. g. green , a. ramos , and h. morgan , _ ac electrokinetics : a survey of sub - micrometre particle dynamics _ , j. phys .
d 33 , 632 ( 2000 ) .
a. b. d. brown , c. g. smith , and a. r. rennie , _ pumping of water with an ac electric field applied to asymmetric pairs of microelectrodes
e 63 , 016305(2000 ) . v. studer , a. ppin , y. chen , and a. ajdari , _ fabrication of microfluidic devices for ac electrokinetic fluid pumping _ , microelectron .
61 - 62 , 915 ( 2002 ) .
n. g. green , a. ramos , a. gonzlez , h. morgan , and a. castellanos , _ fluid flow induced by nonuniform ac electric fields in electrolytes on microelectrodes .
iii.observation of streamlines and numerical simulation _ ,
phys . rev .
e 66 , 026305 ( 2002 ) .
f. nadal , f. argoul , p. kestener , b. pouligny , c. ybert , and a. ajdari , _ electrically induced flows in the vicinity of a dielectric stripe on a conducting plane _ , eur .
j. e 9 , 387 ( 2002 ) .
f. nadal , f. argoul , p. hanusse , b. pouligny , and a. ajdari , _ electrically induced interactions between colloidal particles in the vicinity of a conducting plane _ , phys .
e 65 , 061409 ( 2002 ) . c. marquet , a. buguin , l. talini , and p. silberzan , _ rectified motion of colloids in asymmetrically structured channels _ , phys
88 , 168301 ( 2002 ) .
w. d. ristenpart , i. a. aksay , and d. a. saville , _ electrically guided assembly of planar superlattices in binary colloidal suspensions _
90 , 128303(2003 ) .
m. mpholo , c. g. smith , and a. b. d. brown , _ low voltage plug flow pumping using anisotropic electrode arrays _ ,
actuators b 92 , 262 ( 2003 ) .
a. ramos , a. gonzlez , a. castellanos , n. g. green , and h. morgan , _ pumping of liquids with ac voltages applied to asymmetric pairs of microelectrodes _
e 67 , 056302 ( 2003 ) .
j. w. perram , r. j. hunter , and h. j. l. wright , _ charge and potential at the oxide / solution interface _ , chem .
phys . lett .
23 , 265(1973 ) .
s. h. behrens , and m. borkovec , _ exact poisson - boltzmann solution for the interaction of dissimilar charge - regulating surfaces_,phys .
e 60 , 7040(1999 ) .
j. a. davis , r. o. james , and j. o. leckie , _ surface ionization and complexation at the oxide / water interface : i. computation of electrical double layer properties in simple electrolytes _ , j. colloid interface sci .
63 , 480(1978 ) .
j. sonnefeld , a. gobel , and w. vogelsberger , _ surface charge density on spherical silica particles in aqueous alkali chloride solutions .
part i. experimental results _ , colloid polym .
273 , 926(1995 ) .
j. israelachvili , _ intermolecular and surface forces _ , 2nd .
( academic press , london , 1992 ) .
k. bhatt , o. velev , and s. grego , _ an ac electrokinetic technique for collection and concentration of particles and cells on patterned electrodes _ ,
langmuir 21 , 6603(2005 ) .
p. s. dittrich , k. tachikawa , and a. manz , _ micro total analysis systems . latest advancements and trends_,anal .
78 , 3887(2006 ) .
s. ghosal , _ fluid mechanics of electroosmotic flow and its effect on band broadening in capillary electrophoresis _ , electrophoresis 25 , 214 ( 2004 )
. j. r. macdonald , _ impedance spectroscopy : old problems and new developments _
, electrochim .
acta 35 , 1483(1990 ) .
l. a. geddes , _ historical evolution of circuit models for the electrode - electrolyte interface _ , ann .
25 , 1 ( 1997 ) .
m. z. bazant , k. thornton , and a. ajdari , _ diffuse - charge dynamics in electrochemical systems _ , phys .
e 70,021506 ( 2004 ) .
z. xu , j. y. miao , n. wang , w. wen , and p. sheng , _ maximum efficiency of the electro - osmotic pump _ , phys .
e 83 , 066303 ( 2011 ) .
z. xu , j. y. miao , n. wang , w. wen , and p. sheng , _ digital flow control of electroosmotic pump : onsager coefficients and interfacial parameters determination _ , solid state commun .
151 , 440 ( 2011 ) .
r. hidalgo - alvarez , _ on the conversion of experimental electrokinetic data into double layer characteristics in solid - liquid _ , adv .
34 , 217(1991 ) .
brian j. kirby , ernest f. hasselbrink jr , _ zeta potential of microfluidic substrates : 1 .
theory , experimental techniques , and effects on separations _
, electrophoresis 25 , 187(2004 ) .
sven h. behrens and david g. grier , _ the charge of glass and silica surfaces _ , journal of chemical physics , 115,6716(2001 ) .
li wan , shixin xu , maijia liao , chun liu , and ping sheng , _ self - consistent approach to global charge neutrality in electrokinetics : a surface potential trap model _
x 4 , 011042 ( 2014 ) .
j. c. fogarty , h. m. aktulga , a. y. grama , a. c. t. vanduin , and s. a. pandit , _ a reactive molecular dynamics simulation of the silica - water interface _ , j. chem .
phys . , 132 , 174704 ( 2010 ) .
lee , h. lee , y. k. hyon , t .- c .
lin , and c. liu , _ new poisson boltzmann type equations : one - dimensional solutions _ , nonlinearity 24 , 431 ( 2011 ) . | iso - electric point(iep ) is the ph , at which the @xmath0 potential is measured to be zero .
the occurrence of iep has been understood due to the neutralization of surface charge density ( scd ) at the solid - liquid interface . in this work
, we use the potential trap model to study the sources of the surface charge density at verious pc and ph , by taking the water - silica system as an example .
it is revealed that in the case of @xmath1 , the scd is mainly originated from the dissociation of water molecules . and the bulk ions trapped at the interface can dominate the scd when @xmath2 . due to the mass action law , the dissociation of water molecules is suppressed at the ph close to iep , leading to a zero surface charge density . in this way , zero @xmath0 potential is obtained at the iep .
it has also been obtained that the increase of the salt concentration in the water can decrease the @xmath0 potential , but increase the surface charge density .
pacs numbers : : 47.57.jd , 47.56.+r , 68.08.-p , 66.10.ed , 73.25.+i |
in the following we describe the simulation methods employed as well as various technicalities pertaining to the calculation of some quantities
. we also comment on the determination and quantitative accuracy of the temperature at which the domain wall - confined ferroelectric transition takes place , and on previous first - principles literature on the dws in pbtio@xmath0 .
for most simulations we used the first - principles model for pto described in ref . , where it is labeled `` @xmath3 '' .
this model can be viewed as a taylor series of the energy , around the ideal cubic perovskite structure , as a function of all possible atomic distortions and strains .
the series was truncated at 4th order and only pairwise interaction terms were included .
the potential parameters were computed by using the local density approximation ( lda ) to density functional theory ( dft ) . to compensate for lda s well - known overbinding problem
, we simulate the model under the action of a tensile hydrostatic pressure of 14.9 gpa .
we ran some direct first - principles simulations to verify the model predictions at low temperatures .
for that task , we used the pbesol@xcite approach to dft as implemented in the vasp package.@xcite note that the pbesol functional has been shown to render accurate results for the equilibrium structures of several fe perovskite oxides@xcite including pto ( which we checked ourselves ) .
we used the `` projector augmented wave '' method to represent the ionic cores,@xcite solving for the following electrons : ti s 3@xmath51 , 3@xmath52 , and 4@xmath53 ; pb s 5@xmath52 , 6@xmath53 , and 6@xmath51 ; and o s 2@xmath53 and 2@xmath51 .
wave functions were represented in a plane - wave basis truncated at 500 ev , and a 6@xmath66@xmath61 @xmath54-point grid was used for integrations within the brillouin zone corresponding to a 1@xmath61@xmath620 supercell . to better compare our model - potential and first - principles results , we conducted pbesol structural relaxations in both an unconstrained way [ `` pbesol ( i ) '' results in fig . 1 of our manuscript ] and by imposing lattice vectors obtained from model - potential calculations [ `` pbesol ( ii ) '' ] . for the sake of comparison
, we also ran some simulations at the lda level , all other technicalities being identical to those of our pbesol calculations .
we computed the electric dipoles caused by the displacement of individual atoms within a linear approximation , using the corresponding dynamical charge tensors as computed for the cubic phase . to compute ti - centered local polarizations associated to specific cells , as those shown in fig . 1
, we proceeded in the following way : we summed up the dipoles associated to the central titanium and to its nearest - neighboring atoms ( i.e. , 6 oxygens and 8 leads ) , weighting the contribution of the neighbors appropriately ( e.g. , since an oxygen atom is shared by two ti atoms , it contributes one half of its associated dipole to the polarization centered at each of its two neighboring titaniums ) .
we proceeded analogously to compute pb - centered polarizations .
this is a rather standard scheme@xcite related with the local - mode construction in classic first - principles models of ferroelectrics.@xcite finally , we used a standard monte carlo ( mc ) scheme to account for the effect of temperature in our system , working with a periodically repeated box of 12@xmath612@xmath620 pto unit cells .
at all temperatures we initialized our mc simulations from the multi - domain ground state configuration described in fig . 1
. we typically ran 25,000 mc sweeps for thermalization , followed by at least 25,000 additional sweeps for computing thermal averages . in the region of the dw - confined phase transition , marked in red in fig . 3 of our manuscript , we did production runs of up to 150,000 mc sweeps , as needed to obtain acceptable statistics for @xmath55 . note that , because of the quasi - two - dimensional and strongly - fluctuating nature of the polar order at the walls , this was an especially challenging task ; in particular , as discussed below , our ability to get quantitatively accurate results in the critical region is limited by the size of the considered supercell , which is the largest we can afford with our current simulation tools . we also ran a simulation of the fe phase transition of pto , always in a mono - domain configuration , using a 12@xmath612@xmath612 simulation box periodically repeated . in this case
, we always started the simulations from a perfectly cubic structure ; then we ran 10,000 mc sweeps for thermalization , followed by 50,000 sweeps for computing averages , which was enough to obtain well converged results at all temperatures .
we checked that the temperature dependent mono - domain polarization thus computed is essentially identical to the local polarization obtained at the middle of the domains in our multi - domain simulations .
whenever we needed to run a structural relaxation with our model potential , we performed mc simulated annealings in which we reduced the temperature down to essentially 0 k , allowing the simulation to evolve until the system would get _ frozen _ at an energy minimum . in some cases we started the annealings at very low temperatures , which allowed us to investigate local ( meta - stable ) minima of the energy , as the lack of thermal activation prevents the simulated system from evolving to its most stable configuration .
this was important , for example , to compute the hysteresis loops of fig .
2(a ) of our manuscript , as we needed to follow the evolution of the polar states , as a function of the magnitude of an opposing electric field , up to their meta - stability limit .
beyond that limit , the local energy minimum is lost and the relaxation ( i.e. , the simulated annealing ) naturally evolves towards the polar state favored by the applied field . using the same strategy , we were also able to quantify the energy gap between the states with parallel and anti - parallel dw polarizations , and thus determine that the dw
dw interaction is negligible for the separating distance ( i.e. , 9 unit cells or about 35 ) corresponding to our simulation box . ,
the absolute value of the @xmath13 component of the dw polarization . from the inflection point of this curve , obtained numerically , we can estimate @xmath56 k. ]
as already mentioned in the manuscript , at temperatures around the dw - confined transition our mc results are affected by the fact that our simulation supercell is finite in size .
more specifically , we simulate dws that are composed of 12@xmath612 cells periodically repeated in the dw plane .
while we have checked that a lateral size of 12 cells is sufficient to describe the bulk fe transition of pto with good accuracy , our results suggest that we may not be converged size - wise as regards the dw - confined transformation ( see fig . 3 of our manuscript ) .
whenever one deals with low - dimensional orders , large amplitude fluctuations are more likely to occur , which complicates convergence with size .
unfortunately , at present our software does not allow us to further increase the size of the simulation box , as the calculations become too heavy computationally .
nevertheless , an analysis of the results obtained for 12@xmath612 dws allows us to bracket @xmath24 quite precisely .
first of all , let us stress that the uncertainty caused by these size effects is of little practical importance .
the affected temperature range is relatively small , approximately from 320 k to 350 k. hence , from the point of view of comparing our value of @xmath24 with an eventual experimental determination , the error coming from size effects can be considered negligible .
there are simple and usually effective strategies to estimate the temperature dependence of the order parameter , from our mc data , in ways that are robust against size effects .
thus , for example , if we are interested in the thermal average of the dw polarization @xmath57 , it is a good idea to compute the average of its absolute value , @xmath58 , instead . in this way ,
spurious fluctuations ( i.e. , _ jumps _ between equivalent energy wells ) of the dw polarization , permitted by the finite size of the simulation box , will not result in a vanishing order parameter .
an obvious drawback of this approach is that @xmath59 will be different from zero even in the disordered phase , which is not correct . yet , monitoring the temperature dependence of this quantity is usually enough to locate a phase transition , and even permits a quantitative estimation of the transition temperature .
figure s1 shows the results for @xmath58 from our mc simulations .
as expected , we obtain a smooth temperature dependence that clearly points at a transition occurring in the range between 320 k and 350 k. further , we can estimate the transition temperature by numerically calculating the inflection point of this curve , which happens at about 332 k. we were not fully satisfied by this simple estimate , and tried to analyze our mc data in more detail , proceeding as follows .
figure s2(a ) shows the computed @xmath18 evaluated at @xmath60 and as a function of temperature .
there we can appreciate that at @xmath28 k the dws do not visit the @xmath60 state , as we have @xmath61 .
hence , it is clear that our simulations predict @xmath62 k. above that temperature , there is a range in which we get very small but non - zero values of @xmath63 , as well as a bimodal shape for @xmath18 .
do we really have disordered dws at these temperatures ?
to gain insight into the behavior of the system in this temperature range , we considered several ways to compute the dielectric constant @xmath34 .
let us recall that @xmath64 , and that we have @xmath65 where @xmath66 is the applied electric field , @xmath67 is the supercell volume , @xmath68 , @xmath69 is the boltzmann constant , and @xmath70 denotes a thermal average as computed from our mc simulations .
now , instead of using all the sweeps of our production mc runs to compute @xmath71 , it was instructive to consider smaller sets of data that we label by @xmath72 , compute the corresponding @xmath73 for each set , and perform a statistical analysis of the results .
figure s2(b ) summarizes the outcome of such an exercise .
we show the results obtained by considering sets of data composed of 16,000 mc sweeps each .
( we checked that 16,000 sweeps are typically sufficient to get well - converged thermal averages of our quantities of interests . ) if we approximate @xmath71 by @xmath74 where @xmath75 is the number of 16,000-sweep sets that we have , we obtain the results shown by a solid black line and triangles in fig .
interestingly , @xmath76 presents two peaks located at 320 k and 335 k , respectively . if , alternatively , we approximate @xmath71 by @xmath77 we get the results shown by a dashed black line and circles in fig .
s2(b ) . in this case
, we get a single broad peak with a maximum at @xmath78 k. to understand the origin of these results , it is useful to look at the evolution of @xmath11 during the mc runs , as shown in figs .
s2(c ) and s2(d ) for 335 k and 320 k , respectively . at 335 k
we observe frequent fluctuations in the value of @xmath11 , suggesting that the dws are not ordered .
( note that we have two dws in the simulation cell ; at 335 k we get @xmath79 c / m@xmath10 when the two dw polarizations are parallel , and @xmath80 when they are anti - parallel . )
the fluctuations are large , which leads to a large response according to eq .
( [ eq : chi ] ) and explains the peak that @xmath76 presents at that temperature . at the same time
, it is possible to find relatively long segments of the mc run in which @xmath11 fluctuates around a constant value .
hence , the computed @xmath81 is much smaller than @xmath76 .
figure s2(d ) shows that the situation at 320 k is very different . in this case , our dws clearly present a stable polarization .
nevertheless , from time to time we do observe a jump in @xmath11 , indicating a switch in the polarization of one of the dws .
these infrequent jumps lead to a very large value of @xmath76 , which peaks at that temperature .
in contrast , we find that @xmath81 , which is obviously determined by a segment of the mc trajectory that is free of jumps in @xmath11 , takes a relatively small value . from this analysis
, we can conclude that the peak observed at 320 k in @xmath76 is an artifact associated to the finite size of our simulation supercell , and that our results suggest that @xmath24 is about 335 k or 340 k. to generate the figures in our manuscript , we assumed @xmath82 k and processed our mc data accordingly .
finally , let us note another feature of our results that is related with the finite size of our simulation box . in fig .
4(b ) of our manuscript , the results for @xmath34 with a clear dw contribution extend only up to @xmath83 k , where a discontinuity can be appreciated .
the reason is that above that temperature , because of the finite size our supercell , the initial multi - domain configuration evolves during the mc run into a mono - domain state , which is more stable for the elastic ( unconstrained ) and electrical ( short - circuit ) boundary conditions considered in our simulations .
nevertheless , we do not observe any sort of structural order at the dws for @xmath84 , and have no reason to expect a different behavior at higher temperatures .
hence , our current inability to investigate the dws up to @xmath38 is not troublesome .
the largest quantitative uncertainty affecting our value for @xmath85 is surely related with the inaccuracies inherent to our first - principles model potentials . as shown in ref .
, our potential for pto reproduces the energetics of the fe instabilities with very good accuracy , and it also accounts well for the lowest - energy perturbations of the relevant stable structures .
hence , as far as the homogeneous fe phases of the material are concerned , our potential can be considered an accurate approximation to density functional theory ( dft ) results obtained using the local density approximation ( lda ) .
however , the computed @xmath37 k turns out to be significantly smaller than the experimental value of 760 k , which suggests there is a problem with the accuracy of the lda and employed first - principles methods . as discussed in ref .
, tracking down the origin of this error is not trivial ; the lda could be giving us ( 1 ) fe instabilities that are too weak , ( 2 ) afd instabilities that are too strong , ( 3 ) a fe - afd competition that is too strong , or ( 4 ) a combination of all these problems .
all such errors will affect our calculation of @xmath24 in essentially the same way as they affect the calculation of @xmath38 .
hence , we have reasons to believe that our model may underestimate significantly the temperature of the dw - confined transition .
additionally , one may wonder whether our model reproduces the dft energetics associated with the dws as well as it captures those related with the bulk fe state .
our model renders a dw energy @xmath86 mj / m@xmath10 when we allow for a full structural relaxation of the dw , and @xmath87 mj / m@xmath10 when we constrain the dw to remain in the unpolarized high - symmetry state . using the pbesol approach to dft [ which we denote `` pbesol ( i ) '' in our manuscript ; see methods section ] , we get 149 mj / m@xmath10 and 152 mj / m@xmath10 , respectively , for these two quantities . when we run the pbesol relaxation imposing the cell strain obtained from our model calculations [ `` pbesol ( ii ) '' ] , we get 225 mj / m@xmath10 and 250 mj / m@xmath10 , respectively . note that meyer and vanderbilt reported a value of 132 mj / m@xmath10 for this quantity,@xcite which they computed at the lda level and considering an unpolarized dw .
thus , we can conclude that our model gives acceptable results for the dw energy , even though this piece of information was not used when calculating the potential parameters .
( for reference , we ran lda calculations considering an unpolarized dw , as done in ref . , and got a dw energy of 97 mj / m@xmath10 . ) for us it is more relevant to consider the energy difference between the paraelectric and ferroelectric states of the dw , i.e. , the depth of the energy well corresponding to case iii in fig .
2(b ) of our manuscript .
for this quantity we get 86 mev / cell with our model , 4 mev / cell at the `` pbesol ( i ) '' level , and 26 mev / cell at the `` pbesol ( ii ) '' level .
it is interesting to note that , despite the good agreement for the dw polarization and structure between our model and the pbesol calculations [ see fig .
1(c ) ] , the differences in energy are large .
this comparison suggests that , with respect to the result one would obtain from a pbesol simulation , our model may overestimate significantly the value of @xmath24 . [ for reference , we also computed the depth of the energy well corresponding to the dw - confined fe instability at the lda level , and obtained 6 mev / cell when a full structural relaxation is allowed , and 45 mev / cell when the cell obtained from our model - potential simulations is imposed . hence , lda renders slightly stronger dw - confined instabilities than pbesol . ]
how worrisome are the differences between our model and pbesol energetics as regards the dw order ?
how well would a pbesol simulation reproduce the @xmath38 of the fe transition occurring in the bulk of pto ? is pbesol the best dft flavor available for this task ?
it is not possible to answer those questions at present . as an indication ,
let us mention that our pbesol calculations render a value of 67 mev / cell for the energy difference between the bulk - paraelectric and bulk - ferroelectric phases of pto .
this value corresponds to the depth of the energy well of case i in fig .
2(b ) , for which we get 189 mev / cell using our model .
hence , since our model underestimates pto s experimental @xmath38 quite significantly , we can tentatively conclude that a pbesol simulation would lead to very small curie temperature for this material .
hence , despite its accuracy at predicting equilibrium structures , it is not clear at all whether pbesol should be our dft method of choice for calculating the energetics of pto s fe instabilities .
we hope this discussion illustrates the difficulties involved in assessing the accuracy of dft , and dft - based methods , as regards the calculation of transition temperatures in materials with non - trivial structural and lattice - dynamical effects like pto .
as mentioned , we have reasons to think that our result for @xmath24 might be smaller than the experimental value , and we also have reasons to think it might be larger . at any rate , the existence of a dw - confined polar order and temperature - driven transition remains a solid prediction .
as mentioned in our paper , there are several first - principles studies of the 180@xmath2 dws of pto available in the literature @xcite . interestingly , except for the earliest one,@xcite all of them suggest that no polar order occurs within the dw plane
. we can not be sure about the reasons why these previous works did not find polarized dws , but some possibilities are worth mentioning .
first , it is usual in studies of this sort to assume a simplification of the dw structure , typically imposing that the symmetry operations that are common to the two neighboring domains be preserved in the structural relaxations .
such an approximation , which is adopted to alleviate the computational cost of the simulations , may well render a null dw polarization .
second , we obtained our polar dw configurations from mc annealings based on our pto potential , which allowed us to explore the configuration space very efficiently , and used such structures as the starting point of the first - principles relaxations . in our experience , one
is not guaranteed to reach such a distorted structure in a typical first - principles optimization , which would start from the non - polar dw state , even if all symmetries are broken .
( note that the information on the energetics of the dw instability , given in section iii above , may be relevant to this issue . )
let us also note that , recently , wei _
et al_. @xcite have found polar distortions at the structural ( antiphase ) dws of non - polar compound pbzro@xmath0 .
et al_. present first - principles results showing the bi - stability of the polar dw configuration , and thus argue that such a polarization is switchable .
however , the supplemental material of ref . suggests that the switch of the dw polarization also involves the reversal of structural distortions _ inside _ the domains .
this is reminiscent of the so - called _ hybrid improper ferroelectrics _ where the polarization switch must be accompanied by the reversal of one additional order parameter @xcite and of the polar order predicted at the fs walls of catio@xmath0 @xcite .
in contrast , the polarization of our pto dws can switch without affecting the surrounding domains .
49ifxundefined [ 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 10.1002/adma.201102254 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1038/ncomms1261 [ * * ( ) , 10.1038/ncomms1261 ] @noop * * , ( ) link:\doibase 10.1103/physrevb.81.224118 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _
( , ) @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevb.80.060102 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevlett.111.247603 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop @noop @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevb.68.134103 [ * * , ( ) ] @noop link:\doibase 10.1103/physrevlett.89.067601 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevlett.100.136406 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1038/ncomms4031 [ * * ( ) , 10.1038/ncomms4031 ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) | we present a first - principles study of model domain walls ( dws ) in prototypic ferroelectric pbtio@xmath0 . at high temperature
the dw structure is somewhat trivial , with atoms occupying high - symmetry positions .
however , upon cooling the dw undergoes a symmetry - breaking transition characterized by a giant dielectric anomaly and the onset of a large and switchable polarization .
our results thus corroborate previous arguments for the occurrence of ferroic orders at structural dws , providing a detailed atomistic picture of a temperature - driven dw - confined transformation . beyond its relevance to the field of ferroelectrics ,
our results highlight the interest of these dws in the broader areas of low - dimensional physics and phase transitions in strongly - fluctuating systems .
the structural domain walls ( dws ) occurring in ferroelectric ( fe ) and ferroelastic ( fs ) materials have become a focus of attention .
recent studies show that the dws can present a variety of properties , from conductive @xcite and optical @xcite to magnetic @xcite , that differ from those of the neighboring domains , which suggests that they could be the active element in nano - technological applications @xcite . elucidating the dw behavior
poses major experimental challenges , and the origin of most of the newly discovered effects remains unclear .
in fact , we still lack a detailed structural and dynamical picture of the dws , and in many cases we can only speculate about the structure property relationships at work within them .
hence , there is a pressing need for predictive theoretical studies tackling the dws at an atomistic level and at the relevant conditions of temperature , etc .
the dw structure , and even the possible occurrence of dw - confined ferroic orders , have been discussed theoretically for decades , usually in the framework of continuum ginzburg - landau or phenomenological model theories @xcite .
materials with competing structural instabilities have been a focus of attention , a good example being perovskite srtio@xmath0 ( sto ) .
sto undergoes a fs transition driven by an anti - ferrodistortive ( afd ) mode that involves concerted rotations of the o@xmath1 octahedra in the perovskite structure .
this mode competes with a fe instability that is suppressed by the onset of the afd distortion @xcite . yet
, there are both theoretical and experimental indications that a polar order occurs at low temperatures _ within sto s fs dws _
@xcite , i.e. , in the region where the otherwise dominant afd distortions vanish . in this context , it is worth noting recent first - principles studies predicting that pbtio@xmath0 ( pto ) @xcite and related compounds @xcite present a fe - afd competition that is even stronger than the one occurring in sto .
these are the ideal conditions to obtain interesting effects at structural dws , and motivated this work .
_ low - temperature study_. we employed the tools of ref . , which permit large - scale simulations with first - principles predictive power , to investigate an ideal version of the simplest dws occurring in pto , namely , 180@xmath2 boundaries separating regions of opposed polarization and being perfectly planar .
we used the model potential for pto labeled `` @xmath3 '' @xcite , which we briefly describe in @xcite . as shown in fig .
[ fig1](a ) , we set the polarization of the first domain @xmath4 parallel to the [ 100 ] direction of the perovskite lattice , and took @xmath5 $ ] for the second one ; the dw in between was assumed to reside in a ( 001 ) plane .
our supercell , which contains 12@xmath612@xmath620 perovskite units ( 14400 atoms ) , is periodically repeated and holds two dws .
we investigated the ground state structure of this multi - domain configuration by means of monte carlo ( mc ) simulated annealings @xcite .
figure [ fig1](b ) shows the @xmath7-component of the polarization ( @xmath8 ) as we move along @xmath9 .
we observe two domains within which pto adopts the structure of its homogeneous ground state , with an associated polarization of about 0.99 c / m@xmath10 and a cell aspect ratio of about 1.07 .
the domains are separated by a dw centered at a pbo plane and presenting a thickness of about one unit cell .
our dws do not display any rotations of the o@xmath1 octahedra .
this result lends itself to a simple explanation : because the dws are ultra - thin , hypothetical dw - localized afd modes would _ overlap _ with the neighboring fe distortions , and thus be penalized by the fe - afd competition . as a result
, the absence of localized afd modes seems rather natural . nevertheless , the structure of the dws is far from being trivial . as shown in fig .
[ fig1](c ) , a non - zero @xmath11 polarization appears at the dw plane and rapidly vanishes as we move into the domains .
this dw polarization is switchable , as evidenced by the hysteresis loop in fig .
[ fig2](a ) .
further calculations show that the polarizations of neighboring dws couple , and tend to align in a parallel configuration when the walls are sufficiently close . however , the dw
dw interaction quickly decreases with the separation distance ; for example , for our 12@xmath612@xmath620 supercell , the energy split between the parallel and anti - parallel states is about 0.01 mev _ per _ dw cell , which is negligible .
hence , the anti - parallel configuration shown in fig . [ fig1 ] is a stable state ( i.e. , a local minimum of the energy ) , but not more significant than the also stable , quasi - degenerate parallel configuration .
our results show how strain and the reduced dimensionality determine the energetics of the dw - confined fe distortion .
the onset of the multi - domain structure for @xmath8 implies a strong deformation of the perovskite lattice : it becomes tetragonal and acquires an aspect ratio of 1.07 , the long lattice vector coinciding with the polar axis @xmath7 . in the case of our simulated system , the @xmath12 plane is homogeneously strained throughout the supercell ( stretched along @xmath7 , shrunk along @xmath13 ) .
hence , even if we have @xmath14 at the dws of fig .
[ fig1 ] , the strain disfavors the occurrence of a dw polarization along the @xmath13 direction , which is subject to a compression .
the ensuing effect can be appreciated in the potential wells of fig .
[ fig2](b ) : case i corresponds to the full development of the fe distortion of pto , as it happens within the domains . case ii corresponds to the development of a three - dimensional @xmath13-polarized state when we constrain the cell to be strained as in the @xmath7-polarized fe state ; the equilibrium polarization and associated energy gain get clearly reduced , and we obtain a value of @xmath11 ( about 0.75 c / m@xmath10 ) that is not far from our result at the dw center ( about 0.65 c / m@xmath10 ) . additionally , fig .
[ fig2](b ) shows a case iii corresponding to the condensation of @xmath11 at our dws ; the energy well becomes shallower than in case ii , indicating a further weakening of the polar instability caused by the spatial confinement ( i.e. , by the truncation of interactions favoring the three - dimensional homogeneous polar state ) and the competition with the @xmath8 distortion of the neighboring domains .
nevertheless , the obtained well depth ( 86 mev / cell ) is sizable , which suggests that the predicted dw instability should occur at relatively high temperatures .
-polarized multi - domain state of fig .
[ fig1 ] and for an electric field along the @xmath13 direction . for these simulations we used a 1@xmath61@xmath620 supercell , thus assuming an homogeneous switch , and performed the calculations in the limit of 0 k ( details in @xcite ) .
the total response ( black line and circles ) is split in two parts : ( 1 ) the response of an @xmath7-polarized mono - domain state ( dashed blue line ) ; note that this response does not saturate , as the @xmath8 polarization will eventually rotate to align with the field .
( 2 ) the difference between the total and the mono - domain results ( solid red line ) , which captures the dw response . in order to observe the switch at the dws
, we imposed in the simulations the strain of the multi - domain ground state , and thus prevented the domain polarizations from fully aligning with the applied field for relatively small field values .
such a constraint is similar to the epitaxial one characteristic of thin films .
panel ( b ) : energy wells corresponding to fe instabilities in various situations described in the text . note that the @xmath15 state corresponds to a different atomistic configuration in each case . in the case of the dw polarization ( red solid line )
the energy is given _
per _ cell within the dw plane . ]
we checked the correctness of our model - potential predictions by running direct first - principles calculations of our multi - domain structure , using a 1@xmath61@xmath620 cell .
as shown in figs .
[ fig1](b ) and [ fig1](c ) , the agreement between our model - potential and first - principles results is very good , and the fe character of pto s dws is confirmed @xcite .
we should note that there are several first - principles studies of the 180@xmath2 dws of pto in the literature @xcite , and the consensus is that no dw - confined polarization occurs .
we can not be sure about the reasons why these previous works did not find polarized dws ; some possibilities are discussed in @xcite .
we checked whether this confined polarization occurs in other pto dws .
we found that 180@xmath2 dws lying in other planes
e.g. , a ( 011 ) boundary separating domains with @xmath16 $ ] and @xmath17 $ ] present polar distortions analogous to the one just described . in contrast , we found that 90@xmath2 dws do not present any fe instability , a result probably related with the fact that these boundaries are considerably more distorted than their 180@xmath2 counterparts , or to the elastic ( epitaxial tensile ) constraints we had to impose in order to stabilize them .
pto s 90@xmath2 dws will be discussed elsewhere . finally , let us note that the low - temperature configuration of our pto dws can be described as being bloch - like .
first - principles theory has predicted the occurence of bloch - like dw configurations in materials like linbo@xmath0 @xcite and batio@xmath0 in its rhombohedral phase @xcite . _ behavior with temperature_. we studied pto dws as a function of increasing temperature by running mc simulations as described in @xcite .
figure [ fig3 ] shows the obtained probability density , @xmath18 , for the @xmath13-component of the polarization at the dw plane .
we observe three distinct regions : ( 1 ) for @xmath19 k the dw presents a stable and large polarization , and the equilibrium state resembles the one discussed above .
( 2 ) for @xmath20 k the dw gets disordered and we have a null thermal average @xmath21 . at those temperatures the system presents a mirror symmetry plane perpendicular to the @xmath13 axis , and we could say that the dws are in a _ paraelectric
_ state @xcite .
( 3 ) finally , for 320 k @xmath22 350 k we have a narrow region in which , during the course of the mc simulation , @xmath23 occasionally switches between equivalent polar states .
clearly , the finite size of our simulation supercell is partly responsible for such fluctuations , which should be considered a spurious finite size effect below a certain transition temperature @xmath24 . at any rate ,
the presence of a phase transition is obvious from these results , and the analysis of the mc data suggests that we have @xmath25 k. ( see @xcite for details . ) for the @xmath13-component of the polarization at the center of the dw and as a function of temperature .
the three temperature regions mentioned in the text are marked with different colors : white for @xmath26 k ( where we clearly have @xmath27 ) , red for 320 k @xmath22 350 k ( critical region ) , and blue with stripes for @xmath28 k ( where we clearly have @xmath29 ) .
the left side of the histograms for @xmath30 k and @xmath31 k is drawn using dashed black lines ; this is to emphasize that , at these temperatures , we attribute the @xmath32 fluctuations from positive to negative values to finite size effects ( see @xcite ) . ] , blue circles ) and at the center of the dws ( @xmath33 , red triangles ) .
panel ( b ) : diagonal components of the dielectric permittivity tensor .
the discontinuity for @xmath34 ( triangles ) at @xmath35 k is related with the finite - size effects described in @xcite . ] the phase transitions in our simulated system are clearly appreciated in fig
. [ fig4 ] , which shows the evolution of the relevant order parameters and dielectric response .
as the system cools down from high @xmath36 , the diagonal components of the dielectric tensor increase sharply , revealing a fe transition at @xmath37 k. [ the quantitative disagreement between our computed @xmath38 and the experimental result for pto ( 760 k ) is discussed in ref . .
] note that for @xmath39 k we have @xmath40 , reflecting the cubic symmetry of the paraelectric phase .
then , for @xmath41 the @xmath34 component [ triangles in fig .
[ fig4](b ) ] behaves in an anomalous way .
as regards the dielectric response along the @xmath13 direction , our multi - domain configuration can be viewed as a set of parallel capacitors , so that the total dielectric constant can be split into domain and dw contributions , @xmath42 , where @xmath43 and @xmath44 are the respective volume fractions .
noting that @xmath45 and @xmath46 in our case , and that @xmath47 is featureless for @xmath48 ( which we checked by running mc simulations of the mono - domain case ) , it is clear that the increase in @xmath34 comes from a very large dw response @xmath49 .
hence , the dw - confined transition is driven by a fe soft mode and characterized by very strong fluctuations of the order parameter , as consistent with the reduced dimensionality .
_ final remarks_. as mentioned , the possibility of having polar orders and dielectric anomalies associated with dws has been discussed in the theoretical literature , typically employing phenomenological continuum models @xcite .
our first - principles work corroborates that such effects can indeed occur , providing for the first time a realistic atomistic picture of a @xmath36-driven dw - confined ferroic transformation .
one might describe the found transition as a change in the dw character , from bloch to ising , upon heating ; in fact , some authors have discussed similar effects in these terms @xcite .
however , we think that our result is better described as a proper fe phase transition confined to the dw , to emphasize that it results in a switchable dw polarization . for the same reasons
, we would rather denote the low - temperature state as a ferroelectric dw , and not simply as a dw with bloch - like character .
note that , as in the case of linbo@xmath0 @xcite , bloch - like dws may not display a net polarization .
the present results are the first step in the investigation of such an interesting phenomenon .
aspects for future work include : the possible critical behavior of the transition , the pre - transitional dynamics , the internal structure of the dws ( can we have multi - domain states within them ? ) , and the role of dw dw interactions ( can it affect the dimensionality and features of the transition ? ) .
we thus believe our findings will open exciting research avenues in the fields of phase transitions and low - dimensional physics .
it may seem surprising that the predicted effect has not been reported experimentally .
however , note that observing such a fe order may require some uncommon measurements .
ideally , one would like to work with samples presenting a pattern of highly - ordered stripe domains separated by 180@xmath2 dws , as displayed by suitably grown pto films @xcite and pto / sto superlattices @xcite .
high - resolution x - ray measurements at low temperatures may reveal the dw order . additionally , by measuring the dielectric response along the in - plane direction of the dws , one should be able to observe a clear feature around @xmath50 .
we hope our results will motivate further experimental work to characterize fe dws and the transitions that may occur within them .
work supported by mineco - spain ( grants no .
mat2010 - 18113 and no .
csd2007 - 00041 ) and csic [ jae - doc program ( jcw ) ] .
some figures were prepared using vesta @xcite .
we thank ekhard salje , david vanderbilt , and pavlo zubko for their useful comments . |
@figure#1fig . s@xmath114
( a ) pair distribution function of @xmath31 particles @xmath72 , ( b ) density of vibrational states @xmath73 , and ( c ) participation ratio @xmath74 of the @xmath0 wca glasses generated from equilibrated states at different initial temperatures .
the density @xmath2 is @xmath115 .
the legend shows the values of the initial temperatures.,scaledwidth=45.0% ] ( a ) intermediate scattering function @xmath116 for a particles of wca liquids with @xmath117 .
the density @xmath2 is @xmath115 . from the left to the right ,
the temperatures of the liquids are @xmath118 , @xmath119 , @xmath120 , @xmath121 and @xmath122 , respectively .
the horizontal dashed line marks @xmath123 at which the relaxation time @xmath46 is determined .
( b ) temperature dependence of the relaxation time .
the solid curve is the fit to @xmath46 with vogel - fulcher function @xmath49 . for the case shown here , @xmath124 , @xmath125 , and @xmath126.,scaledwidth=48.0% ] as described in the paper , we generate the @xmath0 glasses by quickly quenching equilibrated high temperature states to local potential energy minima using fire minimization method .
the same approach has been widely applied in the study of disordered solids . in order to check
if structural and vibrational properties of the @xmath0 glasses depend on initial condition or quench rate , we prepare equilibrated states at different temperatures above the glass transition temperature and quench them to local potential energy minima . as shown in fig .
s[sfig : fig1 ] , the pair distribution function @xmath72 , density of vibrational states @xmath73 , and participation ratio spectrum @xmath74 of the @xmath0 glasses do not show apparent initial condition dependence .
figure s[sfig : fig2 ] illustrates an example about how the glass transition temperature is measured .
figure s[sfig : fig2](a ) shows the intermediate scattering function @xmath127 of wca liquids measured at different temperatures , from which the relaxation time @xmath46 is obtained . as shown in fig .
s[sfig : fig2](b ) , the relaxation time increases rapidly with decreasing the temperature , which can be well fitted with the vogel - fulcher function as described in the paper .
the glass transition temperature is defined as the extrapolated temperature at which the relaxation time diverges .
it has been shown that , for strong correlating lennard - jones ( lj ) liquids with the correlation coefficient between the virial @xmath128 and potential energy @xmath129 greater than 0.9 , where the sum is over all pairs of interacting particles , fluctuations of the virial and potential energy , @xmath130 and @xmath131 , are linearly correlated : @xmath132 .
therefore , the pair potential can be well approximated into an inverse power law ( ipl ) @xcite @xmath133 where @xmath15 , @xmath16 , and @xmath13 are defined in the paper , and @xmath102 is a constant . as a consequence , dynamics such as the relaxation time @xmath46 of such strong correlating liquids shows density scaling : @xmath134 with @xmath135 , which is strictly valid in a pure ipl potential system . using this ipl potential approximation , a density scaling with @xmath136 has been proposed , which works well in the vicinity of @xmath57 @xcite . ( a ) correlation between the virial @xmath128 and potential energy @xmath137 of 1000 @xmath0 lj glass states at @xmath138 .
the solid line is the linear fit with a slope of @xmath139 .
( b ) density dependence of @xmath107 for the @xmath0 wca ( circles ) and lj ( squares ) glasses .
the triangles are measured from the very repulsive part of the interaction for the lj glasses .
the vertical dashed line marks @xmath140 . , scaledwidth=48.0% ] the success of the ipl potential approximation in the vicinity of @xmath57 for lj systems may make people believe that behaviors of lj systems are indeed determined by the approximated ipl potential and @xmath141 is the correct density scaling exponent at even higher densities
. however , there are some doubts that make the above conclusions questionable .
firstly , in the infinite density limit , the exponent of the very repulsive part of the interaction in the first several coordination shells should be strictly @xmath142 with @xmath143 . apparently , @xmath144 would not collapse relaxation times in the vicinity of @xmath57 .
this simple thinking at least implies that @xmath107 is not constant in density , while it varies from around @xmath145 near @xmath57 to @xmath146 in the large density limit .
in fact , the density variation of @xmath107 has been demonstrated by berthier and tarjus @xcite .
however , they did not move forward to question the validity of the ipl potential approximation . secondly
, the interaction of the repulsive counterparts of the lj systems , weeks - chandler - andersen ( wca ) systems , can also be well approximated by the ipl potential .
however , it has been shown that density scaling fails to describe the dynamics of wca systems @xcite .
next we will show that the ipl potential approximation indeed has its limitations to predict correct density scaling over a wide range of densities .
previously reported exponent @xmath136 is only approximately correct in a narrow density range around @xmath57 . at higher densities ,
the correct density scaling exponent is @xmath108 obtained from our dimensional analysis of the @xmath0 glasses discussed in the paper , which can not be correctly predicted from the ipl potential approximation .
scaling of the relaxation time in terms of the temperature and density for lj liquids .
the scaling exponents used are ( a ) @xmath147 and ( b ) @xmath108 .
the legend shows values of density .
, scaledwidth=43.0% ] figure s[sfig : fig3](a ) shows that fluctuations of the virial and potential energy for the @xmath0 lj glasses are linearly correlated , from which @xmath107 is calculated . in the paper , we find a crossover density @xmath148 above which the structural differences between the @xmath0 lj and wca glasses vanish , so do the vibrational differences .
it is then easy to assume that interactions of lj ( or its very repulsive part ) and wca glasses can be approximated into the same ipl potential at @xmath1 .
however , fig .
s[sfig : fig3](b ) indicates that this is not the case . just like what has been discussed above , @xmath107 for both lj and wca glasses indeed vary continuously with the density .
wca glasses show a @xmath107 larger than lj glasses . only at densities
much higher than @xmath1 , lj and wca glasses have the same @xmath107 .
the density variation of @xmath107 thus questions the validity of applying ipl potential approximation and assuming @xmath149 to find the correct density scaling . this problem has also been realized by bhling and coworkers who studied densities up to @xmath150 @xcite .
they therefore proposed a new approach instead of the simple ipl potential approximation to describe the dynamics of lj liquids .
if @xmath136 is indeed a robust density scaling , it should work over a wide range of densities .
however , as shown in fig .
s[sfig : fig4](a ) , the density scaling with @xmath151 as proposed by @xcite can only collapse relaxation times in a narrow density range @xmath105 .
it stops working otherwise .
in contrast , fig .
s[sfig : fig4](b ) shows that the density scaling with @xmath152 collapse the data at @xmath91 nicely .
therefore , @xmath152 obtained from our dimensional analysis is the correct density scaling exponent covering a wide range of densities . since @xmath107 is not constant in density
, @xmath152 is not a direct result of the ipl potential approximation .
( a ) illustration of the definition of @xmath153 and @xmath154 from the pair distribution function @xmath72 . the dashed and dot - dashed lines mark @xmath153 and @xmath154 at the first peak and minimum of @xmath72 .
( b ) density dependence of @xmath153 ( lj , stars ; wca , squares ) and @xmath154 ( lj , circles ; wca , triangles ) . the horizontal dot - dashed line marks @xmath109 .
the vertical dashed line marks @xmath1 .
the solid lines have a slope of @xmath155.,scaledwidth=48.0% ] for lj liquids , it has been claimed in @xcite that in order to get accurate simulation results it is enough to take into account merely the interactions within the fcs if and only if the liquid is strongly correlating " . here
, the first coordination shell ( fcs ) is defined as the intermolecular interaction range corresponding to the first minimum of the pair distribution function . since
above @xmath156 the structural and vibrational differences between the @xmath0 lj and wca glasses vanish , is it possible that @xmath1 is also the crossover density above which the fcs for lj glasses is within the range of repulsion , so that behaviors of the systems are completely determined by repulsion ? we measure the location of the first minimum of @xmath72 of the @xmath0 glasses and denote it as @xmath154 , as illustrated in fig .
s[sfig : fig5](a ) .
meanwhile , we locate the first peak of @xmath72 and denote it as @xmath153 . as shown in fig .
s[sfig : fig5](b ) , fcs is within the range of repulsion ( @xmath157 ) when @xmath158 , which is higher than @xmath1 .
therefore , there is no necessary connection between @xmath1 and the condition that fcs is within the range of repulsion .
figure s[sfig : fig5](b ) reveals another interesting phenomenon : both @xmath153 and @xmath154 are scaled well with @xmath159 over the whole range of densities studied . in the paper
, we obtain the same density scaling of the characteristic length from scaling collapse of @xmath72 at @xmath4 [ see eq .
( 2 ) of the paper ] . here , fig .
s[sfig : fig5 ] demonstrates that this characteristic length scale is present even below @xmath1 .
this structural scaling becomes one of the three elements for us to propose the empirical expression of the glass transition temperature [ eq .
( 6 ) of the paper ] . | we find that the density dependence of the glass transition temperature of lennard - jones ( lj ) and weeks - chandler - andersen ( wca ) systems can be predicted from properties of the zero - temperature ( @xmath0 ) glasses . below a crossover density @xmath1 , lj and
wca glasses show different structures , leading to different vibrational properties and consequently making lj glasses more stable with higher glass transition temperatures than wca ones . above @xmath1 ,
structural and vibrational quantities of the @xmath0 glasses show scaling collapse . from scaling relations and dimensional analysis
, we predict a density scaling of the glass transition temperature , in excellent agreement with simulation results .
we also propose an empirical expression of the glass transition temperature using structural and vibrational properties of the @xmath0 glasses , which works well over a wide range of densities . according to the theory of liquids ,
repulsive particle interaction determines the structure of dense liquids , while attraction acts as a perturbation @xcite .
the perturbative role of attraction has been questioned by recent simulations and experiments through the fact that a glass former with attraction can exhibit slower dynamics and consequently have a higher glass transition temperature than that with purely repulsive interaction even though their structures are similar @xcite .
recent theoretical approaches such as the mode coupling theory have failed to explain such nonperturbative effects of attraction from the liquid side above the glass transition temperature @xcite .
alternatively , some dynamical behaviors of supercooled liquids @xcite can be viewed from the glass side .
for instance , recent studies have shown that the heterogeneous dynamics of supercooled liquids and the glass transition temperature to a great extent reflect the quasi - localized nature of low - frequency normal modes of vibration of the zero - temperature ( @xmath0 ) glasses @xcite .
can we also gain any insight from studying properties of the @xmath0 glasses to understand the strong dynamical differences between glass formers with attractive and purely repulsive interactions ?
we achieve the positive answer to the above question by studying both lennard - jones ( lj ) and weeks - chandler - andersen ( wca ) @xcite systems over a wide range of densities .
we demonstrate that lj systems with attraction can have higher glass transition temperatures than wca ones with pure repulsion because lj glasses have either higher characteristic frequency or weaker quasi - localization of low - frequency modes and are thus more stable @xcite . the @xmath0 lj and wca glasses have indistinguishable vibrational properties only when their structures are identical . in order for lj and wca systems to have the same glass transition temperature , the @xmath0 glasses must have identical structures , which happens when the density @xmath2 is above a crossover value @xmath1 . at @xmath3
where lj and wca liquids can have similar structures but different dynamical behaviors , the structures of their @xmath0 glasses are still distinct .
when @xmath4 , pair distribution functions of the @xmath0 glasses measured at different densities show scaling collapse .
correspondingly , vibrational quantities such as the density of vibrational states and mode participation ratio spectrum also show scaling collapse . from scaling relations and dimensional analysis
, we predict a density scaling of the glass transition temperature , which agrees surprisingly well with simulation results .
more interestingly , the dimensional analysis implies an empirical expression of the glass transition temperature in terms of the structural and vibrational information of the @xmath0 glasses .
this expression fits the glass transition temperatures well , even in the density regime where density scaling breaks down .
our systems are three - dimensional cubic boxes with side length @xmath5 consisting of @xmath6 spheres with periodic boundary conditions in all directions . a mixture of @xmath7 and @xmath8 spheres with equal mass @xmath9 is employed .
we have verified that there is no significant system size effects .
the potential between particles @xmath10 and @xmath11 is @xcite @xmath12 + f(r_{ij}),\ ] ] when their separation @xmath13 is smaller than the potential cutoff @xmath14 , and zero otherwise .
here , @xmath15 and @xmath16 depend on the type of interacting particles : @xmath17 , @xmath18 , @xmath19 , and @xmath20 .
@xmath21 guarantees that @xmath22 .
@xmath23 tunes the range of interaction .
here we compare systems with @xmath24 ( wca ) and @xmath25 ( lj ) .
we set the length , mass , and energy in units of @xmath26 , @xmath9 , and @xmath27 , respectively .
the frequency is thus in units of @xmath28 .
the temperature is in units of @xmath29 with @xmath30 the boltzmann constant .
we generate the @xmath0 glasses by quickly quenching equilibrated high - temperature states to local potential energy minima using fire minimization algorithm @xcite .
it is verified in fig .
s5 of the supplemental material ( sm ) that our results do not depend on quench rate .
normal modes of vibration are obtained by diagonalizing the hessian matrix using arpack @xcite .
the glass structure is characterized by the pair distribution function of @xmath31 particles : @xmath32 , where @xmath33 is the number of @xmath31 particles and the sums are over all @xmath31 particles .
vibrational properties are characterized by the density of vibrational states @xmath34 and participation ratio @xmath35 , where the sums are over all modes , and @xmath36 is the participation ratio of mode @xmath37 with @xmath38 the polarization vector of particle @xmath10 .
here , @xmath39 denotes the average over @xmath40 independent configurations . for liquids ,
we apply molecular dynamics simulations in canonical ensemble and use gear predictor - corrector algorithm to integrate the equations of motion @xcite .
we measure the intermediate scattering function of @xmath31 particles , @xmath41)\right>$ ] , where the sum is over all @xmath31 particles , @xmath42 is the location of particle @xmath11 at time @xmath43 , @xmath44 is chosen in the @xmath45direction with the amplitude approximately equal to the value at the first peak of the static structure factor , and @xmath39 denotes ensemble average .
the relaxation time @xmath46 is determined by @xmath47 .
we estimate the glass transition temperature @xmath48 by fitting the relaxation time using vogel - fulcher function , @xmath49 , where @xmath50 and @xmath51 are fitting parameters .
fig .
s6 of the sm shows an example about how @xmath48 is measured .
( color online ) .
( a ) density dependence of the glass transition temperatures for wca ( circles ) and lj ( triangles ) systems , crossover temperature @xmath52 ( diamonds ) , and fits to the glass transition temperature using eq .
( [ tg_0 ] ) for wca ( crosses ) and lj ( pluses ) systems .
the solid line shows the density scaling of the glass transition temperature using eq .
( [ tg ] ) .
the dashed curve is the fit to @xmath52 by @xmath53 , where @xmath54 and @xmath55 .
the vertical dot - dashed line marks @xmath1 .
( b ) structural difference @xmath56 between the @xmath0 lj and wca glasses .
( c ) temperature evolution of @xmath56 between lj and wca systems . from the top to the bottom , the solid curves are for @xmath57 , @xmath58 , @xmath59 , @xmath60 , @xmath61 , and @xmath62 .
the dashed line marks @xmath63 , at which @xmath52 in ( a ) is estimated.,scaledwidth=45.0% ] in fig .
[ fig : fig1](a ) , we compare the glass transition temperatures of lj and wca systems over a wide range of densities . as already discussed in @xcite , @xmath64 .
the gap between @xmath65 and @xmath66 shrinks with increasing the density and becomes zero when the density is above a crossover value @xmath67 .
it has been shown that at fixed density structural differences between lj and wca systems decrease with increasing the temperature @xcite .
we thus calculate @xmath68 @xcite to quantify this difference . in fig .
[ fig : fig1](b ) , we first show @xmath56 at @xmath0 .
interestingly , with increasing @xmath2 , @xmath56 decays to a plateau ( @xmath69 , probably due to noises ) right at @xmath1 .
this indicates that @xmath1 is the crossover density above which the @xmath0 lj and wca glasses have identical structures .
figure [ fig : fig1](c ) shows the temperature evolution of @xmath56 .
as expected , @xmath56 decreases with increasing the temperature .
we estimate a temperature @xmath52 at which @xmath70 and define it as the crossover temperature above which lj and wca systems have identical structures @xcite . as shown in fig .
[ fig : fig1](a ) , @xmath52 decreases with increasing the density and can be approximately fitted into @xmath71 with @xmath54 and @xmath55 .
therefore , @xmath52 decays to zero approximately at @xmath1 , which implies that above @xmath1 lj and wca systems have identical structures at all temperatures . on the contrary ,
distinct structures of the @xmath0 lj and wca glasses at @xmath3 may thus explain why @xmath64 .
( color online ) .
pair distribution function @xmath72 , density of vibrational states @xmath73 , and participation ratio @xmath74 of the @xmath0 glasses , and temperature evolution of the relaxation time @xmath46 of supercooled liquids .
the solid and dashed curves are for wca and lj systems , respectively .
the left and right columns are for @xmath75 and @xmath76 , respectively .
the inset to the top left panel compares @xmath72 of wca and lj liquids at @xmath77.,scaledwidth=47.0% ] figure [ fig : fig2 ] compares the pair distribution function @xmath72 , density of vibrational states @xmath73 , and participation ratio @xmath74 at @xmath0 , and relaxation time @xmath46 at @xmath78 between lj and wca systems on both sides of @xmath1 .
the left column is for @xmath79 .
although @xmath72 of lj and wca liquids look identical ( see the inset to the top left panel ) , the @xmath0 structures are apparently different .
the structure plays a key role in determining vibrational properties of glasses . since @xmath72 are different , both @xmath73 and @xmath74 of lj glasses are different from wca glasses . in contrast , the right column indicates that @xmath72 of the @xmath0 lj and wca glasses at @xmath80 are identical , so do @xmath73 and @xmath74 . in glasses ,
there are excess low - frequency modes beyond debye s prediction , which form the boson peak @xcite .
the boson peak frequency @xmath81 at which @xmath82 exhibits a peak has been shown to closely correlate to the stability of glasses @xcite : a glass with lower @xmath81 would be less stable .
modes near @xmath81 are usually quasi - localized with small participation ratio @xcite , _ i.e. _ a small fraction of less constrained particles vibrate more strongly than the others in the mode . it has been shown that quasi - localization of low - frequency modes is also strongly responsible to the stability of glasses @xcite
: the energy barrier is lower along the direction of a more localized low - frequency mode .
therefore , we believe that boson peak frequency and low - frequency quasi - localization interplay to determine the stability of glasses .
( color online ) .
( a ) boson peak frequency @xmath81 and ( b ) average participation ratio of the lowest @xmath83 modes @xmath84 of the @xmath0 wca ( circles ) and lj ( triangles ) glasses .
the solid line in ( a ) is the fit to the boson peak frequency using eq .
( [ boson_peak ] ) . the vertical dot - dashed line marks @xmath1.,scaledwidth=45.0% ] in fig .
[ fig : fig3 ] , we compare the boson peak frequency @xmath81 and average participation ratio over the lowest @xmath83 modes @xmath85 for lj and wca glasses . at low densities , _ e.g. _
@xmath57 as commonly studied , lj glasses have larger @xmath81 and @xmath84 than wca glasses , so it is expected that @xmath86 . when @xmath87 , @xmath81 looks almost identical for both glasses .
however , lj glasses still have larger @xmath84 than wca glasses . in this density regime
, low - frequency quasi - localization plays the key role in determining the difference in stability between lj and wca systems . as illustrated in the bottom left panel of fig .
[ fig : fig2 ] , supercooled lj liquids relax more slowly than wca ones and @xmath86 . when @xmath4 , fig .
[ fig : fig3 ] indicates that the differences in @xmath81 and @xmath84 between both glasses vanish .
consequently , @xmath88 and supercooled lj and wca liquids exhibit the same dynamics , as shown in the bottom right panel of fig .
[ fig : fig2 ] .
the existence of @xmath1 clarifies the dynamical differences between lj and wca systems .
moreover , @xmath1 is also the crossover density above which @xmath72 of the @xmath0 lj and wca glasses show scaling collapse . as shown in fig .
[ fig : fig4 ] , @xmath72 measured at @xmath4 collapse onto the same master curve when plotted against @xmath89 .
similar scaling collapse has also been observed recently for lj liquids @xcite .
the scaling collapse implies a characteristic length scale @xmath90 ( color online ) scaling collapse of the pair distribution function @xmath72 , density of vibrational states @xmath73 , and participation ratio @xmath74 for the @xmath0 wca ( left column ) and lj ( right column ) glasses.,scaledwidth=47.0% ] as discussed above , the structure determines vibrational properties of the @xmath0 glasses .
consequently , both @xmath73 and @xmath74 may show scaling collapse as well .
figure [ fig : fig3](a ) indicates that at @xmath91 @xmath92 which should be the correct scaling relation of the frequency at @xmath4 . in fig .
[ fig : fig4 ] , we plot @xmath93 and @xmath74 against @xmath94 , and indeed collapse all curves at @xmath4 .
the scaling collapse at @xmath4 may contain interesting implications to the potential energy landscape . in harmonic approximation , frequencies of vibrational modes
measure local curvatures of basins of attraction of the @xmath0 glasses . at @xmath4 ,
potential energy landscapes for lj and wca systems may look similar especially near local minima with the average curvatures scaled with @xmath95 . from dimensional analysis
, we obtain @xmath96^{1/2 } , \label{dimension}\ ] ] where @xmath97 is a characteristic energy . at @xmath4 , because @xmath84 is constant in density , @xmath81 is the key to determine the stability of glasses , so it is plausible to assume that @xmath98 .
then , the combination of eqs .
( [ length])-([dimension ] ) leads to @xmath99 interestingly , eq .
( [ tg ] ) is exactly the density scaling of the glass transition temperature at @xmath4 , as illustrated by the excellent agreement between the the solid line [ plot of eq .
( [ tg ] ) ] and simulation results of @xmath48 in fig .
[ fig : fig1 ] . for lj systems
, this scaling works even below @xmath1 , because eq . ( [ boson_peak ] ) still holds and @xmath84 still remains constant , as shown in fig .
[ fig : fig3 ] .
in contrast , fig . [
fig : fig3 ] indicates that changes of @xmath81 and @xmath84 in density for wca glasses at @xmath3 are more pronounced , leading to a faster deviation of the glass transition temperature from eq .
( [ tg ] ) .
now let us move one step further .
seen from the top panels of fig .
[ fig : fig4 ] and fig .
s9(b ) of the sm , the locations of the first peak and minimum of @xmath72 still follow the scaling described by eq .
( [ length ] ) when @xmath3 , so eq . ( [ length ] ) is the characteristic length scale for all densities . combining eqs . ( [ length ] ) and ( [ tg ] ) and taking into account the effects of @xmath84
, we propose an empirical expression of the glass transition temperature purely based on structural and vibrational properties of the @xmath0 glasses : @xmath100 as shown in fig .
[ fig : fig1](a ) , eq .
( [ tg_0 ] ) fits all the glass transition temperatures nicely , even at low densities ( especially for wca systems ) where density scaling stops working @xcite .
equation ( [ tg_0 ] ) condenses the major findings of our work .
it explains why supercooled lj and wca liquids exhibit distinct dynamics from the @xmath0 glass perspective : vibrational properties including the boson peak frequency and low - frequency quasi - localization are distinct , which intrinsically originate from structural differences .
equation ( [ tg_0 ] ) thus makes a direct quantitative connection between the glass transition and @xmath0 glasses .
the derivation of eq .
( [ tg_0 ] ) does not require any information of particle interaction , so eq . ( [ tg_0 ] )
may be generalized to other systems not showing any density scaling at all , which will be discussed elsewhere @xcite .
although the crossover density @xmath1 is too high to be realized in experiments at present @xcite , the physical pictures that it brings about deserve more discussions .
recent studies have made an interesting point that potential between particles @xmath10 and @xmath11 of lj liquids can be approximated into an inverse power law @xmath101 with @xmath102 a constant varying with @xmath2 @xcite .
based on this assumption , @xmath103 and relaxation times of lj liquids can collapse in terms of @xmath104 , which work well at @xmath105 with @xmath106 @xcite .
however , as shown in figs .
s7 and s8 of the sm and discussed in @xcite , @xmath107 in fact varies with @xmath2 , so @xmath106 is an approximation , only working at @xmath105 .
when @xmath91 , our study indicates that @xmath108 is the correct density scaling exponent , which arises from dimensional analysis and scaling relations of the @xmath0 glasses and can not be predicted by the inverse power - law potential approximation .
moreover , fig .
s7 of the sm shows that the @xmath0 lj and wca glasses have different values of @xmath107 , so the inverse power - law potential approximation can not be simply applied to predict the existence of @xmath1 .
more details can be found in the sm . as suggested by a recent study @xcite , the first coordination shell of lennard - jonesian liquids is enough to give correct physics .
one may expect that the first coordination shell is inside the repulsive part of the interaction when @xmath4 .
we find that it is not exactly the case . at @xmath1 ,
the first minima of @xmath72 of both lj and wca glasses are still beyond the potential minimum at @xmath109 [ see fig .
s9 of the sm ] . at densities higher than @xmath110 , it is true that the first minimum of @xmath72 lies within the repulsive part of the interaction . in that regime ,
behaviors of lj systems are determined by the repulsive part of the potential , and to distinguish attractive and repulsive inter - particle potentials is not meaningful any more @xcite .
we are grateful to ludovic berthier , andrea j. liu , peng tan , gilles tarjus , and lei xu for helpful discussions .
this work was supported by national natural science foundation of china no .
91027001 and 11074228 , national basic research program of china ( 973 program ) no .
2012cb821500 , cas 100-talent program no .
2030020004 , and fundamental research funds for the central universities no .
2340000034 . 50 electronic mail : [email protected] + j. p.
hansen and i. r. mcdonald , theory of simple liquids ( elsevier , amsterdam , 1986 ) .
j. d. weeks , d. chandler , and h. c. andersen , j. chem . phys . * 54 * , 523 ( 1971 ) .
l. berthier and g. tarjus , phys .
rev .
lett . * 103 * , 170601 ( 2009 ) ; j. chem . phys . * 134 * , 214503 ( 2011 ) ; phys . rev .
e * 82 * , 031502 ( 2010 ) .
z. zhang , p. j. yunker , p. habdas , and a. g. yodh , phys .
rev .
lett . * 107 * , 208303 ( 2011 ) .
u. r. pedersen , t. b. schrder , and j. c. dyre , phy .
rev .
lett . * 105 * , 157801 ( 2010 ) .
p. g. debenedetti and f. h. stillinger , nature ( london ) * 410 * , 259 ( 2001 ) .
l. berthier and g. biroli , rev .
mod .
phys . , * 83 * , 587 ( 2011 ) .
m. d. ediger and p. harrowell , j. chem .
phys . * 137 * , 080901 ( 2012 ) .
s. karmakar , e. lerner , and i. procaccia , physica ( amsterdam ) * 391a * , 1001 ( 2012 ) ; g. biroli , s. karmakar , and i. procaccia , phys .
rev .
lett . * 111 * , 165701 ( 2013 ) .
a. widmer - cooper , h. perry , p. harrowell , and d. r. reichman , nature phys .
* 4 * , 711 ( 2008 ) .
h. shintani and h. tanaka , nature mater .
* 7 * , 870 ( 2008 ) .
l. wang , y. duan , and n. xu , soft matter * 8 * , 11831 ( 2012 ) .
m. l. manning and a. j. liu , phys .
rev .
lett . * 107 * , 108302 ( 2011 ) .
k. chen , m. l. manning , p. j. yunker , w. g. ellenbroek , z. x. zhang , a. j. liu , and a. g. yodh , phys .
rev .
lett . * 107 * , 108301 ( 2011 ) .
n. xu , v. vitelli , a. j. liu , and s. r. nagel , europhys .
lett .
* 90 * , 56001 ( 2010 ) .
w. kob and h. c. andersen , phys .
rev .
lett . * 73 * , 1376 ( 1994 ) .
e. bitzek , p. koskinen , f. ghler , m. moseler , and p. gumbsch , phys .
rev .
lett . * 97 * , 170201 ( 2006 ) . http://www.caam.rice.edu/software/arpack .
m. p. allen and d. j. tildesley , _ computer simulation of liquids _
( oxford science publications , new york , 1987 ) .
t. s. ingebrigtsen , t. b. schrder , and j. c. dyre , phys .
rev .
x * 2 * , 011011 ( 2012 ) . due to noises and
accumulated numerical errors , it is impossible and meaningless as well to locate exactly at what temperature @xmath111 .
we thus use the plateau value of @xmath56 at @xmath0 to define @xmath52 .
it is already a rather strict criterion . below @xmath52 with higher values of @xmath56 , @xmath112 and
@xmath113 already look very similar by eye .
a. p. sokolov , u. buchenau , w. steffen , b. frick , and a. wischnewski , phys .
rev .
b * 52 * , r9815 ( 1995 ) ; t. nakayama , rep . prog . phys . * 65 * , 1195 ( 2002 ) ; t. s. grigera , v. martin - mayor , g. parisi , and p. verrocchio , nature ( london ) * 422 * , 289 ( 2003 ) .
s. singh , m. d. ediger , and j. j. de pablo , nature mater .
* 12 * , 139 ( 2013 ) . h. r. schober and g. ruocco , phil . mag . * 84 * , 1361 ( 2004 ) .
n. gnan , t. b. schrder , u. r. pedersen , n. p. bailey , and j. c. dyre , j. chem . phys .
* 131 * , 234504 ( 2009 ) .
l. wang and n. xu , in preparation .
d. coslovich and c. m. roland , j. phys .
chem .
b * 112 * , 1329 ( 2008 ) ; d. coslovich and c. m. roland , j. chem .
phys . * 131 * , 151103 ( 2009 ) ; n. gnan , c. maggi , t. b. schrder , and j. c. dyre , phy .
rev .
lett . * 104 * , 125902 ( 2010 ) ; u. r. pedersen , n. gnan , n. p. bailey , t. b. schrder , and j. c. dyre , j. non - cryst .
solids * 357 * , 320 ( 2011 ) .
l. bhling , t. s. ingebrigtsen , a. grzybowski , m. paluch , j. c. dyre , and t. b. schrder , new j. phys . * 14 * , 113035 ( 2012 ) . l. bhling , a. a. veldhorst , t. s. ingebrigtsen , n. p. bailey , j. s. hansen , s. toxvaerd , t. b. schrder and j. c. dyre , j. phys .
cond .
matt . * 25 * , 032101 ( 2013 ) . |
ce- and yb - based @xmath18-electron kondo - lattice ( kl ) systems have shown the most drastic forms of non - fermi - liquid ( nfl ) behaviour @xcite . in these heavy - fermion ( hf ) materials ,
the ground state sensitively depends on the balance between two competing interactions , which are both determined by the strength of the @xmath19-conduction electron hybridisation @xmath20 : whereas the kondo interaction leads to a screening of the local moments below a kondo temperature @xmath21 , resulting in a paramagnetic ( pm ) ground state with itinerant @xmath19-electrons , the indirect exchange coupling ( rkky interaction ) can mediate long - range magnetic ordering @xcite .
+ one of the explanations for such nfl phenomena is the presence of a quantum critical point ( qcp ) at a particular @xmath22 : if the transition temperature @xmath23 of the long - range magnetic order is continuously shifted to zero by an external parameter @xmath24 , e.g. pressure , magnetic field or chemical substitution , a @xmath25 order quantum phase transition ( qpt ) takes place at @xmath26 and @xmath27 , to which a qcp is associated . here ,
the typical length and time scales of order parameter fluctuations diverge when approaching the transition point .
these fluctuations are believed to be responsible for the observed nfl corrections to the fl prediction for the heat capacity @xmath28 , magnetic susceptibility @xmath29 , electrical resistivity @xmath30 etc .
qcps are not the only mechanism to provide nfl behaviour .
if the magnetic order changes from long - range to short - range and disorder comes into play , spatial regions ( also called `` rare regions '' ) can show local magnetic order , although the bulk system is in a pm state @xcite .
the order parameter fluctuations of these regions can become strong enough to destroy the qpt and can give rise to nfl effects @xcite . here
, the length scale of order parameter fluctuations becomes finite , while the time scale still diverges . as a consequence in ,
e.g. , the quantum griffiths phase ( qgp ) scenario , power - law corrections , @xmath31 and @xmath5 with @xmath7 , are expected in a broad region across the qcp and not only at the qcp itself @xcite .
the global phase transition is then smeared @xcite , as observed in doped ferromagnetic ( fm ) materials , as , e.g. , the itinerant zr@xmath2nb@xmath3zn@xmath32 @xcite , the @xmath33-based hf system urh@xmath2ru@xmath3ge @xcite and the ce - based cepd@xmath2rh@xmath3 @xcite .
the ground state of such systems depends on several factors and can be very exotic , as in ceni@xmath2cu@xmath3 , where a percolative cluster scenario has been proposed @xcite .
+ recently , we reported on the formation of a `` kondo - cluster - glass '' state in cepd@xmath2rh@xmath3 for @xmath34 , caused by the freezing of clusters with predominantly fm coupling : they form below a temperature @xmath35 and freeze at a lower temperature @xmath36 @xcite . in cepd@xmath2rh@xmath3 ,
the chemical substitution of the ce - ligand pd by rh induces not just a negative volume effect , but , more importantly , it locally increases the hybridisation strength @xmath20 of the cerium @xmath19 electrons , leading to a strong enhancement of the local @xmath21 .
simultaneously , disorder is introduced to the system , which induces a statistical distribution of @xmath21 .
both effects create regions where @xmath18 moments are still unscreened and can form fm clusters because of the rkky interaction @xcite . in the temperature region @xmath37 we have observed power - law behaviour of several thermodynamic quantities , i.e. @xmath38 , @xmath12 , @xmath14 @xcite , and we concluded that the quantum griffiths phase scenario might be realised in cepd@xmath2rh@xmath3 for rh content @xmath39 . for @xmath40
the system is too close to the fm long - range order to be considered in the qgp . in this article
we take a closer look at the magnetic susceptibility @xmath9 and magnetisation @xmath0 in single crystals with rh content @xmath41 within the frame of the qgp scenario . + before starting with the analysis of the results , we have to consider that the qgp is restricted to a small region of the temperature - magnetic field ( @xmath42 ) phase diagram .
this region is limited to the ranges @xmath37 and @xmath43 , where @xmath44 and @xmath45 are , respectively , the magnetic fields necessary to destroy the glass state and to remove the effect of the cluster formation ( cf .
inset of fig .
[ fig1 ] ( b ) ) . both @xmath36 and @xmath44
have very low values of the order of mk and mt . to study the dynamic processes in the region close to @xmath46 , @xmath9 was measured down to 0.02 k ( fig . [ fig1 ] ) , at a frequency of 113 hz , with a modulation field of @xmath47 t .
the dc magnetisation @xmath0 was measured with a squid ( quantum design ) at high temperatures and with a high - resolution faraday magnetometer , in magnetic fields as high as 11 t and at temperatures down to 0.05 k ( fig . [ fig2 ] and [ fig3])@xcite . the magnetic field has been applied along the @xmath48 axis in all the measurements presented here .
it is worth mentioning that the magnetic anisotropy of the system is low for @xmath49 @xcite and poly- and single crystals exhibit similar behaviour @xcite .
+ in frame ( a ) of fig .
[ fig1 ] , the real part of the ac susceptibility is plotted as a function of @xmath10 in a double - logarithmic scale for four crystals of cepd@xmath2rh@xmath3 with @xmath50 . the peak temperature is @xmath36 . in frame ( b )
, @xmath9 is shown only for the crystal with @xmath46 at different fields .
k has been determined as the temperature where the field - cooled ( fc ) and zero - field - cooled ( zfc ) measurements of @xmath0 split , as illustrated in the inset of the same frame . as in polycrystals
, @xmath9 follows a @xmath52 power - law function above @xmath36 where the exponent @xmath11 varies systematically with @xmath13 between -0.12 and 0.45 @xcite , as expected in the qgp scenario .
the value of the exponents seems to be independent of the impurity phases observed at about 3 k ( see curve for @xmath53 ) .
the negative @xmath11 value for the sample with @xmath46 , different from the one measured in a polycrystal with the same @xmath13 ( @xmath54 ) , suggests that this sample is still too close to the fm instability .
moreover , the error in estimating the composition by energy dispersive x - ray spectrometry amounts to 1 at .
taking into account that the sample with slightly higher rh content @xmath53 shows @xmath55 , it is not possible from this measurement to discern whether the sample with @xmath46 can be considered in the qgp or not .
furthermore , the exponent is affected by the freezing which already takes place at @xmath56 k. focussing on the temperature range between @xmath36 and @xmath35 , the @xmath9 vs @xmath10 plot changes slope slightly at about 3.5 k , and again at about 1.5 k ( indicated by arrows ) .
we have fitted this range to extract @xmath11 , considering that in the 10 mt curve the fit range can be expanded to even lower temperatures .
+ to verify the presence of a qgp in this sample , we have measured @xmath0 vs @xmath8 at different temperatures . in fig .
[ fig2 ] the isotherms are shown : in frame ( a ) , the curve at 2 k is plotted together with those for two fm concentrations , @xmath57 and @xmath58 , to emphasise the strong decreasing of the magnetic moment with @xmath13 ; in frame ( b ) , the isotherms at 0.05 k and 2 k are compared .
the magnetisation at 0.05 k shows a very small hysteresis ( inset of fig .
[ fig2 ] ) , due to freezing at @xmath56 k , and it is far from reaching saturation at 10 t. we consider the value of the coercive field to be @xmath59 t. since @xmath51 k @xmath60 , we can assume @xmath45 to be close to 12 t , as the magnetic moment @xmath61 at 12 t is only @xmath62 .
it is thus plausible to look for qgp in @xmath0 vs @xmath8 for fields between 0.1 and 12 t. @xmath0 vs @xmath63 is plotted in fig .
[ fig3 ] at three temperatures ; below @xmath36 ( 0.05 k ) , just above it ( 0.5 k ) and close to @xmath35 ( 2 k ) .
the curves follow a power - law behaviour , with an almost constant value of @xmath17 , close to those observed in @xmath12 for the other crystals , as expected in the qgp .
since @xmath0 vs @xmath8 plots are less sensitive to paramagnetic impurities , and for @xmath64 the freezing does not affect the power law , these plots can be considered as signatures of a qgp in cepd@xmath15rh@xmath16 . + as discussed in the introduction , there is a fundamental difference between the nfl behaviour given by long - range and short - range order fluctuations .
the presence of clusters and the power - law corrections to susceptibility and magnetisation in cepd@xmath2rh@xmath3 indicate that the expected qpt at @xmath41 is replaced by disordered phases , possibly like the griffiths one .
+ we wish to thank t. vojta for helpful discussions .
supported by the dfg research unit 960 `` quantum phase transition '' .
9 von lhneysen h , rosch a , vojta m and wlfle p 2007 _ rev .
phys . _ * 79 * 1015 and references therein doniach s 1977 _ physica b _ * 91 * 231 vojta t 2006 _ j. phys . a : math .
gen . _ * 39 * r143-r205 miranda e , dobrosavljevi v and kotliar g 1997 _ phys . rev . lett . _ * 78 * 290 castro neto a h , castilla g and jones b a 1998 _ phys .
lett . _ * 81 * 3531 castro neto a h and jones b a 2000 _ phys .
b _ * 62 * 14975 vojta t and schmalian j 2005 _ phys . rev .
b _ * 72 * 045438 hoyos j a and vojta t 2008 _ phys .
lett . _ * 100 * 240601 ; hoyos j. a. , kotabage c. and vojta t 2007 _ phys .
lett . _ * 99 * 230601 sokolov d a , aronson m c , gannon w and fisk z 2006 _ phys .
lett . _ * 96 * 116404 huy n t , gasparini a , klaasse j c p , de visser a , sakarya s and van dijk n h 2007_phys .
b _ * 75 * 212405 sereni j g , westerkamp t , kchler r , caroca - canales n , gegenwart p and geibel c 2007 _ phys . rev . b _ * 75 * 024432 marcano n , gmez sal j c , espeso j i , de teresa j m , algarabel p a , paulsen c and iglesias j r 2007 _ phys .
lett . _ * 98 * 166406 westerkamp t , deppe m , kchler r , brando m , geibel c , gegenwart p , pikul a p and steglich f 2009 _ phys .
lett . _ * 102 * 206404 dobrosavljevi v and miranda e 2005 _ phys .
lett . _ * 94 * 187203 pikul
a p , caroca - canales n , deppe m , gegenwart p , sereni j g , geibel c and steglich f _ j. phys . condens . matter _
* 18 * l535 sakakibara h , mitamura h , tayama t and amitsuka h 1994 _ jpn .
phys . _ * 33 * 5067 deppe m , pedrazzini p , caroca - canales n , geibel c and sereni j g 2006 _ physica b _ * 378 - 380 * 96 | the magnetic field dependence of the magnetisation ( @xmath0 ) and the temperature dependence of the ac susceptibility ( @xmath1 ) of cepd@xmath2rh@xmath3 single crystals with @xmath4 are analysed within the frame of the quantum griffiths phase scenario , which predicts @xmath5 and @xmath6 with @xmath7 .
all @xmath0 vs @xmath8 and @xmath9 vs @xmath10 data follow the predicted power - law behaviour .
the parameter @xmath11 , extracted from @xmath12 , is very sensitive to the rh content @xmath13 and varies systematically with @xmath13 from -0.1 to 0.4 .
the value of @xmath11 , derived from @xmath14 measurements on a cepd@xmath15rh@xmath16 single crystal , seems to be rather constant , @xmath17 , in a broad range of temperatures between 0.05 and 2 k and fields up to about 10 t. all observed signatures and the @xmath11 values are thus compatible with the quantum griffiths scenario . |
lattice qcd is a powerful method to study quantum chromodynamics ( qcd ) in a nonperturbative way . in lattice qcd
, a path integral is directly evaluated on a discrete space - time lattice by means of the monte carlo method . as computer technology advances pc clusters
can also be used for lattice qcd simulations as well as a number of commercial supercomputers .
since lattice qcd simulations demand huge computer power , it is very important to optimize the simulation codes so as to exploit the full potential of the processor .
thus we optimize the hot spots of the codes such as the operation of a dirac operator to a spinor ( referred as @xmath0 hereafter ) and linear algebra of spinors , e.g. * @xmath1 : scalar product of two spinors * @xmath2 : norm square of a spinor * @xmath3 : add two spinors with a constant factor and assign the result to one of the source spinors in the simulation , a spinor is defined as a vector with 12 complex components on each grid point ( site ) of the lattice , and a gauge field is defined as a complex @xmath4 matrix on a link , which connects nearest neighbor sites .
numerically the operation @xmath0 is a combination of a complex @xmath4 matrix times a complex vector . from a technical point of view
, we should optimize these two applications separately since @xmath0 is processor limited as it needs more than one thousand floating point operations per site , while the linear algebra routines demand more data transfer than actual computations ( memory limited ) . in this note
we report our experience of the optimization of the lattice qcd codes for the new pc cluster with the amd opteron processor @xcite which is recently installed at desy hamburg .
of course the parallel computing with multiple processors is of our interest , however , we have optimized here only single processor version as a first step .
we also compare benchmarks with an older pc cluster with the intel xeon processors ( see table [ tbl : spec ] for the specification of these two processors ) .
.comparison of the specification of the two processors which we use [ cols="^,^,^",options="header " , ] [ tbl : comp ] for linear algebra we have more gain as expected from the difference of the processor clock speed . the larger data cache and
the improvement of memory bandwidth may reflect the result since linear algebra routines are rather memory limited .
in this report , we discuss the details of the optimization of the lattice qcd codes for the amd opteron processor .
tuning of prefetch distance for each application can improve performance by more than a factor of two . the prefetch instruction ,
however , may cause significant slowing down in some linear algebra routines at certain lattice sizes due to bank conflict .
we found that this slowing down can be cured by adopting two different prefetch distances for each source field .
the effect of the modification in sse instructions can be observed only after the adjustment of the prefetch instructions .
large cache effect is observed not only for the linear algebra routines but also for the @xmath0 routines when the source fields can fit into relatively large l2 data cache of the opteron processor .
this suggests that the pc cluster of the opteron processor can achieve high performance when the sublattice size becomes small enough . as a next step of this work
, we will develop a parallel version of the codes .
the author would like to thank h. wittig and y. koma for a number of useful discussions .
the author appreciate the original codes and the benchmark programs written by l. giusti , m. lscher and h. wittig , which is the basis of the improvement reported here . 9 software optimization guide for amd athlon 64 and amd opteron processors ( publication # : 25112 , issue date november 2004 )
amd64 architecture programmer s manual vol .
4 ( publication # : 24592 issue date september 2003 ) sandeep.s , gcc - inline - assembly - howto ( ` http://www.ibiblio.org/gferg/ldp/ gcc-inline-assembly-howto.html ` ) + v0.1 , 01 march 2003 . | we report our experience of the optimization of the lattice qcd codes for the new opteron cluster at desy hamburg , including benchmarks .
details of the optimization using sse / sse2 instructions and the effective use of prefetch instructions are discussed . |
to understand the evolution and appearance of galaxies it is crucial to study the ism / star formation interplay in galaxies .
most of our information about the ism relates to its massive component , the gas and its various phases .
its interplay with star formation can be addressed through the famous kennicutt - schmidt relation ( schmidt 1959 , kennicutt 1998 ) between the rate of star formation and the gas density .
the herschel and spitzer space telescopes have made major breakthroughs in this area , by mapping the emission from dust and the gas content of galaxies .
however , not much is known about the connection of star formation and the most energetic ism components , the cosmic rays and magnetic fields , and their role in structure formation in galaxies . as an extinction - free tracer of star formation
, the radio continuum emission provides crucial complementary information about magnetic fields and cosmic ray electrons ( cres ) .
new developments in radio correlators and polarimeters will revolutionize radio astronomy in the next years .
radio continuum observations with the forthcoming ska and its pathfinders such as vla , gmrt , and lofar are paramount to study the production and propagation of cres , the star formation
ism interplay , and the role of magnetic fields in structure formation and energy balance in galaxies .
+ in this proceeding , i present parts of our studies performed on resolved scales as well as globally in nearby galaxies as ideal laboratories for studying astrophysical processes that regulate the evolution of galaxies , among them the interplay between the star formation and the ism .
the radio continuum emission from galaxies observed in cm - waves is mainly due to two different radiation mechanisms , the nonthermal synchrotron emission and the thermal free - free emission .
the nonthermal emission depends on the magnetic fields and cre number density .
therefore , separating the thermal and nonthermal emission is key to understand the sources of cres and their distribution in galaxies and in estimates of the magnetic field strength .
the classical method requires accurate absolute measurements of the brightness temperature at high and low radio frequencies and needs a prior knowledge of the nonthermal spectral index ( @xmath0 , s@xmath1 ) .
propagating and interacting with matter and energy in the ism , cres cool and their energy spectrum changes depending on the environment .
hence , a fixed @xmath0 is not a correct assumption when obtaining the distributions of the nonthermal emission and the cres / magnetic fields across galaxies .
for m33 , tabatabaei et al .
( 2007 ) developed a thermal / nonthermal separation method based on a thermal radio template ( trt ) leading to the first map of the pure nonthermal spectral index in a galaxy .
in star forming regions , the nonthermal spectrum is flatter ( @xmath20.6 , typical spectral index of supernova remnants ) than in the inter - arm regions and outer parts of m33 ( @xmath21.0 , typical spectral index of synchrotron / inverse compton loss ) , in agreement with the energy loss theory of cres propagating from their places of birth ( fig . 1 ) .
the mean @xmath0 across m33 is 0.95 . separating the thermal and nonthermal emission in other nearby galaxy with higher sfr , ngc 6946 ,
a flatter spectrum is obtained ( @xmath3 , tabatabaei et al . 2013 ) .
if this is due to the star formation feedback and if a general dependency exists in galaxy samples have been impossible to address based on the classical separation method .
this is investigated in detail in our global study ( see sect . 4 ) . )
map obtained from the nonthermal radio fluxes at 3.6 and 20 cm .
the spiral arms and giant hii regions are indicated by contours of the total radio emission at 3.6 cm superimposed .
_ right _ : ngc 6946- @xmath4 mapped using the same method as in m33 .
the 9 giant hii regions are indicated .
the bars show the @xmath4 values .
, title="fig : " ] ) map obtained from the nonthermal radio fluxes at 3.6 and 20 cm .
the spiral arms and giant hii regions are indicated by contours of the total radio emission at 3.6 cm superimposed .
_ right _ : ngc 6946- @xmath4 mapped using the same method as in m33 .
the 9 giant hii regions are indicated .
the bars show the @xmath4 values .
, title="fig : " ]
the equipartition magnetic field strength depends on the nonthermal intensity and spectral index @xmath0 ( beck & krause 2005 ) and hence its local variation can be investigated using the maps of the pure nonthermal emission . the trt separation technique has allowed mapping the magnetic field strength in m33 ( tabatabaei et al . 2008 ) and ngc 6946 ( tabatabaei et al . 2013 ) . in both galaxies ,
the magnetic field strength is correlated with the sfr surface density ( fig . 2 ) .
this correlation is more likely due to the turbulent component of the magnetic field , as the sfr surface density has a tight correlation with the turbulent field .
this shows an enhancement of the turbulent magnetic field due to star formation feedback , such as supernova and strong shocks . as a possible mechanism
, small - scale magnetic fields could be amplified in star forming regions by a turbulent dynamo mechanism where kinetic energy converts to magnetic energy ( e.g. , beck & hoernes 1996 , gressel et al .
+ we note that there is no correlation between the large - scale , ordered magnetic field and sfr surface density in galaxies , as in the case of m33 and ngc 6946 ( fig . 2 ) .
studying a sample of non - interacting / non - cluster galaxies , tabatabaei et al .
( 2016 ) found a correlation between the large - scale magnetic field strength and the rotation speed of galaxies ( fig .
3 ) which shows the effect of the gas dynamics in ordering the magnetic fields in galaxies .
studying the spectral energy distribution ( sed ) provides significant information on the origin , energetics , and physics of the electromagnetic radiation in general . we have performed a coherent multi - band survey of the mid - radio ( 1 - 10ghz , mrc ) seds in a statistically meaningful galaxy sample , the kingfish ( key insights on nearby galaxies ; a far - infrared survey with herschel , kennicutt et al .
2011 ) with the 100-m effelsberg telescope ( pi : e. schinnerer ) . the effelsberg observations at 1.4ghz , 4.8ghz , 8.5ghz combined with archive data at 10.5ghz ( niklas et al .
1997 ) allow us , for the first time , to determine the mrc bolometric luminosities and further present calibration relations for the mrc versus the monochromatic radio luminosities , and as a sfr tracer .
the radio sed is fitted using a bayesian marchov chain monte carlo ( mcmc ) technique leading to measurements for the global nonthermal spectral index @xmath0 and the thermal fraction ( tabatabaei et al . submitted ) .
@xmath0 is found to decrease with increasing the sfr surface density in galaxies , in agreement with our local studies .
hence , due to the star formation feedback , the cre population is dominated by young and energetic particles in galaxies with high sfr .
fst acknowledges financial support from the spanish ministry of economy and competitiveness ( mineco ) under grant number aya2013 - 41243-p . | recent studies show the importance of the star formation feedback in changing the energetic and structure of galaxies .
dissecting the physics of the feedback is hence crucial to understand the evolution of galaxies .
full polarization radio continuum surveys can be ideally performed to trace not only star formation but also the energetic components of the interstellar medium ( ism ) , the magnetic fields and cosmic ray electrons . using the ska precursors
, we investigate the effect of the massive star formation on the ism energy balance in nearby galaxies .
our multi - scale and multi - frequency surveys show that cosmic rays are injected in star forming regions and lose energy propagating away from their birth place . due to the star formation feedback ,
cosmic ray electron population becomes younger and more energetic .
star formation also amplifies the turbulent magnetic field inserting a high pressure which is important in energy balance in the ism and structure formation in the host galaxy . |
a cognitive radio ( cr ) @xcite@xcite system may coexist with a primary network on an either interference - free or interference - tolerant basis @xcite . for the former case
, the cr system only exploits the unused spectra of the primary network . while , for the latter case , the cr system is allowed to share the spectra assigned to primary network under the condition that it must not impose detrimental interference on the primary network .
therefore , the interference from the cr network to the primary system should be carefully managed and cancelled in order to protect the operation of the primary system . various interference mitigation ( i m )
techniques applicable to cr networks have been reported in @xcite . as for multiple - antenna cr networks , transmit beamforming ( for single - data - stream transmission ) @xcite@xcite or precoding ( for multiple - data - stream transmission ) @xcite@xcite is an effective approach to proactively cancel interference from cr transmitters to the primary network . on one hand , it steers the cr transmission to avoid interfering with the primary network . on the other hand
, it exploits the diversity or the multiplexing gain of the multiple - antenna cr system to enhance the reliability or efficiency of the cr system .
however , in the works @xcite@xcite , perfect or partial csi of cr interference channels to primary network ( cr - primary interference channels ) is required at the cr transmitter side to guarantee no / constrained interference to the primary system .
therefore , extra signaling between primary and cr networks is inevitable to obtain the csi , which jeopardizes the applicability of these beamforming and precoding schemes . a more practical precoding scheme - sensing - projection ( sp)-based precoding , which learns the csi using subspace estimation @xcite and does not require
a priori csi , has been proposed for a cr multiple - input multiple - output ( mimo ) transmitter - receiver pair coexisting with a primary time - division - duplexing ( tdd ) system in @xcite .
however , such precoding scheme does not account for the interference from primary transmitters to the cr receiver ( primary - cr interference ) , which leads to a cr throughput loss . in @xcite and @xcite ,
it is proposed to remove the primary - cr interference at the cr receiver via null - space receiver beamforming , which sacrifices the cr throughput as well . moreover
, the cr network in @xcite has to work in a tdd mode aligned with the primary system in order to facilitate the null - space receiver beamforming . in this paper , two enhanced sp - based precoding schemes ,
namely , full - projection ( fp)- and partial - projection ( pp)-based precoding , are proposed for cr mimo systems by incorporating the primary - cr interference .
as the name suggests , the fp - based scheme nulls the cr transmission by fully projecting the transmission onto the estimated null space of the cr - primary interference channels . instead of removing the primary - cr interference using null - space receiver beamforming , the proposed precoding schemes account for the primary - cr interference via sensing .
this , on one hand , improves the cr throughput , and on the other hand , introduces more flexibility into the cr deployment , i.e. , the cr network does not have to work in a tdd mode as in @xcite .
the pp - based precoding can further improve the cr throughput by projecting the cr transmission onto a subspace that partially spans the estimated null space of the cr - primary interference channels .
as a result , the cr throughput is further improved at the cost of introducing extra interference to the primary network .
the remainder of this paper is organised as follows .
the system model is given in section ii .
the principle of the sp - based precoding is introduced in section iii .
the new precoding schemes are proposed in section iv .
the performance of the proposed precoding schemes is evaluated in section v. finally , we conclude the paper in section vi .
we consider a cr system shown in fig . 1 , where a cr transmitter - receiver pair shares the same spectrum with a primary tdd network .
multiple antennas are mounted at the cr nodes and possibly at each of the primary users .
the cr transmitter , cr receiver , primary base station ( bs ) and the @xmath0th primary user are equipped with @xmath1 , @xmath2 , @xmath3 and @xmath4 ( @xmath5 ) antennas , respectively .
block - fading channels are assumed for the primary and cr systems . for a narrowband transmission
, the received symbol at the cr receiver can be expressed as @xmath6 where @xmath7 is the received signal vector at the cr receiver , @xmath8 and @xmath9 are the transmit information vector with @xmath10 and precoding matrix of the cr transmitter , respectively , @xmath11 is the channel matrix from the cr transmitter to cr receiver , whose elements are independent and identically distributed ( i.i.d . )
complex gaussian random variables with zero mean and variance @xmath12 , and @xmath13 stands for the additive white gaussian noise ( awgn ) vector with zero mean and covariance matrix @xmath14 .
moreover , @xmath15 denotes the interference from the primary network to cr receiver .
it can be expressed as @xmath16 the overall cr system is shown in fig . 1 .
the matrices @xmath17 in and @xmath18 ( see fig . 1 ) represent the interference channels from all the @xmath19 active primary users to cr receiver and to cr transmitter , respectively , during primary uplink .
similarly , @xmath20 in together with @xmath21 ( see fig . 1 ) stand for the interference matrices from the primary bs to cr receiver and to cr transmitter during primary downlink .
all these interference matrices ( @xmath22 , @xmath23 , @xmath24 and @xmath25 ) have i.i.d .
complex gaussian random elements with zero mean and covariances @xmath26 , @xmath27 , @xmath28 and @xmath29 , respectively .
moreover , @xmath30 and @xmath31 are the transmitted signal vectors of all the @xmath19 primary users and primary bs , respectively .
we define the covariance matrix of the interference in as @xmath32 .
the precoding problem for cr transmission can be expressed as the following optimisation problem @xcite @xmath33 in , @xmath34 is the channel matrix from the cr transmitter to the @xmath0th primary user .
thus , the channel matrix from the cr transmitter to all primary users becomes @xmath35^t$ ] , where channel reciprocity is assumed .
the constraints on the cr transmission power and the maximum allowed interference perceived at each primary user are given by and , respectively .
the projected channel singular value decomposition ( svd ) or p - svd precoding has been proposed in @xcite as a suboptimal solution for the optimisation problem . it can be expressed as @xmath36^{\frac{1}{2}}\ ] ] where @xmath37 , @xmath38 denotes the power level for a water - filling ( wf ) algorithm , and @xmath39 and @xmath40 originate from the svd of the effective cr channel matrix @xmath41 @xmath42 its svd is expressed as @xmath43 , and @xmath44 in is from another svd @xmath45 , which is estimated via sensing in the sp precoding @xcite as shown in fig . 2 . by analogy with the multiple signal classification technique @xcite
, the signal covariance matrix is decomposed into signal and noise subspaces to estimate @xmath44 , which can be mathematically expressed as @xmath46 in , @xmath47 is the @xmath48th received symbol at the cr transmitter , and @xmath49 denotes the average covariance matrix of the received symbols .
an eigenvalue decomposition is then performed on @xmath49 in , where @xmath50 is a diagonal matrix with descendingly ordered eigenvalues of @xmath49 and @xmath51 contains the corresponding eigenvectors .
the matrix @xmath49 is further decomposed into interference and noise components in with @xmath52 and @xmath53 being the first @xmath54 and the remaining @xmath55 columns of @xmath56 , respectively , and @xmath57 and @xmath58 being their corresponding eigenvalue matrices .
in this section , we elaborate the cr precoding during the primary downlink . a similar precoding for the primary uplink can be easily obtained , which is ignored here for brevity .
when incorporating the primary - cr interference , the precoding problem for the cr transmitter during the primary downlink can be expressed as the following optimisation problem @xmath59 then , the precoding matrix for cr transmission during the downlink can be written as @xmath60^{\frac{1}{2}}\ ] ] where @xmath61 is the power level for the wf algorithm similar to that in and @xmath62 is obtained through the following eigenvalue decomposition @xmath63 it can be seen from and that in order to obtain the cr precoding matrix , the interference - plus - noise covariance matrix @xmath64 needs to be estimated at the cr receiver , besides the estimation of the interference subspace @xmath65 at the cr transmitter . to enable the estimation of @xmath65 and @xmath66
, we propose an enhanced precoding scheme .
the system diagram for this scheme is demonstrated in fig . 2 .
each cr cycle consists of sensing and transmission phases .
we name the cr transmission during the primary downlink transmission as t1 and uplink transmission as t2 . for t1 , the space @xmath67 is estimated at the cr transmitter during the primary uplink according to over @xmath68 symbols .
the estimation of @xmath66 is performed at the cr receiver at the beginning of the primary downlink for a batch of @xmath69 symbols via a procedure similar to . after obtaining these two estimates ,
the cr transmitter starts transmission t1 using the precoding matrix obtained by .
then t2 follows immediately after t1 but right before the sensing phase for the next cr cycle .
the cr precoding matrix for t2 can be obtained by other two sensing sessions concurrent with the sensing phase for t1 .
the fp - based channel projection is employed in this proposed precoding scheme .
therefore , it is termed as fp precoding .
it can be seen from fig . 2 that the proposed fp precoding scheme shifts the cr cycle of the sp precoding rightwards in time .
by doing this , several benefits are obtained .
firstly , introducing cr receiver sensing phases during both the primary downlink and uplink improves the cr instantaneous throughput by incorporating the interference - plus - noise covariance matrix into precoding , and consequently improves the cr throughput .
secondly , shifting the cr cycle diverts part of the cr transmission from the primary downlink to the uplink which reduces the time that primary receivers expose themselves to interference from the cr transmitter .
this is beneficial to the primary network , since primary users are usually more susceptible to cr interference than the primary bs .
theoretically , the proposed fp precoding can completely cancel the cr - primary interference if there is no error in the interference space estimation .
however , the i m ability of the proposed fp precoding degrades rapidly when the cr interference - to - noise ratio , inr@xmath70 , drops below a threshold .
this is due to the fact that in some components in the noise subspace may swap with those in the interference subspace when the noise amplitude @xmath71 is relatively large compared to the interference channel gain @xmath72 .
this phenomenon is known as a _ _ subspace swap _
_ @xcite . for low inr
, the interference subspace has a high probability to swap with the noise subspace .
when a subspace swap happens , can be rewritten as @xmath73 which means that the precoding matrix @xmath74 and @xmath75 span the same space . therefore ,
when the cr inr is low the average interference power received at primary users can be expressed as @xmath76 this suggests that the average interference power at primary users is proportional to the channel gain between cr and primary users at low cr inr .
the average cr - primary interference in the large cr inr regime becomes @xmath77 where is due to the fact that @xmath78 ; is obtained using the fact that @xmath79 for high inr @xcite with @xmath80 $ ] , and @xmath81 $ ] ; follows from the independence of @xmath82 and @xmath83 and @xmath84 for any constant matrix @xmath85 .
note that @xmath86 in is the transmit covariance matrix for the primary user .
an interesting fact can be observed from that at high cr inr the average received interference at primary users does not depend on the interference channel @xmath75 .
the average interference is proportional to the channel noise @xmath87 and inversely proportional to the sensing length @xmath68 . to further improve the throughput of the cr link , the cr transmitter may null its transmission to a subspace partially spanning the interference space .
therefore , we introduce another precoding scheme , namely , the pp precoding
. the pp precoding works in a similar manner to the above proposed fp precoding except for the selection of the interference space . for the downlink cr precoding
, the cr transmitter first obtains @xmath88 and @xmath56 via eigenvalue decomposition in during uplink sensing .
then , a subspace @xmath89 partially spanning the interference space is obtained by choosing @xmath90 eigenvectors corresponding to the first @xmath90 largest eigenvalues of @xmath88 , where @xmath90 can be determined according to various criteria .
one candidate criterion is @xmath91 with @xmath92 .
we call @xmath93 the trivial over dominant interference ratio ( tdir ) .
this selection process chooses @xmath90 dominant interference subchannels to form an estimate of the interference space and ignores the other @xmath94 trivial ones .
when equal power is assigned to each cr antenna , @xmath93 stands for the maximum ratio of the resulting and nullified interference to the primary receiver .
finally , substituting the estimated subspace @xmath89 for @xmath95 in , the precoding matrix @xmath74 for the downlink cr transmission can be obtained via .
however , we may fail to find a value of @xmath90 satisfying . in this case
, the proposed fp precoding is used .
the uplink cr precoding can be performed in the similar manner as the downlink counterpart .
the joint probability density function ( pdf ) of the ordered eigenvalues @xmath96 $ ] of @xmath49 , with @xmath97 is @xcite @xmath98 where @xmath99 is the transmission power of each primary user antenna and @xmath100 with @xmath101 is given by @xmath102}{\prod^{m_{\rm{min}}}_{i=1}(m_{\rm{max}}-i ) !
\prod^{m_{\rm{min}}}_{i=1}(m_{\rm{min}}-i)!}\ ] ] with @xmath103 .
therefore , the probability for the occurrence of is @xmath104 where @xmath105
. in other words , for the pp precoding scheme the probabilities of using the ` real ' pp ( @xmath90 satisfying exists ) and using fp are @xmath106 and @xmath107 , respectively .
therefore , the cr transmitter uses @xmath108 and @xmath109 degrees of freedom ( dof ) for interference mitigation in the pp and fp precoding schemes , respectively .
meanwhile , the dof for cr transmission for the pp and fp precoding are @xmath110 and @xmath111 , respectively .
this means that compared to the proposed fp precoding the pp precoding scheme transfers @xmath112 dof from interference mitigation to cr transmission , which leads to a higher throughput for the cr link .
it can be seen from that in the large inr regime , @xmath106 is fixed for a given noise power @xmath87 and @xmath99 , i.e. , the probability of the ` real ' pp and fp does not change with the interference channel @xmath113 . considering the fact from that at high inrs the average interference power of fp @xmath114 is fixed and the average interference power resulting from ` real ' pp @xmath115 is proportional to the square of the interference channel gain @xmath27 , the overall average interference of the pp precoding @xmath116 is linearly proportional to @xmath27 for large inrs .
in this section , the performance of the proposed precoding schemes is evaluated via simulations .
we consider a scenario where a cr mimo system coexists with a primary tdd system which has one 2-antenna bs and two single - antenna users .
each cr node is equipped with four antennas , i.e. , @xmath117 , @xmath118 , @xmath119 and @xmath120 . the primary network works as a downlink - broadcast and an uplink multiple - access system .
the primary bs uses perfect zero - forcing beamforming at both the primary downlink and uplink .
the transmission power of the cr and primary networks is @xmath121 .
all the results are obtained by averaging over 2000 simulation runs .
first , we evaluate the throughput of the cr system with the proposed precoding schemes over different values of signal - to - noise ratios , snr@xmath122 . in fig . 3 ,
the throughputs ( mutual information in averaged over a cr cycle ) of the two proposed precoding schemes are compared with that of the sp precoding of @xcite , @xcite and the p - svd precoding with perfect csi of @xcite .
the system setup is as follows : @xmath123 , and @xmath124 .
it can be seen that the proposed fp / pp precoding schemes lead to higher cr throughput than the sp precoding , and the throughput gain becomes larger as the snr increases .
4 evaluates the impact of cr inr on the cr throughput and the resulting cr - primary interference under different precoding schemes .
it has the same setup as that of fig . 3 with @xmath125 .
by comparing fig .
4(a ) with fig .
4(b ) , it can be seen that the proposed fp / pp precoding schemes outperform the sp counterpart at low inrs , since they lead to higher cr throughput without introducing extra interference . at high inrs , both the proposed fp and sp precoding schemes have fixed interference , and there is a fairly good agreement between the derived and simulated interference of the fp precoding . another phenomenon which can be seen from fig .
4(b ) is that the interference of the sp precoding is slightly smaller than that of the fp precoding .
this is due to the fact that the sensing of the sp precoding is longer than the uplink sensing of the fp precoding .
moreover , at high inrs the interference of the proposed pp precoding is linearly proportional to the cr inr , which supports our analysis in section iv.b .
in this paper , two sp - based precoding schemes , namely , fp and pp precoding , have been proposed for cr mimo systems to mitigate the interference to the primary network and improve the cr throughput .
these two precoding schemes are capable of estimating the csi of interference channels between primary and cr networks and can account for the interference from the primary system via a novel sensing approach .
therefore , no extra signaling is required between primary and cr systems , which consequently eases the deployment of cr networks .
the performance of the proposed precoding schemes has been evaluated and compared to that of the existing precoding approaches .
it has been demonstrated that the fp precoding can boost the cr throughput and does not introduce extra interference to the primary system in the low inr regime .
the pp precoding can further improve the cr throughput if the primary system can tolerate some extra interference .
i. f. akyildiz , w. y. lee , m. c. vuran , and s. mohanty , `` next generation / dynamic spectrum access / cognitive radio wireless networks : a survey , '' _ computer networks _ , vol .
50 , no . 13 , pp .
21272159 , sept .
2006 .
wang , x. hong , h .- h .
chen , and j. s. thompson , `` on capacity of cognitive radio networks with average interference power constraints , '' _ ieee trans .
wireless commun . , _ vol
4 , pp . 16201625 , apr .
2009 .
t. k. phan , s. a. vorobyov , n. d. sidiropoulos , and c. tellambura , `` spectrum sharing in wireless networks via qos - aware secondary multicast beamforming , '' _ ieee trans .
sig . process .
57 , no . 6 , pp .
23232335 , june 2009 .
l. bixio , g. oliveri , m. ottonello , m. raffetto , and c. s. regazzoni , `` cognitive radios with multiple antennas exploiting spatial opportunities , '' _ ieee trans .
sig . process .
_ , vol.58 , no . 8 , pp .
44534459 , aug .
2010 . h. yi , h. hu , y. rui , k. guo , and j. zhang , `` null space - based precoding scheme for secondary transmission in a cognitive radio mimo system using second - order statistics , '' in _ proc .
ieee icc09 , _ dresden , germany , june 2009 .
r. zhang , f. gao , and y .- c .
liang , `` cognitive beamforming made practical : effective interference channel and learning - throughput tradeoff , '' _ ieee trans .
commun . , _ vol .
706718 , feb .
2010 .
o. somekh , o. simeone , y. bar - ness , and w. su , `` detecting the number of transmit antennas with unauthorized or cognitive receivers in mimo systems , '' in _ proc .
ieee milcom07 , _ orlando , usa , oct . 2007 .
d. w. tufts , a. c. kot , and r. j. vaccaro , `` the analysis of threshold behavior of svd - based algorithms '' , in _ proc .
xxist annu .
asilomar conf . signals .
comput . , _
monterey , usa , nov .
1987 , pp . 24162419 .
f. li , h. liu , and r. j. vaccaro , `` performance analysis for doa estimation algorithm : unification , simplification and observations , '' _ ieee trans . aerospace and elec .
sys . , _ vol .
4 , pp . 11701184 , apr . 1993 .
m. chiani , m. z. win , a. zanella , r. k. mallik , and j. h. winters , `` bounds and approximations for optimum combining of signals in the presence of multiple co - channel interferers and thermal noise , '' _ ieee trans .
2 , pp . 296307 , feb . 2003 . fig . 4 . ( a )
cr throughput and ( b ) resulting interference of different precoding schemes ( @xmath126 @xmath127 @xmath128 @xmath129 , @xmath130 @xmath131 @xmath132 @xmath133 @xmath134 @xmath124 , and @xmath125 ) . | in this paper , we propose two subspace - projection - based precoding schemes , namely , full - projection ( fp)- and partial - projection ( pp)-based precoding , for a cognitive radio multiple - input multiple - output ( cr - mimo ) network to mitigate its interference to a primary time - division - duplexing ( tdd ) system .
the proposed precoding schemes are capable of estimating interference channels between cr and primary networks , and incorporating the interference from the primary to the cr system into cr precoding via a novel sensing approach .
then , the cr performance and resulting interference of the proposed precoding schemes are analyzed and evaluated . by fully projecting the cr transmission onto a null space of the interference channels ,
the fp - based precoding scheme can effectively avoid interfering the primary system with boosted cr throughput . while , the pp - based scheme is able to further improve the cr throughput by partially projecting its transmission onto the null space . + _ * index terms * _ cognitive radio , interference mitigation , mimo , precoding . |
one of the major challenges in the analysis of nonlinear pdes is to understand oscillations and concentrations in weakly converging sequences of functions since these phenomena distinguish weak from strong convergence in @xmath1-spaces . in its most simple form , this observation is already expressed in vitali s classical convergence theorem , by which a norm - bounded , non - oscillating ( contained in the requirement of convergence in measure ) , non - concentrating ( equiintegrable ) sequence is seen to converge strongly .
this connection between `` abstract '' compactness and `` concrete '' oscillation and concentration effects also supports the view that questions of compactness have a `` physical '' meaning and should be studied for their own sake , cf .
@xcite .
another motivation for such a study is that often sequences of interest are constrained by a linear differential constraint ( curl - freeness for a sequence of gradients for instance ) and/or pointwise constraints .
then , compensated compactness theory , see for example @xcite , considers whether these additional restrictions allow one to improve weak to strong compactness ; in fact , in this vein , the term `` compensated compactness '' should probably be replaced by `` compactness by compensation '' ( which is also closer to the french original `` compacit par compensation '' ) , but we here stick to the more conventional terminology . at the heart of this investigation
is an analysis of _ compatibility _ between these constraints and different shapes of oscillations and concentrations .
if together the constraints rule out any such phenomena , then we indeed have the sought improvement of weak to strong compactness . starting with simple scalar defect measures , more and more refined tools have been developed to study these questions .
the basic philosophy here is to retain as much information as possible about the sequence in an `` extended limit '' because weak convergence always involves a lossy averaging operation .
perhaps the most well - known objects of this kind are young measures , introduced by young in @xcite .
they represent the asymptotic value distribution of oscillations in sequences : a bounded sequence @xmath2 ( @xmath3 ) allows for the selection of a subsequence ( not relabeled ) that generates a young measure @xmath4 , defined to be a parametrized family of probability measures on @xmath5 , with the property that @xmath6 for any carathodory integrand @xmath7 such that the sequence @xmath8 is equiintegrable .
intuitively , the family @xmath9 contains the asymptotic value distribution of the sequence @xmath10 and thus reflects oscillations . as an important extension , diperna and majda @xcite compactified the target space to describe concentration effects as well .
the generalized young measure ( or diperna majda measure ) generated by @xmath10 now adds to the classical young measure @xmath4 a positive radon measure @xmath11 on @xmath12 and another parametrized family @xmath13 of probability measures on the unit sphere @xmath14 of @xmath5 ( @xmath15 always denotes the unit ball in @xmath5 ) .
it allows us to compute limits of a larger class of nonlinear quantities with @xmath16-growth : @xmath17 whenever @xmath18 has at most growth of order @xmath16 and the @xmath16-recession function @xmath19 ( describing the behavior of @xmath18 at infinity ) exists in a sufficiently strong sense : @xmath20 here , @xmath11 and @xmath21 respectively describe the location and value distribution of concentration effects in the sequence .
recent works on this topic include @xcite .
a different set of ideas , which could be considered an `` analysis '' , is expressed in tartar s theory of h - measures @xcite or , equivalently , grard s microlocal defect measures @xcite .
switching to a phase - space or microlocal perspective , these objects enable us to quantify not only the location but also the _ direction _ of oscillations and concentrations .
h - measures are not able to retain the asymptotic value distribution expressed in the young measure , and are furthermore restricted to the ; however , some extensions to an @xmath1-setting are possible , see @xcite .
the question of whether it is possible to combine the complementary approaches of young measures and h - measures and to find a tool allowing one to study _ at the same time _ the location , value distribution , and direction of oscillations and concentration effects in @xmath1-bounded sequences of functions , has remained open for some time ( cf .
chapters 38 , 39 in @xcite and @xcite for some remarks on this problem ) .
the principal aim of this work is to introduce precisely such a tool , called the * microlocal compactness form * ( mcf ) generated by an @xmath1-bounded sequence , where @xmath22 .
the following are some of its main features : * mcfs allow one to represent limits for a large class of nonlinear integral functionals ; in particular , the young measure is ( essentially ) contained . *
the directional information about oscillations and concentrations in the generating sequence is preserved and the h - measure can be reconstructed ( in an @xmath23-setting ) .
* differential and pointwise constraints on the sequence are reflected in the generated mcf and in fact hold for _ any _ generating sequence ( contrary to the situation for young measures , for which generating sequences can have very different properties ) .
consequently , mcfs provide new `` geometric '' proofs in compensated compactness theory .
* mcfs can be defined for all @xmath1-spaces , @xmath22 . *
they enjoy the same compactness properties as the weak convergence , i.e. @xmath1-boundedness is enough for the extraction of a generating subsequence . *
the @xmath16-compactness wavefront set of an mcf is defined as the analogue of the wavefront set in classical microlocal analysis , but with respect to weak
strong compactness and not @xmath24-regularity like in the classical notion .
* when considering a laminate ( nested microstructure ) , its hierarchy is reflected in the generated mcf .
* mcfs contain enough information to serve as the foundation of a relaxation theory for integral functionals in the presence of anisotropy .
* for some hyperbolic systems , one can derive equations for mcfs that express the propagation of singularities .
since we are interested in directional properties of oscillations and concentrations , the definition of microlocal compactness forms involves the fourier transform .
in fact , it is well - known that by switching to a fourier point of view we can understand the intuition behind weak convergence phenomena much better : fourier - transforming a sequence of functions @xmath25 in @xmath23 that oscillate with increasingly higher frequencies ( e.g. @xmath26 ) , one observes that the fourier transforms @xmath27 push mass out to infinity . applying parseval s theorem ,
the duality pairings with a test function @xmath28 can be written as @xmath29 since @xmath30 decays at infinity , integrating @xmath27 against @xmath30 corresponds to `` band - limiting '' the sequence @xmath10 and consequently , infinitely high frequencies are lost from the weak limit @xmath31 .
these considerations also persist for concentration effects , which are highly localized in space , and hence , by the uncertainty principle , are strongly smeared out in fourier space .
the band - limiting effect of a test function then implies that this fourier mass is again lost in the limit .
we conclude from this informal discussion that a tool aiming to improve the study of oscillations and concentrations must look at `` infinite frequencies '' only microlocal compactness forms will do just that .
another conclusion that can be drawn from this heuristic point of view on weak convergence is that for _ infinite _ frequencies it is possible to `` circumvent '' the uncertainty principle for the fourier transform : localizing in real space corresponds to smearing out the fourier transform , but when a packet of fourier mass wanders out to infinity , it does not matter how much smeared out it is ( we always look at conical neighborhoods around directions ) .
this simple observation will enable us to construct a theory fulfilling all the aforementioned properties .
after these preliminary considerations , we can now state the main idea of microlocal compactness forms : consider for a bounded sequence @xmath2 the double limit @xmath32 } { \;\mathrm{d}}x.\ ] ] here , @xmath33 , is a test function with sufficient regularity and we further require that its recession function @xmath34 exists in the sense @xmath35 this choice for the class of test functions will capture the location and value distribution of oscillations and concentrations . most often , we will not refer to @xmath36 directly , but only through @xmath37 for reasons of consistency with the notation for generalized young measures .
next , @xmath38 is a positively @xmath39-homogeneous fourier multiplier with sufficient smoothness , serving to represent directional information as well as differential constraints .
the family @xmath40 consists of smooth functions with @xmath41 on the ball @xmath42 and support inside @xmath43 , whose purpose it is to cut off all finite frequencies .
finally , by @xmath44 we denote the fourier multiplier operator with symbol @xmath45 .
the fundamental existence theorem in section [ ssc : def ] will establish that for all admissible @xmath46 the limit in can be expressed as @xmath47 which is a sesquilinear form in @xmath48 and @xmath49 .
here , @xmath50 and @xmath51 are parametrized sesquilinear forms and @xmath52 is a finite borel measure on @xmath12 .
the first part corresponds to oscillations , the second part to concentrations in the generating sequence @xmath10 .
in the course of this paper we will show that this object indeed quantifies the difference between weak and strong compactness of the sequence @xmath10 and has all the properties alluded to before .
in particular , we will consider applications to compensated compactness theory ( section [ ssc : comp_compact ] ) , laminates ( section [ ssc : laminates ] ) , relaxation ( section [ ssc : relax ] ) , and to the propagation of singularities in hyperbolic systems ( section [ ssc : propagation ] ) .
besides the theories of young measures and h - measures , also classical microlocal analysis has served as inspiration for the present work . following the groundbreaking work by hrmander , kohn , nirenberg ( see chapter viii of @xcite for historical references )
, this theory has been very successfully applied to linear pdes @xcite and to a degree also to nonlinear pdes @xcite . however , for nonlinear equations , microlocal analysis suffers from the defect that it measures @xmath24-regularity , which is often not well adapted to nonlinear pdes .
this paper is organized as follows : after recalling some preliminaries and fixing notation in section [ sc : setup ] , we move to the heart of the matter in section [ sc : mcf ] and prove the main existence theorem as well as many basic properties of mcfs , including a result on how the information of both the young measure and the h - measure can be extracted from an mcf .
we consider concrete oscillations and concentrations in section [ sc : osc_conc ] and give a few basic examples .
then , section [ sc : diff_constr ] examines how differential constraints on a sequence are reflected in the generated mcf and establishes a general compensated compactness theorem .
finally , section [ sc : appl ] concludes the paper by considering the applications mentioned above .
the author wishes to thank , among others , giovanni alberti , camillo de lellis , sebastian heinz , dorothee knees , jan kristensen , alexander mielke , ayman moussa , stefan mller , and richard nickl for stimulating discussions related to the topic of the paper , partly during the oberwolfach workshops on `` variational methods for evolution '' in december 2011 and on the `` calculus of variations '' in july 2012 .
by @xmath53 we denote the open ball around a point @xmath54 with radius @xmath55 ; the special notation @xmath15 will be used for the unit ball in @xmath5 .
the unit sphere in @xmath5 is @xmath14 .
we reserve the notation @xmath56 for the ( real ) unit sphere in @xmath57 .
the scalar product @xmath58 between two vectors @xmath59 will always be defined as @xmath60 and we write out any complex conjugation in the second variable . often , we use the space @xmath61 of all @xmath62-matrices ( or restricted to @xmath63 ) in place of @xmath5 .
we equip this matrix space with the hilbert structure generated by the scalar product @xmath64 , whereby we obtain the frobenius norm @xmath65 .
the open unit ball and unit sphere with respect to this norm are denoted by @xmath66 and @xmath67 , respectively . as usual , @xmath68 .
the space of all @xmath5-valued ( vector ) radon measures on a borel set @xmath69 is denoted by @xmath70 , the sets of positive radon measures and probability measures are @xmath71 and @xmath72 , respectively .
we use the lebesgue spaces @xmath73 , @xmath74 , @xmath75 , ( where @xmath76 and @xmath77 is a positive measure on @xmath78 ) , and the sobolev spaces @xmath79 , with their usual meanings ; the corresponding norms are @xmath80 , @xmath81 , .
continuous and continuously differentiable functions are contained in the spaces @xmath82 with norms @xmath83 .
the spaces `` @xmath84 '' and `` @xmath85 '' additionally entail that the contained functions vanish at the boundary / infinity or have compact support . in order to capture the behavior of functions defined on @xmath5 at infinity we need to consider a suitable compactification of @xmath5 .
the sphere compactification to be defined shortly , has turned out to be a good compromise between simplicity and generality and is used widely , see for example chapter 32 of @xcite or @xcite in an h - measure context and @xcite in the context of ( generalized ) young measures .
in contrast to the one - point compactification @xmath86 ( `` @xmath87 '' refers to the disjoint union ) it retains information about the `` directions of infinity '' . on the other hand , it is still sufficiently concrete to be useful in our analysis ( but more general compactifications up to the stone
ech compactification @xmath88 could in principle also be of interest ) .
formally , the * sphere compactification * @xmath89 of @xmath5 is the set @xmath90 with the topology such that @xmath91 is homeomorphic to the closed unit ball @xmath92 under @xmath93 we follow @xcite , where also proofs of the following facts can be found , and define the fourier transform and the inverse fourier transform of functions @xmath94 via @xmath95(\xi ) : = \int u(x ) { \mathrm{e}}^{-2 \pi{\mathrm{i}}x \cdot \xi } { \;\mathrm{d}}x , \qquad \xi \in { \mathbb{r}}^d , \\
\check{v}(x ) & = { \mathcal{f}}^{-1}[v](x ) : = \int v(\xi ) { \mathrm{e}}^{2 \pi{\mathrm{i}}x \cdot \xi } { \;\mathrm{d}}\xi , \qquad x \in { \mathbb{r}}^d.\end{aligned}\ ] ] we observe the following fact , which will be needed in the proof of lemma [ lem : osc ] : if @xmath96 is a tempered distribution on @xmath97 , define the * standing wave * @xmath98 with profile @xmath99 and direction @xmath100 by @xmath101 for its fourier transform @xmath102 we get @xmath103 for @xmath104 ( @xmath105 ) , that is , @xmath106 as a particular case we have that if @xmath107 , then @xmath108 , where @xmath109 is the line through the origin in direction @xmath110 .
for the proof consider without loss of generality the case @xmath111 and choose @xmath48 to be a tensor product of test functions .
then use the fact that the fourier transform of a tensor product is the tensor product of the fourier transforms and conclude by virtue of @xmath112 = \delta_0 $ ] .
let @xmath113 satisfy mihlin s condition @xmath114 for all multi - indices @xmath115 with @xmath116 and a constant @xmath117 .
define @xmath118 to be the smallest @xmath117 for which holds .
it follows from the mihlin multiplier theorem , see for instance theorem 5.2.7 in @xcite , that the corresponding multiplier operator @xmath119 : = { \mathcal{f}}^{-1}[\psi{\mathcal{f}}[u ] ] , \qquad u \in { \mathcal{s}}({\mathbb{r}}^d),\ ] ] has a bounded extension from @xmath120 to itself , where here and in the following @xmath22 , and @xmath121 with a dimensional constant @xmath122 .
define the set @xmath123 to contain all @xmath113 that additionally are positively @xmath39-homogeneous .
owing to this homogeneity , all such @xmath124 automatically satisfy the estimate with @xmath125 . also , @xmath118 is a norm on the space @xmath123 .
we will also continually use the space @xmath126 of @xmath127-valued multipliers , which we usually abbreviate by @xmath128 .
the associated multiplier operators @xmath129 , @xmath130 , are defined as before , but with @xmath131 acting on @xmath132 via matrix - vector multiplication ( if @xmath31 is a rapidly decaying function and by duality otherwise ) .
these matrix - multipliers satisfy the same boundedness assertions as before .
we consider scalar - valued @xmath133 to be part of @xmath126 , identifying @xmath124 with @xmath134 , where @xmath135 the @xmath136-identity matrix . alternatively
, one can view @xmath137 as the convolution ( or singular integral ) operator with the tempered distribution @xmath138 .
since @xmath139 is positively @xmath39-homogeneous , we may appeal to proposition 2.4.7 in @xcite to represent @xmath140 where @xmath141 and @xmath142 is given as the tempered distribution @xmath143 for a @xmath24-function @xmath78 on @xmath56 with vanishing integral .
the following result will be used countless times in the sequel : [ lem : commutation ] let @xmath144 and @xmath130 . define @xmath145 \bigr ] ,
\qquad v \in { \mathrm{l}}^2({\mathbb{r}}^d;{\mathbb{c}}^n),\ ] ] to be the multiplication and fourier multiplier operators , respectively
. then , the commutator @xmath146 : = m_{\varphi}t_\psi - t_\psi m_{\varphi},\ ] ] is smoothing of order 1 , that is , @xmath147 $ ] extends to a bounded operator from @xmath148 to @xmath149 for all @xmath22 .
in particular , @xmath147 $ ] is compact as an operator from @xmath149 to itself
. a direct proof can be found in lemma 1.7 of @xcite ( also see @xcite for a more advanced version with @xmath150 ) .
for smooth @xmath151 , the result is in fact well known in the theory of fourier multipliers and pseudo - differential operators , cf . 7.3 in chapter vi of @xcite . the present result can be reduced to that result by a smoothing argument .
another useful compactness result is the following : [ lem : bandlim_compact ] let @xmath152 and let @xmath76 be a bounded open set .
then , the fourier multiplier operator @xmath153 is compact from @xmath74 to @xmath154 , that is , @xmath155
\to t_\eta[u ] \quad\text{in $ { \mathrm{l}}^p_{\mathrm{loc}}({\mathbb{r}}^d;{\mathbb{c}}^n)$ } \qquad\text{whenever}\qquad u_j { \rightharpoonup}u \quad\text{in $ { \mathrm{l}}^p(\omega;{\mathbb{c}}^n)$.}\ ] ] we consider functions in @xmath74 to be implicitly extended by zero outside @xmath78 .
the operator @xmath156 can be written as a convolution , i.e.@xmath157(x ) = \bigl[\check{\eta } { \star}u \bigr](x ) = \int \check{\eta}(y ) u(x - y ) { \;\mathrm{d}}y , \qquad u \in { \mathrm{l}}^p(\omega;{\mathbb{c}}^n),\ ] ] with @xmath158 . from @xmath25 in @xmath73
it then immediately follows that @xmath159 \to t_\eta[u]$ ] pointwise and also that @xmath160\|}_\infty \leq { \|\check{\eta}\|}_\infty \cdot { \|u_j\|}_1 \leq c < \infty$ ] by minkowski s inequality .
standard theory further yields @xmath159 { \rightharpoonup}t_\eta[u]$ ] in @xmath149 and hence in @xmath161 . as the pointwise and the weak limit agree
, it follows that @xmath159 \to t_\eta[u]$ ] strongly in @xmath161 , see for instance theorem iv.8.12 in @xcite ( p. 295 ) . by the uniform boundedness of the sequence
, this implies convergence in @xmath154 as well .
we will also need : [ lem : tetar_conv ] let @xmath162)$ ] with @xmath163 and define @xmath164 for @xmath165 and any @xmath166 .
then , the fourier multipliers @xmath167 are uniformly @xmath168-bounded and @xmath169 \to u$ ] ( strongly ) in @xmath74 for all @xmath170 as @xmath171 .
this lemma follows immediately from @xmath172 = \check{\eta}_r { \star}u$ ] and the fact that @xmath173 is an approximation of the identity as @xmath171 , see for example section 1.2.4 in @xcite .
we remark that multipliers can also be defined for less smooth ( but not merely continuous ) symbols , see @xcite for a detailed study .
recall that a family @xmath174 , where @xmath175 is called * @xmath16-equiintegrable * if it is @xmath1-bounded and @xmath176 there are several other equivalent ways of characterizing @xmath16-equiintegrability , see for example theorem 2.29 in @xcite .
we here only record the following simple fact , which follows from hlder s inequality : [ eq : equiint_product ] let @xmath174 , @xmath175 , be a @xmath16-equiintegrable family and @xmath177 be uniformly @xmath178-bounded , where @xmath179 .
then , the family @xmath180 is ( @xmath181-)equiintegrable . besides `` horizontal '' test functions that are used to localize in space
, we also need our test functions to depend on the `` vertical '' argument , so that they allow us to localize in the target space as well .
spaces of such test functions necessarily depend on the exponent of the @xmath1-space , in which we are working .
first , for @xmath182 and functions @xmath183 , @xmath184 ( recall that @xmath15 denotes the unit ball in @xmath5 ) define @xmath185 it is easy to check that @xmath186 and @xmath187 .
now let @xmath22 and define the space @xmath188 of * test functions * , where @xmath76 is open , as follows ( it will become clear in section [ ssc : def ] why the definition is chosen as such ) : @xmath189 the norm of an element @xmath190 with @xmath37 is @xmath191 clearly , for every @xmath192 , the associated @xmath36 has a * ( strong ) @xmath193-recession function * @xmath194 notice that @xmath195 for all @xmath196 .
furthermore , @xmath197 and @xmath198 finally , we remark that for @xmath192 we can consider @xmath36 also as a function on the extended compactified space @xmath199 .
define this extended function @xmath200 as follows : @xmath201 one can then check that the assumptions on @xmath48 entail that @xmath200 is continuous ( with respect to the relevant topologies ) and @xmath202 .
in all of the following let @xmath22 and @xmath76 an open bounded set ( for unbounded sets @xmath78 a largely equivalent theory is available as long as all spaces are replaced by their respective local versions ) .
we will work with parametrized sesquilinear forms @xmath203 where as before @xmath204 .
we usually write the application of this sesquilinear form to the pair @xmath205 in the `` tensor '' notation @xmath206 let now @xmath207 be a positive finite borel measure . then , we say that the family @xmath208 is * weakly * measurable * with respect to @xmath209 if for every fixed @xmath130 and every fixed carathodory - type function @xmath210 with the properties that @xmath211 is @xmath209-measurable for all fixed @xmath212 and @xmath213 is continuous for all fixed @xmath214 , the compound function @xmath215 is @xmath209-measurable . for a family @xmath216
we define weak*-measurability in a completely analogous manner .
[ def : mcf ] let @xmath217 be a triple consisting of * a parametrized family @xmath218 of continuous sesquilinear forms @xmath219 * a positive and finite measure @xmath220 , and * a parametrized family @xmath221 of continuous sesquilinear forms @xmath222 also assume that @xmath223 satisfies the following properties : * the family @xmath224 is weakly * measurable with respect to @xmath225 and the family @xmath226 is weakly * measurable with respect to @xmath52 .
* for every fixed @xmath130 , the function @xmath227 lies in @xmath228 and the function @xmath229 ( with the function @xmath230 ) lies in @xmath231 .
then , @xmath223 is called a * @xmath16-microlocal compactness form ( mcf ) * on @xmath78 with target space @xmath5 .
the set of all such microlocal compactness forms is denoted by @xmath232 .
let @xmath233 , @xmath190 with @xmath37 , and @xmath130 .
then , we define the * duality product * @xmath234 if we want to stress the domain @xmath78 , we write @xmath235 . a justification for the above tensor - product notation will be provided in section [ ssc : ext_repr_distrib ] below .
notice that according to the properties of @xmath223 , we may also consider @xmath236 as a sesquilinear form on @xmath237 and then @xmath238 to state the main existence theorem for microlocal compactness forms , we first fix @xmath152 with @xmath163 .
set @xmath164 for @xmath165 and any @xmath166 .
it will be seen shortly that the theory is independent of the choice of @xmath239 .
[ thm : omega ] let @xmath2 be a norm - bounded sequence , where @xmath240 . then ,
after selecting a subsequence ( not relabeled ) , there exist a @xmath16-microlocal compactness form @xmath233 such that for all @xmath190 with @xmath37 and @xmath130 it holds that @xmath241 } { \;\mathrm{d}}x,\ ] ] where @xmath44 is the fourier multiplier operator with symbol @xmath242 .
moreover , the following properties hold : * * independence . *
the mcf @xmath223 is independent of the choice of @xmath239 and the choice of the sequence @xmath243 ( that is , every sequence @xmath244 with @xmath245 as @xmath246 gives the same result ) . * * basic estimate . * for all @xmath46 as above
it holds that @xmath247 where @xmath248 with a dimensional constant @xmath249 . *
* canonical form . *
it is possible to choose @xmath250 in the situation of the preceding theorem we say that the ( sub)sequence @xmath10 * generates * @xmath223 .
let us make a few remarks about the definition of mcfs : 1 .
the definition is tailored so that we can localize in both the @xmath214- and in the @xmath251-variable ( at least to some degree ) . at first sight
this appears to contradict the uncertainty principle for the fourier transform , but as we here only look at oscillations and concentrations that have asymptotically _ infinite _ frequency , the uncertainty principle does not apply ( as was already the case for h - measures ) .
2 . further to the first point , the purpose of the outer limit @xmath171 is precisely for the microlocal compactness form to `` forget '' finite frequencies .
this is important both from a conceptual and from a technical point of view , and exploited for example in lemma [ lem : phi_exchange ] below .
3 . for @xmath252-a.e .
@xmath253 , @xmath254 can also be considered as an anti - linear bounded function from @xmath128 to the space of @xmath5-valued radon measures on @xmath5 and likewise , for @xmath52-a.e .
@xmath255 , @xmath256 can be considered as an anti - linear bounded function from @xmath128 into the space of @xmath5-valued radon measures on the unit sphere @xmath14 .
so , mcfs can be considered as `` multiplier - parametrized '' measures .
while the family @xmath224 is easily seen to be uniquely determined , the choice of @xmath226 and @xmath52 is of course _ not _ unique : multiplying @xmath52 by any measurable function @xmath257 and then multiplying @xmath256 correspondingly by @xmath258 yields the same mcf .
however , in this paper we will always use the canonical choice , whereby also @xmath259 is determined @xmath52-almost everywhere . 5 . one can also define a local version of mcfs , where @xmath1 is replaced by @xmath260 . then
, @xmath52 is only @xmath261-finite and all estimates are only valid when restricted to compact subsets of @xmath78 ( with constants depending on that compact subset ) .
the above definition sheds some light on why we chose to work with test functions @xmath192 instead of using @xmath262 directly : in the above definition , we plug @xmath263}$ ] into the argument @xmath264 , so @xmath36 by itself is only `` half '' the test function .
also , it is our aim to investigate @xmath16-growth concentrations and @xmath36 only has @xmath193-growth .
7 . for the special case @xmath265 see sections [ ssc : ext_repr_distrib ] and [ ssc : mcf_ym_hm ] .
[ rem : omega_alt ] using parseval s formula , we can rewrite the integral in : @xmath266 } { \;\mathrm{d}}x & = \int_\omega t_{(1-\eta_r)\psi^ * } \bigl [ h({\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\,},u_j ) \bigr ] \cdot \overline{u_j } { \;\mathrm{d}}x \\ & = { \bigl\langle ( 1-\eta_r)\psi^ * { \mathcal{f}}\bigl [ h({\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\,},u_j ) \bigr ] , \overline{{\mathcal{f}}[u_j ] } \bigr\rangle},\end{aligned}\ ] ] where @xmath267 as usual . in the last line one of the fourier
transforms is applied to a function in @xmath268 with @xmath269 and one to a function in @xmath270 with @xmath271 so that @xmath272 , which ( unless @xmath273 ) is only defined as a tempered distribution .
the bracket @xmath274 therefore is to be understood in an appropriate distributional sense .
if additionally @xmath275 and @xmath276 with @xmath179 , we can further express this as @xmath277(\xi ) \cdot \overline{{\mathcal{f}}[u_j](\xi ) } { \;\mathrm{d}}\xi \label{eq : mcf_l2 } \\ & = \int_{{\mathbb{r}}^d } \int_\omega \int_\omega ( 1-\eta_r(\xi))\psi^ * \biggl(\frac{\xi}{{|\xi|}}\biggr ) h(x , u_j(x ) ) \cdot \overline{u_j(y ) } \ , { \mathrm{e}}^{2\pi{\mathrm{i}}(y - x ) \cdot \xi } { \;\mathrm{d}}y { \;\mathrm{d}}x { \;\mathrm{d}}\xi .
\label{eq : full_integral}\end{aligned}\ ] ] we divide the proof into several steps .
_ for any @xmath190 with @xmath37 , we have by hlder s inequality , , and the usual mihlin multiplier estimates , cf . , that @xmath278}\bigr| } { \;\mathrm{d}}x \notag\\ & \qquad \leq { \|t_{1-\eta_r}\|}_{{\mathrm{l}}^p\to{\mathrm{l}}^p } \cdot { \|t_\psi\|}_{{\mathrm{l}}^p\to{\mathrm{l}}^p } \cdot { \|f\|}_{{\mathbf{f}}^p } \cdot { \|(1+{|u_j|})^{p-1}\|}_{p/(p-1 ) } \cdot { \|u_j\|}_p \notag\\ & \qquad \leq c_{d , n } \max\{p,(p-1)^{-1}\ } \cdot { \|1+{|u_j|}\|}_p^p \cdot { \|f\|}_{{\mathbf{f}}^p } \cdot { \|\psi\|}_{\mathcal{m}}. \label{eq : omega_n}\end{aligned}\ ] ] it can be calculated that @xmath279 and hence this norm does not depend on @xmath280
. we will show below that @xmath239 can be chosen arbitrarily without changing the definition of mcfs , hence we can pick one and absorb @xmath281 into @xmath249 .
now take countable dense subsets @xmath282 , @xmath283 ( see the proof of lemma [ lem : countable_test ] below ) such that all @xmath284 are smooth , and use a diagonal construction to select a subsequence of @xmath10 ( not relabeled ) such that all the limits @xmath285 } { \;\mathrm{d}}x , \qquad m , n \in { \mathbb{n}},\ ] ] and @xmath286 exist . selecting another ( not relabeled ) subsequence if necessary , we may furthermore assume that @xmath287 we have for every @xmath288 , @xmath289 } { \;\mathrm{d}}x \\ & \qquad = \lim_{j\to\infty } \int h_m({\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\,},u_j ) \cdot \overline{t_{\psi_n}[u_j ] } { \;\mathrm{d}}x - \lim_{j\to\infty } \int h_m({\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\,},u_j ) \cdot \overline{t_{\eta_r\psi_n}[u_j ] } { \;\mathrm{d}}x \\ & \qquad = i_{m , n } - \int h_m \cdot \overline{t_{\eta_r\psi_n}[u ] } { \;\mathrm{d}}x,\end{aligned}\ ] ] where the last equality holds as @xmath290 is a compact operator by lemma [ lem : bandlim_compact ] . letting @xmath171 , we get by virtue of lemma [ lem : tetar_conv ] , @xmath291 } { \;\mathrm{d}}x = i_{m , n } - \int h_m \cdot \overline{t_{\psi_n}[u ] } { \;\mathrm{d}}x\ ] ] for any sequence @xmath171 .
this shows the independence from the choice of the sequence @xmath171 in ( a ) , at least for all @xmath292 , the general assertion then follows by density ( see below ) .
we next show that the value of the expression in does not depend on the choice of @xmath239 .
for two such choices @xmath293 we observe @xmath294 } { \;\mathrm{d}}x = \int_\omega h_m \cdot \overline{t_\psi t_{\zeta_r-\eta_r}[u ] } { \;\mathrm{d}}x,\ ] ] since @xmath295 is a convolution operator with a rapidly decaying function and so @xmath296 \to t_{\zeta_r-\eta_r}[u]$ ] strongly ( cf .
the proof of lemma [ lem : bandlim_compact ] ) . both @xmath239 and @xmath297 are identically @xmath181 on the unit ball @xmath298 .
thus , @xmath299 vanishes on the ball @xmath42 and so , the right hand side of tends to zero as @xmath171 by lemma [ lem : tetar_conv ] .
this establishes the second part of ( a ) .
we can now unambiguously define for all @xmath300 , @xmath301 } { \;\mathrm{d}}x.\ ] ] from we now immediately see for @xmath292 . by density
we can then extend this definition to all @xmath302 .
in particular , one can check , using an easy cauchy sequence argument , that the limits in the definition exist and are independent of the choice of @xmath239 and the sequence @xmath171 .
subsequently , ( b ) holds for all @xmath303 .
_ in the following we establish the properties ( i)(v ) from the definition of microlocal compactness forms .
fix @xmath130 , set @xmath304(x)}$ ] , and define the positive measures @xmath305 via @xmath306 by hlder s inequality and a multiplier estimate similar to , @xmath307 thus , for all @xmath192 , @xmath308 } { \;\mathrm{d}}x \notag\\ & \qquad = \int_\omega h({\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\,},u_j ) \cdot v_{j , r } { \;\mathrm{d}}x \notag\\ & \qquad = \int_\omega ( 1+{|u_j|})^{p-1 } ( s^{p-1}h)\biggl(x , \frac{u_j}{1+{|u_j| } } \biggr ) \cdot v_{j , r } { \;\mathrm{d}}x \notag\\ & \qquad = \int ( s^{p-1}h)(x , w ) \cdot r \,{\;\mathrm{d}}\mu_{j , r}^\psi(x , w , r ) .
\label{eq : mu_int}\end{aligned}\ ] ] now fix a sequence @xmath243 and select further subsequences such that @xmath309 as @xmath310 ( for all @xmath280 in the chosen sequence ) and @xmath311 as @xmath243 , where both limits are to be understood in @xmath312 . passing to the limit @xmath310 and then @xmath243 in the last expression , @xmath313 from the previous proof step ,
we also conclude that we in fact did not need to select any further subsequences above . in the following we suppress any dependence on @xmath49 , which for the moment we consider as fixed .
disintegrating @xmath314 , for instance using theorem 2.28 in @xcite , we infer the existence of @xmath315 and a family @xmath316 of probability measures on @xmath92 that is weakly * measurable with respect to @xmath317 , and is such that @xmath318 i.e. @xmath319 for all @xmath320 .
apply the disintegration theorem again to decompose @xmath321 for @xmath322 and a family @xmath323 of probability measures on @xmath92 that are weakly * measurable with respect to @xmath324 .
moreover , @xmath325 taking another @xmath326-subsequence if necessary , we can define @xmath327 let @xmath328 be relatively open in @xmath12 .
then , using the same fourier multiplier estimates as before and standard results in measure theory , we infer @xmath329 this estimate implies that @xmath324 is absolutely continuous with respect to @xmath209 ( note that the borel sets @xmath328 with @xmath330 generate the borel @xmath261-algebra on @xmath12 ) .
thus , by the radon nikodm theorem there exists a density @xmath331 such that @xmath332 denote furthermore the lebesgue decomposition of @xmath209 by @xmath333 combining all the previous arguments , we have decomposed @xmath334 as follows : @xmath335 define @xmath336 , @xmath337 and also let @xmath338 be defined through its action on @xmath339 as follows ( using the weak * measurability ) : @xmath340 then employ , and the fact that @xmath195 for all @xmath341 , to infer @xmath342 we will show next that in fact the last integral in vanishes . to this aim take any @xmath343 ( @xmath344 , @xmath345 ) with the property that @xmath346 and hence @xmath347 .
the family of functions @xmath348 is uniformly bounded and thus , by virtue of lemma [ eq : equiint_product ] , the functions @xmath349}\ ] ] are ( @xmath181-)equiintegrable ( with respect to both @xmath326 and @xmath280 ) .
hence , this family is @xmath350-weakly compact by the dunford pettis theorem .
let @xmath351 in @xmath352 as @xmath310 and @xmath243 ( in this order ) .
then , @xmath353 this has to equal the expression for @xmath354 and , differentiating the integral with respect to the domain ( varying @xmath355 ) , we get @xmath356 finally , varying @xmath357 over all of @xmath358 , we may conclude @xmath359 which shows the claim .
_ we now establish that one can replace @xmath209 by @xmath52 as defined in .
for this , observe that if @xmath262 is zero on a neighborhood of @xmath360 , we can replace @xmath361 by @xmath362 analogously to , we have for all @xmath192 , @xmath363 } { \;\mathrm{d}}x = \int \frac{(s^{p-1}h)(x , w)}{{|w|}^{p-1}}\cdot r \,{\;\mathrm{d}}\nu_{j , r}^\psi(x , w , r).\ ] ] now carry out the same procedure as in the first two steps of the proof ; note in passing that still holds with @xmath209 replaced by @xmath52 .
we define @xmath364 in the same way as @xmath365 . transforming @xmath366 via @xmath367
, we arrive at @xmath368 for different @xmath369 , @xmath370 of course .
we now compare this to for all integrands @xmath371 , where @xmath372 is positively @xmath193-homogeneous and @xmath373)$ ] with @xmath374 . letting @xmath375
, we get @xmath376 thus , we can just replace @xmath209 by @xmath52 and @xmath377 by @xmath370 in ; we drop the bar in the following .
_ so far the proof has established @xmath378 for all @xmath190 and a fixed @xmath130 .
we now re - introduce the dependence on @xmath49 and immediately observe that @xmath379 and @xmath380 must depend anti - linearly on @xmath49 . indeed
, the left hand side is anti - linear in @xmath49 by and the measure @xmath52 does not depend on @xmath49 by construction .
thus , we can in fact write @xmath381 where now for @xmath252-a.e . fixed @xmath253 , @xmath254 is an anti - linear bounded function from @xmath128 to the space of @xmath5-valued radon measures on @xmath15 ; for the boundedness we also used the estimate , whereby @xmath382 is finite almost everywhere with respect to @xmath52 . likewise , for @xmath52-a.e .
@xmath255 , @xmath256 is an anti - linear bounded function from @xmath128 into the space of @xmath5-valued radon measures on the unit sphere @xmath14 .
this shows properties ( i)(iii ) from the definition of mcfs .
property ( iv ) follows by tracing the weak * measurability properties of the disintegrated measures in the above decomposition , whereas ( v ) holds by construction .
this finishes the proof .
not every object satisfying definition [ def : mcf ] is generated by at least one sequence , see example [ ex : mcf_nogen ] for a counterexample .
we close this section by a remark about how _ not _ to define microlocal compactness forms : instead of , one might be tempted to define @xmath223 via @xmath383 { \;\mathrm{d}}x,\ ] ] where @xmath36 is assumed to have @xmath16-growth . unfortunately , @xmath384 in general is only in @xmath350 and @xmath44 is not a bounded operator from @xmath350 to itself ; this follows from the characterization of @xmath350-fourier multipliers as _ finite _ complex - valued measures ( see for example theorem 2.5.8 in @xcite ) .
thus , for a correct definition we need to weaken the concentration effects inside of @xmath129 and this is exactly what is accomplished by our definition .
this section collects a few elementary properties of mcfs , which will be used countless times in the sequel .
we start with the following two useful technical lemmas : [ lem : pq_conc ] let @xmath2 be a norm - bounded sequence that generates a @xmath16-microlocal compactness form @xmath385 , where @xmath76 is bounded and @xmath240 .
then , for all @xmath386 the sequence @xmath10 also generates an @xmath387-microlocal compactness form @xmath388 with @xmath389 furthermore , if @xmath10 is norm - bounded in @xmath390 , then @xmath391 has compact support for @xmath252-a.e .
@xmath253 and @xmath392 for @xmath393-a.e .
@xmath255 .
since @xmath74 embeds into @xmath394 on our bounded @xmath78 , the generation of @xmath395 follows directly from theorem [ thm : omega ] .
let @xmath396 and @xmath130 .
let us denote by @xmath397 the @xmath398-recession function of @xmath36 as in .
clearly , @xmath399 exists as well and is identically zero . as @xmath400
, we immediately get @xmath401 varying @xmath36 and @xmath49 , this implies the conclusions of the lemma .
the additional statement is evident from the definition since for @xmath37 such that @xmath402 , where @xmath403 , we have @xmath404 .
[ lem : countable_test ] there exists a countable set @xmath405 with @xmath406 and a countable set @xmath407 with @xmath408 , such that for any two mcfs @xmath409 , @xmath410 in view of the linearity and the fundamental estimate , it suffices to construct dense subsets @xmath405 and @xmath407 with the stated properties . for the second set we may simply use a smooth dense subset of @xmath411 . for the first set , we choose a countable dense subset @xmath412 and a countable dense subset @xmath413 consisting only of smooth functions .
then , define @xmath414 and set @xmath415 . the next result concerns
the choice of `` good '' generating sequences : [ lem : better_generation ] let @xmath2 generate @xmath233 .
then , there exists a sequence @xmath416 that also generates @xmath223 .
take countable families @xmath417 and @xmath407 as in lemma [ lem : countable_test ] .
for each fixed @xmath418 let @xmath419 with @xmath420 in @xmath74 and almost everywhere as @xmath421 ; this can be constructed by mollification .
we can furthermore require @xmath422 } { \;\mathrm{d}}x - \int_\omega h_m({\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\,},v_j^{(n ) } ) \cdot \overline{t_{(1-\eta_r)\psi_n}[v_j^{(n ) } ] } { \;\mathrm{d}}x\biggr| } \leq \frac{1}{j}\ ] ] whenever @xmath423 .
indeed , the preceding expression can be estimated by @xmath424 \cdot \overline{t_{(1-\eta_r)\psi_n}[u_j ] } \\ & \qquad\qquad + h_m({\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\,},v_j^{(k ) } ) \cdot \overline{t_{(1-\eta_r)\psi_n}[u_j - v_j^{(k ) } ] } { \;\mathrm{d}}x\biggr| } \\ & \qquad \leq c { \bigl\|h_m({\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\,},u_j)-h_m({\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\,},v_j^{(k)})\bigr\|}_{p/(p-1 ) } + c { \bigl\|t_{(1-\eta_r)\psi_n}[u_j - v_j^{(k)}]\bigr\|}_p.\end{aligned}\ ] ] the first term goes to zero as @xmath421 by pratt s theorem since @xmath420 a.e .
and we have the converging @xmath425-integrable majorants @xmath426 .
the second expression goes to zero by a standard multiplier estimate .
we conclude the proof via a standard diagonal argument .
next , the commutation lemma [ lem : commutation ] implies the following important technical result : [ lem : phi_exchange ] let @xmath2 generate @xmath233 .
then , for all @xmath190 with @xmath37 , all @xmath130 , and all @xmath144 , it holds that @xmath427 } { \;\mathrm{d}}x \\ & = \lim_{r\to\infty } \lim_{j\to\infty } \int_\omega h({\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\,},u_j ) \cdot \overline{t_{(1-\eta_r)\psi}[\overline{{\varphi } } u_j ] } { \;\mathrm{d}}x.\end{aligned}\ ] ] it suffices to show that @xmath428 - t_{(1-\eta_r)\psi}[\overline{{\varphi } } u_j]\bigr ] } { \;\mathrm{d}}x = 0.\ ] ] without loss of generality assume that @xmath429 in @xmath430 and that @xmath25 in the space @xmath74 .
this may involve taking a subsequence , but since we will prove the convergence to zero for all such subsequences , the result then also applies to the original sequence of @xmath326 s .
so , the commutation lemma [ lem : commutation ] ( which also holds with the symbol @xmath431 instead of @xmath49 ) implies @xmath432 - t_{(1-\eta_r)\psi}[\overline{{\varphi } } u_j ] \\ & \qquad = \overline{{\varphi } } t_{(1-\eta_r)\psi}[u_j - u ] - t_{(1-\eta_r)\psi}[\overline{{\varphi } } ( u_j - u ) ] \\ & \qquad\qquad + \overline{{\varphi } } t_{(1-\eta_r)\psi}[u ] - t_{(1-\eta_r)\psi}[\overline{{\varphi } } u ] \\ & \qquad \to 0 + \overline{{\varphi } } t_{(1-\eta_r)\psi}[u ] - t_{(1-\eta_r)\psi}[\overline{{\varphi } } u ] \qquad\text{as $ j \to \infty$.}\end{aligned}\ ] ] by lemma [ lem : tetar_conv ] , the remainder @xmath433 - t_{(1-\eta_r)\psi}[\overline{{\varphi } } u ] \bigr ] } { \;\mathrm{d}}x\ ] ] converges to zero as @xmath171 . the last result in particular implies : [ lem : locality ] let the sequences @xmath434 generate @xmath409 , respectively . if @xmath435 on an open subdomain @xmath436 then for all @xmath437 , @xmath190 , @xmath130 , @xmath438 and we write `` @xmath439 on @xmath440 '' . assume without loss of generality that @xmath441 , otherwise consider separately the regions where @xmath355 has one sign .
then , by the previous lemma , @xmath442 } { \;\mathrm{d}}x \\ & = \lim_{r\to\infty } \lim_{j\to\infty } \int_\omega \sqrt{{\varphi } } h({\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\,},u_j ) \cdot \overline{t_{(1-\eta_r)\psi}[\sqrt{{\varphi } } u_j ] } { \;\mathrm{d}}x \\ & = \lim_{r\to\infty } \lim_{j\to\infty } \int_\omega \sqrt{{\varphi } } h({\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\,},v_j ) \cdot \overline{t_{(1-\eta_r)\psi}[\sqrt{{\varphi } } v_j ] } { \;\mathrm{d}}x \\ & = \cdots = { \bigl\langle\hspace{-2.5pt}\bigl\langle { \varphi}f \otimes \overline{\psi},\omega_2 \bigr\rangle\hspace{-2.5pt}\bigr\rangle},\end{aligned}\ ] ] which proves the claim .
the following lemma will be used many times in the sequel : [ lem : eliminate_rlim ] let @xmath2 generate @xmath233 and also assume @xmath25 in @xmath74 .
then , for all @xmath190 with @xmath37 , and all @xmath130 , it holds that @xmath443 } { \;\mathrm{d}}x.\ ] ] we have @xmath444 } { \;\mathrm{d}}x \\ & = \lim_{j\to\infty } \int_\omega h({\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\,},u_j ) \cdot \overline{t_\psi[u_j - u ] } { \;\mathrm{d}}x\\ & \qquad + \lim_{r\to\infty } \lim_{j\to\infty } \int_\omega h({\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\,},u_j ) \cdot \overline{t_{(1-\eta_r)\psi}[u ] } { \;\mathrm{d}}x,\end{aligned}\ ] ] because @xmath445 \to t_{\eta_r\psi}[u]$ ] strongly in @xmath74 , see lemma [ lem : bandlim_compact ] . if @xmath446 is the weak limit of @xmath447 ( up to a subsequence ) , @xmath448 } { \;\mathrm{d}}x = \lim_{r\to\infty } \int_\omega h \cdot \overline{t_{(1-\eta_r)\psi}[u ] } { \;\mathrm{d}}x = 0,\ ] ] where we also used @xmath449 { \rightharpoonup}0 $ ] for @xmath171 , see lemma [ lem : tetar_conv ] , to conclude .
the final result in this section shows that @xmath450 is equivalent to strong convergence of any generating sequence , which expresses that the mcf quantifies the difference between weak and strong convergence .
[ lem : omega0_strong_conv ] let @xmath2 generate @xmath233 and also @xmath25 in @xmath74
. then @xmath451 strongly if and only if @xmath450 let @xmath190 and @xmath130 . if @xmath451 strongly in @xmath74 , then by the preceding lemma , @xmath452 } { \;\mathrm{d}}x = 0.\ ] ] now assume @xmath450 .
according to proposition [ prop : equiint ] in section [ ssc : conc ] below we have that @xmath10 is @xmath16-equiintegrable .
define @xmath453 and set @xmath454 further , by lemma [ lem : eliminate_rlim ] and the extended representation result stated in proposition [ prop : ext_repr ] in the next section , @xmath455 where @xmath456 is a term that vanishes as @xmath457 ; this comes from the @xmath16-equiintegrability of @xmath10 whereby the integrand in the first line is also equiintegrable and @xmath458 as @xmath457 .
thus , using the equiintegrability again , @xmath459 . in this section
we show how mcfs allow us to represent limits for a wider class of integrands , consider mcfs as bilinear forms , and we also briefly touch on the complicated topic of `` compound '' symbols , i.e. symbols that can not be written as a tensor product @xmath303 .
we first define the set of * representation integrands * @xmath460 in this context recall that a function @xmath461 is carathodory if @xmath462 is @xmath252-measurable for all fixed @xmath463 and @xmath464 is continuous for all fixed @xmath255 .
the main result for this class of integrands is that even if only @xmath465 , the representation of limits via the mcf still holds : [ prop : ext_repr ] let @xmath2 generate @xmath233 .
then , also for @xmath465 and @xmath130 it holds that @xmath466 } { \;\mathrm{d}}x \\ & \qquad = { \bigl\langle\hspace{-2.5pt}\bigl\langle f \otimes \overline{\psi},\omega \bigr\rangle\hspace{-2.5pt}\bigr\rangle } \\ &
\qquad = \int_\omega { \bigl\langle h(x,{\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\ , } ) \otimes \overline{\psi},\omega_x \bigr\rangle } { \;\mathrm{d}}x + \int_{{\overline{\omega } } } { \bigl\langle h^\infty(x,{\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\ , } ) \otimes \overline{\psi},\omega_x^\infty \bigr\rangle } { \;\mathrm{d}}\lambda_\omega(x).\end{aligned}\ ] ] the fact that the expression on the right hand side is well - defined and equals the left hand side can be proved by a strategy similar to the one employed in the proof of the corresponding result for generalized young measures , see proposition 2 in @xcite . next , we investigate how an mcf can be considered a bilinear form : we start with the hilbert space case @xmath265 , where the strongest assertions can be made ( as is to be expected ) : since all @xmath467-functions give rise to a bounded @xmath23-fourier multiplier , we get the following improved basic estimate ( cf . ):
@xmath468 and we can allow @xmath469 .
this shows that @xmath223 can be seen as a bilinear form ( see for the definition of @xmath200 ) @xmath470 note that we can not allow @xmath471 because there is no countable dense subset , which we need to construct a generic limit for all @xmath49 . by a similar argument for @xmath472 and using the embedding @xmath473 for bounded @xmath78
, we can furthermore consider @xmath223 as a bilinear form @xmath474 where @xmath475 note that here we need @xmath476-growth of @xmath36 , translating into the above condition on @xmath200 , in order for the integral in the definition of the mcf @xmath223 to be bounded . for @xmath477
, however , more regularity in @xmath49 beyond mere continuity seems to be necessary .
we now turn to the question of `` compound '' symbols .
let @xmath190 with @xmath37 .
for a sequence @xmath478 with the property that additionally @xmath276 for all @xmath418 ( @xmath179 ) , we can write @xmath479 } { \;\mathrm{d}}x \\ & = \lim_{r\to\infty } \lim_{j \to \infty } \int_{{\mathbb{r}}^d } \int_\omega \int_\omega ( 1-\eta_r(\xi))\psi^ * \biggl(\frac{\xi}{{|\xi|}}\biggr ) h(x , u_j(x ) ) \cdot \overline{u_j(y ) } \ , { \mathrm{e}}^{2\pi{\mathrm{i}}(y - x ) \cdot \xi } { \;\mathrm{d}}y { \;\mathrm{d}}x { \;\mathrm{d}}\xi,\end{aligned}\ ] ] see . one can therefore consider @xmath480 to be the `` symbol '' of the above double fourier integral ( similarly to the theory of pseudo - differential operators ) .
we can thus extend the class of admissible `` symbols '' @xmath303 to all @xmath481 of the form @xmath482 for @xmath33 a carathodory function and @xmath483 such that @xmath484 is continuous .
we call these symbols @xmath18 * compound symbols * and define @xmath485 the question whether one can define @xmath223 for even more general compound symbols depending on @xmath214 , @xmath212 , and @xmath486 , and how much smoothness is minimally necessary , is rather intricate . in this context
it is well known that for symbols of pseudo - differential operators some smoothness is indispensable , even for @xmath487-boundedness ; a classical example for this can be found in chapter vii of @xcite .
this necessary regularity also persists for the subclass of symbols that are positively @xmath39-homogeneous in @xmath251 , see theorem 3 and proposition 1 of @xcite .
a first , rather abstract , approach to such compound symbols in @xmath488 is through the theory of tensor products of banach spaces , see for example @xcite or chapter viii of @xcite . as it turns out there are many ways to define the tensor product of two banach spaces @xmath489 . for our purposes
the most relevant is the * projective tensor product * @xmath490 , which is the completion of @xmath491 with respect to the * projective crossnorm * @xmath492 with this largest reasonable crossnorm , @xmath490 is the smallest tensor product completion of @xmath491 that yields a banach space structure .
the projective tensor product @xmath490 has the special property that its dual @xmath493 is simply the space of bounded bilinear forms on @xmath494 and therefore is ideally suited to our purposes . in our concrete situation , we have @xmath495 , @xmath496 ( we use @xmath497 instead of @xmath49 because of the anti - linearity ) , and @xmath498 if it is at all possible to define the pairing between @xmath233 and a compound symbol @xmath499 for which @xmath500 , then , for reasons of consistency ( cf . )
, this pairing must be @xmath501 where @xmath502 is again the @xmath193-recession function , defined just like in .
then , the above discussion shows that @xmath18 is an admissible symbol if @xmath503 finally , we show how an mcf @xmath233 can be considered as a distribution , and this will give a second way to treat compound symbols in @xmath488 .
recall the definition of the sphere compactification @xmath89 of @xmath5 in section [ ssc : sphere_compact ] and that of @xmath200 in , and for the moment consider @xmath223 as a bilinear form taking as its arguments @xmath504 and @xmath505 .
of course , we may restrict to the subspace of such @xmath200 with infinitely many derivatives and then we can consider @xmath223 as a bilinear form @xmath506 here , @xmath507 means that @xmath508 , with one - sided derivatives on @xmath14 . in view of the schwartz kernel theorem ,
see for instance chapter v in @xcite or section 4.6 in @xcite , we infer that @xmath223 extends to a ( complex - valued ) distribution @xmath509 using the distribution @xmath510 , we can `` extend '' the representation to the symbol class @xmath511 .
this approach , however , has a significant drawback , namely that now some @xmath214-regularity is genuinely needed ( as can be seen from the proofs of the schwartz kernel theorem ) , so the interpretation of the duality pairing between a symbol and an mcf is less clear .
we noted in the introduction that the microlocal compactness form generated by a sequence contains ( essentially ) all the information of the young measure and the h - measure ( or microlocal defect measure ) . in this section
, we make these statements precise and give more or less explicit ways of extracting the young measure and h - measure . in the following
, @xmath78 is always a _ bounded _ lipschitz domain . * tartar s h - measure * @xcite ( also see @xcite for variants and historical information ) or * grard s microlocal defect measure * @xcite generated by a sequence @xmath512 with @xmath513 is the matrix - valued measure @xmath514 , where like for ordinary matrices @xmath515 and @xmath516 denote the row and column index , respectively , defined via @xmath517(\xi ) \cdot \overline{{\mathcal{f}}[{\varphi}_2 u_j^l](\xi ) \cdot \psi \biggl ( \frac{\xi}{{|\xi| } } \biggr ) } { \;\mathrm{d}}\xi\ ] ] for all @xmath518 with @xmath519 , all @xmath520 , and all @xmath521 . the fact that the h - measure is completely contained in the microlocal compactness form is essentially a triviality and follows directly from the definition by restricting to test functions of the form @xmath522 , taking into account remark [ rem : omega_alt ] , in particular , and lemma [ lem : eliminate_rlim ] .
thus , for @xmath523 ( or only @xmath524 , see the previous section ) we get @xmath525(\xi ) \cdot \overline{{\mathcal{f}}[{\varphi}_2 u_j^l](\xi ) \cdot \psi^k_l \biggl ( \frac{\xi}{{|\xi| } } \biggr ) } { \;\mathrm{d}}\xi \\ & = \lim_{j\to\infty } \ , \int_\omega { \mathcal{f}}[{\varphi}_1 u_j](\xi ) \cdot \overline{\psi \biggl ( \frac{\xi}{{|\xi| } } \biggr ) { \mathcal{f}}[{\varphi}_2 u_j](\xi ) } { \;\mathrm{d}}\xi \\ & = { \bigl\langle\hspace{-2.5pt}\bigl\langle { \varphi}\otimes \overline{\psi } , \omega \bigr\rangle\hspace{-2.5pt}\bigr\rangle},\end{aligned}\ ] ] where @xmath526 is the mcf generated by @xmath10 . in the same vein
, we can also write for @xmath527 and @xmath528 ( for notational simplicity we only consider the case @xmath529 ) @xmath530 where @xmath531 ( @xmath532 ) now is the h - measure associated with ( a subsequence of ) the sequence @xmath533 . for the correspondence of the mcf with the ( generalized ) young measure , let us first recall some basic facts and fix notation , see @xcite for details : the * ( generalized ) young measure * @xmath534 consists of the * oscillation measure * @xmath535 , which is a parametrized ( and weakly*-measurable ) family of probability measures on @xmath5 , the * concentration measure * @xmath536 , and the * concentration - angle measure * @xmath537 , which is a @xmath11-weakly * measurable family of probability measures on the unit sphere @xmath14 of @xmath5 . if the young measure @xmath538 is generated by a sequence @xmath2
, then we have the representation of limits ( cf .
theorem 7 in @xcite , and @xcite for the case @xmath539 ) @xmath540 for all @xmath541 with the property that ( cf . ) @xmath542 whereby @xmath19 is well - defined in the sense @xmath20 the classical young measure is just the oscillation part @xmath9 of the generalized young measure ; in this case the limit representation above only holds if @xmath543 is an equiintegrable sequence ( in this case @xmath544 ) . the fact that the ( generalized ) young measure can be computed from the mcf
is contained in the following proposition .
[ prop : mcf_ym ] let @xmath2 generate the microlocal compactness form @xmath233 and also the ( generalized ) young measure @xmath545
. assume further that @xmath25 .
then , the knowledge of @xmath223 and @xmath31 completely determines @xmath538 .
the fact that besides @xmath223 we also need to know @xmath546 ( @xmath255 ) is owed to the definition of mcfs being tailored to quantify the _ difference _ between weak and strong convergence ( or compactness ) . for instance , any strongly converging sequence generates the zero mcf , but obviously in general not the same young measure .
as recalled before , the generalized young measure @xmath538 allows us to compute the limits @xmath547 for all integrands @xmath18 as above .
it is well known in young measure theory that all these limits together also determine @xmath538 ( for the classical young measure we only need to test with @xmath548 , so that @xmath549 is an equiintegrable sequence ) .
we need to show that all these limits can be expressed through @xmath223 and @xmath31 .
let @xmath550 and take a cut - off function @xmath551 such that @xmath552
write @xmath553 \cdot \overline{(z - u(x ) ) } \\ & = : g_{\varepsilon}(x , z ) + h_{\varepsilon}(x , z ) \cdot \overline{(z - u(x))}.\end{aligned}\ ] ] then , @xmath554 and @xmath555 by proposition [ prop : ext_repr ] in conjunction with lemma [ lem : eliminate_rlim ] , we can express the second limit through @xmath223 as follows : @xmath556 for the first limit in we get , exploiting the continuity of @xmath18 in its second argument , @xmath557 for the second term we have , using proposition [ prop : ext_repr ] again , @xmath558 \cdot \overline{(u_j - u ) } { \;\mathrm{d}}x \\ & = : \lim_{j\to\infty } \int_\omega k_{\varepsilon}({\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\,},u_j ) \cdot \overline{(u_j - u ) } { \;\mathrm{d}}x\\ & = \phantom { : } { \bigl\langle\hspace{-2.5pt}\bigl\langle g_{\varepsilon}\otimes i , \omega \bigr\rangle\hspace{-2.5pt}\bigr\rangle},\end{aligned}\ ] ] where @xmath559 and @xmath560 .
consequently , @xmath561 let now @xmath434 with @xmath25 and @xmath562 both generate @xmath233 and the young measures @xmath563 .
then , the previous arguments show @xmath564 for all @xmath550 .
thus , @xmath565 .
this section considers a few standard examples and showcases techniques to explicitly calculate the generated microlocal compactness forms .
we also introduce a shorthand notation for the action of mcfs .
consider the prototypical oscillating sequence @xmath566 where @xmath567 , @xmath100 , @xmath568 is the indicator function of the interval @xmath569 , and @xmath570 denotes the largest integer below @xmath571 .
intuitively , this sequence oscillates between the values @xmath572 and @xmath573 , with volume fractions @xmath574 and @xmath575 , respectively , and in direction @xmath110 .
the following general principle about oscillations shows that all of these facts can be recovered from the mcf generated by the sequence @xmath10 : [ lem : osc ] let @xmath576 be @xmath181-periodic ( the `` profile function '' ) and assume that @xmath577 ( @xmath571 ) generates the homogeneous young measure @xmath578 .
then , the simple oscillation in direction @xmath100 , @xmath579 generates the microlocal compactness form @xmath580 \otimes \overline{\delta}_{\pm n_0 } \quad \in { \mathbf{mcf}}^2(\omega;{\mathbb{c}}^n),\ ] ] where @xmath581 is the average of @xmath99 over one period cell and @xmath582 .
this notation means that for all @xmath583 , @xmath130 , @xmath584 ( z - z_0 ) } { \;\mathrm{d}}\nu(z ) { \;\mathrm{d}}x.\ ] ] of course , the exponent @xmath273 here is chosen arbitrarily and can be replaced by any @xmath22 .
we also remark that the notation @xmath585 indicates that the action of our mcf on @xmath49 is anti - linear ( as opposed to a dirac measure which would be linear ) .
since all statements are of a local nature ( and we will indeed localize using a compactly supported cut - off function @xmath355 ) , we assume without loss of generality that @xmath586 . in general we are thus neglecting effects at the boundary @xmath587 , but it can be seen easily that the sequence @xmath10 does not concentrate at all ( @xmath588 ) , whence this is of no concern .
take any @xmath589 with @xmath590 , @xmath345 and @xmath130 from the collection exhibited in lemma [ lem : countable_test ] .
it suffices to show the assertion for such @xmath48 , @xmath49 . for @xmath591 define the cones ( see figure [ fig : osc_lem_proof ] ) @xmath592 now , for @xmath550 arbitrary , choose @xmath591 so small that @xmath593 next , take smooth cut - off functions @xmath594 with @xmath595 for @xmath596 , @xmath597 .
it always holds that @xmath598 .
we will now prove the relations @xmath599 for all @xmath600 with @xmath601 such that the fourier transforms @xmath602 have compact support .
an approximation argument then yields these relations for our original @xmath590 as well .
we first prove . by lemma [ lem : phi_exchange ] and parseval s relation we have @xmath603(\xi ) \cdot \overline{{\mathcal{f}}[{\varphi}_2 u_j](\xi ) } \cdot \zeta_r(\xi ) { \;\mathrm{d}}x,\end{aligned}\ ] ] where @xmath604 the remarks in section [ ssc : fourier ] about fourier transforms of standing waves entail that @xmath27 ( a tempered distribution ) has support in the line @xmath109 .
hence , @xmath605 = \hat{{\varphi}}_2 { \star}{\mathcal{f}}[u_j]$ ] is a @xmath24-function with support in a tubular neighborhood @xmath606 of the line @xmath109 ( @xmath607 ) .
the function @xmath297 has support in the set @xmath608 see figure [ fig : osc_lem_proof ] .
elementary geometry yields that for @xmath609 it holds that @xmath610 ( in fact , @xmath611 ) .
hence we may assume that @xmath280 is so large that @xmath612 does not meet @xmath613 $ ] for any @xmath326 .
thus , holds . for use similar arguments as before to see that @xmath614(\xi ) \cdot \overline{\psi \biggl(\frac{\xi}{{|\xi|}}\biggr ) { \mathcal{f}}[{\varphi}_2 u_j](\xi ) } \cdot \rho_\pm
\biggl(\frac{\xi}{{|\xi|}}\biggr ) ( 1-\eta_r(\xi ) ) { \;\mathrm{d}}\xi \\ & = \lim_{r\to\infty } \lim_{j\to\infty } \int_{{\mathbb{r}}^d } { \mathcal{f}}\bigl [ { \varphi}_1 g(u_j ) \bigr](\xi ) \cdot \overline{\psi(\pm n_0 ) { \mathcal{f}}[{\varphi}_2 u_j](\xi ) } \cdot \rho_\pm \biggl(\frac{\xi}{{|\xi|}}\biggr ) ( 1-\eta_r(\xi ) ) { \;\mathrm{d}}\xi \\ & \qquad + { \bigl\langle\hspace{-2.5pt}\bigl\langle { \varphi}g \otimes \overline{\rho_\pm \cdot ( \psi({\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\,})-\psi(\pm n_0 ) ) } , \omega \bigr\rangle\hspace{-2.5pt}\bigr\rangle}\end{aligned}\ ] ] and the last error term is of order @xmath615 by . a geometric argument analogous to the one employed for establishing shows that for @xmath326 sufficiently large , the support of @xmath605 $ ] is contained in @xmath616 and hence the factor @xmath617 in the last integral is equal to @xmath181 .
finally , apply parseval s relation and lemma [ lem : phi_exchange ] again to get @xmath618 } { \;\mathrm{d}}x + { \mathrm{\textup{o}}}({\varepsilon}).\ ] ] an application of lemma [ lem : eliminate_rlim ] and letting @xmath619 then yields . employing ,
, we have shown @xmath620 ( u_j - z_0 ) } { \;\mathrm{d}}x \\ & = \int_{{\mathbb{r}}^d } \int { \varphi}g(z ) \cdot \overline{\bigl [ \psi(+n_0)+\psi(-n_0 ) \bigr ] ( z - z_0 ) } { \;\mathrm{d}}\nu(z ) { \;\mathrm{d}}x,\end{aligned}\ ] ] where the last equality follows from the usual young measure theory , see section [ ssc : mcf_ym_hm ] .
we can now compute the mcf generated by the example sequence at the beginning of this section .
we stress that both the value distribution and the _ direction _ of the oscillation are reflected in the generated mcf .
[ ex : osc ] let @xmath621 be as in . by the preceding oscillation lemma the sequence @xmath10 generates the mcf @xmath622 \otimes \overline{\delta}_{\pm n_0 } \;\in\;{\mathbf{mcf}}^2(\omega;{\mathbb{c}}^n),\ ] ] where @xmath623 . for first - order laminates as in the last example
, we always encounter a characteristic component of the form @xmath624 . for higher - order laminates or general nested microstructures ,
this becomes more complicated .
the mcf corresponding to a second - order laminate is computed later in example [ ex:2nd_order_lam ] using proposition [ prop : lamination ] , which investigates higher - order laminations .
we will see then that the mcf also reflects the _ hierarchy _ of laminations at different scales .
one can extend lemma [ lem : osc ] to @xmath181-periodic profile functions @xmath625 ( possibly unbounded ) .
the result , with the space @xmath626 replaced by @xmath627 , and the proof remain largely the same except for the following two modifications : parseval s theorem only holds if we assume that all @xmath621 are of class @xmath628 , see lemma [ lem : better_generation ] . also , to make the usual young measure theory applicable one needs to use the fact that the family @xmath10 as defined in is equiintegrable . besides oscillations ,
mcfs also represent concentration phenomena .
the following is a simple example : [ ex : conc ] let @xmath629 have compact support and satisfy @xmath630 .
further , let @xmath631 with @xmath632 .
define @xmath633 then , @xmath10 generates the mcf @xmath634 where @xmath635 is the surface measure ( acting on @xmath497 ) @xmath636(t \eta ) \
, \overline{\hat{w}(t \eta ) } \ , t^{d-1 } { \;\mathrm{d}}t \biggr ) \ , \bigl ( { \mathcal{h}}^{d-1 } { \begin{picture}(10,8)\put(2,0){\line(0,1){7}}\put(1.8,0){\line(1,0){7}}\end{picture}}{\mathbb{s}}^{d-1 } \bigr ) ( { \mathrm{d}}\eta).\ ] ] the above expression for @xmath223 is a shorthand notation for the mcf that acts on @xmath192 , @xmath130 , as @xmath637 we only sketch the proof : let @xmath638 be as before and consider @xmath639 } { \;\mathrm{d}}x \\ & \phantom{:}= \int h \bigl ( x , z_0 \ ,
j^{d / p } w(jx ) \bigr ) \cdot \overline{t_{(1-\eta_r)\psi } \bigl [ z_0 \ , j^{d / p } w(jx ' ) \bigr](x ) } { \;\mathrm{d}}x \\ & \phantom{:}\approx \int h \bigl ( 0,z_0 \ , j^{d / p } w(jx ) \bigr ) \cdot \overline{t_{(1-\eta_r)\psi } \bigl [ z_0 \ , j^{d / p } w(jx ' ) \bigr](x ) } { \;\mathrm{d}}x\end{aligned}\ ] ] for @xmath326 large , where `` @xmath640 '' means that the error vanishes as @xmath641 , @xmath171 .
here we used the fact that the support of @xmath642 shrinks to zero .
then , we employ @xmath643(x ) = t_{(1-\eta_{r / j})\psi}[w](jx)$ ] to calculate further @xmath644 } { \;\mathrm{d}}y \\ & \approx \int h^\infty(0,z_0 ) { |w|}^{p-1 } \cdot \overline{t_{(1-\eta_{r / j})\psi}[z_0 w ] } { \;\mathrm{d}}y.\end{aligned}\ ] ] for the last expression we used that @xmath645 for almost every @xmath646 and then in @xmath647 since we have a uniform @xmath647-majorant .
finally , use parseval s formula to conclude @xmath648 } { \;\mathrm{d}}y \\ & = h^\infty(0,z_0 ) \cdot \int { \mathcal{f}}[{|w|}^{p-1}](\xi ) \cdot \overline{\psi(\xi)z_0 \hat{w}(\xi ) } { \;\mathrm{d}}\xi \\ & = h^\infty(0,z_0 ) \cdot \int { \mathcal{f}}\bigl [ { |w|}^{p-1 } \bigr](\xi ) \ , \overline{\hat{w}(\xi ) } \cdot \overline{\psi(\xi ) z_0 } { \;\mathrm{d}}\xi.\end{aligned}\ ] ] transforming to polar coordinates , we arrive at .
we also prove the following interesting converse to lemma [ lem : pq_conc ] : [ prop : equiint ] let @xmath2 generate @xmath233 .
then , the following are equivalent : * @xmath10 is @xmath16-equiintegrable .
* @xmath649 for @xmath52-a.e
. @xmath255 . *
@xmath650 for @xmath52-a.e .
@xmath255 .
we first show ( i ) @xmath651 ( ii ) .
let @xmath652 , set @xmath653 and observe @xmath654 .
then define @xmath655 through @xmath656 where @xmath657 , and for which we may compute @xmath658 .
using lemma [ lem : eliminate_rlim ] we have @xmath659 we claim that @xmath660 this is equivalent to the equiintegrability of the functions @xmath661 , which can be seen as follows : @xmath662 is uniformly bounded in @xmath663 , thus , up to a subsequence , @xmath664 . in particular , @xmath665 weakly in @xmath352 .
hence , @xmath661 is weakly relatively compact in @xmath352 and consequently , the dunford
pettis criterion implies that this sequence is equiintegrable and holds .
thus , using an analogous argument for the upper bound , we arrive at @xmath666 and together with @xmath667 the equivalence of ( i ) and ( ii ) follows .
since ( iii ) clearly implies ( ii ) , it remains to show ( i ) @xmath668 ( iii ) .
for this , let @xmath192 and by the @xmath16-equiintegrability of @xmath10 take @xmath652 so large that @xmath669 thus , if we define @xmath670 with @xmath671 defined as before , then @xmath672 and letting @xmath375 while employing @xmath673 , we arrive at @xmath674 varying @xmath36 and @xmath49 , the sought implication follows .
in this section we investigate how differential constraints on the generating sequence are reflected in the corresponding microlocal compactness form .
we mostly work in the full - space case @xmath586 , which will suffice for our later aims ( but see remark [ rem : domain ] below for the case @xmath675 ) .
we let @xmath676 and @xmath677 be the fourier multiplier operators with symbols @xmath678 and @xmath679 , respectively . then
, @xmath680 is the space of all distributions @xmath681 in @xmath57 such that @xmath682 ; as norm we use @xmath683 . in turn , @xmath684 is defined to be the dual space to @xmath685 , where @xmath179 .
let @xmath686 be a * linear constant - coefficient partial differential operator of order @xmath687 * , @xmath688 where we have employed the usual multi - index notation .
a function @xmath170 with @xmath689 in the @xmath690-sense , that is @xmath691 for all @xmath692 , where @xmath179 and @xmath693^ * \partial^\alpha$ ] , is called * @xmath686-free*. the operator @xmath686 has the * symbol * @xmath694 moreover , we define the * bounded symbol * @xmath695 and the * homogeneous symbol * @xmath696 as follows : @xmath697 the corresponding operators @xmath698 and @xmath699 are bounded from @xmath1 to itself by the usual multiplier theorems and we have the identities @xmath700 which we will use to factor @xmath686 into an @xmath1-bounded and a differential part . as an example , consider the operator @xmath701 ; then @xmath702 is the @xmath326th riesz transform , i.e. the multiplier operator with symbol @xmath703 , cf . section 4.1.4 in @xcite
. we will sometimes assume the following * constant - rank property * on @xmath686 , which was introduced by murat @xcite ( also see @xcite ) : @xmath704 it is well - known that many useful operators satisfy this constant - rank property , among them @xmath705 , @xmath706 , the static maxwell operator @xmath707 , and the operator corresponding to the constraints @xmath708 , @xmath709 ( an example by tartar ) .
when we impose differential constraints on sequences @xmath710 with @xmath513 , we will do so in the usual way in compensated compactness theory @xcite , i.e. we require that @xmath711 using the decomposition @xmath712 we get @xmath713 hence , is equivalent to @xmath714 we now prove that linear pde constraints on a sequence are reflected in the generated mcf in a simple way : [ thm : diff_omega ] let @xmath710 generate @xmath715 , let @xmath686 be a linear constant - coefficient pde operator of order @xmath687 ( not necessarily satisfying the constant - rank property ) , and @xmath716 then , `` @xmath717 '' in the sense that @xmath718 for all @xmath190 and all @xmath719 . here , @xmath49 has a different signature , because @xmath696 takes values in @xmath720 .
an @xmath223 satisfying is called * @xmath686-free*. in the special case that @xmath721 , i.e.@xmath722_{i , j , k } , \qquad \text{$w \colon { \mathbb{r}}^d \to { \mathbb{r}}^{m \times d}$}\ ] ] ( @xmath723 , @xmath724 ) we also call @xmath223 a * gradient mcf*. since @xmath725 and @xmath726 are asymptotically the same as @xmath727 , we have @xmath728 } { \;\mathrm{d}}x & = \int
h({\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\,},u_j ) \cdot \overline{t_{(1-\eta_r ) \psi { \mathbb{a}}_b}[u_j ] } { \;\mathrm{d}}x + e(r ) { \|u_j\|}_p \\ & = \int h({\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\,},u_j ) \cdot \overline{t_{(1-\eta_r ) \psi}[{\mathcal{a}}_b u_j ] } { \;\mathrm{d}}x + e(r ) { \|u_j\|}_p\end{aligned}\ ] ] and the error term @xmath729 vanishes as @xmath243 . using , the last expression converges to zero as @xmath310 and @xmath243 ( in this order ) .
[ rem : domain ] our results are not directly applicable when @xmath675 , because we employed the fourier characterization of @xmath686-freeness , which is only valid in the whole space .
if we want to work in an arbitrary lipschitz domain @xmath76 , there are two options : 1 . if @xmath686 is a first - order operator , then under the additional assumption of the natural boundary condition @xmath730 the preceding result holds in @xmath78 as well , because then @xmath689 in @xmath78 is equivalent to @xmath689 in @xmath57 ( both in the @xmath690-sense ) .
2 . assume that for all @xmath731 there exists an @xmath1-bounded extension @xmath732 such that @xmath733 , @xmath734 , @xmath735 has compact support , and @xmath736 in @xmath57 whenever @xmath689 in @xmath78 .
then it can be seen that the previous result continues to hold . for example , if @xmath721 , such an extension is available and can be constructed via the potential . the following is a converse to theorem [ thm : diff_omega ] in the @xmath23-case : [ thm : diff_genseq ] let @xmath737 be generated by a sequence @xmath738 with @xmath25 in @xmath739 .
furthermore , let @xmath686 be a linear constant - coefficient pde operator of order @xmath687 ( not necessarily satisfying the constant - rank property ) and assume that @xmath740 which is in particular implied by the condition , and also that @xmath689 ( in the @xmath741-sense ) . then
, @xmath742 before we come to the proof , a few remarks are in order : 1 . the fact that @xmath223 is generated by _
some _ sequence is a genuine assumption , see example [ ex : mcf_nogen ] below .
the simplicity of the expression of pde - constraints should be compared to the situation for ( generalized ) young measures , where the characterization problem is in general a difficult one . for classical young measures , this is the objective of the celebrated kinderlehrer pedregal theorem @xcite ; the recent extension to generalized young measures can be found in @xcite .
remarkably , @xmath743 in @xmath741 holds for _ all _ generating sequences .
this is a much stronger conclusion than what can be obtained in young measure theory , where for example a gradient constraint on the young measure in the sense of the aforementioned kinderlehrer
pedregal theorem is far from implying that all generating sequences ( asymptotically ) consist only of gradients .
in fact , generically there also exist divergence - free generating sequences for gradient young measures , and divergence - free sequences are in a sense `` orthogonal '' to sequences of gradients .
the conclusion is equivalent to @xmath744 in @xmath745 , and so we compute using parseval s theorem , @xmath746 } { \;\mathrm{d}}x + \int u_j \cdot \overline{t_{\eta_r { \mathbb{a}}_b^ * { \mathbb{a}}_b}[u_j ] } { \;\mathrm{d}}x.\end{aligned}\ ] ] now let first @xmath310 and then @xmath243 to arrive via lemmas [ lem : bandlim_compact ] , [ lem : tetar_conv ] at @xmath747 } { \;\mathrm{d}}x.\ ] ] the first term is zero by and the second term is equal to @xmath748 and hence is also zero by assumption .
[ rem : afree_proj ] if @xmath686 satisfies the constant - rank property , we can project the elements of a generating sequence @xmath10 with @xmath743 in @xmath749 onto the set of @xmath686-free functions , see lemma 2.14 in @xcite , to obtain a generating sequence that is actually @xmath686-free and not just asymptotically @xmath686-free .
we now give two examples for the preceding results : [ ex : osc_curl0 ] consider again the situation of example [ ex : osc ] , but let now @xmath750 . as we saw in that example
, the laminar oscillation between @xmath572 and @xmath573 in direction @xmath100 generates the mcf @xmath622 \otimes \overline{\delta}_{\pm n_0}.\ ] ] take @xmath721 as in .
then , condition for @xmath223 translates into @xmath751 where @xmath752 denotes the summation over both signs for @xmath753 .
one can compute @xmath754 and hence condition is true if @xmath755 for some @xmath756 .
on the other hand , from for @xmath757 and @xmath758 , we get that @xmath743 in @xmath690 implies @xmath759 whence holds .
thus , from theorems [ thm : diff_omega ] , [ thm : diff_genseq ] we conclude that @xmath223 is a gradient mcf if and only if @xmath760 as expected , this is a generalization of the usual hadamard jump condition requiring rank - one connectedness across the jump surface ( in this context also see @xcite ) .
again considering the situation of example [ ex : osc ] , but now with @xmath761 and @xmath762 , we get by a similar derivation that @xmath763 if and only if @xmath764 we end this section with an example showing that certain mcfs can not be generated by _ any _ sequence .
this raises the problem of characterizing the subclass of mcfs that are generated by at least one sequence .
we reserve this for future work .
[ ex : mcf_nogen ] let @xmath765 and consider the mcf @xmath766.\ ] ] a quick way to see that this @xmath223 is not generated by any sequence is as follows : theorem [ thm : diff_genseq ] implies that any generating sequence asymptotically would have to be a sequence of gradients .
however , the corresponding young measure is @xmath767 , which is not a gradient young measure .
this follows for example from the kinderlehrer
pedregal theorem @xcite , because @xmath768 . in this section
we introduce a notion of wavefront set of microlocal compactness forms @xmath233 : recall the sphere compactification @xmath89 from section [ ssc : sphere_compact ] and also the extension construction for @xmath200 in .
the * ( @xmath16-compactness ) wavefront set * @xmath769 is the smallest closed subset @xmath572 of @xmath770 with the property that for any @xmath771 with @xmath344 , @xmath345 ( @xmath772 ) , and any @xmath130 , @xmath773 it can be seen that if @xmath774 , then @xmath450 .
since @xmath223 can also be seen as a distribution , see section [ ssc : ext_repr_distrib ] , it is worth pointing out that the wavefront set considered here is different from the one considered by hrmander , cf .
chapter viii of @xcite , applied to this distribution .
another notion with related aims , the `` @xmath23-compactness wavefront set '' , was introduced by grard in @xcite .
the oscillatory mcf @xmath622 \otimes \overline{\delta}_{\pm n_0}\ ] ] from example [ ex : osc ] has the wavefront set @xmath775 a concentration effect is expressed in the mcf @xmath776 from example [ ex : conc ] . for its wavefront
set we get @xmath777 where the support of @xmath635 , which is given in , depends on the cancellation properties of the profile function @xmath99 s fourier transform .
the following lemma describes the effect of pointwise constraints on the wavefront set : [ lem : wf_pointwise_constr ] let @xmath2 generate @xmath233 and let @xmath778 for all @xmath253 satisfy the _ continuity property _ @xmath779
if the pointwise constraint @xmath780 holds , then @xmath781 let @xmath771 and @xmath130 be such that @xmath782 then , @xmath783 } { \;\mathrm{d}}x = 0,\ ] ] because @xmath784 .
the interplay between pointwise and differential constraints , which is related to compensated compactness theory , is more complicated and will be treated in the next section .
our notion of the @xmath16-compactness wavefront set is named in analogy to the classical notion from microlocal analysis , as introduced by hrmander , see for example in chapter viii of @xcite .
however , a @xmath16-mcf measures the difference between weak and strong @xmath1-compactness and not the directions where @xmath24-regularity fails , like the classical wavefront set .
of course , our @xmath16-compactness wavefront set contains more information than a mere analogue of the classical wavefront set , because the asymptotic value distribution of the generating sequence is also preserved .
we remark in passing that in classical microlocal analysis there is no object corresponding to the full mcf and the quantitative information it contains .
the general philosophy of compensated compactness is to employ additional properties of sequences to improve weak to strong compactness .
we here take a slightly broader view and also consider mere restrictions on the direction and value distribution of admissible oscillations and concentrations as `` compensated compactness '' .
we will focus on the interplay between differential and pointwise constraints and how it affects the wavefront set ( and thus compactness ) .
let @xmath686 be a linear constant - coefficient pde operator of order one as before .
from now on we always assume the constant - rank property .
our basic setup follows the general framework of compensated compactness as introduced by tartar @xcite : let @xmath2 be a sequence satisfying the * differential inclusion * @xmath785 it turns out that this formulation is very useful for a variety of problems , for instance in the theory of conservation laws @xcite , rigidity theory @xcite , and fluid mechanics , see @xcite , where the euler equation is written in this form ( a related reference investigating young - measure solutions to the euler equation is @xcite ) .
the aim now is to investigate how we can use mcfs to efficiently exploit the conditions above in our study of weak strong compactness .
having already considered pointwise and differential constraints in isolation , we now aim to analyze the interaction between them .
the following theorem is a general compensated compactness result : [ thm : cc_general ] let @xmath710 generate the microlocal compactness form @xmath715 and let @xmath686 be a constant - coefficient pde operator of order one satisfying the constant - rank property .
we furthermore assume : * @xmath743 in @xmath786 . *
let @xmath787 for all @xmath255 with the property that the orthogonal projection @xmath788 onto @xmath789 is smooth in @xmath214 .
* @xmath790 as @xmath310 .
then , @xmath791 in particular , if @xmath792 , then @xmath450 and @xmath10 is strongly compact .
a useful special case is : [ cor : cc_simple ] let @xmath710 generate @xmath715 and let @xmath686 be as in theorem [ thm : cc_general ] . moreover , assume @xmath793 where @xmath794 is closed .
then , @xmath795 in particular , if @xmath792 , then @xmath450 and @xmath10 is strongly compact .
one can also interpret this result as a statement about `` ellipticity '' : the directions @xmath251 that do not feature in @xmath796 are the directions for which the system `` @xmath717 , @xmath797 '' is elliptic , cf .
@xcite for more on this kind of ellipticity .
the reason why we require the constant - rank property is that we need to ensure that projections onto @xmath798 are continuous , which necessitates that this space has constant dimension as a function of @xmath251 .
not much is known in the situation where the constant - rank property is violated , but see @xcite for a situation in which the constant - rank property is violated , but one can still obtain a result . for all @xmath799 with @xmath800
we denote by @xmath801 the non - orthogonal projection with @xmath802 using some linear algebra in conjunction with ( ii ) , from which it immediately follows that @xmath803 , and also employing the constant - rank property of @xmath686 , one shows @xmath804 moreover , @xmath805 is smooth in @xmath806 and positively @xmath39-homogeneous in @xmath251 ( since @xmath696 is ) .
let @xmath589 and @xmath130 with @xmath807 for some @xmath652 , where @xmath808 we will show in the sequel that @xmath809 proving the theorem . take a smooth cut - off function @xmath810)$ ] such that @xmath811 for all @xmath812 and all @xmath813 with @xmath814 , as well as @xmath815 whenever @xmath816 for any @xmath817 .
we consider @xmath818 to be extended to all @xmath819 by positive @xmath39-homogeneity .
let @xmath820 be the pseudodifferential operator with _ compound _
symbol @xmath821 , i.e. for @xmath822 , @xmath823(x ) = \int \!\ !
\int \rho(y,\xi)\psi(\xi){\mathbb{h}}(y,\xi ) u(y ) \ , { \mathrm{e}}^{2\pi(x - y)\cdot\xi } { \;\mathrm{d}}y { \;\mathrm{d}}\xi.\ ] ] then , our assumptions imply that @xmath820 extends to a bounded operator from @xmath1 to itself , see chapter vi of @xcite ( in particular section vi.6 , which entails that @xmath820 is also a classical pseudodifferential operator ) .
we further assume without loss of generality that @xmath25 in @xmath149 .
then , @xmath824 } { \;\mathrm{d}}x \notag\\ & \qquad = \int g(u_j ) \cdot \overline{t_{(1-\eta_r)\psi}[{\varphi}u_j ] } { \;\mathrm{d}}x + e_1(j , r ) \notag\\ & \qquad = \int g(u_j ) \cdot \overline{t_{(1-\eta_r)\rho\psi}[{\varphi}u_j ] } { \;\mathrm{d}}x + e_1(j ,
r ) \notag\\ & \qquad = \int { \varphi}g(u_j ) \cdot \overline{t_{(1-\eta_r)\rho\psi}[u_j ] } { \;\mathrm{d}}x + e_1(j , r ) + e_2(j , r ) \notag\\ & \qquad = \int { \varphi}g(u_j ) \cdot \overline{t_{(1-\eta_r)\rho\psi{\mathbb{h}}}[u_j ] } { \;\mathrm{d}}x + e_1(j , r ) + e_2(j , r ) + e_3(j , r ) \notag\\ & \qquad = \int { \varphi}g(u_j ) \cdot \overline{t_{(1-\eta_r)\rho\psi{\mathbb{h}}}[{\mathbb{g}}u_j ] } { \;\mathrm{d}}x \label{eq : gcc } \\ &
\qquad\qquad + e_1(j , r ) + e_2(j , r ) + e_3(j , r ) + e_4(j , r ) \notag\\ & \qquad = 0 + e_1(j , r ) + e_2(j , r ) + e_3(j , r ) + e_4(j , r ) \notag\end{aligned}\ ] ] here , the last equality follows because @xmath825 by construction . if we can show that @xmath826 then @xmath827 and the conclusion of the theorem follows . for @xmath828
, @xmath829 this is clear by an argument analogous to the proof of lemma [ lem : phi_exchange ] and for @xmath830 it follows from assumption ( iii ) .
concerning @xmath831 , we define the linear map @xmath832 , @xmath833 , by @xmath834 from our assumptions we get that the symbol @xmath835 is smooth and positively @xmath39-homogeneous in @xmath251 .
hence , @xmath836 is an @xmath837-bounded pseudodifferential operator , and by construction @xmath838 therefore , @xmath839
\bigr\|}_p & = \lim_{j\to\infty } { \bigl\|t_{{\mathbb{j}}{\mathbb{a}}_0}[u_j]\bigr\|}_p = \lim_{j\to\infty } { \bigl\|t_{{\mathbb{j}}{\mathbb{a}}_b}[u_j]\bigr\|}_p \\ & = \lim_{j\to\infty } { \bigl\|t_{{\mathbb{j}}}[{\mathcal{a}}_b u_j]\bigr\|}_p = 0,\end{aligned}\ ] ] because @xmath744 in @xmath840 .
thus , @xmath841 .
this finishes the proof .
we point out that the preceding results still hold in @xmath675 under the same assumptions as in remark [ rem : domain ] . for the second part of that remark , in
we additionally need to use an argument based on lemma [ lem : commutation ] to ensure that @xmath842 is only applied to @xmath843 with @xmath844 .
then , we only get the result for the * restricted wavefront set * : @xmath845 we can put the preceding result to use in the following examples , which are modeled after the situation occurring in blow - up proofs in the calculus of variations , see for example @xcite : [ ex : cc_gradients ] let @xmath846 with @xmath847 for a fixed matrix @xmath848 .
the latter condition here means @xmath849 then , the following conclusions follow from corollary [ cor : cc_simple ] : * if @xmath850 , then any mcf @xmath851 generated by a subsequence of @xmath10 must satisfy @xmath852 where @xmath853 but the set on the right hand side is empty since @xmath854 thus , @xmath450 and @xmath10 is strongly compact ; this result is of course well - known , cf .
lemma 2.7 in @xcite and also section 3.2 in @xcite . *
if @xmath855 for @xmath856 , @xmath100 , then we can still obtain some information : @xmath857 so , while in this case we can not improve @xmath25 to strong convergence , at least we know that oscillations and concentrations can only occur `` in direction @xmath110 '' .
there exists a second - order constant - rank operator @xmath686 acting on @xmath858-valued vector fields that characterizes symmetric gradients , i.e. @xmath859 see example 3.10 ( e ) in @xcite . applying our preceding compensated compactness result to the situation of a sequence @xmath860 with @xmath861 for a fixed matrix @xmath862 .
it turns out that @xmath863 then , if @xmath10 generates an mcf @xmath864 : * if @xmath865 can not be written as a symmetric tensor product , then @xmath450 and @xmath10 is strongly compact . *
if @xmath866 for some @xmath867 and @xmath868 , then @xmath869 this means that oscillations and concentrations can only occur in direction @xmath870 _ or _ @xmath871 and reflects the inherent rigidity in the system , also see @xcite , where a similar rigidity was exploited to prove weak * lower semicontinuity of integral functionals in the space bd of functions of bounded deformation .
let @xmath872 and take @xmath686 to be the constant - rank operator corresponding to tartar s example @xmath873 for which ( @xmath874 ) @xmath875 again consider the situation @xmath876 where @xmath877 is fixed . then : * if @xmath878 , then @xmath450 and @xmath10 is strongly compact . *
if @xmath879 , then @xmath880
after having developed aspects of the general theory of microlocal compactness forms , we now move to present several applications . the first application concerns laminates ( nested microstructures ) .
it will turn out that mcfs not only contain the value distribution and the direction of the various oscillations , but also reflect the hierarchical structure of laminates .
as seen in the previous section , they furthermore allow one to quickly read off which differential constraints ( e.g. curl - freeness ) are satisfied by generating sequences .
first , we need the following definition : @xmath233 is called * homogeneous * if @xmath881 are constant in @xmath214 and @xmath882 for a constant @xmath883 .
equivalently , this is the case if for all @xmath190 of the form @xmath884 and all @xmath130 it holds that @xmath885 for instance , all mcfs generated by the simple ( one - directional ) oscillations in the oscillation lemma [ lem : osc ] are homogeneous .
one of the properties of a homogeneous mcf @xmath223 is that its ( lipschitz ) domain of definition can be chosen arbitrarily and we simply write @xmath886 . a proof sketch of this fact is as follows :
let @xmath2 generate a homogeneous mcf @xmath233 and for simplicity assume @xmath513 .
now , let @xmath887 be another lipschitz domain . choose a vitali cover @xmath888 of @xmath252-almost all of @xmath440 consisting of scaled copies of @xmath78 , where @xmath889 and @xmath890 .
define @xmath891 by @xmath892 since @xmath893 , lemma [ lem : eliminate_rlim ] and the following lemma imply that for all @xmath894 with @xmath884 , and all @xmath130 , there exists an error term @xmath895 with @xmath896 as @xmath310 such that @xmath897 } { \;\mathrm{d}}x \\ & \phantom{:}= \sum_i
\int_{a_i + r_i \omega } { \varphi}(x ) g \bigl(u_j \bigl ( \displaystyle\frac{x - a_i}{r_i } \bigr ) \bigr ) \cdot \overline{t_\psi \biggl [ u_j \bigl ( \displaystyle\frac{x'-a_i}{r_i } \bigr ) \biggr ] } { \;\mathrm{d}}x + e(j).\end{aligned}\ ] ] [ lem : hom_partition ] let @xmath2 generate the homogeneous microlocal compactness form @xmath886 on the domain @xmath78 and also @xmath25 .
moreover , let @xmath898 be a ( countable ) partition of @xmath252-almost all of @xmath78 consisting of open sets @xmath899 ( @xmath900 ) , @xmath192 , and @xmath130 . then , @xmath901 } { \;\mathrm{d}}x\ ] ] by the locality lemma [ lem : locality ] and
the assumption that @xmath902 it suffices to show the assertion on the finite union @xmath903 for all @xmath904 .
indeed , from hlder s inequality together with , we get that the error of restricting to @xmath905 vanishes as @xmath906 .
let @xmath550 and choose @xmath907 , @xmath908 , such that @xmath909 and @xmath910 this is possible again because of .
now use lemmas [ lem : eliminate_rlim ] , [ lem : commutation ] to derive @xmath911 } { \;\mathrm{d}}x \\ & = \lim_{j\to\infty } \sum_i \int_{\omega_i } \zeta_i h({\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\,},u_j ) \cdot \overline{t_\psi[u_j - u ] } { \;\mathrm{d}}x + { \mathrm{\textup{o}}}({\varepsilon } ) \\ & = \sum_i \lim_{j\to\infty } \int_{\omega_i } h({\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\,},u_j ) \cdot \overline{t_\psi[(u_j - u)\zeta_i ] } { \;\mathrm{d}}x + { \mathrm{\textup{o}}}({\varepsilon } ) \\ & = \lim_{j\to\infty } \sum_i \int_{\omega_i } h({\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\,},u_j ) \cdot \overline{t_\psi[(u_j - u){\mathbbm{1}}_{\omega_i } ] } { \;\mathrm{d}}x + { \mathrm{\textup{o}}}({\varepsilon})\end{aligned}\ ] ] letting @xmath619 , the conclusion of the lemma follows .
returning to our previous argument , we get from a short computation using the positive @xmath39-homogeneity of @xmath49 , and changing variables , @xmath912 \bigl ( \frac{x - a_i}{r_i } \bigr ) } { \;\mathrm{d}}x + e(j ) \\ & = \sum_i r_i^d \int_\omega { \varphi}(a_i + r_i y ) g(u_j(y ) ) \cdot \overline{t_\psi[u_j](y ) } { \;\mathrm{d}}y + e(j).\end{aligned}\ ] ] letting @xmath641 , we infer from the homogeneity of @xmath223 ( with @xmath913 and @xmath914 ) , @xmath915 on the other hand , if @xmath916 generates @xmath917 , then @xmath918 . thus , also @xmath919 is homogeneous and @xmath920 which is what we claimed . as building blocks for the laminates considered in this section , we will use the oscillation lemma [ lem : osc ] , also see example [ ex : osc ] . to build up higher - order laminations we will employ the following result : * @xmath921 and @xmath922 in @xmath1 , * @xmath10 , @xmath923 are @xmath16-equiintegrable , * @xmath10 , @xmath923 generate the homogeneous mcfs @xmath924 and the homogeneous young measures @xmath925 , respectively . then , for any @xmath100 , @xmath926 there exists an @xmath1-bounded sequence that generates the homogeneous microlocal compactness form @xmath927 , @xmath928 \otimes \overline{\delta}_{\pm n_0},\ ] ] where @xmath929 . before we come to the proof ,
let us interpret the conclusion .
the microlocal compactness form @xmath919 has two parts : the first is the expected convex combination of @xmath930 and @xmath931 resulting from the `` fast '' oscillations generating @xmath930 and @xmath931 .
in contrast to the corresponding result in young measure theory , however , @xmath919 also contains a second part reflecting the `` slowly '' oscillating lamination construction in direction @xmath110 . in this context recall that the young measures @xmath932 can essentially be calculated from the mcfs @xmath933 and the weak limits , see proposition [ prop : mcf_ym ] .
_ for every @xmath936 divide @xmath382 into @xmath515 strips of length @xmath937 in direction @xmath934 as in figure [ fig : lamination_proof ]
. then subdivide every such strip into a strip of length @xmath938 and one of length @xmath939 .
cover as much as possible of all strips having side length @xmath938 with regular cubes @xmath940 , where @xmath941 , @xmath942 , and @xmath943 .
similarly , cover all strips of side length @xmath939 with cubes @xmath944 , where @xmath945 , @xmath946 , and @xmath947 .
this of course leaves some fraction of @xmath382 uncovered , but the total uncovered volume is less than @xmath937 and so vanishes as @xmath948 .
further , set @xmath949 and define @xmath950 it is easy to see that ( recall @xmath929 ) @xmath951 and @xmath952.\end{aligned}\ ] ] let @xmath953 with @xmath954 , @xmath955 , and @xmath130 be from the collection exhibited in lemma [ lem : countable_test ] .
we have @xmath956 } { \;\mathrm{d}}x \notag\\ & \qquad = \sum_{m=1,2 } \int_{d_k^m } { \varphi}g(w_{j , k } ) \cdot \overline{t_\psi[w_{j , k}-\bar{w}_k ] } { \;\mathrm{d}}x \notag\\ & \qquad\qquad + \sum_{m=1,2 } \int_{d_k^m } { \varphi}g(w_{j , k } ) \cdot \overline{t_\psi[\bar{w}_k - x { \mathbbm{1}}_q ] } { \;\mathrm{d}}x + e(j , k ) \notag\\ & \qquad = : i_1(j , k ) + i_2(j , k ) + j_1(j , k ) + j_2(j , k ) + e(j , k ) .
\label{eq : int_split}\end{aligned}\ ] ] here and in all of the following , @xmath957 denotes a generic error term that may change from line to line and satisfies @xmath958 above , this property is satisfied because @xmath959 } { \;\mathrm{d}}x\biggr| } \\ & \leq c { \|{\varphi}\|}_\infty \cdot { |g(0)| } \cdot { |q \setminus ( d_k^1 \cup d_k^2)|}^{(p-1)/p } \quad \to \quad 0 \qquad\text{as $ k \to \infty$.}\end{aligned}\ ] ] _ step 2 .
_ we first investigate @xmath960 . applying an argument as in lemma [ lem : hom_partition ] , we get @xmath961 } { \;\mathrm{d}}x \\ & = \sum_i \int_{q(a_i , s_k ) } { \varphi}(x ) g \bigl(u_j \bigl(\frac{x - a_i}{s_k}\bigr)\bigr ) \cdot \overline{t_\psi\bigl [ u_j \bigl(\frac{x'-a_i}{s_k}\bigr ) - a \bigr](x ) } { \;\mathrm{d}}x \\ & \qquad + e(j , k ) \\ & = \sum_i s_k^d \int_q { \varphi}(a_i + s_k y ) g(u_j(y ) ) \cdot \overline{t_\psi[u_j - a](y ) } { \;\mathrm{d}}y + e(j , k)\end{aligned}\ ] ] for the last equality we changed variables and used @xmath962(x ) = t_\psi[u_j]\bigl(\frac{x - a_i}{s_k}\bigr),\ ] ] which can be verified by a simple calculation
. then use the uniform continuity of @xmath355 to get @xmath963 } { \;\mathrm{d}}y + e(j , k)\end{aligned}\ ] ] passing to the limits @xmath641 , @xmath948 , we arrive at @xmath964 } { \;\mathrm{d}}y \\ & = \biggl(\lim_{k\to\infty } \sum_i s_k^d { \varphi}(a_i)\biggr ) \cdot { \bigl\langle\hspace{-2.5pt}\bigl\langle g(z ) \cdot q \otimes \overline{\psi } , \omega_1 \bigr\rangle\hspace{-2.5pt}\bigr\rangle}_q.\end{aligned}\ ] ] we further compute the riemann sums @xmath965 thus , also employing , @xmath966 analogously , one shows that @xmath967 _ step 3 .
_ turning to @xmath968 , we first observe @xmath969 } { \;\mathrm{d}}x \\ & = \sum_i \int_{q(a_i , s_k ) } { \varphi}(x ) g \bigl(u_j \bigl(\frac{x - a_i}{s_k}\bigr)\bigr ) \cdot \overline{t_\psi[\bar{w}_k - x { \mathbbm{1}}_q](x ) } { \;\mathrm{d}}x \\ & = \sum_i s_k^d \int_q { \varphi}(a_i + s_k y ) g(u_j(y ) ) \cdot \overline{t_\psi[\bar{w}_k - x { \mathbbm{1}}_q](a_i + s_k
y ) } { \;\mathrm{d}}y.\end{aligned}\ ] ] now let @xmath310 and represent the limit using the young measure from assumption ( iii ) , using the equiintegrability from assumption ( ii ) : @xmath970 \cdot \overline{t_\psi[\bar{w}_k - x { \mathbbm{1}}_q](a_i + s_k y ) } { \;\mathrm{d}}y \\ & \qquad = \int_{d_k^1 } { \varphi}(x ) \cdot \biggl[\int g(z ) { \;\mathrm{d}}\nu_1(z)\biggr ] \cdot \overline{t_\psi[\bar{w}_k - x { \mathbbm{1}}_q](x ) } { \;\mathrm{d}}x.\end{aligned}\ ] ] clearly , an analogous assertion holds for @xmath971 .
hence , with @xmath972 we have @xmath973 } { \;\mathrm{d}}x.\ ] ] by a geometric argument in fourier space analogous to the one in the proof of the oscillation lemma [ lem : osc ] , we may further derive @xmath974 \cdot \overline{(\psi(+n_0)+\psi(-n_0 ) ) ( a - x ) } { \;\mathrm{d}}x \notag\\ & \qquad\quad + \lim_{k\to\infty } \int_{d_k^2 } { \varphi}\cdot \biggl[\int g { \;\mathrm{d}}\nu_2\biggr ] \cdot \overline{(\psi(+n_0)+\psi(-n_0 ) ) ( b - x ) } { \;\mathrm{d}}x \notag\\ & \qquad = \theta \int_q { \varphi}{\;\mathrm{d}}x \cdot \int g { \;\mathrm{d}}\nu_1 \cdot \overline{(\psi(+n_0)+\psi(-n_0 ) ) ( a - x ) } \notag\\ & \qquad\quad + ( 1-\theta ) \int_q { \varphi}{\;\mathrm{d}}x \cdot \int g { \;\mathrm{d}}\nu_2 \cdot \overline{(\psi(+n_0)+\psi(-n_0 ) ) ( b - x)}. \label{eq : i34_lim}\end{aligned}\ ] ] this corresponds to the action of the microlocal compactness form @xmath975 \otimes \overline{\delta}_{\pm n_0}.\ ] ] _ step 4 . _
now select a diagonal subsequence @xmath976 of @xmath977 such that @xmath978 } { \;\mathrm{d}}x \\ & \qquad = \lim_{k\to\infty } \lim_{j\to\infty } \int_q { \varphi}g(w_{j , k } ) \cdot \overline{t_\psi[w_{j , k } - x { \mathbbm{1}}_q ] } { \;\mathrm{d}}x.\end{aligned}\ ] ] combining this with , , , and , we have @xmath978 } { \;\mathrm{d}}x \\ &
\hspace{-2pt}\qquad = \theta { \bigl\langle\hspace{-2.5pt}\bigl\langle f \otimes \overline{\psi},\omega_1 \bigr\rangle\hspace{-2.5pt}\bigr\rangle } + ( 1-\theta ) { \bigl\langle\hspace{-2.5pt}\bigl\langle f \otimes \overline{\psi},\omega_2 \bigr\rangle\hspace{-2.5pt}\bigr\rangle } + { \bigl\langle\hspace{-2.5pt}\bigl\langle f \otimes \overline{\psi},\omega_\mathrm{mix } \bigr\rangle\hspace{-2.5pt}\bigr\rangle } = { \bigl\langle\hspace{-2.5pt}\bigl\langle f \otimes \overline{\psi},\bar{\omega } \bigr\rangle\hspace{-2.5pt}\bigr\rangle},\end{aligned}\ ] ] with @xmath919 given in .
an application of lemma [ lem : eliminate_rlim ] shows that @xmath979 generates @xmath919 and the proof is finished . as mentioned before
, the preceding result is particularly useful in combination with the oscillation lemma [ lem : osc ] as we can then compute the mcf generated by laminates of any order .
we only illustrate this with the following example of a second - order laminate . since all sequences will be uniformly bounded , @xmath22 ( and the domain ) can be chosen arbitrarily .
[ ex:2nd_order_lam ] let @xmath980 , @xmath981 and @xmath982 .
we want to construct the laminate shown schematically in figure [ fig : lamination_ex ] .
define @xmath983 and @xmath984 constant .
it follows from lemma [ lem : osc ] ( also see example [ ex : osc ] ) that @xmath10 generates the homogeneous mcf @xmath985 \otimes \overline{\delta}_{\pm n_1},\ ] ] where @xmath986 trivially , @xmath923 generates the zero mcf @xmath987 .
it can also be shown easily that the sequences @xmath10 and @xmath923 generate the young measures @xmath988 and @xmath989 , respectively . therefore , with @xmath990 an application of proposition [ prop : lamination ] yields the existence of the mcf @xmath991 \otimes \overline{\delta}_{\pm n_2 } \\ & = { \mathcal{l}}^d \otimes \bigl\ { \bigl [ \theta_2 \theta_1 \overline{(a - m ) } \delta_a + \theta_2 ( 1-\theta_1 ) \overline{(b - m ) } \delta_b \bigr ] \otimes \overline{\delta}_{\pm n_1 } \\ & \qquad\quad\hspace{2pt } + \bigl [ \theta_2 \overline{(m - x ) } ( \theta_1 \delta_a +
( 1-\theta_1 ) \delta_b ) + ( 1-\theta_2 ) \overline{(c - x ) } \delta_c\bigr ] \otimes \overline{\delta}_{\pm n_2 } \bigr\}.\end{aligned}\ ] ] consequently , from this mcf we can not only read off the distribution of values and the oscillation directions , but also the _ hierarchy _ of the laminate .
finally , if @xmath992 and we impose the constraint that @xmath993 , we infer from theorems [ thm : diff_omega ] , [ thm : diff_genseq ] , also see example [ ex : osc_curl0 ] , the necessary and sufficient conditions @xmath994 for @xmath995 to be asymptotically a sequence of gradients .
this conclusion is of course in agreement with hadamard s jump condition ( in conjunction with a boundary adjustment ) , also see @xcite .
just like in young s original works @xcite , we now consider as an application the relaxation of integral functionals .
more precisely , we extend integral functionals depending on @xmath1-gradients to functionals on @xmath16-growth gradient mcfs and study microstructures of minimizing sequences . in all of the following , @xmath76 is an open lipschitz domain .
let @xmath996 , where `` @xmath997 '' denotes the scalar product between matrices ( when considered as vectors in @xmath998 ) .
we consider the functional @xmath999 : = \int_\omega f(x , \nabla u(x ) ) { \;\mathrm{d}}x , \qquad u \in { \mathrm{w}}^{1,p}(\omega;{\mathbb{r}}^m).\ ] ] for relaxation problems involving non-(quasi)convex @xmath48 , usually the following three questions are of interest : * what is the _ relaxation _ of the functional , i.e. the @xmath1000-weakly lower semicontinuous envelope of @xmath1001 ( defined to be the largest @xmath1000-weakly lower semicontinuous function below @xmath1001 ) ? * can one extend the functional @xmath1001 to a larger space , in which the minimization problem always has a solution ?
* what is the _ microstructure _ that develops in minimizing sequences to reach that relaxed value ? as is well - known ,
see for example chapter 9 of @xcite , the * relaxation * is given by @xmath1002 = \int_\omega qf(x , \nabla u(x ) ) { \;\mathrm{d}}x , \qquad u \in { \mathrm{w}}^{1,p}(\omega;{\mathbb{r}}^m),\ ] ] where @xmath1003 denotes the quasiconvex envelope of @xmath48 with respect to the second argument .
we recall that a locally bounded borel function @xmath1004 is called * quasiconvex * if @xmath1005 for all @xmath1006 and all @xmath1007 .
it can be shown that the domain @xmath298 can equivalently be replaced by any other bounded lipschitz domain and , if @xmath1008 has @xmath16-growth , @xmath355 can be chosen from @xmath1009 instead .
the * quasiconvex envelope * @xmath1010 is defined as the largest quasiconvex function below @xmath1008 . if @xmath1008 is continuous and bounded from below ( which we assume ) , it can be expressed via dacorogna s formula @xmath1011 we here use the convention that @xmath1010 and hence also @xmath1012 are allowed to take the value @xmath1013 .
see chapters 5 , 6 in @xcite for details on these notions .
this classical relaxation formula answers the first question ( q1 ) : the relaxation of @xmath1001 is simply @xmath1012 . to investigate the second question ( q2 ) , we define @xmath1014 to denote the subset of all @xmath851 with the additional property that there exists a sequence @xmath1015 generating @xmath223 ( this is not automatic , see example [ ex : mcf_nogen ] ) . then we consider the following relaxation problem : [ pr : gamma ] find an `` extension '' @xmath1016 of @xmath1001 onto the larger space @xmath1017 . more precisely ,
find a functional @xmath1018 such that @xmath1019 \ \ \textup{\textbf{:}}\ \
\text{$(w,\omega ) \in ( { \mathrm{l}}^p \times { \mathbf{mcf}}_\mathrm{gen}^p)(\omega;{\mathbb{r}}^{m \times d})$ , $ \operatorname{curl}\ , w = 0 $ , and}\\ & \hspace{115pt}\text{$\omega$ is a gradient mcf in the sense of~\eqref{eq : aomega0_ext},~\eqref{eq : a_curl } } \,\bigr\ } } \\ & \qquad = \inf \ , { \bigl\{\ , { \mathcal{j}}[\nabla u ] \ \ \textup{\textbf{:}}\ \ u \in { \mathrm{w}}^{1,p}(\omega;{\mathbb{r}}^m ) \,\bigr\}}\end{aligned}\ ] ] and @xmath1020 \to \bar{{\mathcal{j}}}[\nabla u,\omega ] \text { whenever $ \nabla u_j { \rightharpoonup}\nabla u$ and $ ( \nabla u_j)$ generates the mcf $ \omega$.}\ ] ] we compute , using lemma [ lem : bandlim_compact ] and employing the ( classical ) young measure @xmath1021 generated by the sequence @xmath1022 , @xmath1023 & = \lim_{r\to\infty } \lim_{j\to\infty } \int_\omega h({\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\,},\nabla u_j ) : \overline{t_{\eta_r}[\nabla u_j ] } { \;\mathrm{d}}x \\ & \qquad + \lim_{r\to\infty } \lim_{j\to\infty } \int_\omega h({\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\,},\nabla u_j ) : \overline{t_{1-\eta_r}[\nabla u_j ] } { \;\mathrm{d}}x \biggr ] \\ & = \int_\omega \int h(x , z ) : \overline{\nabla u(x ) } { \;\mathrm{d}}\nu_x(z ) { \;\mathrm{d}}x + { \bigl\langle\hspace{-2.5pt}\bigl\langle f \otimes i,\omega \bigr\rangle\hspace{-2.5pt}\bigr\rangle } \\ & = \int_\omega { \bigl\langle h(x,{\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\ , } ) : \overline{\nabla u(x ) } , \nu_x \bigr\rangle } + { \bigl\langle h(x,{\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\ , } ) \otimes i,\omega_x \bigr\rangle } { \;\mathrm{d}}x \\ & \qquad + \int_{{\overline{\omega } } } { \bigl\langle h^\infty(x,{\,{\begin{picture}(1,1)(-0.5,-2.5)\circle*{2}\end{picture}}\ , } ) \otimes i,\omega_x^\infty \bigr\rangle } { \;\mathrm{d}}\lambda_\omega(x).\end{aligned}\ ] ] by proposition [ prop : mcf_ym ] , the young measure @xmath1024 can be computed from @xmath1025 and @xmath223 , whence we infer the existence of @xmath1016 that only depends on @xmath1025 and @xmath223 . in principle , we can also write down an expression for @xmath1016 only depending on @xmath1025 and @xmath223 , using the procedure from the proof of proposition [ prop : mcf_ym ] . however , in general this is rather cumbersome , so we stick to the above expression involving both the mcf and the young measure .
trivially , @xmath1026 . on the other hand , the minimization of @xmath1016 can not yield a value strictly lower than the infimum of @xmath1001 , because , by assumption all admissible mcfs are generated by sequences of gradients ( after projecting , see remark [ rem : afree_proj ] ) . finally , if @xmath1027 and @xmath1022 generates the mcf @xmath223 , then @xmath1028 \to \bar{{\mathcal{j}}}[\nabla u,\omega]$ ] .
thus , for the extended functional @xmath1016 the minimum value is always attained .
this shows that @xmath1016 satisfies all the requirements of problem [ pr : gamma ] turning to ( q3 ) , we follow an approach to study the formation of microstructure when approaching the minimum popularized in mller s influential lecture notes @xcite , and investigate the following related problem involving _ pointwise _ minimizers : define the set of pointwise minimizers @xmath1029 of @xmath48 , i.e.@xmath1030 moreover , assume that @xmath1031 satisfies the continuity condition ( ii ) in theorem [ thm : cc_general ] , which is for example trivially satisfied if @xmath1032 for a.e .
now assume that a sequence @xmath1033 is not only minimizing , but even satisfies the following stronger differential inclusion : @xmath1034 alternatively , we can assume the weaker condition ( iii ) from theorem [ thm : cc_general ] .
by said theorem ( also cf .
example [ ex : cc_gradients ] ) , we can conclude @xmath1035 this in particular means that oscillations and concentrations can only occur _ in fixed directions _ or even not at all if @xmath789 does not contain any rank - one matrix .
the considerations so far in particular imply that the theory of microlocal compactness forms retains enough information to study relaxations in the presence of anisotropic effects .
this will be demonstrated in the following example , which shows that the mcf allows us to infer directional properties of microstructure : let @xmath100 be a unit vector and consider the following integrand with critical anisotropy : @xmath1036 note that this is the only integrand in the family @xmath1037 , @xmath1038 , that is both bounded from below and has non - trivial pointwise minima .
alternatively , one can write @xmath48 in the form @xmath1039 and so it follows that @xmath1040 . we can also estimate that @xmath1041 and by an elementary computation , @xmath1042 vanishes if and only if @xmath1043 for some @xmath856 .
assume that @xmath1044 such that @xmath1022 generates the ( generalized ) young measure @xmath1045 as well as the mcf @xmath1046 .
then , we can write @xmath1016 as a function of @xmath538 , @xmath1047 & = \int_\omega \int_{{\mathbb{r}}^{m \times d } } { |a|}^2 - n_0^t a^t a n_0 { \;\mathrm{d}}\nu_x(a ) { \;\mathrm{d}}x\\ & \qquad + \int_{{\overline{\omega } } } \int_{\partial { \mathbb{b}}^{m \times
d } } 1 - n_0^t a^t a n_0 { \;\mathrm{d}}\nu_x^\infty(a ) { \;\mathrm{d}}\lambda_\nu(x).\end{aligned}\ ] ] to investigate minimizing microstructure , assume furthermore that @xmath1022 satisfies the differential inclusion with @xmath1048 then , the wavefront condition entails @xmath1049 this allows us to draw the following conclusions : 1 .
oscillations and concentrations asymptotically have to be in direction @xmath110 .
2 . since the fourier support of a concentration profile ( cf . example [ ex : conc ] ) is asymptotically close to the set @xmath1050 , concentrations have to look like `` ridges '' with normal @xmath110 ; single poles for example are not possible .
we conclude the discussion by noting that the theory of mcfs provides a way to assign a precise meaning to statements about the `` asymptotic direction '' of oscillations and concentrations , which otherwise is only an intuitive notion . in this last section
we briefly explore an application to propagation of singularities in semilinear pde systems , which was also one of tartar s main motivations behind h - measures @xcite , also cf .
@xcite .
define the linear pde operator @xmath1051 and consider for @xmath1052 ( @xmath78 a lipschitz domain ) the semilinear system @xmath1053 where @xmath1054 is a given semilinearity , which we assume to be a carathodory function with @xmath1055 for a constant @xmath1056 and all @xmath1057 .
the aim of this section is to `` replace '' @xmath31 by an mcf @xmath223 in the above system and derive the `` equations '' that @xmath223 satisfies .
this then allows us to understand better the propagation of singularities along the flow governed by .
let @xmath1060 be of class @xmath1061 and @xmath130 .
abbreviating the fourier multiplier @xmath1062 by @xmath1063 , and suppressing the @xmath214-dependence of various quantities for ease of notation , we compute for @xmath1064 : @xmath1065 } { \;\mathrm{d}}x { \;\mathrm{d}}t \notag\\ & \qquad = \int \!\ ! \int { \varphi}\ , \bigl ( dh(u ) \partial_t u \bigr ) \cdot \overline{t[u ] } + { \varphi}\ , h(u ) \cdot \overline{t[\partial_t u ] } { \;\mathrm{d}}x { \;\mathrm{d}}t \notag\\ & \qquad = \int \!\ ! \int { \varphi}\ , \bigl ( dh(u ) { \mathcal{a}}u \bigr ) \cdot \overline{t[u ] } + { \varphi}\ , \bigl ( dh(u ) g(u ) \bigr ) \cdot \overline{t[u ] } \label{eq : semlin_pde_omega_int}\\ & \qquad\qquad\qquad + { \varphi}\ , h(u ) \cdot \overline{t[{\mathcal{a}}u ] } + { \varphi}\ , h(u ) \cdot \overline{t[g(u ) ] } { \;\mathrm{d}}x { \;\mathrm{d}}t . \notag\end{aligned}\ ] ] for the third term we get @xmath1066 } { \;\mathrm{d}}x { \;\mathrm{d}}t = - \sum_k \int \!\ ! \int \partial_k \bigl ( { \varphi}\ , h(u ) \bigr ) \cdot \overline{a^{(k ) } t[u ] } { \;\mathrm{d}}x { \;\mathrm{d}}t \\ & \qquad= - \sum_k \int \!\
! \int ( \partial_k { \varphi } ) \ , h(u ) \cdot \overline{a^{(k ) } t[u ] } + { \varphi}\ , \bigl ( dh(u ) \partial_k u \bigr ) \cdot \overline{a^{(k ) } t[u ] } { \;\mathrm{d}}x { \;\mathrm{d}}t.\end{aligned}\ ] ] at this point of the discussion we further need to assume on @xmath36 the structural * commutation relations * @xmath1067^ * dh(u ) = dh(u ) a^{(k ) } \qquad\text{for $ k = 1,\ldots , d$.}\ ] ] note that these are clearly satisfied if all @xmath1068 are real and scalar , i.e. @xmath1069 for @xmath1070 .
then , with , we may continue the preceding transformations to arrive at @xmath1071 } { \;\mathrm{d}}x { \;\mathrm{d}}t = - \int \!\ !
\int ( { \mathcal{a}}^ * { \varphi } ) \ , h(u ) \cdot \overline{t[u ] } + { \varphi}\ , \bigl ( dh(u ) { \mathcal{a}}u \bigr ) \cdot \overline{t[u ] } { \;\mathrm{d}}x { \;\mathrm{d}}t,\ ] ] where @xmath1072^ * \partial_k.\ ] ] plugging this into and observing the surprising fact that we may cancel terms , we get @xmath1073 } { \;\mathrm{d}}x { \;\mathrm{d}}t \\ & \quad\;\ ; = \int \!\ ! \int { \varphi}\ , \bigl ( dh(u ) g(u ) \bigr ) \cdot \overline{t[u ] } - ( { \mathcal{a}}^ * { \varphi } ) \ , h(u ) \cdot \overline{t[u ] } + { \varphi}\ , h(u ) \cdot \overline{t[g(u ) ] } { \;\mathrm{d}}x { \;\mathrm{d}}t.\end{aligned}\ ] ] now assume we are given a sequence @xmath1074 with every @xmath621 solving in the above sense for a semilinearity @xmath1075 , that is , @xmath1076 we also assume @xmath25 .
in fact , it suffices to assume that the preceding equation is satisfied in the weaker sense that the @xmath1-norm of @xmath1077 vanishes as @xmath641 .
up to a subsequence , the pairs @xmath1078 generate a microlocal compactness form @xmath1079 . write the preceding equation for @xmath621 and let first @xmath641 , then @xmath171 , to derive @xmath1080 \otimes \overline{\psi } , \omega \bigr\rangle\hspace{-2.5pt}\bigr\rangle } -{\bigl\langle\hspace{-2.5pt}\bigl\langle ( { \mathcal{a}}^ * { \varphi } ) \ , h(z_1 ) \cdot q_1 \otimes \overline{\psi } , \omega \bigr\rangle\hspace{-2.5pt}\bigr\rangle}.\end{aligned}\ ] ] here , the splittings @xmath1081 correspond to the two parts of the generating sequence .
[ thm : mcf_system ] the mcf @xmath1082 satisfies the * extended ( mcf ) system * associated to , that is @xmath1083 \otimes \overline{\psi } , \omega \bigr\rangle\hspace{-2.5pt}\bigr\rangle}\ ] ] for all @xmath1064 , @xmath1084 such that @xmath34 exists in the sense of and the commutation relations are satisfied , and all @xmath130 . notice that the right - hand side in this definition takes the role of a semilinear interaction term and is zero if @xmath1085 for all @xmath326 .
this extended system now defines a flow on the level of mcfs and hence describes the propagation of oscillations and concentrations or , equivalently , of ( lack of ) compactness .
consequently , a solution to the extended system could be called a `` compactness flow '' .
if the original system is linear and @xmath1086 is just a function of @xmath214 , we can simplify the above extended system to @xmath1087 where now @xmath1088 is only the @xmath31-part of @xmath223 .
this shows that in this situation oscillations and concentrations are only transported according to a similar law as in the original system and the more complicated interactions between the solution and the semilinearity , which are present in the full extended semilinear system above , do not occur . to illustrate the differences between the preceding result and previous works on `` pdes for measures '' or `` young measure solutions '' to pdes as for instance in theorem 1.5 of @xcite
, we write the young measure system corresponding to our system : consider a sequence of solutions @xmath621 and a corresponding sequence @xmath1089 of semilinearities , which we further assume to generate young measures @xmath1090 and @xmath1091 , respectively ( this can be achieved after selecting a subsequence ) .
we arrive at the system @xmath1092 many works on this topic such as @xcite now focus on the relationship between @xmath1090 and @xmath1091 : indeed , it is one of the main achievements of theorem 1.5 in _ loc .
cit . _ that , in their more restricted situation , this system can be written in terms of @xmath1090 only , thus expressing a compensated compactness fact ( in the sense that products of weakly converging sequences converge to the product of the weak limits , which is the main thrust of @xcite ) .
our result , however , goes further than the above young measure system : we are not only interested in the flow of the _ value distributions _ expressed in the equations for @xmath1090 and @xmath1091 ( and in fact one can find equivalent equations for the _ distribution functions _ ) , but moreover in a flow describing the dynamics of the _ frequency _ and _ directional _ properties of asymptotic oscillations . in the mcf system in theorem [ thm : mcf_system ] this is expressed through the `` frequency / directional test function '' @xmath49 .
we leave finer investigations into the propagation of singularities and compactness to future work .
for now we only remark in closing that in the context of the above extended ( mcf ) system , the compensated compactness theorem [ thm : cc_general ] allows one to prove statements about the `` ellipticity '' present in the system .
j. kristensen and f. rindler , _ characterization of generalized gradient young measures generated by sequences in w@xmath1097 and bv _
197 * ( 2010 ) , 539598 , erratum : vol . 203
( 2012 ) , 693 - 700 .
d. mclaughlin , g. papanicolaou , and l. tartar , _ weak limits of semilinear hyperbolic systems with oscillating data _ , macroscopic modelling of turbulent flows ( nice , 1984 ) , lecture notes in physics , vol . 230 , springer , 1985 , pp .
277289 .
s. mller , _ variational models for microstructure and phase transitions _ , calculus of variations and geometric evolution problems ( cetraro , 1996 ) , lecture notes in mathematics , vol .
1713 , springer , 1999 , pp .
85210 .
l. tartar , _ compensated compactness and applications to partial differential equations _ , nonlinear analysis and mechanics : heriot - watt symposium , vol .
iv , res . notes in math .
39 , pitman , 1979 , pp .
136212 .
to3em , _ the compensated compactness method applied to systems of conservation laws _ , systems of nonlinear partial differential equations ( oxford , 1982 ) , nato adv .
111 , reidel , 1983 , pp .
263285 .
@xmath1093-measures , a new approach for studying homogenisation , oscillations and concentration effects in partial differential equations _ , proc .
edinburgh sect .
a * 115 * ( 1990 ) , 193230 . | this work introduces microlocal compactness forms ( mcfs ) as a new tool to study oscillations and concentrations in @xmath0-bounded sequences of functions .
decisively , mcfs retain information about the location , value distribution , and direction of oscillations and concentrations , thus extending at the same time the theories of ( generalized ) young measures and h - measures . in @xmath0-spaces oscillations and concentrations
precisely discriminate between weak and strong compactness , and thus mcfs allow one to quantify the difference in compactness .
the definition of mcfs involves a fourier variable , whereby also differential constraints on the functions in the sequence can be investigated easily a distinct advantage over young measure theory .
furthermore , pointwise restrictions are reflected in the mcf as well , paving the way for applications to tartar s framework of compensated compactness ; consequently , we establish a new weak - to - strong compactness theorem in a `` geometric '' way . after developing several aspects of the abstract theory , we consider three applications : for lamination microstructures ,
the hierarchy of oscillations is reflected in the mcf .
the directional information retained in an mcf is harnessed in the relaxation theory for anisotropic integral functionals .
finally , we indicate how the theory pertains to the study of propagation of singularities in certain systems of pdes .
the proofs combine measure theory , young measures , and harmonic analysis .
msc ( 2010 ) : 28b05 ( primary ) ; 35b05 , 35a27 , 49j45 , 35l67 .
keywords : microlocal compactness form , mcf , oscillation , concentration , microlocal analysis , relaxation , microstructure , laminate , compensated compactness , hyperbolic system , propagation of singularities . date : ( version 3.0 ) . |
recent advances in experimental techniques have allowed the redefinition of the once supposedly well - understood problem of crystal plasticity . in spite of the traditional picture of a smooth fluid - like motion , acoustic emission ( ae )
measurements on deforming ice crystals @xcite as well as more recent compression tests on metallic specimens @xcite have during the last decade made it clear that plastic deformation of crystalline materials proceeds as a heterogeneous and intermittent adjustment to applied external drives . such intermittence in plastic deformation had been for long observed @xcite and dismissed as considered unworthy of serious consideration .
the recent discoveries challenge this picture since that heterogeneity appears to take over through the emergence of a burst - like scale - free behavior .
plastic deformation is found to develop through strain avalanches , characterized by power law distributed sizes and energy emissions .
slip motion is found to follow fractal patterns @xcite and deformed specimens exhibit self - affine surfaces @xcite .
the observation of scale invariance suggested that fluctuations should not be imputed to the discrete nature of the deforming lattices .
instead , the collective behavior of crystal dislocation was held responsible for the emergence of scale - free patterns , in a close - to - critical behavior fashion . such an observation triggered a number of numerical @xcite and , whenever feasible , analytical @xcite studies so that the validity and the limits of applicability of this picture could be assessed . at present
, there is evidence that the notion of scale invariance in crystal plasticity as a consequence of a close - to - critical state is consistent with observation in single and multi - slip geometries , over a variety of materials , hardening coefficients and loading conditions .
however , the general understanding of these processes is still chiefly phenomenological and several aspects of such phenomenology are still obscure . in particular , it is still not entirely clear to which extent such considerations should be extended to the case of polycrystalline geometries .
grain structure has always been supposed to play a central role in crystal yielding @xcite , and has attracted considerable attention also more recently @xcite .
the so called hall - petch law , which relates the yield stress of a polycrystal to its average grain size , has been recently reformulated in the light of collective dislocation dynamics models @xcite and its breakdown for nanometer sized grains explained as a result of the failure of those systems to initiate or sustain such a process .
these observations pose the general question on how strain avalanches spread through grain boundaries ( gb ) .
the most simplistic approach consists in assuming that the elastic interactions of two dislocations on two sides of a gb are entirely screened by the gb .
such a crude assumption holds only if the different slip geometries of neighboring grains allow that . in general
, however , nothing prevents dislocation stresses to act on neighboring grains , at least in a dampened form due to geometrical incompatibilities .
a reasonably balanced approach consists in postulating that plastic flow ( i.e. dislocation dynamics ) within a certain grain builds up back stresses due to incompatibility of orientations and slip configurations with the neighboring grains .
as soon as the polycrystal reaches the yield point , the accumulated back stresses are redistributed within the neighboring grains .
plastic yield is thus accompanied by a relaxation process by which dislocations effectively interact through grain boundaries by transferring incompatibility stresses .
the advantage of such an approach is that its results indirectly account for experimental observation . by looking at cumulative probability distributions of emission amplitudes in ae experiments @xcite , for example , creeping ice crystals
exhibit a power - law decay with a rather universal exponent @xmath0 .
polycrystalline samples still retain the power - law behavior , however with a lower ( and possibly non - universal ) exponent and , as expected , a large size cut - off due to finite grain sizes .
a natural explanation of the cut - off is that to some extent strain avalanches are confined within grains .
however , explorations of avalanche models @xcite seem to suggest that the decrease of the exponent @xmath0 is possible only if excess stresses within one grain can eventually trigger dislocation avalanches in the neighboring ones .
dislocation interactions through grain boundaries are thus essential to explain plastic flow in polycrystals .
the idea of incompatibility stresses acting on neighboring grains certainly captures part of the physics of the problem but it leaves several questions unanswered , by virtue of its generality .
for instance , the exact mechanism by which stresses are relaxed is not clear , as well as not much can be said about the typical range of stress rearrangements on a quantitative basis . in this paper , we perform an analytical and numerical study of dislocation interactions through grain boundaries , in the light of the above considerations . in order to make the problem treatable
, we consider a relatively simple gb model , that of a low - angle grain boundary , or dislocation wall , in a two - dimensional geometry .
this serves as a starting point to capture the relevant physics of the problem and expose it in a conveniently simple way . in passing
, we also recall that dislocation walls are frequently encountered in deforming crystals as well as in a variety of similar systems such as colloidal crystals @xcite and flux - line lattices in type ii superconductors @xcite . in the following
we discuss how stresses generated by dislocations deform low - angle grain boundaries and how this affects in return the gb stress field perceived by dislocations in their proximity .
theoretical predictions are compared with numerical results .
we find that this perturbative effect accounts for a screening effect for stress deforming the grain boundaries .
the typical strength of this screening proves inversely proportional to the grain size and the perturbation itself is found to be short - ranged and exponentially suppressed beyond a distance that again scales with the grain size .
the analytical and numerical computations are in the next section , while section [ discussion ] is devoted to discussion and conclusions .
let us consider a small - angle grain boundary of length @xmath1 in two dimensions .
the gb is schematized as a discrete linear assembly of @xmath2 edge dislocations distributed along the @xmath3 direction , with burgers vectors @xmath4 parallel to the @xmath5 axis , of sign @xmath6 .
the dislocation spacing is assumed constant and set equal to @xmath7 .
the gb is pinned at the boundaries ( @xmath8 and @xmath9 ) and can be deformed under an applied shear stress , see fig .
[ setup ] . by representing the gb displacement through the vector @xmath10 , where @xmath11 is the displacement of the @xmath12th dislocation along the @xmath5 axis
, the elastic energy cost of a given deformation @xmath10 reads @xmath13=\frac{k}{2}\sum_{m\neq n}\frac{(\eta_m-\eta_n)^2}{(m - n)^2},\ ] ] where @xmath14 . in our notations
, @xmath15 is the shear modulus and @xmath16 the poisson ratio .
sums in equation ( [ energy ] ) run over all couples of dislocations . as a result of long range dislocation interactions , equation ( [ energy ] )
has the form of a non - local elastic functional .
its elastic kernel scales as @xmath17 in fourier space , where @xmath18 is the wave vector associated with the gb deformation profile .
if an external stress field @xmath19 is applied , forces acting on the gb can be calculated in terms of peach - koehler interactions .
the gb begins to deform and settles in a configuration where the applied forces equate the restoring elastic tensions deriving form the term [ [ energy ] ] above .
given the particular geometry of our problem , the gb profile is given by the displacement field @xmath10 which satisfies @xmath20 where the restoring force on the right - hand side is calculated as the ( variational in the continuum case ) derivative of the elastic energy functional in ( [ energy ] ) .
carrying on the above calculation , equation ( [ equilibrium_1 ] ) can be rewritten as a simple linear non - homogeneous problem , in the form @xmath21 by defining @xmath22
let us first review the simple case in which the applied stress is uniform , i.e. @xmath23 @xcite . in order to solve equation ( [ equilibrium_2 ] ) , which can be rewritten in operatorial terms as @xmath24
one needs to diagonalize the matrix of interactions @xmath25 .
this corresponds to solving the eigenvalue problem @xmath26 where we have called @xmath27 each of the @xmath28 eigenvalues and @xmath29 each component of the eigenvector corresponding to @xmath27 .
it can be shown that for large enough @xmath28 , the eigenvalues read @xmath30 and the eigenvectors @xmath31 such an eigenvector representation is possible since the functions @xmath29 match boundary conditions and constitute a _
complete _ set , such that any continuous function with piecewise continuous and differentiable derivative can be expanded in terms of them @xcite .
orthogonality of eigenvectors is ensured by the following condition @xmath32 now @xmath33 and @xmath34 , where we have called @xmath35 the matrix of eigenvectors and @xmath36 the ( diagonal ) matrix of eigenvalues and we have exploited the orthogonality of the transformation @xmath35 , such that @xmath37 .
easily , we derive @xmath38 in the form @xmath39 and replacing the sum over @xmath40 with an integral ( this is done also in the following - see the appendix for a discussion ) @xmath41 where the @xmath42/2 $ ] factor simply limits the sum to odd values of the @xmath43 index .
solving this problem in the case of a more complicated form of the external stress can be a difficult task , instead .
in particular , we are interested in the shear stress generated by an external edge dislocation @xmath44 , placed at @xmath45 .
the dislocation @xmath44 has a burgers vector of modulus @xmath46 and sign @xmath47 . even assuming that the position of such a dislocation is fixed
, the magnitude of the gb - dislocation interaction will necessarily depend on the deformation of the gb itself .
equation ( [ equilibrium_2 ] ) becomes @xmath48 with @xmath49/[x_a^2+(dn - y_a)^2]^2 $ ] and @xmath50 $ ] .
here we have assumed @xmath51 , as the non - local stiffness of the gb would not allow huge deformations to take place .
the equation above can be rewritten in the form of equation ( [ equilibrium_2 ] ) as @xmath52\eta_m =- x_ac_n({\bf r}_a)\ ] ] or , equivalently , @xmath53 where @xmath54 is an operator defined accordingly .
the obvious solution would be @xmath55 however , unlike the case of a uniform stress , the inversion of the @xmath56 matrix is a tough task .
we can not proceed through a straightforward diagonalization as before , since the matrices @xmath57 and @xmath54 are never simultaneously diagonal .
a feasible way to carry on the calculation requires an approximation .
matrix @xmath54 includes corrections coming from the deformation of the gb , which is supposed to be a relatively small effect compared to the main interaction carried by @xmath57 .
these considerations can be easily reformulated in a more quantitative fashion . as a consequence , using simple relations of matrix algebra , we can write @xmath58 , which is a first order expansion in @xmath54 .
the displacement is then @xmath59 or , in the other notation , @xmath60c_m({\bf r}_a).\ ] ] in equation ( [ eta_1 ] ) , the second term within square brackets is considerably smaller than the first and can be neglected .
the deformation of the gb becomes @xmath61 so far we have made basically two assumptions : i ) we deal with a large gb ; ii ) the external dislocation is not in the immediate vicinity of the gb . these choices have allowed us to take continuum limits in sums and to disregard higher order terms . at this point , however , the sum over @xmath40 in equation ( [ eta_2 ] ) can not be calculated easily , unless other approximations are made .
we restrict ourselves to the limit in which the external dislocation sees a quasi - infinite gb , i.e. @xmath62 . the forces exerted by the external dislocation on the gb dislocations can be either attractive or repulsive depending on the distance and the angle between the external dislocation and each gb dislocation . as a result the concavity of the gb profile may change along the @xmath3 direction .
the sum over @xmath40 in equation ( [ eta_2 ] ) can be replaced by an integral from @xmath63 to @xmath64 ( see appendix ) .
the integral is solved calculating residues around the poles of the @xmath65 function .
the displacement thus determined is @xmath66 this result is compared with numerical results from discrete dislocation dynamics simulations of the gb - dislocation system .
these are done with a similar setup as e.g. in ref .
@xcite , with a straight gb as an initial configurations . by letting the gb relax in the stress field of an external dislocation ,
a good agreement is found ( see fig . [ gb_profiles ] ) .
notice that in order for the simulation setup to correspond to eq .
( [ eta_a ] ) ( which was derived by computing the integral from @xmath63 to @xmath64 ) , we have considered a system where in addition to the deformable part of the gb of length @xmath1 , there is a large number ( here 500 ) of pinned dislocations on both sides of the deformable part of the gb .
the agreement with eq .
( [ eta_a ] ) is the better the larger the number of these pinned dislocations .
we are finally able to determine how a dislocation induced deformation affects the stress field generated by the gb .
we consider the case in which @xmath62 , since it matches are simulation setups
. the shear stress at position @xmath67 due to the deformed gb is a functional of the displacement field @xmath11 and can be expressed , by calling @xmath68 the shear stress of the @xmath12th gb dislocation , as @xmath69 where @xmath11 is given by equation [ [ eta_a ] ] .
since we are interested in plastic flow activation well inside each grain , we shall assume @xmath70 , that is we keep far enough from the stress singularities of gb dislocations .
this way we can express stress fields through their taylor expansion around @xmath71 as @xmath72 . equation ( [ stress_functional ] ) will be rewritten as @xmath73 where the @xmath74-th term on the right hand side is obtained summing the @xmath74-th order as in equation ( [ stress_functional ] ) . in passing we
note that , given the nature of the derivation operation , in equation ( [ stress_sum ] ) even terms will be odd functions of @xmath5 , while odd terms will be even functions of @xmath5 .
then it is no coincidence that @xmath75 is the well known stress field generated by a straight gb ( see references @xcite ) . as for the first order @xmath76 calculations are not different from the ones carried on in the previous sections .
we eventually obtain , in the continuum limit of the sum , @xmath77 @xmath78 one can observe that this term is , as expected , an even function of @xmath5 .
it accounts for corrections observed along the @xmath5-direction , where the gb deformation produces an even decrease in the absolute value of the shear stress , see fig .
[ stressfields ] for corresponding results from dislocation dynamics simulations .
this is balanced by an increase along the @xmath79-degree directions along which shear stress is redistributed between adjacent grains .
such corrections are introduced by the @xmath80 term , which can be calculated integrating the second - order term of the expansion of @xmath81 .
the procedure does not differ from what we have seen so far , with the exception of some additional complications arising from the fact that this time the sum over @xmath43 is squared . skipping the details of the calculation
, we obtain @xmath82 @xmath83 @xmath84,\ ] ] where , as usual , @xmath85 .
indeed , this term proves to be an odd function of the variable @xmath5 .
the same method allows the calculation of further orders of the stress expansion ( [ stress_sum ] ) which become relevant as test dislocations are close to the gb , where the anisotropy of stress fields becomes relevant . we should point out that the terms of the sums over @xmath43 ( and @xmath86 ) are exponentially suppressed . as a consequence
, sums can be stopped at the first terms . in particular ,
if the asymmetry or the _ exact _ shape of the gb profile are not a concern , the very first terms provide a fair description of the deformed gb .
furthermore , one should notice that the correction itself decays exponentially with the distance of the external dislocation @xmath44 . this suggests the emergence of a _ screening _ distance , beyond which the effects of gb deformation do not affect dislocation - gb interaction anymore .
this distance is expected to be proportional to @xmath87 , i.e. higher order corrections in @xmath43 are weaker at large distance , but may come into play on a shorter range .
both analytical calculations and numerical studies of the stress fields of a gb - dislocation system show that the gb deformation tends to _ screen _ the dominating stress that is causing the deformation . at the same time , due to the fact that a stress field of an edge dislocation has both positive and negative regions , some parts of such a stress field are also _ enhanced _ in the same process .
one should note , however , that the strongest effect the gb deformation has is screening close to the deformed part of the gb .
this happens on both sides of the gb ( see fig .
[ stressfields ] ) . the implications for collective dynamics in polycrystals with flexible grain boundaries would then be the following : low angle gb s tend to accommodate the excess stresses within a grain by deforming in such a way that such excess stresses are partly screened .
this screening effect is exponentially suppressed beyond a length that is of the order of the typical grain size @xmath88 , and has a magnitude scaling as @xmath89 .
consequently , in addition to acting as boundaries to the dislocations themselves , deformable gb s tend to hinder also stresses from being transferred from one grain to another .
this effect gets stronger with decreasing grain size .
one should note , however , that also various dislocation arrangements present within grains tend to screen the long range stress fields typical to these systems .
the behavior of this simple model system is hence controlled mainly by a single parameter , namely the typical grain size @xmath88 .
in fact , our results suggest that deformable low angle gb s tend to confine avalanches of plastic activity to some extent within single grains .
this could be anticipated to be the case also for general ( high angle ) grain boundaries : even if they can not be thought as deformable arrays of mobile dislocations , the large difference in the respective orientations of neighboring grains means that only a fraction of the stress is effectively transferred over such a gb .
recent experiments on polycrystalline ice reveal an intriguing crossover in the ae amplitude distributions : small and large avalanches appear to be characterized by different power law exponents @xcite .
the observation that small avalanches are described by the same exponent as avalanches in single crystal samples leads to the natural conclusion that such small avalanches do not feel the presence of the gb s . on the other hand
, larger avalanches will necessarily interact with the gb s . in the experiment presented in ref .
@xcite , the exponent describing such large avalanches was found to be smaller as compared to the one describing the statistics of smaller avalanches .
thus , the presence of gb s will change the statistics of avalanches in some way .
one possibility , along similar ideas as presented in ref .
@xcite , is to consider the spreading process on a coarse - grained scale of grains .
an initial avalanche within a grain , if reaching the region close enough to a grain boundary , may spread to neighboring grains thus triggering avalanches there .
triggering is governed by an effective spreading probability @xmath90 that depends ( among other things ) on the details of the dislocation - grain boundary interaction discussed above .
hence the ensuing spreading process is characterized by an average @xmath90 and geometrical factors ( correlations in gb orientations , grain sizes and their correlations etc . ) .
this leads to a number @xmath28 of grains that participate in the `` total '' avalanche .
the statistics of @xmath28 will follow at the simplest a mean - field branching process for which one in general has three possible scenarios : _ i _ ) a sub - critical process with @xmath91 , _ ii _ ) an exactly critical process with @xmath92 , _ iii _ ) a supercritical process which never stops .
clearly , the last case can be excluded by common sense . as for the the first two options
, it appears much more likely that the grain - to - grain spreading is actually sub - critical .
this would be in agreement with the observation that in experiments of ref .
@xcite a grain - size dependent cut - off for the acoustic emission amplitudes was observed . on the other hand , this would indicate that the observed avalanche distributions would not change from what one sees on a single grain level . in conclusion
, we have considered the gb mediated dislocation interactions for a toy " system in two dimensions .
this has allowed us to make analytical progress , to compare with numerics , and to understand the nature of screening .
regardless of the obvious limitations of the simple model studied here , working out these problems leads to an increased understanding of the physics of avalanches in the plasticity of polycrystalline materials .
clearly , there is much room for short - term investigations in three separate directions at least .
one is the accumulation of further experimental data on these phenomena .
the second is the influence of the geometry ( dimension , gb angle , grain size , correlations ) on these processes , which can be studied partly experimentally and partly by numerical , atomistic , simulations .
finally , there seems to be room for the development of coarse - grained models which would match their predictions with experimental data .
_ acknowledgements - _ the authors would like to acknowledge the support of the center of excellence -program of the academy of finland , and the financial support of the european commissions nest pathfinder programme trigs under contract nest-2005-path - com-043386 .
mja and ll would like to thank the isi institute ( turin , italy ) for hospitality .
pm is grateful to the laboratory of physics , helsinki university of technology ( finland ) for kind hospitality and acknowledges financial support from the ministerio de educacin y ciencia ( spain ) , under grant fis2007 - 66485-c02 - 02 .
in several cases in the derivations , sums have been replaced by integrals to infinity .
while the validity of the _ infinity _ assumption is quite intuitive if @xmath1 is taken significantly larger than @xmath93 , the transition to continuum requires a few words of discussion .
provided that the sum is extended to infinity , we calculate it using the poisson integral relation ( see @xcite ) @xmath94 calculating the integral first , we always obtain an exponential @xmath95 term , which rapidly goes to zero if @xmath96 is increased , as we always assumed @xmath97 . as a result
, the @xmath98 term of the sum over @xmath96 alone provides a good estimate of the initial sum .
but this is equivalent , as it can be easily noticed , to replacing the sum with an integral in the first place
. 99 j. weiss , and j. r. grasso , j. phys .
* 101 * , 6113 ( 1997 ) .
c . miguel _
et al . _ ,
nature * 410 * , 667 ( 2001 ) .
m. d. uchic _ et al .
_ , science * 305 * , 986 ( 2004 ) .
d. m. dimiduk _
_ , science * 312 * , 1188 ( 2006 ) .
r. f. tinder , and j. p. trzil , acta metall .
* 21 * , 975 ( 1973 ) .
h. potthoff , phys .
( a ) * 77 * , 215 ( 1983 ) .
j. weiss , and d. marsan , science * 299 * , 89 ( 2003 ) .
m. zaiser _
et al . _ ,
lett * 93 * , 195507 ( 2004 ) .
t. richeton _
et al . _ , nature mater . * 4 * , 465 ( 2005 ) .
f. f. csikor _
_ , science * 318 * , 251 ( 2007 ) .
m. koslowski _
et al . _ ,
. lett . * 93 * , 125502 ( 2004 ) .
m. zaiser , and p. moretti , j. stat .
p08004 ( 2005 ) .
n. zhang , wei tong , int .
j. plasticity * 20 * , 523 ( 2004 ) .
v. yamakov _
_ , nature mater . * 1 * , 1 ( 2002 ) .
v. b. shenoy _
et al . _ ,
lett . * 80 * , 742 ( 1998 ) .
z. shan , _ et al .
_ , science * 305 * , 654 ( 2004 ) .
f. leoni and s. zapperi , j. stat .
p12004 ( 2007 ) .
r. bruinsma , b. i. halperin , and a. zippelius , phys .
b * 25 * , 579 ( 1982 ) a. l. fetter , and j. d. walecka , theoretical mechanics for particles and continua ; mcgraw - hill companies ( february 1 , 1980 ) m .- c .
miguel , _ et al . _ ,
* 89 * , 165501 ( 2002 ) .
j. p. hirth , and j. lothe , theory of dislocations , krieger publishing company ; ( may 1992 ) l. d. landau , e. m. lifshitz , theory of elasticity , butterworth - heinemann ; ( january 1 , 1986 ) | the dynamics of dislocation assemblies in deforming crystals indicate the emergence of collective phenomena , intermittent fluctuations and strain avalanches . in polycrystalline materials ,
the understanding of plastic deformation mechanisms depends on grasping the role of grain boundaries on dislocation motion . here
the interaction of dislocations and elastic , low angle grain boundaries is studied in the framework of a discrete dislocation representation .
we allow grain boundaries to deform under the effect of dislocation stress fields and compare the effect of such a perturbation to the case of rigid grain boudaries .
we are able to determine , both analytically and numerically , corrections to dislocation stress fields acting on neighboring grains , as mediated by grain boundary deformation .
finally , we discuss conclusions and consequences for the avalanche statistics , as observed in polycrystalline samples . |
the topology of generic intersections of quadrics in @xmath0 of the form : @xmath1 where @xmath2 , @xmath3 has been studied for many years : for @xmath4 they are diffeomorphic to a triple product of spheres or to the connected sum of sphere products ( @xcite , @xcite ) ; for @xmath5 this is no longer the case ( @xcite , @xcite ) but there are large families of them which are again connected sums of spheres products ( @xcite ) .
+ let @xmath6 .
the generic condition , known as _ weak hyperbolicity _ and equivalent to regularity , is the following : + _ if @xmath7 has @xmath8 or fewer elements then the origin is not in the convex hull of the @xmath9 with @xmath10 . _
+ a crucial aspect of these varieties is that they admit natural group actions .
all of them admit @xmath11 actions .
their complex versions in @xmath12 @xmath13 ( now known as _ moment angle manifolds _ ) admit natural @xmath14 actions .
the quotient is in both cases a polytope @xmath15 that determines completely the manifolds and the actions .
we will use both notations @xmath16 and @xmath17 for these manifolds to the notations @xmath18 and @xmath19 to specify that some _ facets _ may be empty . in theorem 1
this would mean that the binding might be empty .
in all our examples this will be explicit . ] .
they both fall under the general concept of _ generalized moment angle complexes _
( @xcite and @xcite ) .
+ an open book construction with these manifolds was used to provide a description of the topology of some cases not covered by the main theorem in @xcite ( see remark on page 281 ) . in @xcite
it is a principal technique for proving the results about some families when @xmath5 .
+ in section 2 we recall this construction , underlining its consequences for moment - angle manifolds : + if @xmath15 is a simple convex polytope and @xmath20 is one of its facets , there is an open book decomposition of @xmath21 with binding @xmath22 and trivial monodromy .
+ when @xmath4 , we also give a topological description of the binding and leaves of the decomposition , in terms of the _ odd cyclic partitions _ of @xmath23 that classify the varieties .
this description is almost complete in the case of moment - angle manifolds for @xmath4 : the leaf is the interior of a manifold that can be : + * a product @xmath24 , * a connected sum along the boundary of products of the form @xmath25 , * in some cases , in the connected sum there may appear summands of the form @xmath26 or of the exterior of an embedded @xmath27 in @xmath28 .
the specific products that appear in the above formulae will be described precisely in [ pagec ] in section 3 .
the proofs will be postponed to section 7 where they will follow from a more general theorem for the corresponding real manifolds that requires , as in @xcite , additional dimensional and connectivity hypotheses .
the corresponding result in the real case amounts to the topological description of the _ half _ manifolds @xmath29 , complementing the results in @xcite .
parts of these theorems follow directly from the results in @xcite , other parts require the use of many of the technical lemmas proved there .
all these manifolds with boundary are also generalized moment - angle complexes . + in section 4 we recall some results in @xcite which state that the @xmath30-@xmath31 complex manifolds @xmath32 which correspond to configurations @xmath33 which satisfy an arithmetic condition ( condition ( * k * ) ) fibre ( possibly as seifert fibrations ) over toric varieties ( or orbifolds with simple singularities ) with fibre a complex torus . from this we conclude that the pages of the open books described in section 2 are complex manifolds of the form @xmath34 where @xmath35 is obtained from @xmath33 by suppressing one element of the configuration .
+ in section 5 we show that every moment - angle manifold @xmath36 admits an almost contact structure and , as a consequence , these manifolds also admit a quasi - contact structure .
we recall that there is the conjecture @xcite that every almost contact manifold is in fact a contact manifold . + in section 6
we describe a new construction of compact contact manifolds @xmath37 in arbitrarily large odd dimensions .
these manifolds are generalizations of the moment - angle manifolds @xmath36 that have been studied by v. gmez gutirrez and the second author @xcite .
+ even dimensional moment - angle manifolds can be given complex structures but admit symplectic structures only in a few well - known cases @xcite so it is surprising that many of the odd dimensional ones admit contact structures . in this respect
we remark that using results of c. meckert @xcite and y. eliashberg @xcite it is possible to give a contact structure on these manifolds .
see remark [ rmk4 ] in section 6 .
+ however our method is distinct and in some sense explicit without using open books .
what we use is the heat flow method described in @xcite . in other words ,
we construct a positive confoliation @xmath38 which is conductive i.e. every point of the manifold can be connected by a suitable legendrian curve to a point where the form is positive ( see definition 4 in 6.3 in section 6 ) . + in section 7 we recall some old results about the topology of intersections of quadrics and use them to proof a new one from which theorem [ pagec ] of section 2 follows .
let us recall the concept of an _ open book _ , introduced by h. e. winkelnkemper in 1973 ( see @xcite ) .
an open book in a smooth manifold @xmath39 is a pair @xmath40 consisting of the following : * a proper submanifold @xmath41 of codimension two in @xmath39 with trivial normal bundle , so @xmath41 admits a neighborhood @xmath42 diffeomorphic to @xmath43 . * a locally trivial smooth fibration @xmath44 such that there exists a neighborhood @xmath42 of @xmath41 as above , in which @xmath45 is the normal angular coordinate . the submanifold @xmath41 is called the _ binding _ and the fibres of @xmath45 are the _ pages _ of the open book @xmath40 .
+ also we can define an open book as follows : let @xmath20 be a manifold with a boundary and @xmath46 a self diffeomorphism of @xmath20 that is the identity on @xmath47 .
the mapping torus @xmath48 of @xmath46 , i.e. , the quotient of @xmath49 by the equivalence relation generated by @xmath50 , is a manifold with boundary @xmath51 . collapsing each circle @xmath52 , to a point ,
@xmath53 we obtain a manifold @xmath54 without boundary called the _ relative mapping torus _ of @xmath46 . +
this manifold has an obvious open book structure whose binding is a copy of @xmath47 , the collapsed @xmath55 , and whose mapping to the circle is induced by the projection @xmath56 and so the pages are copies of @xmath20 .
+ we call _ monodromy _ of an open book @xmath40 in a manifold @xmath39 to any self diffeomorphism of a page @xmath20 , such that there is a diffeomorphism @xmath57 taking the obvious open book to @xmath40 .
+ passing to our varieties , @xmath58 admits a @xmath11 action obtained by changing the signs of the coordinates . the quotient is a simple polytope @xmath15 which can be identified with the intersection of @xmath58 and the first orthant of @xmath0 .
it follows that @xmath58 can be reconstructed from this intersection by reflecting it on all the coordinate hyperplanes .
+ by a simple change of coordinates @xmath59 , this quotient can be identified with the @xmath60-dimensional convex polytope given by @xmath61 @xmath62 let @xmath63 be obtained from @xmath33 by adding an extra @xmath64 which we interpret as the coefficient of a new extra variable @xmath65 , so we get the manifold @xmath66 : @xmath67 @xmath68 let @xmath69 be the intersection of @xmath58 with the half space @xmath70 and @xmath71 the intersection of @xmath66 with the half space @xmath72 .
+ the boundary of @xmath71 is @xmath58 ( this shows that @xmath58 is always the boundary of a parallelizable manifold ) .
+ @xmath69 admits an action of @xmath73 by changing signs on all the variables except @xmath74 and the quotient is again @xmath15 . in other words
, @xmath69 can be obtained from @xmath75 by reflecting @xmath15 on all the coordinate hyperplanes except @xmath76 .
+ consider also the manifold @xmath77 which is the intersection of @xmath58 with the subspace @xmath76 .
@xmath77 is the boundary of @xmath69 .
+ @xmath78 acts on @xmath66 ( rotating the coordinates @xmath79 ) with fixed set @xmath77 .
its quotient can be identified with @xmath69 .
the map @xmath80 is a retraction from @xmath66 to @xmath69 which restricts to the retraction from @xmath58 to @xmath69 @xmath81 observe further that this retraction restricted to @xmath82 is homotopic to the identity : the homotopy preserves the coordinates @xmath83 , @xmath84 and folds gradually the half space @xmath85 of the @xmath65 , @xmath74 plane into the ray @xmath86 , @xmath70 preserving the distance to the origin .
+ so @xmath58 is the double of @xmath69 and @xmath66 is the double of @xmath82 , and @xmath66 is the open book with binding @xmath77 , page @xmath69 and trivial monodromy .
[ bookr ] every manifold @xmath66 is an open book with trivial monodromy whose binding is @xmath77 and page @xmath69 .
since the manifold @xmath19 can be considered for each @xmath87 as a manifold @xmath66 with repeated coefficient @xmath9 , then it is an open book with binding the manifold obtained by taking @xmath88 .
this proves : [ bookc ] if @xmath15 is a simple convex polytope and @xmath20 is one of its facets , there is an open book decomposition of @xmath19 with binding @xmath22 and trivial monodromy . when @xmath4 the topology of @xmath89 can be described precisely ( see @xcite and section 8) and that includes also that of all the bindings .
it can be expressed in terms of the _ cyclic partition _ associated to @xmath89 .
we have a precise description of the topology of the leaves in most cases . for our purposes we will only need the case where the total manifold is a moment - angle manifold which is described in the following : [ pagec ] let @xmath4 , and consider the manifold @xmath89 corresponding to the cyclic partition @xmath90 . consider the open book decomposition of @xmath89 corresponding to the binding at @xmath91 , as given by theorem [ bookc ]
. then the leaf of this decomposition is diffeomorphic to the interior of : * if @xmath92 , the product @xmath93 * if @xmath94 and @xmath95 , the connected sum along the boundary of @xmath96 manifolds : @xmath97 * if @xmath98 and @xmath99 , the connected sum along the boundary of @xmath100 manifolds : @xmath101 @xmath102 * if @xmath103 and @xmath104 , a connected sum along the boundary of two manifolds : @xmath105 where @xmath106 is the exterior of an embedded @xmath107 in @xmath28 .
the proof of this theorem and of its real version will appear in section 7 .
similar results can be given when @xmath5 for large families in the spirit of @xcite .
to illustrate the variety of decompositions obtained we give now some examples , by direct application of the previous theorem ( which gives always the decomposition with binding at @xmath91 ) .
the reader may get a better feeling if she or he looks at these examples in the light of the proof of the theorem in section 7 .
an additional feature of these examples is that we obtain three different open book decompositions of the same moment - angle manifold : + let @xmath108 .
the @xmath109-tuple in @xmath110 corresponding to the five roots of unity ( @xmath33 satisfies the weak hiperbolicity condition ) . 1 .
consider @xmath111 where @xmath64 has multiplicity 3 .
we obtain the moment - angle manifold @xmath112 + when @xmath91 we have a configuration @xmath63 , where now the multiplicity of @xmath64 is two .
then the binding is @xmath113 + the page , by theorem [ pagec](b ) , is in this case the manifold @xmath114 2 .
consider @xmath111 where now @xmath115 has multiplicity 3 .
the moment - angle manifold is the same : @xmath116 + when @xmath91 the binding is , + @xmath117 and the page , by theorem [ pagec](d ) , is @xmath118 where @xmath106 is the complement of an embedding @xmath119 in @xmath120 .
consider now @xmath111 where now @xmath121 has multiplicity 3 .
the moment - angle manifold is again @xmath122 when @xmath91 the binding is @xmath123 and the page , by theorem [ pagec](d ) , is @xmath124 where @xmath106 is the complement of the ( unique , up to isotopy ) embedding of @xmath125 in @xmath120 . when we take @xmath126 or @xmath127 with multiplicity 3 we obtain , by symmetry the same open book decompositions as in examples @xmath128 and @xmath129 above , respectively .
let : @xmath33 be an admissible configuration , + @xmath36 the moment - angle manifold corresponding to @xmath33 , + @xmath130 the page and @xmath131 the binding of the open book decomposition . [
cols="^,^,^,^ " , ]
if we take the quotient of @xmath36 by the scalar ( or diagonal ) action of @xmath78 : @xmath132 we obtain a compact , smooth manifold @xmath133 .
these manifolds are called by some authors @xmath30-@xmath31 manifolds .
+ it is known that various of these objects admit natural complex structures ( see , for example , @xcite , theorem 12.2 ) : when @xmath8 is even , both @xmath32 and @xmath134 have natural complex structures .
when @xmath8 is odd , @xmath36 itself has a natural complex structure .
+ let @xmath135 denote the canonical projection .
+ consider now the open book decomposition of @xmath36 described above , corresponding to the variable @xmath136 .
let @xmath35 be obtained from @xmath33 by removing @xmath64 .
it is clear that the diagonal @xmath78-action on @xmath36 has the property that each orbit intersects the page @xmath130 in a unique point and every point of this page is intersected tranversally by the orbits .
this implies that the restriction of the canonical projection @xmath137 to each page is a diffeomorphism onto its image @xmath34 . + for @xmath8 even we therefore obtain , by pulling - back the complex structure of @xmath34 : + [ rmk1 ] _ the page of the open book decomposition of @xmath36 with binding @xmath138 admits a natural complex structure which makes it biholomorphic to the complex manifold @xmath34 . _ + we can go a step further by using the results in @xcite : + it is shown there that for every @xmath33 the manifold @xmath32 has a holomorphic and locally - free action of @xmath139 whose orbits determine a tranversally khler foliation @xmath140 of complex dimension @xmath141 .
the condition for the leaves of @xmath140 to be compact is the following : + an admissible configuration @xmath142 fulfills condition @xmath143 if and only if we may choose , for the ( real ) space of solutions of the system @xmath144 a basis with integer coordinates .
+ also it was shown in @xcite that every configuration @xmath33 can be made to satisfy condition @xmath143 by an arbitrarily small perturbation .
the main result proved in @xcite is the following : [ cef ] let @xmath33 be an admissible configuration that satisfies condition @xmath143 .
then + ( a ) the leaves of the foliation @xmath140 of @xmath32 are compact complex tori of complex dimension @xmath145 .
+ ( b ) the quotient space of @xmath32 by @xmath140 is a projective toric variety of complex dimension @xmath146 .
we denote it by @xmath147 .
+ ( c ) the toric variety @xmath147 comes equipped with an equivariant orbifold structure .
+ ( d ) the natural projection @xmath148 is a holomorphic principal seifert bundle , with compact complex tori of complex dimension @xmath145 as fibers .
+ ( e ) the transversely khlerian form @xmath149 of @xmath32 projects onto a khlerian form @xmath150 of @xmath147 .
the bundle @xmath151 is called a _ generalized calabi - eckmann fibration _ over @xmath147 .
+ if @xmath33 satisfies condition @xmath143 ( and therefore @xmath35 also ) the manifolds @xmath147 and @xmath152 are both toric orbifolds ( possibly singular ) and therefore both are algebraic varieties . as a consequence , @xmath153 is a quasi - projective variety , which ( when is nonsingular ) is a khler manifold and , in particular , symplectic .
+ therefore , we have shown : assume @xmath8 is even .
then the leaf of the open book structure on @xmath36 is naturally a complex manifold .
after a small perturbation of @xmath33 , this leaf admits a holomorphic ( seifert ) fibration over a quasitoric variety , with fibre a compact complex torus .
the even dimensional moment - angle manifolds and the @xmath30-manifolds have the characteristic that , except for a few , well - determined cases , they do not admit symplectic structures .
it was , therefore , surprising to us that the odd dimensional ones can have contact structures .
in fact we can conjecture that they are all contact manifolds : assume @xmath154 is even . then , for every admissible configuration @xmath155 , the odd dimensional manifold @xmath36 is a contact manifold .
here is a first example : the next theorem was proved by f. bourgeois in @xcite ( it is a corollary of theorem 10 in @xcite ) .
[ bour ] if a closed manifold @xmath156 admits a contact structure , then so does @xmath157 .
therefore , _ for @xmath158 , moment - angle manifolds such as @xmath159 admit a contact structure .
_ + we will get closer to our conjecture by showing that odd - dimensional moment - angle manifolds admit structures that are weaker versions of the contact structure .
a @xmath160-dimensional manifold @xmath156 is called _ almost contact _ if its tangent bundle admits a reduction to @xmath161 .
it is known that every contact manifold is an almost contact manifold ( see @xcite ) .
+ on the other hand , d. martnez , v. muoz and f. presas in @xcite defined a quasi - contact manifold as follows . a @xmath160-dimensional manifold @xmath156 is called _ quasi - contact _ if it admits a closed @xmath129-form @xmath149 such that @xmath162 is a non - zero @xmath163-form all over the manifold .
they proved that given an almost contact manifold @xmath156 and given @xmath164 , there exist a quasi - contact structure @xmath149 in @xmath156 such that @xmath165=\gamma$ ] .
they also conjectured that every almost contact manifold is in fact a contact manifold .
+ now , we have : if @xmath8 is even , @xmath36 is an almost contact manifold and also a quasi - contact manifold .
consider the fibration @xmath166 with fibre the circle , given by taking the quotient by the diagonal action .
since @xmath167 is a complex manifold we have that the foliation defined by the diagonal circle action is transversally holomorphic .
therefore , @xmath36 has an atlas modeled on @xmath168 with changes of coordinates of the charts of the form @xmath169 where @xmath170 and @xmath171 where @xmath172 is an open set in @xmath173 and , for each fixed @xmath174 the function @xmath175 is a biholomorphism onto an open set of @xmath176 .
this means that the differential , in the given coordinates , is represented by a matrix of the form @xmath177,\ ] ] where @xmath178 denotes a column @xmath179-real vector and @xmath180 .
the set of matrices of the above type form a subgroup of @xmath181 . by gram - schmidt
this group retracts onto @xmath182 .
+ this shows that @xmath36 is an almost contact manifold and therefore also a quasi - contact manifold by the result in @xcite .
@xmath183 1 .
the conjecture in @xcite would imply our conjecture that every odd dimensional moment - angle manifold admits a contact structure .
we are working on a direct proof of this .
it would give support to the conjecture in @xcite .
in section [ s6 ] we will construct contact structures in some manifolds related to moment - angle manifolds .
the construction in @xcite is based on an open book decomposition of the manifold .
we have still not related this open book structure with the one we have in the case of a moment - angle manifold .
the definition of _ overtwistedness _ in higher dimensions was given by k. niederkrger based in the existence of a _ plastikstufe _ ( see @xcite ) .
let @xmath184 be a cooriented @xmath160-dimensional contact manifold , and let @xmath185 be a closed @xmath179-dimensional manifold .
a _ plastikstufe _
@xmath186 with singular set @xmath185 in @xmath156 is an embedding of the @xmath23-dimensional manifold @xmath187 that carries a ( singular ) legendrian foliation given by the @xmath188-form @xmath189 satisfying : * the boundary @xmath190 of the plastikstufe is the only closed leaf .
* there is an elliptic singular set at @xmath191 . *
the rest of the plastikstufe is foliated by an @xmath78-family of stripes , each one diffeomorphic to @xmath192 , which are spanned between the singular set on one end and approach @xmath193 on the other side asymptotically .
a contact manifold is called _ @xmath194-overtwisted _ if it admits the embedding of a plastikstufe . in dimension three ,
the definition of @xmath194-overtwisted is identical to the standard definition of overtwistedness .
following the ideas of f. presas in the article @xcite , k. niederkger and o. van koert proved the next interesting results ( see @xcite ) : + let @xmath195 be the unit sphere in @xmath12 with coordinates @xmath196 and let @xmath45 be the polinomial @xmath197 the @xmath188-form @xmath198 defines a contact structure on @xmath195 ( see ( * ? ? ?
* proposition 7 ) ) .
* corollary 4 ) every sphere @xmath199 with @xmath200 supports a ps - overtwisted contact structure .
more precisely , on @xmath199 , with @xmath200 , exists a contact structure which admits the embedding of a plastikstufe ps@xmath201 ( with @xmath202 and @xmath203 ) . as a consequence of theorem [ bour ]
, we have the next corollary : for @xmath158 , moment - angle manifolds such as @xmath204 are @xmath194-overtwisted contact manifolds .
theorem 2 of @xcite was extended in @xcite to the case where the manifold is given by two quadratic forms which are not necessarily simultaneously diagonalizable .
this includes the manifolds we construct now : + let @xmath205 and @xmath206 be two integers and @xmath142 be an admissible configuration with @xmath207 .
now consider the manifold @xmath37 define by the following equations : @xmath208 where @xmath209 .
@xmath37 has real dimension @xmath210 .
+ the topology of the manifolds @xmath37 is related to the topology of the associated moment - angle manifolds @xmath36 as shown by the following theorem .
+ the manifold @xmath37 is diffeomorphic to : @xmath211 where @xmath212 .
we will show that for every set of @xmath213 positive numbers @xmath214 the @xmath188-form @xmath215+\sum_{j=1}^n\left[b_j\left(z_j\bar{z}_j-\bar{z}_jdz_j\right)\right]\right]\ ] ] can be deformed into a contact form on the manifold @xmath37 , by an arbitrarily small @xmath216 perturbation . for all @xmath218 the @xmath188-form @xmath38 restricted to the tangent space
@xmath219 is a nontrivial form .
we will only consider the case where @xmath220 and @xmath221 since the other cases are completely analogous . for @xmath218 ,
the linear function @xmath222 is trivial if and only if there exist @xmath223 and @xmath224 such that : @xmath225\\ \nonumber = t\left[2w_1dw_1 + \sum_{j=1}^n\left[\lambda_j\left(\bar{z}_jdz_j+z_jd\bar{z}_j\right)\right]\right]\\ \nonumber+\overline{t}\left[2\bar{w}_1d\bar{w}_1+\sum_{j=1}^n\left[\bar{\lambda}_j\left(z_jd\bar{z}_j+\bar{z}_jdz_j\right)\right]\right]\\ + \mu\left[\left(w_1d\bar{w}_1+\bar{w}_1dw_1\right)+\sum_{j=1}^n \left(z_jd\bar{z}_j+\bar{z}_jdz_j\right)\right ] .
\end{aligned}\ ] ] comparing coefficients in equation we conclude that the unique point satisfying the last equations is the origin @xmath226 .
however , the origin is not in @xmath227 .
we conclude that @xmath38 is nontrivial .
we will denote by @xmath228 the kernel of @xmath38 at the point @xmath229 .
+ the only vectors @xmath230 where @xmath231 are of the form : @xmath232 where @xmath223 and @xmath233 .
+ if @xmath234 is tangent to the sphere @xmath195 and @xmath235 then it follows that @xmath236 must be of the form @xmath237 with @xmath233 .
+ the condition @xmath238 , when @xmath239 and @xmath240 implies : @xmath241 which implies since @xmath239 that @xmath242 if @xmath243 we have : @xmath244 therefore , if @xmath245 implies @xmath246 since , by hypothesis , what is inside the parentheses is positive .
+ vectors of the form @xmath247 which are in @xmath248 , imply that @xmath245 .
hence , if @xmath239 we must have that @xmath246 also .
+ for @xmath249 the vectors of the form @xmath234 with @xmath223 and @xmath224 are in @xmath250 . hence @xmath250 has real dimension three .
+ on the other hand , on the set of points with @xmath249 , the vectors of the form : @xmath251 are all in @xmath252 .
+ hence @xmath253 is a two dimensional vector space .
we will denote this 2-dimensional vector space at the point @xmath254 by @xmath255 .
+ therefore , in the set of points of @xmath227 such that @xmath239 , the form @xmath38 is a contact form .
+ for a generic set of admissible configurations @xmath256 the intersection of @xmath227 with the complex hyperplane with equation @xmath249 is a real codimension - two submanifold @xmath257 of @xmath227 ( i.e. a submanifold of @xmath258 of dimension 2n-3 ) and @xmath259 .
@xmath257 is essentially the moment - angle manifold @xmath260 where @xmath261 . for all @xmath263 the @xmath188-form @xmath38 restricted to the tangent space
@xmath264 is a non trivial form . as in the case @xmath217
, we will consider @xmath265 for all @xmath266 and @xmath267 since the other cases are completely analogous . for @xmath263 ,
the linear function @xmath268 is trivial if and only if there exist @xmath269 and @xmath224 such that @xmath270\\ \nonumber = t\left[\sum_{r=1}^s 2w_rdw_r+\sum_{j=1}^n\left[\lambda_j\left(\bar{z}_jdz_j+z_j\bar{z}_j\right)\right]\right ] \\
\nonumber + \overline{t}\left[\sum_{r=1}^s 2\bar{w}_rd\bar{w}_r+\sum_{j=1}^n\left[\lambda_j\left(z_jd\bar{z}_j-\bar{z}_jdz_j\right)\right]\right ] \\ + \mu \left[\sum_{r=1}^s \left(w_rd\bar{w}_r+\bar{w}_rdw_r\right)+\sum_{j=1}^n\left(z_jd\bar{z}_j+\bar{z}_jdz_j\right)\right].\end{aligned}\ ] ] comparing coefficients we have that the points satisfying the last equations are of the form @xmath271 where at least two @xmath272 , @xmath273 not zero , @xmath274 , @xmath275 . in this cases , the equations @xmath276 and @xmath277 defines a brieskorn manifold and @xmath38 is a contact form on that manifolds @xcite , in particular is non trivial on it . + we conclude that @xmath38 is non trivial on @xmath37 .
we will be denote by @xmath228 the kernel of @xmath38 at the point @xmath278 .
+ the unique vectors @xmath279 where @xmath280 are of the form : @xmath281 @xmath282 where @xmath223 and @xmath224 .
+ if @xmath234 is tangent to the sphere @xmath283 and @xmath284 for all @xmath266 then it follows that @xmath236 must be of the form @xmath285 the condition @xmath286 when @xmath284 for all @xmath287 and @xmath288 implies : @xmath289 which implies , since @xmath290 , that @xmath291 if @xmath292 we have : @xmath293 therefore , if @xmath245 implies @xmath246 since , by hypothesis , @xmath294 is positive .
+ vectors of the form @xmath234 which are in @xmath295 imply that @xmath245 .
hence , if @xmath284 we must have that @xmath246 also .
+ when @xmath296 for all @xmath297 , the vectors of the form @xmath234 with @xmath223 and @xmath224 are in @xmath250 . hence @xmath250 has real dimension three .
+ on the other hand , on the set of points with @xmath296 for all @xmath297 , the vectors of the form @xmath298 are all in @xmath253 .
+ hence @xmath299 is a two dimemsional space .
we will denote this @xmath129-dimensional vector space at the point @xmath300 by @xmath255 .
+ therefore , in the set of points of @xmath37 such that @xmath290 for all @xmath297 , the form @xmath38 is a contact form .
+ for a generic set of admissible configurations @xmath301 the intersection of @xmath37 with the complex hyperplanes @xmath302 is a submanifold @xmath257 of real codimension @xmath303 of @xmath37 .
we introduce some definitions following the ideas of s. j. altschuler and l. f. wu in @xcite : the space of _ conductive confoliations _ , @xmath304 ,
is defined to be the subset of @xmath305 such that * @xmath38 is a positive confoliation : @xmath306 , where @xmath178 denotes de hodge operator ; * every point @xmath307 is accessible from a contact point @xmath308 of @xmath38 : there is a smooth path @xmath309\to { { \pmb{z}}^{^{\mathbb{c}}}_{({\pmb{\lambda}},n , s)}}$ ] from @xmath310 to @xmath311 with @xmath312 in the orthogonal complement of @xmath313 for all @xmath314 .
[ aw ] if @xmath315 then @xmath38 is @xmath316 close to a contact form . from the above and the fact that @xmath257 does not separate @xmath37 , since it is of codimension two , it follows : let @xmath178 denote the hodge operator for a given riemannian metric on @xmath37 , then for the appropriate orientation of @xmath37 one has that @xmath317 for @xmath318 and if @xmath319 @xmath320 therefore @xmath38 is a positive confoliation on @xmath37 .
we must show that every point is accessible from a contact point of @xmath38 .
let @xmath321 .
then , there exists a smooth parametrized curve @xmath309\to{{\pmb{z}}^{^{\mathbb{c}}}_{({\pmb{\lambda}},n , s)}}$ ] such that @xmath322 , @xmath323 if @xmath324 $ ] and @xmath312 is a nonzero vector such that @xmath325 .
let us fix a riemannian metric @xmath326 . for @xmath327 let @xmath328 denote the @xmath129-dimensional subespace of @xmath329 which is orthogonal to @xmath330 at @xmath331 .
+ let us first show that there exists an open neighborhood @xmath332 of @xmath319 and a smooth and non - vanishing vector field @xmath333 defined on @xmath334 such that 1 .
@xmath335 , 2 .
@xmath336 for all @xmath337 , 3 .
@xmath338 for all @xmath337 .
indeed , let @xmath339 .
then @xmath340 has dimension two if @xmath341 or @xmath340 has dimension one if @xmath342 is transverse to @xmath228 .
+ let @xmath343 be a non zero vector . extend this vector anchored at @xmath344 to a smooth vector field @xmath345 defined in a neighborhood @xmath346 of @xmath319
let @xmath347 be the orthogonal projection .
consider the vector field defined on @xmath41 by @xmath348 .
then @xmath349 is a smooth vector field and by continuity @xmath349 satisfies all the required properties in a possible smaller neighborhood @xmath334 .
+ to finish the proof of the lemma we have that by standard extension theorems ( partition of unity ) there exists an extension of @xmath349 to a nonsingular vector field @xmath350 defined on an open neighborhood @xmath351 of @xmath344 . the vector field defined by @xmath352
has the property that @xmath353 .
+ by multiplying the vector field @xmath354 by a positive constant @xmath355 , if necessary , we can assume that all the solutions of the differential equation defined by the vector field @xmath356 on @xmath357 are defined in the interval @xmath358 .
+ if @xmath309\to{\mathcal{u}}$ ] is the solution of the differential equation determined by @xmath356 and satisfying the initial condition @xmath322 then this parametrized curve satisfies the requirements of the theorem [ aw ] if @xmath359 is sufficiently small .
the proposition implies that every point of @xmath257 can be joined , by a legendrian path of finite length , to a point where the form @xmath360 is positive .
+ since @xmath38 is a contact form on @xmath361 we have that any two points of @xmath361 can be joined by a legendrian curve .
therefore every point of @xmath37 is _ accessible _ and @xmath38 defines a _ conductive confoliation _ in the sense of j. s. altschuler and l. f. wu ( @xcite ) .
+ these forms are also called _ transitive confoliations _ by y. eliashberg and w. p.
thurston @xcite , since we can connect any point of the manifold to a point where the form @xmath38 is contact by a legendrian path of finite length .
+ therefore applying theorem [ aw ] we can deform @xmath38 to a contact form @xmath362 .
furthermore @xmath362 can be chosen arbitrarily close to @xmath38 in the @xmath316 topology .
we have the theorem : for a generic set of admissible configurations @xmath363 , the manifold @xmath37 is a contact manifold .
in other words : let @xmath158 , @xmath364 and let @xmath33 be an admissible configuration .
the manifolds @xmath365 where @xmath212 admit contact structures .
[ rmk4 ] it was shown by c. meckert @xcite that the connected sum of contact manifolds of the same dimension is a contact manifold .
it was pointed to us by dishant pancholi that this implies the manifolds @xmath37 have a contact structure .
+ indeed , the manifolds @xmath37 are connected sums of products of the form @xmath366 with @xmath23 even and @xmath145 odd , and @xmath367 .
whitout loss of generality , we suppose that @xmath368 ( the other case is analogous ) then @xmath369 is an open book with binding @xmath370 and page @xmath371 . hence @xmath366 is an open book with binding @xmath372 and page @xmath373 .
the page @xmath374 is paralellizable since it embeds as an open subset of @xmath375 , therefore , since @xmath376 is even it has an almost complex structure .
furthermore , by hypothesis , @xmath377 hence by a theorem of y. eliashberg @xcite the page is stein and is the interior of an compact manifold with contact boundary @xmath372 .
hence by a theorem of e. giroux @xcite @xmath366 is a contact manifold .
however our construction is in some sense explicit since it is the instantaneous diffusion through the heat flow of an explicit 1-form which is a positive confoliation . + another
interesting fact is that the manifolds @xmath37 also have an open book decomposition .
however for these open book decompositions there does not exist a contact form which is supported in the open book decomposition like in giroux s theorem because the pages are not weinstein manifolds ( i.e manifolds of dimension @xmath163 with a morse function with indices of critical points lesser or equal to @xmath23 ) .
in this last section we recall some old results about the topology of intersections of quadrics @xmath378 and the ideas behind their proofs .
we use these to proof a new result about the topology of the manifolds with boundary @xmath379 from which theorem [ pagec ] of section 2 follows .
we recall here the results of @xcite , whose proofs are equally valid for any intersection of quadrics and not only for @xmath4 : + let @xmath380 as before , @xmath15 its associated polytope and @xmath381 its facets obtained by intersecting @xmath15 with the coordinate hyperplanes @xmath382 ( some of which might be empty ) .
+ let @xmath383 be the reflection on the @xmath87-th coordinate hyperplane and for @xmath384 let @xmath385 be the composition of the @xmath383 with @xmath10 .
+ let also @xmath386 be the face obtained by intersecting the facets @xmath387 for @xmath388 .
+ the polytope @xmath15 , all its faces , and all their combined reflections on the coordinate hyperplanes form a cell decomposition of @xmath58 . then the elements @xmath389 are generators of the chain groups @xmath390 , where to avoid repetitions one has to ask @xmath391 ( since for any @xmath87 the intersection @xmath383 acts trivially on @xmath387 )
. + a more useful basis is given as follows : let @xmath392 and @xmath393 be the product of the @xmath394 with @xmath10 .
the elements @xmath395 with @xmath391 are also a basis , with the advantage that @xmath396 is a chain subcomplex for every @xmath397 and , since @xmath394 annihilates @xmath387 and all its subfaces , it can be identified with the chain complex @xmath398 , where @xmath399 is the union of all the facets @xmath387 with @xmath10 .
it follows that @xmath400 for the manifold @xmath69 we have the same elements , but we can not reflect them in the subspace @xmath76 .
this means we miss the classes @xmath395 where @xmath401 and we get induces an epimorphism in homology and fundamental group . ]
@xmath402 these splittings are consistent with the ones derived from the homotopy splitting of @xmath403 described in @xcite .
even if it is not clear that they are the _ same _ splitting , having two such with different geometric interpretations is most valuable .
let us recall the results of @xcite for @xmath4 . in this case , every intersection of quadrics is diffeomorphic to one of the following particular forms : + take @xmath90 a partition of @xmath23 , where we will not distinguish between a partition and any of its permutations preserving the cyclic order and we will think of the index @xmath87 in @xmath404 as an interger mod @xmath96 . corresponding to it we have the configuration @xmath33 consisting of the @xmath405-th roots of unity , the @xmath87-th one taken with multiplicity @xmath404 .
in other words , we have @xmath406 in increasing order where the size of @xmath407 is @xmath404 and for every @xmath408 one has @xmath409 is the @xmath87-th of the @xmath405-th roots of unity .
any configuration can be deformed into one of these by concentrating in one point all the coefficients that is possible without breaking the weak hyperboli - city condition .
we think of the @xmath407 as classes of points that can be joined this way .
+ the pairs @xmath410 with non - trivial homology are those where @xmath397 consists of @xmath411 or @xmath412 consecutive classes , that is , those where @xmath397 is either one of the @xmath413 or one of their complements @xmath414 in @xmath415 . in those cases there
is just one dimension where the homology is non - trivial and it is infinite cyclic . in the case of @xmath416 that homology group is in dimension @xmath417 where @xmath418 is the length of @xmath416 . to specify a generator
consider the cell @xmath419 where @xmath420 and @xmath421 , @xmath422 are any two indices in the extreme classes of @xmath414 ( in other words , those contiguous to @xmath416 ) .
+ @xmath419 is non empty of dimension @xmath417 .
it is not in @xmath423 , but its boundary is .
therefore it represents a homology class in @xmath424 , which is actually a generator , and defines a generator @xmath425 of @xmath426 .
since @xmath419 has exactly @xmath427 facets it is a @xmath417-simplex so when reflected in all the coordinate subspaces containing those facets we obtain a sphere , which clearly represents @xmath428 .
+ the class corresponding to @xmath414 is in dimension @xmath429 and is the poincar dual of the the one corresponding to @xmath416 .
one gets again easily a representative which in this case is not a sphere , but which can be turned into one with a good choice and a surgery if @xmath94 . for our purpose
we do not need to say more about this case .
+ the net result is that , if @xmath94 , all the homology below the top dimension can be represented by embedded spheres with trivial normal bundle which can be made disjoint inside @xmath430 and ( since the inclusion @xmath431 induces an epimorphism in homology due to the homotopy equivalence @xmath432 ) we represent all the classes in @xmath433 by spheres , the @xmath434-cobordism theorem shows that this manifold is a connected sum along the boundary of manifolds of the form @xmath435 .
then its boundary @xmath58 is a connected sum of spheres products . knowing its homology we can tell the dimensions of those spheres : + _ if @xmath94 and @xmath58 is simply connected of dimension at least @xmath109 , then : _ @xmath436 when @xmath92 a simple computation shows that @xmath437 the topology of @xmath430 is implicit in the above proof , and since any @xmath58 with @xmath95 is such a @xmath66 so we have : + _ if @xmath77 is simply connected of dimension at least @xmath109 , and @xmath94 , @xmath95 then : _ @xmath438 the case @xmath98 has to be considered separately .
the difference in the topology of @xmath69 between case @xmath439 and @xmath103 can be seen as follows : + as mentioned before , in the first case the map @xmath440 induces an epimorphism in homology . + this is not the case for @xmath98 .
for example , take the case @xmath441 . here
@xmath77 consists of four copies of @xmath78
while @xmath69 is a torus minus four disks and @xmath77 follow easily by looking at its faces . ] or , equivalently , a sphere minus four disks ( where all the homology comes from the boundary ) with a handle attached that carries the homology not coming from the boundary . + the main fact is that @xmath77 is given by @xmath442 classes , and has only @xmath443 homology generators below the top dimension , only half of which survive in @xmath69 . while the latter manifold has @xmath96 homology generators .
+ to be more precise , the removal of the element @xmath444 allows the opposite classes @xmath445 and @xmath446 to be joined into one without breaking the weak hyperbolicity condition .
therefore @xmath77 has fewer such classes and @xmath447 , which gives a generator of @xmath448 , does not give anything in @xmath449 because there _ it is not a union of classes _ ( it lacks the elements of @xmath446 to be so ) .
+ the two classes in @xmath448 missing in @xmath449 are thus those corresponding to @xmath450 and @xmath451 , all the others contain both @xmath445 and @xmath446 and thus live in @xmath449 . +
as shown above , these two classes are represented by embedded spheres in @xmath69 with trivial normal bundle built from the cells @xmath452 and @xmath453 by reflection , where @xmath454 the corresponding spheres are obtained by reflecting in the hyperplanes corresponding to elements in @xmath461 and @xmath462 , respectively . since these sets are disjoint , the spheres also intersect in a single point .
+ now , a neighborhood of the vertex @xmath457 in @xmath15 looks like the first orthant of @xmath463 where the faces @xmath452 and @xmath464 correspond to complementary subspaces . when reflected in all the coordinates hyperplanes of @xmath463 , one obtains a neighborhood of @xmath457 in @xmath69 where those subspaces produce neighborhoods of the two spheres .
therefore the spheres intersect transversely in that point .
( this shows again that those two classes do not come from the boundary @xmath77 : any homology class coming from the boundary can be separated from any other homology class in @xmath69 and so has trivial homology intersection with it ) .
+ a regular neighborhood of the union of those spheres is diffeomorphic to their product minus a disk , specifically to @xmath465 if @xmath99 the rest of the classes coming from @xmath77 can be represented again by disjoint products @xmath466 so finally the @xmath434-cobordism theorem gives + _ if @xmath58 is simply connected of dimension at least @xmath467 , and @xmath98 , @xmath99 then : _ when @xmath98 and @xmath104 we have the additional complication that making @xmath76 we pass from the _ pentagonal _ @xmath69 to the _ triangular _ @xmath77 , which is not a connected sum but a product of three spheres and , furthermore , not all its homology below the middle dimension is spherical .
all we can say from the above is that @xmath470 where the latter manifold carries all the homology coming from the boundary @xmath77 .
+ we need to be more precise : + the homology generators in @xmath69 are those corresponding to @xmath471 , @xmath472 , @xmath473 , @xmath474 and @xmath475 . out of these @xmath471 and @xmath476
give us the handle which is not in @xmath106 , so the classes in this part are @xmath472 , @xmath474 and @xmath475 all coming from its boundary , which is @xmath477 .
they can all be killed by taking the union @xmath478 of @xmath106 with @xmath479 along their common boundary .
this implies ( by the mayer - vietoris sequence ) that @xmath478 is a homology sphere .
it is simply - connected if @xmath69 is so because this implies that @xmath106 is so too and that @xmath77 is connected .
+ so @xmath478 is a homotopy sphere and @xmath106 is the exterior of @xmath257 inside a homotopy sphere . adding the inverse of that homotopy sphere at the interior of @xmath106 does not change its diffeomorphism type but turns @xmath478 into a standard sphere
therefore @xmath106 is diffeomorphic to the exterior of a @xmath482 in @xmath483 .
it can be expected that this embedding can be assumed to be standard , but for the moment we have the following result : let @xmath4 , and consider the manifold @xmath58 corresponding to the cyclic decomposition @xmath90 and the half manifold @xmath29 .
when @xmath94 assume @xmath58 and @xmath484 are simply connected and the dimension of @xmath58 is at least @xmath467 .
then @xmath69 diffeomorphic to : * if @xmath92 , the product @xmath485 * if @xmath94 and @xmath95 , the connected sum along the boundary of @xmath96 manifolds : @xmath486 * if @xmath98 and @xmath99 , the connected sum along the boundary of @xmath100 manifolds : @xmath487 @xmath488 * if @xmath98 and @xmath104 , a connected sum along the boundary of two manifolds : @xmath489 where @xmath106 is the exterior of an embedded @xmath482 in @xmath483 .
let @xmath4 , and consider the manifold @xmath58 corresponding to the cyclic partition @xmath90 .
when @xmath94 assume @xmath58 and @xmath484 are simply connected and the dimension of @xmath58 is at least @xmath467 . consider the open book decomposition of @xmath66 given by theorem [ bookr ] .
then the leaf of this decomposition is diffeomorphic to the interior of : * if @xmath92 , the product @xmath485 * if @xmath94 and @xmath95 , the connected sum along the boundary of @xmath412 manifolds : @xmath490 * if @xmath98 and @xmath99 , the connected sum along the boundary of @xmath100 manifolds : @xmath487 @xmath488 * if @xmath98 and @xmath104 , a connected sum along the boundary of two manifolds : @xmath489 where @xmath106 is the exterior of an embedded @xmath482 in @xmath483 .
bahri , a. and bendersky , m. and cohen , f. r. and gitler , s. , the polyhedral product functor : a method of decomposition for moment - angle complexes , arrangements and related spaces " . ,
1634 - 1668 , 2010 .
gmez gutierrz , v. and lpez de medrano , s. , intersections of quadrics , 30 years later `` , _ in preparation_. see also lpez de medrano , s. , ' ' singularities of homogeneous quadratic mappings " , to appear in _ revista de la real academia de ciencias exactas , fisicas y naturales . _
serie a. ( doi : 10.1007/s13398 - 012 - 0102 - 6 ) .
martnez , d. and muoz , v. and presas , f. , open book decompositions for almost contact manifolds " , _ proceedings of the xi fall workshop on geometry and physics , publicaciones de la rsme _ , vol . 6 , pp . 131 - 149 , 2004 . | we construct open book structures on moment - angle manifolds and give a new construction of examples of contact manifolds in arbitrarily large dimensions .
* key words : * open book decomposition , moment - angle manifolds , contact structures . |
understanding how chemical evolution has proceeded in the galactic bulge can provide clues for models of bulge formation and evolution .
it is not known , for example , whether the milky way bulge was formed rapidly in a single collapse or via secular dynamical evolution driven by the disk .
certain elemental abundance ratios can be used to infer timescales for chemical enrichment within a particular stellar population .
the most studied of these ratios involves comparing the abundances of the so - called @xmath1-elements ( such as o , mg , or ca ) , which are produced via massive - star core - collapse supernovae of type ii ( snii ) , to abundances of iron , which is produced in sn ia .
probing elemental species that are created in other astrophysical sites , such as agb stars or wr stars , can add further constraints to bulge formation scenarios .
the first study to provide chemical abundance distributions of several elements in a sample of bulge red giants was mcwilliam & rich ( 1994 ) .
it is only within the last few years that additional abundance studies have appeared , all of which rely on the 8 - 10 m class telescopes .
these recent studies include , in the optical , zoccali et al .
( 2006 ) ; fulbright et al .
( 2006 , 2007 ) ; lecureur et al . ( 2007 ) ; mcwilliam et al . ( 2007 ) and , in the infrared , rich & origlia ( 2005 ) ; cunha & smith ( 2006 ) ; rich et al .
( 2007 ) and melendez et al .
although a relatively large number of bulge targets have been studied so far , the abundance patterns of the galactic bulge population continue to be probed in increasing detail .
one element that can add new insight into the nature of chemical evolution in the bulge is fluorine .
understanding the origins of this light element has advanced considerably in recent years , based upon @xmath0f abundances derived from infrared vibration - rotation lines of hf ( jorissen et al .
1992 ; cunha et al . 2003 ; cunha & smith 2005 ; smith et al .
renda et al . ( 2004 ) use the observed abundances to model the galactic chemical evolution of fluorine , with its synthesis occurring primarily in three different astrophysical sites : in agb stars as a result of he - burning ( goriely et al .
1989 ; forestini et al . 1992 ; jorissen et al .
1992 ) , in sn ii via neutrino nucleosynthesis ( woosley et al . 1990 ; woosley & weaver 1995 ) , and in wr stars as a result of he - burning and extensive stellar winds ( meynet & arnould 2000 ) .
renda et al . ( 2004 ) found that neutrino nucleosynthesis was the important source of @xmath0f in the early galaxy ( at low metallicity ) ; however , the fluorine abundances found in near - solar metallicity stars required significant contributions from both agb stars and wr winds .
this paper concentrates on determining fluorine abundances in a sample of red giants of the galactic bulge and these results are combined with previously derived abundances from other elements .
observational evidence for fluorine production in wr stars , compared to agb stars or neutrino nucleosynthesis in sn ii , is discussed as well as the implication for the nature of chemical evolution in the bulge .
the target stars for this analysis of fluorine were taken from our previous infrared high - resolution spectroscopic study of galactic bulge giants ( cunha & smith 2006 ) . the sample , which is composed of 5 k- and 2 m - giants , is presented in table 1 .
details about the nature of these stars , all of which lie in baade s window , can be found in cunha & smith ( 2006 ) .
the spectra were observed in queue mode with the 8.1 m gemini south telescope and the noao spectrograph phoenix ( hinkle et al . 1998 ) at a resolution r@xmath250,000 ; these were centered at 23400 in order to include the hf 1 - 0 r9 line and covered a window of @xmath2120 .
the k - giants in our sample were observed in may and july 2004 ; june and july 2005 ( same spectra were analyzed previously for na in cunha & smith 2006 ) ; while the two m - giant observations were taken more recently during one night in june 2007 .
a description of the phoenix observations and the reduction of the high - resolution spectra can be found in cunha & smith ( 2006 ) and smith et al .
all target stars were previously analyzed in the literature and had stellar parameters and microturbulent velocities ( table 1 ) derived in cunha & smith ( 2006 ) .
the effective temperatures were obtained using calibrations of infrared photometry ( j - k and/or v - k colors ) and extinction maps of stanek ( 1996 ) .
the surface gravities were derived from standard relations between stellar luminosity and mass as defined by isochrones corresponding to 10 gyr by girardi et al .
the microturbulent velocities were estimated from measurements of co molecular lines which are also present in the observed phoenix spectra in the k - band .
more detailed information on the derivation of the stellar parameters can be found in cunha & smith ( 2006 ) .
the fluorine abundances are derived from the hf 1 - 0 r9 line at 23357 .
the reliability of this line as an acurate abundance indicator has been verified in cunha et al .
( 2003 ) from comparisons with other hf lines ( which were analyzed in jorissen et al .
fluorine abundances were obtained from synthetic spectra computed with an updated version of the synthesis code moog ( sneden 1973 ) and adopting marcs model atmospheres ( gustafsson et al .
figure 1 shows both synthetic and observed spectra for one sample star .
the derived fluorine abundances are presented in table 1 in the nomenclature of a(x)=log[n(x)/n(h ) ] + 12.0 .
in addition to the @xmath0f abundances in table 1 , values for a(na ) are also shown , with 5 of the na abundances taken from cunha & smith ( 2006 ) .
they are presented here along with the two new na abundance results for bmb78 and bmb289 ; from the na i line at 23379 .
oxygen abundances are also included for completeness with abundances taken from cunha & smith ( 2006 ) .
the chemical evolution of the galatic bulge has been modelled recently by ballero et al .
( 2007 ) , who focused on the contraints provided by recently published abundances of iron and @xmath1-elements in bulge red - giants .
although at the moment such models do not predict the evolution of the element fluorine in particular , the behavior of the fluorine abundances derived in this study can be used to interpret some aspects of chemical evolution in the bulge population .
this interpretation begins with figure 2 , where the ratio of f / o ( log[n(f)/n(o ) ] ) is plotted as a function of the oxygen abundance , a(o ) , and oxygen is used as a proxy for the overall metallicity .
the five bulge @xmath0f measurements are shown as the red circles , with estimated errors indicated . all results to - date for galactic field stars are also plotted , with these abundances taken from cunha et al .
( 2003 ) , cunha & smith ( 2005 ) , and cunha et al .
the two populations shown in figure 2 ( the galactic field and the bulge ) both exhibit generally increasing values of f / o as the o - abundance increases .
overall , the bulge giants overlap the trend set by the field stars , with the most o - rich bulge star studied ( iv-072 ) apparently defining a smooth extension of the field - star trend to ever increasing oxygen abundances .
one bulge m - giant , bmb78 , defies the general trend by having a relatively low value of f / o given its high oxygen abundance .
the solid line in figure 2 represents the predicted values of @xmath0f/@xmath3o , as a function of metallicity , derived from the woosley & weaver ( 1995 ) sn ii yields , convolved with a salpeter mass function and an upper limit of 40m@xmath4 ; the @xmath0f from these models is produced by neutrino nuclesosynthesis .
more recently , however , heger et al .
( 2005 ) argue that @xmath0f production via neutrino nucleosynthesis should be lowered by about a factor of two , due to reduced cross - sections , and the dashed line in figure 2 is a shift of the woosley & weaver ( 1995 ) yields downward by 0.3 dex as a simple way of viewing these suggested revisions .
it is clear from the results in the figure that at the lowest metallicities , the observed values of f / o for field disk stars tend to approach the values predicted by the yields in which @xmath0f is synthesized via neutrino nucleosynthesis .
the sun and near - solar metallicity field stars , however , fall above the predicted f / o values from neutrino nucleosynthesis and this difference points to significant contributions to @xmath0f production from wr and agb stars , as suggested by renda et al . ( 2004 ) . with four out of the five bulge stars containing larger ratios of fluorine to oxygen than can be accommodated by neutrino nucleosynthesis alone ,
one is left with two possibilities for @xmath0f production at high metallicities , based upon the renda et al .
( 2004 ) model : the agb and wr stars .
can one now attempt to distinguish between these two sites for @xmath0f production in the bulge , keeping in mind that there are no bulge - specific chemical evolution models for @xmath0f ?
looking first at the agb stars , jorissen et al .
( 1992 ) pointed out that there is a positive correlation between f / o with the s - process abundances ( their figure 12 ) and this correlation was modelled by goriely & molawi ( 2000 ) for neutron capture nucleosynthesis in agb stars .
both the model predictions and the observed correlation between fluorine and s - process abundances would suggest that the most fluorine - rich star observed in the bulge , iv-072 , should be heavily enriched in s - process elements at the level of [ s / fe ] @xmath2 + 1.5 dex , if the @xmath0f resulted from agb production .
however , mcwilliam & rich ( 1994 ) derived abudances for two s - process elements in iv-072 and obtained [ y / fe]=-0.02 dex and [ la / fe]=-0.04 dex ; far below what would expected from agb models and observed correlations .
in addition , recent results for heavy - element abundances in three metal - rich bulge dwarfs , whose brightnesses were increased during microlensing events , do not find s - process enrichments : [ s / fe ] @xmath2 + 0.12 dex ( zr , ba and la from cohen et al . 2008 ) ; -0.24 dex ( ba from johnson et al . 2008 ) and -0.28 dex ( ba from johnson et al .
2007 ) . given the apparent lack of s - process enriched high - metallicity bulge stars , the best explanation for the large f / o value in iv-072 may be wr fluorine production .
such a conclusion is reached by renda et al .
( 2004 ) for the metal - rich end of disk chemical evolution .
we note , however , the cautionary points raised by palacios et al .
( 2005 ) in regard to @xmath0f production in wr stars ; rotationally - induced mixing and mass - loss prescriptions can in fact lead to either an order - of - magnitude decrease in @xmath0f production ( for high - mass ( @xmath530 - 80 m@xmath4 ) fast rotators at solar - to - supersolar metallicities ) or an order - of - magnitude increase in @xmath0f production ( for lower mass ( @xmath630 m@xmath4 ) fast rotators at supersolar metallicities ) .
the issue of the @xmath0f(@xmath1,p)@xmath7ne reaction rate uncertainty raised by the downwards revision proposed by lugaro et al ( 2004 ) , appears ameliorated by the recent work of ugalde et al ( 2008 ) , which is consistent with the canonical rate of caughlan & fowler ( 1988 ) .
the large @xmath0f abundance in iv-072 may require a relatively large amount of wr - wind material sculpting the chemical evolution of the metal - rich bulge population .
this conclusion , based on fluorine , agrees with conclusions that are based on the ratios of o to mg in metal - rich bulge and disk stars by mcwilliam et al .
( 2007 ) . while 4 out of 5 bulge fluorine abundances follow an increase in f / o as the stellar metallicity increases , the peculiar position of bmb78 in figure 2 questions whether bulge metallicity increased in a monotonic fashion .
this star is quite oxygen - rich yet has a low fluorine abundance : its value of @xmath0f/@xmath3o is consistent with the yields predicted from neutrino nucleosynthesis only . the low value of f / o in bmb78 does not result from errors within the analysis .
errors in the hf and oh abundances are discussed in detail in cunha et al .
( 2003 ) and smith et al , ( 2003 ) , respectively .
abundance uncertainties are expected to be @xmath80.15 dex for fluorine and @xmath80.20 dex for oxygen . since both hf and oh exhibit similar sensitivities to changes in stellar parameters ,
their ratio is effectively less sensitive to analysis uncertainties . as bmb78 falls about 1.0 dex below the trend defined by the other stars , analysis errors are unlikely to explain its low value of f / o . since the @xmath0f - yield from sn ii neutrinos is sensitive to the metallicity of the supernova progenitor star , it is possible that bmb78 is a star that formed from gas that was substantially enriched by ejecta from a metal - poor supernova .
such a picture would indicate that metallicity in bulge stars proceeded in an inhomogeneous manner at some level .
this scenario can be tested , as @xmath0f is not the only metallicity - dependent element that has been studied in bmb78 .
sodium yields fom sn ii are also metallicity dependent and na has been measured in bmb78 ( table 1 ) .
figure 3 displays results for sodium , where na - to - o ratios are plotted versus the oxygen abundance .
field - star values of na / o and a(o ) are included as the small blue open symbols .
the solid curve contains the massive - star yields from ww95 convolved with a salpeter mass function .
sodium yields are sensitive to stellar metallicity , with the na - to - o ratio increasing with increasing metallicity ( taken here to be mapped by the oxygen abundance ) , and the observed field star values track this curve quite well .
the bulge values of na / o and a(o ) from cunha & smith ( 2006 ) are shown as the large filled symbols with their associated estimated errors : note that na abundances for bmb78 and bmb289 are presented here for the first time .
additional bulge stars from fulbright et al .
( 2007 ) and lecureur et al .
( 2007 ) are shown as the smaller filled symbols .
the agreement in the trend of na / o with a(o ) is similar for all three bulge studies . the sample of bulge red giants included in figure 3 show some peculiarities compared to the field stars .
first there are the two na - rich but o - poor giants from the fulbright et al .
( 2007 ) paper .
the pattern of na / o and a(o ) found in these two stars is very similar to what is found in globular clusters and fulbright et al .
conclude that these two red giants are actually members of the bulge globular cluster ngc6522 located in baade s window .
all three bulge studies also contain a small number of stars that fall to the o - rich side of the distribution , with lower na - to - o ratios .
the star bmb78 is one of these examples , having a low na abundance when compared to its large oxygen abundance . since both f and na have massive - star yields that increase with metallicity , whereas o does not , the low values of f / o and na / o in this star can result from enrichment by a low - metallicity sn ii .
such a picture would suggest that chemical evolution within the bulge population was not homogeneous .
the small number of bulge stars that are found with lower values of na / o may result from inhomogeneous chemical evolution .
a picture of inhomogeneous chemical evolution can be checked for consistency as illustrated in figure 4 , where the abundance ratios of f / ti are plotted versus na / ti .
titanium is chosen as the fiducial element since there is evidence that oxygen yields are being altered at high metallicity by metal - rich wr winds ( mcwilliam et al .
2007 ) and ti typifies an @xmath1-element and thus serves as a monitor of sn ii enrichment . in this diagram
the bulge stars fall along a sequence of increasing values of f / ti with increasing na / ti ; the metallicity sensitive elements f and na increase in lockstep and , in this case , bmb78 exhibits the lowest values of f / ti and na / ti , which is consistent with processed gas from a metal - poor snii .
fluorine abundances are measured for the first time in a sample of red - giants in the galactic bulge .
the fluorine abundances obtained generally define a steady increase in f / o versus a(o ) , which is reminiscent of the disk results and can be explained by production of @xmath0f in a combination of agb and wr stars .
the most oxygen - rich target in this sample has a large fluorine abundance , but no accompanying s - process enhancement , in contrast to the predictions for agb nucleosynthesis by goriely & mowlavi ( 2000 ) .
the abundance pattern observed for this metal - rich bulge target favors @xmath0f production during the wr phase of evolution .
one oxygen - rich giant in this sample , however , fails to follow the disk trend and shows a fluorine abundance , as well as sodium , that is more compatible with pollution from metal - poor sn ii , where @xmath0f is synthesized by neutrino nucleosynthesis .
these results may indicate that there was inhomogeneous mixing in the gas that formed the milky way bulge during its phase of chemical enrichment .
we thank andy mcwilliam for kindly sending us bulge s - process results prior to publication and the referee whose suggestions improved the paper .
this work is supported in part by the nsf ( ast06 - 46790 ) and nasa ( nag5 - 9213 ) .
based on observations obtained at the gemini observatory , which is operated by the assoc .
of univ . for research in astronomy
, under a cooperative agreement with the nsf on behalf of the gemini partnership : the nsf ( united states ) , the stfc ( uk ) , the nrc ( canada ) , conicyt ( chile ) , the arc ( australia ) , cnpq ( brazil ) and secyt ( argentina ) . based on observations obtained with the phoenix spectrograph , developed and operated by noao .
lcccccc i-322 & 4250 & 1.5 & 2.0 & 4.50 & 6.13 & 8.60 + iv-003 & 4500 & 1.3 & 1.8 & ... & 4.23 & 8.05 + iv-167 & 4375 & 2.5 & 2.2 & @xmath66.10 : & 7.30 & 9.10 + iv-072 & 4400 & 2.4 & 2.2 & 5.60 & 7.35 & 9.20 + iv-329 & 4275 & 1.3 & 1.8 & 4.30 & 5.30 & 8.35 + bmb 78 & 3600 & 0.8 & 2.5 & 4.26 & 5.58 & 9.00 + bmb 289 & 3375 & 0.4 & 3.0 & 4.90 & 6.05 & 8.75 + observed ( dotted line ) and synthetic ( solid and dashed lines ) spectra of the star bmb78 in the region of the hf line .
the synthetic spectra were calculated for three fluorine abundances as specified in the figure . ]
ratios of f to o plotted versus the oxygen abundance , a(o ) .
the values of f / o in 4 of the bulge stars track the trend defined for field stars , with the o - rich star iv-072 extending the general field - star trend .
the bulge star bmb78 has a low value of f / o for its o - abundance .
the solid curve illustrates model values of f / o versus a(o ) for neutrino nucleosynthesis from woosley & weaver ( 1995 ) , with the dashed curve representing a downward shift of the values of f / o as suggested by heger et al .
( 2005 ) . ]
the behavior of na / o versus o for bulge stars from this study ( red circles with errorbars ) , fulbright et al .
( 2007 - small red asteriks ) and lecureur ( 2007 - small red crosses ) .
galactic field star results are the small blue open symbols from nissen & schuster ( 1997 ) , fulbright ( 2002 ) , reddy et al .
( 2003 ) , and bensby et al .
the solid line represents yields from woosley & weaver ( 1995 ) convolved with a standard imf .
note the position of bmb78 , with a low ratio of na / o at high metallicity ; this abundance pattern can result from enrichment by metal - poor sn ii . ]
the run of fluorine over titanium versus the abundances of sodium over titanium for the bulge targets stars and the field star @xmath1 boo ( cunha et al .
2003 ; cunha & smith 2006 ) . the position of the sun in this diagram is also shown for comparison . ] | fluorine ( @xmath0f ) abundances are derived in a sample of 6 bulge red giants in baade s window .
these giants span a factor of 10 in metallicity and this is the first study to define the behavior of @xmath0f with metallicity in the bulge .
the bulge results show an increase in f / o with increasing oxygen .
this trend overlaps what is found in the disk at comparable metallicities , with the most oxygen - rich bulge target extending the disk trend .
the increase in f / o in the disk arises from @xmath0f synthesis in both asymptotic giant branch ( agb ) stars and metal - rich wolf - rayet ( wr ) stars through stellar winds .
the lack of an s - process enhancement in the most fluorine - rich bulge giant in this study , suggests that wr stars represented a larger contribution than agb stars to @xmath0f production in the bulge when compared to the disk .
if this result for fluorine is combined with the previously published overall decline in the o / mg abundance ratios in metal - rich bulge stars , it suggests that wr winds played a role in shaping chemical evolution in the bulge .
one star in this study exhibits a very low value of f / o while having a large o - abundance ; this chemical mixture can be understood if this star formed from gas that was enriched by metal - poor core - collapse supernovae and may indicate that chemical evolution in the bulge was inhomogeneous . |
over the past two decades , dynamical simulations of star clusters have become much more realistic .
this realism takes the form of an increasingly complex treatment of individual stars in the cluster . for years , dynamical models only considered stars as single equal - mass points .
the introduction of a mass function into dynamical models quickly necessitated some treatment of stellar evolution .
high mass stars have much shorter lifetimes than low mass stars , mass loss from high mass stars can remove a significant fraction of the mass from the cluster , and very high mass stars can completely dominate the dynamical evolution of the cores of clusters @xcite .
binary stars also have a substantial impact on the cluster dynamics , by acting as energy sinks or sources .
it has been known for a long time that a single massive binary can dominate the evolution of star clusters @xcite .
a hard binary system can produce cluster energy through an encounter with another system that leaves the binary system with a tighter orbit .
encounters between soft binary systems and other objects can release the binary s binding energy to cool the cluster , whereas interactions with hard binaries effectively heat the cluster .
dynamically produced binaries were recognized as a key population for halting core collapse @xcite .
open clusters also have primordial binaries ( * ? ? ?
* ; * ? ? ?
* e.g. ) , and those binary systems can affect the cluster evolution from its birth .
stellar dynamicists realized that the point mass approximation for stars was neglecting a number of dynamically significant processes in clusters .
allowing stars to have radii that change as they evolve was an important next addition to stellar dynamics simulations @xcite .
finite stellar radii are most important for two aspects of these simulations .
first , binary stars can undergo mass transfer as one member of the system fills its roche lobe , either through evolution of the star or dynamical modification of the orbital parameters of the system .
changing binary systems will change how the binaries affect the evolution of the cluster . in the extreme case
, the two components of the binary system can merge .
secondly , stars with finite radii can collide with other stars , either through direct hyperbolic collisions ( in dense clusters ) or in highly eccentric binary systems created after close encounters .
one of the earliest papers to show the dynamical importance of binary - single encounters was written 30 years ago @xcite .
stellar collisions can produce blue stragglers @xcite and other non - standard stellar populations @xcite .
these populations can in turn modify the dynamical evolution of the cluster .
young , open clusters are of particular interest when simulating star clusters . because they tend to have a relatively small number of stars , open clusters provide observations that do not require prohibitive computational expense to reproduce .
they also are well studied and the evolution of their stellar populations ( population i ) are well determined .
some open clusters are young enough to have pre - main sequence stars that are observable ; indeed numerous authors have reported pre - main sequence stars in the pleiades for instance @xcite . for dynamicists ,
open clusters showcase a variety of phenomenon that play an important role in the evolution of the cluster .
the environment that typifies the core of such a cluster is dense enough to provide for a rather high ( and well known ) binary fraction , and collisions , mergers , and other encounters are expected .
all of these factors make open clusters an excellent place to start when integrating stellar evolution with dynamics .
a number of authors have endeavored to push the limits of realism in dynamical simulations .
@xcite start their simulations with 10000 - 15000 stars distributed via plummer and king models with varying fractions of binaries .
they assign masses to these using stars using @xcite initial mass functions , and evolve them from the zero age main sequence using recipes detailed in @xcite .
their model includes the effects of the galactic tidal field .
because their study focused on blue stragglers , they included collisions and mergers as well as close encounters .
@xcite ( hereafter pz01 ) start with 3072 stars all distributed using a king model , and using a @xcite mass function .
their treatment of binary and stellar evolution is somewhat different from @xcite , and will be discussed in more detail in later sections of this paper .
all previous work assumed that all stars began their lives on the zero age main sequence .
however , low mass stars make up the bulk of stars in a cluster for any reasonable initial mass function .
low mass stars also have a pre - main sequence lifetime that is a substantial fraction of the cluster lifetime , significantly influencing a cluster population .
these young stars have radii which can be up to 10 times larger than their main sequence radii .
therefore , some binaries could have undergone an episode of mass transfer that was hitherto not taken into account .
also , larger stars are more likely to have experienced a collision ; those collision products will have been missed in previous simulations .
including pre - main sequence evolution is regarded as a potentially important step in increasing the realism of stellar dynamics simulations @xcite . in this paper , we explore the effect of including the pre - main sequence phase of stellar evolution in a stellar dynamics calculation .
we look at the differences between simulations with and without the pre - main sequence in terms of global cluster properties ( density profile , dissociation time , etc . ) and in terms of the observable effect on the colour - magnitude diagram of the cluster .
we look at the change in number and nature of unusual stellar populations and of the evolved stellar populations . in section 2 ,
we outline the dynamical method used and how we incorporated pre - main sequence evolution . in section 3
we present our results , and discuss their implication in section 4 .
for all our simulations , we use the starlab software environment , featuring the kira integrator @xcite and seba stellar and binary evolution package @xcite .
the simulations were run using the grape-6 hardware @xcite . in the interests of simulating observable results , we choose parameters corresponding to population i , young clusters .
these clusters typically have approximately 1000 stars , and are no older than a few billion years .
we ran three sets of three simulations with different realizations of the input parameters .
our initial conditions mirror those of pz01 , except that in some cases , our stars begin on the pre - main sequence rather than the zero age main sequence . in order to effectively determine the degree that adding pre - main sequence evolution affects a cluster simulation
, we use the same input snapshots for the first three w6 models of pz01 , with various modifications outlined below . for the input snapshots ,
pz01 begin the simulation with 2048 nodes ( single or binary stars ) set up with a @xcite model with a dimensionless depth ( @xmath0 ) of 6 , taking into account the velocity anisotropy and non - spherical shape that cluster would experience in the galactic tidal field similar to that of @xcite .
the initial virial cluster radius was taken to be 2.5 pc .
they then add a binary companion to every second node , for a total of 3072 stars .
this is comparable to young clusters such as ngc 2516 and the pleiades .
masses were applied to the nodes using a initial mass function prescribed by @xcite with masses ranging from 0.1 to 100 , and a mean mass @xmath1 0.6 .
this yields an initial total mass of @xmath2 , which is similar to estimates for the hyades of @xcite .
the masses for the secondary stars were selected randomly between 0.1 and the mass of the primary .
the orbital eccentricities were selected from a thermal distribution between 0 and 1 .
the orbital separation @xmath3 was selected with a uniform probability in log @xmath3 , with a minimum separation of roche lobe contact or 1 , whichever is smaller .
the maximum separation of the binaries was taken to be @xmath4 ( about 0.02 pc ) .
note that the condition of roche lobe contact implicitly includes some dependence the initial radius of the star .
finally , each star was assigned a radius , luminosity , and temperature based on their current evolutionary state .
main sequence initial prescriptions for radius , temperature , and luminosity are given by @xcite .
we added pre - main sequence attributes ( where applicable ) using the tracks obtained from @xcite .
we labeled our simulations as ` pz - ms ' , ` pz - pms ' or ` rw - pms ' , with differences as follows . for our pz - ms runs ,
the snapshots were identical to those used in pz01 the positions , velocities , masses and other stellar and binary parameters were not changed .
these runs were used for comparison purposes , since all stars in this set of simulations began their lives on the zero age main sequence . for our pz - pms runs ,
the positions , velocities , masses , and binary parameters were the same as those used in pz01 but all low mass stars ( @xmath5 ) were started on the pre - main sequence birthline rather than on the zero age main sequence .
essentially the only initial condition that differed between the pz - ms and pz - pms runs was the stellar radius for low mass stars .
the condition for chosing binary orbit separations was a limit based on roche lobe contact , but pre - main sequence stars have much larger radii than zero age main sequence stars .
therefore , many of the binary systems in the pz - pms simulations were initially in contact . to address this issue , we also ran a third set of simulations ( rw - pms ) in which the positions , velocities and masses of stars were identical to pz01 , but the binary semi - major axes were re - assigned , with a limit based on roche lobe contact as determined by the pre - main sequence radii .
for each set of initial conditions , we ran three different realizations , corresponding to the first three realizations reported in pz01 .
where relevant , we label individual realizations as ` pz - pms1 ' , etc . adding the pre - main sequence phase of evolution to the stars in our cluster highlights some confusion in dating clusters .
what exactly is meant by an age of zero ?
if the pre - main sequence phase is included , then low mass stars will reach the zero age main sequence at different times , all of which are later than the start of the simulation , which we define as @xmath6 . in pz01 , it was clear that @xmath6 corresponded to the moment when all stars are on the zero age main sequence . for our pre - main sequence simulations ,
we take @xmath6 to be the moment when all stars are on the deuterium - burning birthline as defined by @xcite ( see section [ pmsev ] ) .
the difference between these two implementations is dependent on the mass of star in question . when comparing ages from dynamical simulations to those determined from observations , it is important to understand how the observed ages were derived .
the entire system is evolved , including the effects of dynamics , individual stellar evolution , and binary evolution . for a complete description of all the computational techniques ,
the reader is directed to the descriptions of starlab , kira , and seba @xcite . in this section
, we will concentrate on describing the numerical methods and parameters that were modified when we introduced the pre - main sequence phase of evolution .
single star evolution from the zero age main sequence is based on the time - dependent mass - radius and mass - luminosity relations given by @xcite .
these relations are valid for the evolution of solar - metallicity stars on the main sequence , subgiant branch or hertzsprung gap , giant branch , horizontal branch and asymptotic giant branch .
stars are immediately turned into inert remnants ( white dwarf , neutron star or black hole ) at the end of the asymptotic giant branch .
the effective temperature and bolometric correction can be calculated for each star from the mass , radius and luminosity .
quantities such as core mass and mass loss due to a stellar wind are also parameterized , as outlined in @xcite .
the evolution of the stars and binaries are computed with the seba binary population synthesis package ( see portegies zwart & verbunt , 1996 , sect . 2.1 with changes made in model b of portegies zwart & yungelson , 1998 ) . in this synthesis model the two stars in a binary are evolved synchronously updating each of the stars at regular intervals while keeping track of the changes to the orbital period and eccentricity .
the simulation model includes various stability criteria for mass transfer , tidal circularization , the emission of gravitational waves , supernova explosions , etc . for our pre - main sequence recipes
, we looked to the evolutionary tracks of @xcite , a selection of which are shown in figure [ pmstracks ] .
they give pre - main sequence tracks that begin with a fully convective protostar , and end just after hydrogen burning begins .
their tracks have resolution in mass of 0.1 from 0.3 to 2 , higher mass resolution between 0.1 and 0.3 , and lower resolution between 2 and 7 .
we used the tracks for a metallicity of @xmath7 ( i.e. solar metallicity , appropriate for young open clusters ) and used the portions of their tracks which begin at the pre - main sequence birthline described by @xcite and end at the zero age main sequence .
we did not implement the pre - main sequence phase for stars with masses greater than 7 because these stars spend little or no time on the pre - main sequence .
we combined all the tracks to create a lookup table in which we interpolated linearly through mass and age to find radius and temperature .
although other pre - main sequence tracks exist @xcite , we chose to use the @xcite tracks .
they have sufficient resolution and span the entire range of masses to which pre - main sequence evolution applies . since theoretical tracks from different authors differ amongst themselves it was important to be able to have enough data to infer the properties of the star for a given mass and age .
the range of masses is quite appropriate since our initial mass function has a lower limit of 0.1 , and stars with masses higher than about 7 do not have a significant pre - main sequence phase .
another reason we chose these tracks was that they had data for stellar radius and temperature that were easily manipulated .
@xcite compare their tracks with others and find good correspondence with most other tracks , although there is still some debate concerning the form of very low ( @xmath8 ) mass pre - main sequence evolution . beyond temperature ,
radius and luminosity , a number of other parameters need to be specified for pre - main sequence stars .
these parameters come in two categories : those dealing with single - star evolution , and those dealing with binary evolution .
unfortunately , many of these parameters have not been determined for pre - main sequence stars . to make reasonable choices
, we had to look to stars with similar composition ( main sequence ) or similar structure ( the large convective envelope of a giant ) to give us these other characteristics .
for instance , the way that we chose to calculate the mass transfer and age rejuvenation from accreting binary star systems in starlab is similar to the main sequence recipes , but the parameters controlling mass transfer relations are based on the values for giant stars .
similarly , we took the value for the radius of gyration to be that of giant stars . since pre - main sequence stars do not exhibit mass loss ( especially stars with masses less than 7 ) , we neglected stellar winds on the pre - main sequence .
one of the major benefits of using the starlab environment is that there are prescriptions for mergers or collisions of varying types of stars .
collisions and mergers can be distinguished in that a merger is a result of unperturbed binary evolution , and is preceded by a period of mass transfer .
currently , the code does not treat collisions that do not result in a merged object , so at the moment when the two stars are replaced by a single star the result will depend only on the types of stars and their mass
. encounters involving main sequence and post - main sequence stars are described in pz01 .
the merger of two main sequence stars is treated as conservative mass accretion from the less massive secondary to the more massive primary .
this results in a rejuvenation of the star which usually is observed as a blue straggler if the masses of the stars are large enough , or as a reasonably normal main sequence stars if the total mass of the new object is less than the current turnoff mass .
the merger of other types of stars result in evolved stars or unusual objects , as warranted by the structure of the two stars involved in the collision .
the treatment of encounters involving pre - main sequence stars was similar to the treatment of main sequence stars
when a pre - main sequence star collided with a more evolved star , the result was taken to be similar to the result of a main sequence star colliding with the same kind of star .
for pre - main sequence / pre - main sequence collisions , the merged object was returned to the pre - main sequence birthline as its evolutionary state will be completely disrupted .
these choices are in agreement with the results of the hydrodynamic simulations of @xcite of collisions involving pre - main sequence stars .
two of the most general functions that represent the time evolution of star cluster are its total mass and total number of particles versus time . the processes which can decrease the total number of cluster members are a merger between two stars , type ia supernovae ( which do not leave a remnant ) and the escape of a star from the system .
stars may escape as a result of a mixture of influences : dynamical encounter with a binary , supernova kick , or removal by the galactic tidal field .
figure [ mnvt ] shows the total number of stars ( where a binary system counts as 2 stars ) versus time for each set of runs .
the largest effect is the large initial drop in the number of stars in the pz - pms runs ( dashed line ) .
this has mainly to do with early mergers that occur within these clusters . in these runs ,
all of the binaries that had a small orbital separation in the main sequence configuration are now in contact . as a result ,
a large number of binaries merge immediately after initialization of the simulation .
eventually , the evidence of this event is erased , and the number of remaining cluster members approaches that of the other series .
the rw - pms runs experience a decreased number of mergers throughout the life of the cluster since the radii of the stars become smaller throughout the pre - main sequence lifetime , and may only reach a point of roche lobe overflow when one of the stars becomes a giant . as a result ,
the rate of decrease of the number of stars for the rw - pms runs ( @xmath9 stars / myr ) is lower than the rate the pz - ms runs ( @xmath10 stars / myr ) , but with a very similar slope to the pz - pms runs ( @xmath11 stars / myr ) .
the slopes were measured between 500 and 1500 myr .
this dependence of the dynamical results on the initial binary merger rate highlights the most important conclusion of this paper .
the inclusion of pre - main sequence stars in dynamical simulations affects only the binary properties of the system , and a consistent treatment of the pre - main sequence binary population is necessary .
the evolution of the total mass of the cluster is driven by two processes .
one is mass loss from stars via stellar winds , and the other is the escape of stars from the cluster .
the differences caused by including the pre - main sequence phase are quite small , as shown in figure [ mnvt ] .
the maximum difference between the total mass of the pz - ms run and both pre - main sequence runs is at most 6% .
.number loss and mass loss at 1.5 gyr for all runs [ cols="<,>,>,>",options="header " , ]
in this paper , we investigated the effects of allowing stars to begin their lives on the pre - main sequence in dynamical simulations .
we added pre - main sequence evolutionary tracks which begin at the deuterium - burning birthline , and end at the zero - age main sequence .
stars with masses less than 7 in the dynamical simulations have their radii , luminosities , temperatures , and other properties determined by these tracks starting from the @xmath6 point of the dynamical simulation .
we compared these simulations with standard ones in which all stars begin their lives on the zero age main sequence at @xmath6 .
the dominant characteristic which distinguishes between our different initial conditions can be traced back to the number of mergers in the early stage of the cluster evolution . in the pz - pms runs ,
a lot of the binaries begin in roche lobe contact , since the binary orbits are based on zero age main sequence radii , and pre - main sequence stars have much larger radii initially .
these early pre - main sequence / pre - main sequence mergers show up as a decrease in the total number of stars , but an increase in average mass .
this effect changes the mass function of the cluster , not only initially , but the effect continues to be noticed for quite a large portion of the cluster s lifetime ( many tens of initial half - mass relaxation times ) . on the other hand , since the stars in the rw - pms runs start at a greater separation , and since the initial pre - main sequence radius is the largest radius for a given star until it becomes a giant star , there is an absence of mergers in these runs .
this increases the binary fraction , and affects the mass function as well . in spite of the initial drop of total number of objects in the pz - pms models ,
the time evolution of the total mass is not noticeably affected by the starting point of the stars .
similarly , the total number of stars in both pms simulations seem to decrease at a slower rate than their pz - ms counterparts .
this is due to a smaller number of escapers ejected from the cluster through an encounter .
the inclusion of pre - main sequence evolution causes the stellar interactions to be less violent . during a binary star / single star interaction , accretion during the stellar encounter affects the ensuing binary parameters which in turn affects the resulting ejection velocity .
indeed , the merger of close binaries ( as in the pz - pms runs ) or the absence of close binaries in the rw - pms runs should affect the cluster in the same way since binaries act as a heat sink for the total energy of the cluster .
since the predominant form of binary that stars in these models will encounter is not very hard , the heat sinks of the cluster can absorb more , thus decreasing the energy available for other purposes . in the rw - pms models very little circularization
is observed .
this is because the stars are contracting away from each other , and therefore can not maintain close proximity as required for circularization .
the pz - pms runs contain the largest number of circularized binaries and has a tidal circularization period which best matches observations of the hyades .
another item of note is the frequency of the different types of mergers .
almost all of the mergers were due to the normal binary evolution in which a star s radius becomes large than its roche lobe .
the two most frequent mergers in the pms runs are pre - main sequence / pre - main sequence and pre - main sequence / giant star mergers ( with the latter being more dominant in the rw - pms case ) .
hydrodynamic simulations of such collisions @xcite suggest that the result will be a larger pre - main sequence star in the former case , and a larger giant star in the latter case .
there is a significant amount of future work that could be done in this area and could involve simulating other cluster configurations ( for instance , with more stars or in different tidal field ) .
since pre - main sequence stars in very young clusters have been studied ( e.g. , @xcite ) , a focus on the first 100 myr of a simulation could be insightful .
another avenue of research would be to perform population synthesis in order to better determine which initial distribution of binary parameters will result in the observed distribution of these parameters , along the lines of @xcite .
both the choices for initial binary orbits in the rw - pms and pz - pms runs were quite simplistic and should be improved upon .
there are two noteworthy pieces of astrophysics that we have completely neglected from these simulations .
both are particularly relevant to the study of young clusters , particularly clusters significantly younger than the 600 myr hyades analogue we concentrated on here .
the first topic is the inclusion of the gas out of which the stars in these clusters formed .
we know from observations of star forming regions that there is a reasonably long period of time when both stars and gas co - exist , and are presumably interacting dynamically .
we neglect that stage completely by starting with a stars - only king model .
the second topic is the effect of circumstellar ( and circumbinary ) disks around the young stars in the cluster .
interactions between stars with disks should be slightly different than interactions between stars without disks , and could modify our results . in this paper
, we have demonstrated that the careful treatment of cluster initial conditions is important , and there are clearly other avenues for improvement as well .
our conclusion from this study is that inclusion of the pre - main sequence phase of stellar evolution is critical for any simulations that wish to understand the properties of binary stars in stellar systems .
the choice of initial binary parameters and their subsequent evolution are strongly modified by the properties of pre - main sequence stars . rw acknowledges support from sharcnet . as is supported by nserc and the canadian foundation for innovation .
spz acknowledges support from the dutch royal academy of science ( knaw ) , dutch organization for scientific research ( nwo ) and dutch research school for astrophysics ( nova ) .
sills a. , deiters s. , eggleton p. , freitag m. , giersz , m. , heggie d. , hurley j. , hut p. , ivanova n. , klessen r. s. , kroupa p. , lombardi j. c. jr . , mcmillan s. , portegies zwart s. , & zinnecker h. 2003 , new astronomy , 8 , 605 | we investigate the effects of the addition of pre - main sequence evolution to star cluster simulations .
we allowed stars to follow pre - main sequence tracks that begin at the deuterium burning birthline and end at the zero age main sequence .
we compared our simulations to ones in which the stars began their lives at the zero age main sequence , and also investigated the effects of particular choices for initial binary orbital parameters .
we find that the inclusion of the pre - main sequence phase results in a slightly higher core concentration , lower binary fraction , and fewer hard binary systems . in general
, the global properties of star clusters remain almost unchanged , but the properties of the binary star population in the cluster can be dramatically modified by the correct treatment of the pre - main sequence stage . |
jahn - teller model as a protopype model for phonon removal of degeneracy of electron levels in complex molecules @xcite was investigated mainly in its rotational symmetric e@xmath0e local version with electron coupling to two degenerate intramolecular phonon modes , one antisymmetric and one symmetric against reflection.importance of focusing on the lattice version of the model increased due to jt - based structural phase transitions in some recently discovered high-@xmath5 superconductors and manganese - based perovskites @xcite , @xcite . for the reflection symmetric two - level electron - phonon models with linear coupling to one phonon mode ( exciton , dimer ) shore et al .
@xcite introduced variational wave function in a form of linear combination of the harmonic oscillator wave functions related with two levels .
two asymmetric minima of effective polaron potential turn coupled by a variational parameter respecting its anharmonism by assuming two - center variational phonon wave function .
this approach was shown to yield the lowest ground state energy for the two - level models @xcite , @xcite .
the peculiarities due to reflection phenomena are also expected in the case of linear coupling with two phonon modes .
in contrast to the model with coupling to one phonon mode , the oscillations mediated by the antisymmetric mode are nonlinearly coupled to the symmetric phonon mode .
therefore , in order to improve the harmonic oscillator variational ansatz for this model , it was necessary to introduce additional variational parameters ( vp ) considering _ correlation _ of both phonon modes .
this was performed by lo @xcite for the e@xmath0e jt model and by lo et al .
@xcite ) for the dimer model .
the correlation was found most effective in the region of parameters where the competing localization ( polaron ) energy and the delocalization transfer energy ( tunneling ) were comparable . in crystals exhibiting high space anisotropy the rotational symmetry of jahn - teller molecules
can be broken .
therefore it is reasonable to investigate jt model generalized by breaking the rotational symmetry ( while saving the reflection symmetry ) : we assume different coupling strengths for the onsite ( intralevel ) and interlevel electron - phonon couplings .
such a model can also be considered as a generalization of the exciton - phonon or the dimer - phonon model by assuming the electron tunneling phonon assisted .
jt model with the broken rotational symmetry will involve the coupling of two minima as well as the coupling of the symmetric and antisymmetric phonon modes . in the lattice model , however , in contrast to the local case , the mode correlation appears to be of a marginal importance ( section iii ) . as a consequence ,
our ansatz for the variational wave function of the local model will involve ( i ) the reflection vp introduced by the construction of the two - center phonon wave function , and ( ii ) correlation vp respecting coupling of the symmetric and antisymmetric phonon modes .
the question arises of the importance of these vps and respective effects regarding their relevance in different regions of the model parameters .
formulation of the variational ansatz of the local case is presented in the section iv .
calculation of the ground state energy of the lattice model and related discussions are the contents of the sect.iii . and v. reliability of different variational alternatives was checked by comparison with results of exact numerical simulations for the local model .
we investigate 1d lattice of spinless double degenerated electron states linearly coupled to two intramolecular phonon modes described by hamiltonian @xmath6 where @xmath7 are phonon annihilation operators , and the pauli matrices @xmath8 represent two - level electron system .
they satisfy identities @xmath9=i\sigma_{kn } , \
l = x , y , z$ ] , representing @xmath10-pseudo - spins related to the electron densities in a usual way , i.e. @xmath11 @xmath12@xmath13 @xmath14 is a unit matrix , and @xmath15 are electron annihilation operators .
the operator @xmath16 of the displacement by a lattice constant @xmath17 acts in both the electron and phonon space , @xmath18 . in terms of the creation - annihilation electron and phonon operators the hamiltonian can be cast as follows : @xmath19\ , .\qquad
\label{1}\end{aligned}\ ] ] for @xmath20 , the interaction part of ( [ hams1 ] ) @xmath21 yields the rotationally symmetric @xmath22 form @xcite with a pair ( an antisymmetric and a symmetric under reflection ) of double degenerated vibrations .
this is , e.g. , the case of @xmath23 ions with @xmath24 configurations in high-@xmath5 cuprates @xcite,@xcite .
taking @xmath25 removes the degeneration of the vibronic states breaking the rotational symmetry of the electron - phonon interactions , the model still staying within the class of jt models @xcite , @xcite .
the dispersionless optical phonon mode @xmath26 splits the degenerated unperturbed electron level ( @xmath27 ) while the mode @xmath28 mediates the electron transitions between the levels .
this latter term represents phonon - assisted tunneling , a mechanism of the nonclassical ( nonadiabatic ) nature as well as is the pure tunneling in related exciton and dimer models .
evidently , hamiltonian ( [ hams1 ] ) ( @xmath29 ) is reflection - symmetric , @xmath30 , @xmath31 where @xmath32 is the phonon reflection operator .
while the phonon @xmath33 is antisymmetric under the reflection , phonon @xmath34 remains symmetric .
in addition , the transfer part of ( [ 1 ] ) exhibits @xmath35 symmetry of the left- and right - moving electrons ( holes ) .
let us note that the quantum phonon assistance of the electron tunneling ( @xmath36-term in ( [ 1 ] ) and ( [ hams1 ] ) ) constitutes the difference of the model from the related dimer and exciton quantum models where instead of @xmath37 of ( [ hams1 ] ) there stands @xmath38 , where @xmath39 is the distance between the levels @xcite,@xcite .
the local part of ( [ hams1 ] ) can be diagonalized in electron subspace by the fulton - gouterman unitary operator @xcite @xmath40 , where @xmath41 as follows @xmath42i_n . \label{diagh1}\end{aligned}\ ] ] on the other hand , in the transfer term @xmath43 there appears a nondiagonality @xmath44\nonumber\\ \times\left [ ( 1+g_{2n}g_{2n \pm 1})i_n+ ( 1-g_{2n}g_{2,n\pm 1 } ) \sigma_{xn}\right ] .
\qquad \label{v}\end{aligned}\ ] ] here , pauli matrices transform as @xmath45 , @xmath46 , and @xmath47 .
the diagonal terms of ( [ v ] ) represent the polaron transfer within one level while the off - diagonal ones represent the interlevel polaron transfer through the lattice .
evidently , the contribution of the off - diagonal terms proportional to @xmath48 is much smaller when compared with those proportional to @xmath49 . because of nonconservation of the number of coherent phonons , they are able even in the ground state to assist electron transitions between the levels . in the hamiltonian ( [ diagh1 ] ) ,
the operator @xmath50 ( [ un ] ) , highly nonlinear in the phonon-@xmath33 appears mediated by phonons @xmath34 .
it introduces multiple electron oscillations between the split levels mediated by _ continuous virtual absorption and emission of the phonons @xmath33_. the effect is analogous to rabi oscillations in quantum optics due to photons @xcite .
let us note that rabi oscillations assist both the interlevel onsite and intersite electron transitions mediated by the electron transfer @xmath51 .
further we shall investigate by variational approach the two variants of the model ( [ hams1 ] ) - the lattice and the local one ( considering @xmath52 ) .
shore et al @xcite and wagner et al @xcite proved for exciton or dimer models coupled to one phonon mode that the two - center wave function generalized to an asymmetric nonunitary ansatz with a variational parameter @xmath53 of a form @xmath54 ( where @xmath55 is a normalization constant ) , yields lower ground state @xcite , @xcite .
so , rather generally we define phonon wave functions @xmath56 @xmath57 here , the generators of variational displacements are defined @xmath58 , \label{8}\ ] ] and those of squeezing @xmath59 , \label{9}\ ] ] for @xmath60 . in eq .
( [ diagh1 ] ) coupling of the phonon modes @xmath33 and @xmath34 occurs , therefore one includes into the ansatz ( [ 7 ] ) also the correlation generator @xmath61 .
\label{10}\ ] ] with a correlation vp @xmath62 .
the form ( [ 7 ] ) is written here seemingly for the local case , while the most general form of it should contain the dependencies of all variational parameters on @xmath63 ( site number in the lattice ) : @xmath64 etc , showing the displacement of the mode @xmath33 at the site @xmath65 due to the electron at the site @xmath63 , @xmath66 .
we take @xmath67 where @xmath68 are independent of @xmath63 .
( [ q ] ) indicates that the phonon displacement accompanies the electron at the site @xmath63 .
the same is valid for the squeezing , @xmath69 and mixing parameter @xmath62 .
when we consider the ground state of the lattice model , we omit the electron and phonon dynamic terms by setting @xmath70 , @xmath71 , thus even in the lattice case we are left with the effective local molecule hamiltonian , the only difference being the additional contribution of the transfer terms ( those containing @xmath51 ) .
as it is shown later , in the lattice model we can set @xmath72 without worsening considerably the variational results , while for the local case @xmath52 introducing this parameter considerably improves the results . in what follows the ground state energy as function of optimized variational parameters @xmath53 , @xmath73 will be determined .
first we consider the lattice model setting @xmath72 . by averaging @xmath74 ( ( [ diagh1 ] ) , ( [ diagh2 ] ) ) over the phonon wave functions ( [ 6 ] ) with ( [ 7 ] ) one obtains for the site hamiltonian ( [ diagh1 ] ) the local part in the form @xmath75 where @xmath76 @xmath77 } { 1+\eta^2 + 2\eta\exp(-2|\tilde \gamma_{1}|^2 ) } \gamma_{2 } i \
, , \label{hav12}\end{aligned}\ ] ] where @xmath78 and @xmath79 are @xmath80- and @xmath36-dependent parts of the interaction @xmath81 ( [ hav12 ] )
. from the transfer hamiltonian @xmath82 ( [ diagh2 ] ) there remains @xmath83 where @xmath84 : @xmath85 .
\left [ ( 1+e_2 ^ 2 ) i + ( 1-e_2 ^ 2)\sigma_{x } \right ] , \label{t2}\end{aligned}\ ] ] and @xmath86 are given by @xmath87 expressions @xmath86 ( [ ei ] ) in the ground state are independent of @xmath88 and @xmath63 because of the form of @xmath89 ( [ q ] ) .
this substantially simplifies subsequent calculations leaving @xmath90 and @xmath91 as functions of @xmath92 independent on @xmath63 .
the transfer terms @xmath93 in ( [ havt ] ) with ( [ t2 ] ) are expressed as @xmath94 here , @xmath95 and @xmath96 are diagonal and off - diagonal terms of ( [ havt ] ) , respectively . the effective polaron potential in ( [ hd ] ) ( with ( [ hav2])-([ei ] ) ) as a highly nonlinear function of @xmath97 and @xmath98 is visualized in fig . 1 . the potential exhibits two sets of minima related to two competing ground states : + ( i ) two nonequivalent broad minima related to both the levels ( [ 7 ] ) at @xmath99 and @xmath98 close to @xmath100 ; + ( ii ) one narrow minimum at @xmath97 close to @xmath100 , @xmath101 , where both levels approach close together .
this minimum develops at growing @xmath36 ; evidently its behaviour depends also on @xmath51 value because of nonlinearity of the debye - waller factor @xmath84 ( ( [ t2 ] ) and ( [ ei ] ) ) .
the ground state energies related to these two sets of competing minima were calculated numerically .
the result of the numerical minimalization of the diagonalized form of energy ( [ hd ] ) @xmath102^{1/2 } \label{ef}\end{aligned}\ ] ] can be written symbolically as @xmath103 here , the index @xmath104 denotes the optimized values .
the model parameters @xmath105 are parameters of the effective interaction , asymmetry and nonadiabaticity , respectively
. energy @xmath106 ( [ hg ] ) is in scaled units , renormalized by @xmath51 .
the results of the numerical evaluations of the ground state energy ( [ hg ] ) are depicted in figs .
2a , 2b different model parameters .
one can distinguish there two regions depending on the value of @xmath107 with different behaviour of ground state in each of them : + ( i ) the ground state pertaining to the lower broad minimum at @xmath108 and @xmath109 with a small reflection part at @xmath110 is referred to as a `` heavy '' region .
it corresponds to predominantly intralevel `` heavy '' polaron which is represented by a two - peak wave function , both peaks representing a harmonic oscillator ( @xmath111 $ ] ) displaced by the value @xmath112 ( as it is seen from the form of the ansatz ( [ 7 ] ) ) .
the `` heavy '' polaron is dominant at @xmath113 where the broad minimum @xmath114 is dominating .
+ ( ii ) close to @xmath115 , the energies of two minima go very close together ( their difference is of the order of the phonon energy @xmath116 ) .
they drop to one narrow minimum which represents a new ground state . _ close to @xmath115 , continuous transition to a new ground state occurs_. it stabilizes at @xmath117 , where ( optimized ) @xmath97 value is close to @xmath100 and @xmath118 .
this region is referred to as a `` light '' polaron region because , owing to the abrupt decrease of @xmath119 at @xmath117 , the effective mass of the intralevel polaron drops to almost its free - electron value , i.e. , _ the polaron selflocalization vanishes in the `` light '' region_. because of this `` undressing '' , the transport characteristics of the excited electron would increase . moreover , due to the tiny distance between the levels , their coupling takes place by the exchange of virtual phonons @xmath33 . at suitable conditions , in the excited state , when both levels are occupied by electrons of opposite spins ,
_ the mechanism of virtual phonon exchange implies the pairing of electrons , i.e. formation of `` light '' bipolarons _ .
the ground state energy , especially its behavior dependent on pairs of parameters @xmath107 , @xmath116 and @xmath107 , @xmath120 is illustrated in details in fig .
2 . while being weakly @xmath116-dependent , the energy strongly decreases with @xmath107 inside the `` light '' phase ( fig .
the position of the phase line is slightly shifted from @xmath115 at @xmath121 to higher values of @xmath107 with increasing @xmath116 .
this is consistent with the fact that the phonon fluctuations are most effective when the difference of the energies of the phases is of the order of the phonon energy .
the ground state energy in the `` heavy '' region ( @xmath113 ) is independent of @xmath107 , its decrease inside the `` light '' region ( @xmath117 ) is dependent on the effective coupling @xmath120 ( fig .
the energy decrease due to @xmath120 is stronger in the `` light '' phase ( depending on @xmath107 ) than in the `` heavy '' phase .
the region of importance of reflection measure @xmath122 and @xmath123 is illustrated in the fig .
we show the difference between the ground state energy with @xmath123 and @xmath53 omitted and ground state with four variational parameters calculated numerically . however , contribution due to @xmath123 @xmath124 of the maximum value in fig .
the said parameters turn out to be most relevant in e@xmath0e jt case ( @xmath115 ) .
the effects of @xmath122 and @xmath123 on the ground state is apparent only at moderate couplings in the `` heavy '' region , reaching their maximum at @xmath115 .
the narrow minimum ( the `` light '' region ) is resistant against @xmath53 . in order to demonstrate important properties of displacements ( @xmath125 and @xmath116 dependence ) it is sufficient , except of the region of importance of the reflection ( fig .
3 ) inside the `` heavy '' region apparent for @xmath98 , to calculate them in the limit @xmath126 . we shall use eq .
( [ ef ] ) approximated for the cases of both `` heavy '' and `` light '' polaron and obtain implicit expressions for @xmath127 which however are good to visualize their behaviour in both regions : \(i ) `` heavy '' polaron ( @xmath98 small , @xmath128 ) : @xmath129 } , % \frac{\beta}{2 t e_2 ^ 2 } \frac{e_1 % } { \left ( 1+e_1 ^ 2+\frac{\omega}{t e_2 ^ 2}\right ) } ; \label{ie1}\end{aligned}\ ] ] @xmath130 is small except of the region of fluctuations visualized in fig .
4 . one can see , that the selflocalization due to phonons @xmath131 in the `` heavy '' region @xmath132 clearly implies the `` cave '' in the ground state energy due to the reflection measure @xmath53 ( anharmonicity ) of the ground state ( fig .
\(ii ) `` light '' polaron ( @xmath97 small , @xmath133 ) : @xmath134 in the `` light '' region the fluctuations of @xmath97 are missing as well as the fluctuations of the energy .
this is consistent with the above result of the resistance of the narrow minimum against @xmath53 . for both `` heavy '' and `` light '' polaron a dependence of @xmath135 on the nonadiabaticity parameter @xmath136 appears .
it implies the dependence of the debye - waller factor and consequently of the polaron mass on the phonon frequency @xmath116 .
this can be thought of as an analogy of the _ isotope effect _ at zero temperature . in the `` heavy '' phase ,
electron transitions mediated by phonons @xmath34 to the upper level enhance fluctuations @xmath137 which mix phonons @xmath34 with phonons @xmath33 and contribute to the fluctuations of the ground state energy of the heavy region .
this is the reason for the similarity of the results in fig .
now we consider the same hamiltonian , but taken for the local molecule ( @xmath52 ) . from the outset
we include the additional variational mixing parameter @xmath62 into the ansatz .
the understanding of the importance of introducing the mixing parameter in the local model in the contrast to the lattice one can be gained if one examines the expressions for hamiltonians , in particular , those terms which contain the mixing of two phonon modes .
the mixing is contained mostly in the `` local '' term with @xmath36 , @xmath138 the transfer term @xmath51 also contains the mixing , but it always enters the expression in the form @xmath139 , thus for the transfer term its contribution is always shadowed by the larger term @xmath140 .
thus , if @xmath141 and large enough , its contribution is always dominant over mixing term @xmath36 .
but if @xmath52 , we are left with the only nonlinear coupling term @xmath142 and accurate accounting for mode mixing becomes crucial . the variational mean value of the part of the hamiltonian ( [ diagh1 ] ) in the state ( [ 6 ] ) renormalized by @xmath116 yields ( appendix a ) @xmath143}{\cosh 2\lambda } \left\ { -\tanh 2\lambda\sinh 2\lambda \cosh 2(r_1+r_2 ) \cosh 2(r_1-r_2)\right.\nonumber\\ & & \left.+\tilde\zeta_1 ^ 2[(\exp 2(r_1+r_2)-\exp(-2(r_1+r_2))\cosh 4\lambda ) ( 1+\tanh^2 2\lambda)\right.\nonumber\\ & & \left.+2 \exp(-2(r_1+r_2))\sinh 4\lambda\tanh 2\lambda]\cosh 2(r_1-r_2)\right.\nonumber\\ & & \left.+\frac{\tilde\zeta_1 ^ 2}{\cosh^2
2\lambda}(\sinh 4r_1-\sinh 4 r_2 ) -2\zeta_1(\tanh ( 2\lambda ) e^{2(r_2-r_1)}\zeta_2+\zeta_1)\right\ } \\ & & + \frac{2\alpha}{\omega c } ( 1-\eta^2 ) \zeta_1 \nonumber \\ & & -\frac{2\beta}{\omega c } \left\ { ( 1+\eta^2)[\zeta_2 -\zeta_1\tanh 2\lambda \exp 2 ( r_2-r_1)]\frac{\exp \left
( -\frac{2\tilde \zeta_1 ^ 2}{\cosh 2\lambda } \right ) } { \cosh 2\lambda } + 2 \eta \zeta_2 \right\ } \nonumber \ , , \label{11}\end{aligned}\ ] ] where @xmath144 from eq . ( [ 11 ] ) one can see that @xmath62 causes correlations of the selftrapping and tunneling dominated regions . for @xmath72 , eq .
( [ 11 ] ) turns to @xmath145\zeta_2 \label{13}\end{aligned}\ ] ] where @xmath146 for @xmath126 , @xmath147 and ( [ 11 ] ) becomes @xmath148 the ground states of ( [ 11 ] ) , ( [ 13 ] ) and ( [ 15 ] ) were found by minimalization of the expressions against the involved vps .
the respective ground state energies @xmath149 , @xmath150 , @xmath151 and @xmath152 will be compared mutually and with the exact value from numerical simulation in order to find out importance of the variational parameters @xmath153 in different regions of model parameters @xmath154 ( reflection ) and @xmath155 ( effective e - ph coupling ) .
the place of e@xmath0e jt model will clearly come out as an important special case .
in lattice electron - phonon models the nonadiabaticity parameter is the ratio of phonon frequency and the band width , @xmath156 .
the energy is scaled by @xmath51 and the effective coupling parameter ( ratio of the polaron energy and of the phonon frequency ) is @xmath157 ) . in the present local model ,
the nonadiabaticity parameter is the ratio of the frequency and the coupling parameter @xmath158 .
the energy was scaled by @xmath116 and the ratio of polaron energy and the frequency is then @xmath159 .
therefore , the reduction of the ground state energy of the local model due to selflocalization is stronger than that of the lattice model by the factor @xmath160 .
the energy of the tunneling @xmath161 , is comparable with the polaron energy at @xmath162 . because close to @xmath115 there is @xmath163 , the effects of all variational parameters for @xmath117 except of @xmath164 vanish .
quantum fluctuations due to @xmath165 and anharmonicity due to @xmath62 are concentrated in the crossing region close to @xmath166 and @xmath115 .
the ground state in the phase plane @xmath167 and @xmath120 shown in fig.4 exhibits two phases separated by the crossover line close to @xmath115 .
it means , that the effective polaron potential exhibits two minima which reflect positions of two levels governed by the model parameters @xmath125 and @xmath107 .
the minima coincide within the border of the regions close to @xmath115 .
the phase @xmath113 is @xmath107 independent , selftrapping dominated , with quantum fluctuations due to @xmath168 .
the phase @xmath117 is the phonon-2-assisted tunneling dominated region with continuum emission and absorption of virtual phonons-@xmath33 .
the phonon exchange couples the levels within one minimum .
the minimum is much more sensitive to the change of model parameters @xmath169 as well as to quantum fluctuations @xmath170 and @xmath62 .
electron in the selftrapping dominated region is trapped by the phonons-@xmath33 but due to the interactions mediated by phonons-2 the electron can fluctuate to the higher level .
due to the reflection symmetry of the phonons-2 continuum oscillations of the electrons at simultaneous emission and absorption of phonons-1 occurs .
these oscillations couple the levels and so the electrons into pairs .
this mechanism was described in a recent paper @xcite for a lattice model . from ( [ 15 ] ) , for small @xmath167 , one can approximately take @xmath171 , @xmath172 and for @xmath173 we are left with the result @xmath174 .
this dependence can be recognized in fig.5 .
comparing figs .
6a and 6b one can see that the mode correlation induced anharmonicity @xmath62 ( fig.6a ) is by one order larger than the contribution to the anharmonicity of the reflection level mixing @xmath53 fig.6b .
the correlation is most effective for weak effective couplings @xmath120 at @xmath175 . for large @xmath120
it contributes only close to @xmath115 , where it reveals maximum for all @xmath120 . in order to check the validity of the calculations using variational approach we performed also numerical diagonalization of the hamiltonian in the phonon-@xmath176 space .
we limit ourselves in @xmath177 phonon-1 states and @xmath178 phonon-2 states , thus state vector is @xmath179 dimensional . as numerical diagonalization results show , about 20 - 50 phonon states are sufficient for convergence .
we show the results of numerical diagonalization of hamiltonian matrix as function of @xmath107 for @xmath166 and @xmath180 . in the first case we took 20@xmath18120 state vector , while in the latter case to achieve satisfactory convergence ( especially for the tunneling - dominated region when @xmath117 ) we had to increase the number of phonons-2 up to 50 .
fluctuations of phonon-1 coordinate @xmath182 , @xmath183 and the conjugate momentum @xmath184 for @xmath126 can be easily calculated analytically .
the wave function of the phonon-@xmath33 has the gaussian form @xmath185.\ ] ] in fact , it is squeezed displaced harmonic oscillator if one does not take into account the parameter @xmath53 ; introducing @xmath62 does not invoke higher order nonlinearities with respect to phonon-@xmath33 coordinates .
thus @xmath186 , like it should be for an harmonic oscillator , and the expressions for second momenta have the form : @xmath187 and @xmath188 the shape of fluctuations due to the correlation @xmath62 ( [ 21 ] ) closely follows the shape of the curve in fig.7 , having its maximum close to @xmath189 and @xmath190 .
the similarity of the product of uncertainties of phonon coordinate and momentum displayed in fig.7 .
illustrates anharmonic source of the contribution due to the mode correlation .
the correlations @xmath173 manifest themselves in ( [ 20a ] ) and ( [ 21 ] ) as a phonon anharmonicity .
it is noteworthy to emphasize its quantum origin ( compare the note below eq .
( [ 1 ] ) ) .
the problem with @xmath191 needs special consideration and will be considered elsewhere .
a suitable choice of the variational wave functions for various electron - phonon two - level systems is a long - standing problem in solid state physics as well as in quantum optics . for two - level reflection symmetric systems with intralevel electron - phonon interaction
the approach with a variational two - center squeezed coherent phonon wave function was found to yield the lowest ground state energy .
the two - center wave function was constructed as a linear combination of the phonon wave functions related to both levels introducing new vp .
this symmetry implies coupling of the levels mediated by continuous virtual emission and absorption of phonons accompanying the electron tunneling between the levels .
this is an analogy of rabi oscillations in quantum optics .
investigation of referring properties ( tunneling rates , debye - waller factors ) was performed mostly for the models with onsite ( holstein ) electron - phonon interactions in dimer or exciton - phonon models and also for e@xmath0e jt model .
the methods used there were based on either combination of variational approach and unitary transformations or numerical diagonalization . in two - level models with phonon assisted tunneling
there appears coupling of both phonon modes mediated by rabi oscillations .
therefore , also the interlevel quantum correlations of the phonons must be taken into account .
this brings into the variational ansatz additional quantum vp .
the two - level system is characterized by strongly nonlinear effective potential provided by phonons .
the nonlinearity is governed by the bare hamiltonian parameters exhibiting a crossover from an asymmetric double potential well with two broad nonequivalent minima pertaining to the levels to a regime of one narrow minimum when the broad minima coincide into one potential well .
this effect is due to the effective attraction of the levels by virtual exchange of phonons .
much effort was expended also to improve one - center variational approach by including correlations of the concerning two squeezed coherent phonon modes of the jt model .
when compared to the lattice case , the local version of the model provides qualitatively similar effective potential .
this conclusion was to be expected : in the symmetric case minima of the nonlinear effective potential related with two levels are close together and so the quantum fluctuations there are most effective .
we considered both the mixing and squeezing correlation vp in order to find region of relevance of both parameters as functions of the interaction constants .
+ the support from the grant agency of the czech republic of our project no .
202/01/1450 is highly acknowledged .
we thank also the grant agency vega ( no .
2/7174/20 ) for partial support .
we used following formulas @xcite @xmath192 \label{23 } , \\ & & \langle 0|s_1(\zeta)d_{1}(\eta)b_1^{\dag m } \rangle=\frac{d^m}{d\lambda^m } \langle 0|s_1(\zeta)d_{1}(\eta ) \exp(\lambda
b_1^{\dag})\rangle |_{\lambda=0 } , \label{24}\\ & & s_{12}(\lambda)= t^{\dag}\left ( \frac{\pi}{4 } \right ) s_1 ( \lambda/2 ) s_2 ( -\lambda/2 ) t\left ( \frac{\pi}{4 } \right ) \label{25 } \\ & & t\left(\frac{\pi}{4}\right ) \left ( \matrix { b_1\cr b_2}\right ) t^{\dag}\left(\frac{\pi}{4}\right ) = \frac{1}{\sqrt 2 } \left ( \matrix { b_1-b_2\cr b_1+b_2}\right ) , \label{26 } \\ & & t^{\dag}|0\rangle = t|0\rangle = |0\rangle \label{27}\end{aligned}\ ] ] 10 m. c. m. obrien and c. c. chancey , am.j.phys .
61 ( 8) , 688 ( 1993 ) m. d. kaplan and b. g. vekhter , `` cooperative phenomena in jahn - teller crystals '' ( ed .
j. p. fackler , plenum , new york and london 1995 ) .
o. gunnarson , phys .
lett . 74 , 1875 ( 1995 ) ; o. gunnarson , rev .
69 , 575 ( 1997 ) .
h. b. shore and l. m. sander , phys .
b 7 ( 10 ) , 4537 ( 1973 ) m. sonnek , t. frank and m. wagner , phys .
b 49 ( 22 ) , 15637 ( 1994 ) c.f .
lo , phys .
rev . a 43 ( 9 ) , 5127 ( 1991 ) c.f .
lo and r. sollie , phys.rev .
b 44 ( 10 ) , 5013 ( 1991 ) r.l .
fulton and m. gouterman , j. chem .
35 , 1059 ( 1961 ) .
b. w. shore , and p. l. knight , j. mod .
40 , 1195 ( 1993 ) ; i.i .
rabi , phys.rev .
49 , 324 ( 1936 ) ; phys .
51 , 652 ( 1937 ) .
e. majernkov , j. riedel and s. shpyrko , phys .
b 65 ( 17 ) , 174305 ( 2002 ) b. l. schumaker , phys .
133 ( 6 ) , 317 ( 1986 ) | ground state of the quantum jahn - teller model with broken rotational symmetry was investigated by the variational approach in two cases : a lattice and a local ones .
both cases differ by the way of accounting for the nonlinearity hidden in the reflection - symmetric hamiltonian . in spite of
that the ground state energy in both cases shows the same features : there appear two regions of model parameters governing the ground state : the region of dominating selftrapping modified by the quantum effects and the region of dominating phonon - assisted tunneling ( antiselftrapping ) . in the local case ( i )
the effect of quantum fluctuations and anharmonicity due to the two - mode correlations is up to two orders larger than contributions due to the reflection effects of two - center wave function ; ( ii ) the variational results for the ground state energy were compared with exact numerical results . the coincidence is the better the more far away from the transition region at the e@xmath0e symmetry where the variational approach fails .
* ground state of quantum jahn - teller model : * * selftrapping vs correlated phonon - assisted * * tunneling * eva majernkov@xmath1 , and s. shpyrko@xmath2 @xmath3department of theoretical physics , palack university , t .
17 .
listopadu 50 , cz-77207 olomouc , czech republic @xmath4institute of physics , slovak academy of sciences , dbravsk cesta , sk-84 228 bratislava , slovak republic * list of content * 1 .
introduction 2 .
extended ( lattice ) generalized jahn - teller model 3 .
ground state of the lattice model 4 .
ground state of the local model 5 .
discussion of the numerical results 6 .
quantum fluctuations in the local model 7 . |
relativistic shock waves are widely thought to generate energetic particles producing high - energy emission in many astrophysical sources .
the example of such sources can be hot spots in radio galaxies , jets in blazars and in galactic microquasars " , gamma - ray burst sources , and pulsar winds .
it is also believed that high - energy cosmic rays have their origin in astrophysical environments in which relativistic shocks occur .
the aim of this paper is to present the current status of cosmic ray acceleration theory at relativistic shock waves and to briefly review and comment on the models proposed for ultra - high - energy cosmic ray production at such shocks .
the basic acceleration mechanism discussed in the context of cosmic ray production at shocks is the first - order fermi process .
this process can be applied to sufficiently energetic particles , with gyroradii much larger than the shock thickness , defined by gyroradii of `` thermal '' ions present in the plasma .
such energetic particles see the shock front as a sharp discontinuity in the plasma flow . in the following , we review the results obtained mainly in the test particle approach , in which the influence of accelerated particles on the physical conditions at the shock is not taken into account . in the first - order fermi acceleration process particles gain their energies by consecutive crossings of the shock front . in order to do so
, they have to be confined near the shock .
the particle confinement is provided by the turbulent magnetic fields which perturb particle trajectories leading to particle diffusion in pitch - angle . during the acceleration process some fraction of particles transmitted downstream escapes from the vicinity of the shock .
the competition between particle energy gains at shock interactions and the escape process determines the stationary particle spectrum , often of the power - low form . in the case of a nonrelativistic shock , where the fluid speeds are much lower than the energetic particle velocity ,
the resulting particle spectrum is independent of the background conditions near the shock , including the configuration of the regular magnetic field , the spectrum and the amplitude of mhd turbulence .
this is mainly because of a nearly isotropic form of the particle distribution function at the shock , in the conditions where magnetic field perturbations near the shock are sufficient to ensure efficient particle scattering . in such conditions , the spectral index @xmath0 for the phase - space distribution function
is given exclusively by the shock compression ratio @xmath1 , and @xmath2 , where @xmath3 is given in the shock normal rest frame , and @xmath4 and @xmath5 are the shock velocities with respect to the upstream and downstream plasma rest frames , respectively .
the physical picture is much more complicated for relativistic shocks , where the shock velocity or its projection along the upstream mean magnetic field @xmath6 ( @xmath7 the upstream magnetic field inclination angle to the shock normal ) is comparable to the speed of light .
the energetic particle distribution becomes anisotropic near the shock , and anisotropy increases with growing shock lorentz factor @xmath8 .
this fact substantially influences the resulting particle spectrum , which is then very sensitive to the background conditions .
studies of particle acceleration processes at mildly relativistic shock waves started with the works of @xcite and @xcite for parallel shocks ( @xmath9 ) .
they showed that the spectral indices are different from the @xmath10 value obtained in the nonrelativistic case , and depend on the form of the wave power spectrum of the magnetic field perturbations ( see also * ? ? ? * ) .
acceleration processes in oblique subluminal ( @xmath11 ) shock waves were analyzed by @xcite under the assumption of magnetic moment conservation for particles interacting with the shock .
this assumption restricted validity of their considerations to the case of a weakly perturbed magnetic field , where cross - field diffusion does not play a significant role .
they showed that in such conditions particle spectra can be even as flat as @xmath12 in cases where @xmath13 is close to the speed of light .
this feature results from effective multiple reflections of anisotropically distributed upstream particles from the compressed field downstream of the shock .
the feature discussed is also visible in fig .
1 , which shows results of monte carlo particle simulations performed by @xcite . in the figure ,
one may note the lack of results for superluminal ( @xmath14 ) shocks propagating in a weakly perturbed magnetic field .
in such conditions , particles are tied to the field lines .
therefore , the upstream particles can be only transmitted downstream with no possibility of returning to the shock , so the first - order fermi process does not operate in this case and the power - law particle spectrum can not be formed . nevertheless , because of anisotropy of particle distribution at a relativistic shock , particle energy gains in a transmission process can be much larger than those resulting from the adiabatic compression at the shock @xcite . in the effectively accelerating astrophysical shocks
one may expect large - amplitude mhd waves to occur .
the role of such finite - amplitude magnetic field perturbations in forming a particle spectrum was investigated by a number of authors using monte carlo simulations ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the derived power - law particle spectra directly depend on the conditions near the shock
. they can be either very steep or very flat for different configurations of the mean magnetic field with respect to the shock normal and different amplitudes of perturbations , as shown in fig .
1 . note also , that the changes of the particle spectral index can be nonmonotonic with an increasing field perturbations amplitude @xcite . in the case of a highly perturbed magnetic field
, power - law spectra can be formed for superluminal shocks .
they are , however , very steep for mildly perturbed conditions at the shock .
the first - order fermi process at ultrarelativistic shocks has been recently discussed by a number of authors ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this topic is of a great importance for gamma - ray burst external shocks astrophysics and uhecr generation in these sources .
for @xmath15 the energetic particle distribution function is highly anisotropic at the shock in the upstream plasma rest frame .
this stems from the fact that when a given particle crosses the shock upstream , its momentum direction has an opening angle around the shock normal @xmath16 , thus it is almost aligned with the shock propagation direction . in such conditions ,
a small perturbation to the particle trajectory , provided by its momentum deflection in the upstream regular magnetic field or scattering off the mhd fluctuations , allows for its being overtaken by the shock and transmitted downstream . as a result ,
the mean particle energy gain in a single acceleration loop upstream - downstream - upstream is @xmath17 @xcite , which restricts seriously the efficiency of the acceleration process .
efficient particle acceleration is possible when particle distribution upstream is close to isotropy .
however , such conditions can be met at the ultrarelativistic shock wave only at the particle first shock crossing .
then , the mean energy gain for particles reflected from the shock is very large , @xmath18 , but it should be stressed that the expected efficiency of such reflection is rather low @xcite . as explained above , the conditions at the shock do not allow for particle distribution isotropization in subsequent shock crossings , even if a highly perturbed medium exists upstream of the shock .
the important topic in the context of uhecr production at high-@xmath8 shock waves is the existence of the asymptotic spectral index for relativistic shock acceleration .
all the above mentioned works on the first - order fermi process at such shocks yield consistent estimates of the accelerated particle spectral index @xmath19 , in the limit of the very high shock lorentz factors ( @xmath20 ) .
the models of the burst afterglow spectra often give results pointing to essentially the same value of @xmath0 for synchrotron radiating electrons , which is sometimes interpreted as observational confirmation of the correctness of theoretical models proposed for ultrarelativistic shock acceleration .
however , in all discussed derivations there is explicit or implicit assumption of a highly turbulent conditions near the shock ( see * ? ? ?
* ) . it is uncertain whether such conditions are indeed met in astrophysical shocks . on the other hand , in conditions with medium - amplitude magnetic field perturbations ,
particle spectra generated at oblique realistic shocks with @xmath21 can be much steeper than those obtained in the asymptotic limit , as seen in the simulations of @xcite ( see also * ? ? ?
in such conditions , the shocks will be unable to efficiently accelerate particles in the first - order fermi process .
the factors important for the first - order fermi processes at relativistic shocks are the presence and the configuration of the mean magnetic field , the structure of the turbulent field component , and also the presence of the long - wave magnetic field perturbations .
the acceleration studies presented above apply very simple approaches for the modeling of the turbulent mhd medium near the shock .
their perturbed magnetic field models either lack some of the factors mentioned above or take them into account in a very simplified way . in some models , the continuity of the perturbed field across the shock is also not preserved .
this factor is , however , very important for the acceleration process since it can introduce the correlations in particle motion on both sides of the shock .
there is thus a need for a more detailed analysis of the first - order fermi process at relativistic shocks , which would include all the important factors to yield the better insight into the real physical situation at the astrophysical shocks .
an attempt at doing such analysis for mildly relativistic shocks has been recently performed by @xcite .
we considered the first - order fermi process at shocks propagating in `` more realistically '' modeled perturbed magnetic fields .
the model assumes a wide wavevector range turbulent field component with the power - law spectrum imposed on the uniform magnetic field component upstream of the shock .
the continuity of the full perturbed field across the shock is preserved by the use of the respective shock jump conditions . in our modeling ,
the particle spectra were derived with the method of monte carlo simulations by integrating particle trajectories in such magnetic field , with a hybrid method enabling to treat both long- and short - wave perturbations . for ( a ) the flat and ( b ) the kolmogorov wave spectrum of the magnetic field turbulence .
the upstream perturbation amplitude @xmath22 is given near the respective curves .
linear fits to the spectra are also presented and values of the spectral indices @xmath0 are given .
the spectra have vertical shifts for clarity .
particles with energies in the range @xmath23 can effectively interact with the field perturbations . ]
figure 2 summarizes the results for oblique superluminal shocks , where the particle spectra for the shock with @xmath24 @xmath25 are shown as an example .
this is a representative example for relativistic shocks , because most field configurations in such shocks lead to analogous superluminal conditions @xcite .
one can note that particle spectra diverge from a simple power law in the full energy range .
for the small turbulence amplitude the particle spectrum is essentially the compressed upstream injected distribution .
power - law sections in the spectra are produced at larger turbulence amplitudes , but in our simulations they are formed by a small number of particles .
the spectra steepen with growing energy , but still within the resonance range , and are followed by the cutoffs .
these features arise due to the finite wavevector range of the turbulence considered in the simulations . in the perturbed magnetic field with the limited dynamic range of field perturbations , scattering conditions vary with particle energy .
low - energy particles have their resonance wavevectors close to the maximum wavevector value @xmath26 , so that most of the turbulence reside in the long - wave range for this particles .
these long - wave , high - amplitude magnetic field perturbations form _ locally _ oblique , subluminal field configurations at the shock , which enable acceleration of particles interacting with the shock in the locally subluminal conditions ( see also * ? ? ?
the acceleration process becomes less efficient for high - energy particles due to decreasing amount of the respective long waves for these particles , which results in the spectrum steepening and the cutoff .
this feature is independent of the initial energy of particles injected at the shock @xcite , suggesting that actually the turbulent magnetic field structure does not allow for particle acceleration to higher energies . from the above considerations it is evident that the critical component of any shock acceleration study is the applied model of the perturbed magnetic field near the shock .
the real magnetic field structure and its generation mechanism at a relativistic shock front are poorly known .
therefore , an understanding of these and other related issues ( e.g. , particle injection , the shock structure ) is the basic condition of reaching the real progress in the study of cosmic ray acceleration at relativistic shock waves .
a noticeable advance in this field results from the application of particle - in - cell ( pic ) simulations , developed to study `` microphysical '' properties of relativistic shocks @xcite .
as the theoretical considerations show @xcite , the turbulent magnetic field can be generated locally at the shock via relativistic two - stream ( weibel ) instability .
the source and the mechanism of the field generation is kinetic in nature , as the instability is driven by the anisotropy of particle distribution function at the shock associated with a two - stream motion of the two plasma particle populations : the first formed by the upstream particles inflowing onto the shock front and the second composed of particles reflected from the shock . the more detailed insight into the dynamics of the weibel instability give us particle - in - cell simulations , in which two relativistic plasma collide with each other to form a shock .
the structure of the small - scale strong turbulent magnetic field , generated downstream of the shock , is very complicated spatially and , in addition , changes with time .
the field is predominantly transversal and the total magnetic energy becomes a substantial fraction of the equipartition field energy .
the initial perturbations grow in size and the generated field structures are transported downstream . at the given point downstream of the shock and a given time long after collision , the power spectrum of the turbulent magnetic field is a power - law , with the most power concentrated at long wavelengths .
the scales associated with the magnetic fields thus generated are , however , small compared to gyroradii of particles undergoing the first - order fermi acceleration .
nonetheless , the magnetic field structure evolution is probably accompanied by plasma heating and non - thermal particle acceleration , thus possibly providing the seed particles for the fermi process .
the pic simulations can be therefore treated as a first step towards understanding of processes important for the study of cosmic ray acceleration at relativistic shocks . in summary ,
the present theoretical knowledge of cosmic ray production at relativistic shocks is incomplete .
existing theoretical models explain only some features of the observed cosmic rays . because of the strong dependence of particle spectra on conditions at the shock
, the models do not allow for reliable modeling of astrophysical objects in which relativistic shock waves occur .
further progress requires an increase in the number and the quality of observational data revealing the physical parameters near the shock . on the other hand ,
there is a need for advance in theoretical studies , which most probably is to be made through advanced numerical simulations .
particle - in - cell methods are appropriate to study the microphysics of relativistic shocks .
realistic modeling of particle acceleration at astrophysical shocks has to incorporate results of such studies and by itself requires a full plasma nonlinear description .
this should take into account appropriate boundary conditions , modification of the shock structure by backreaction of accelerated particles ( see * ? ? ?
* ) and second - order acceleration processes .
extragalactic astrophysical objects harboring relativistic shock waves are the likely sources of ultra - high - energy cosmic rays , with energies in the range @xmath27 ev . in the light of the apparent drawbacks in the theory of particle acceleration at relativistic shocks ,
it is clear that it is now vary hard to judge with certainty which mechanism proposed for uhecr production is in fact responsible for the observational data on cosmic rays .
nevertheless , it is worthy to comment on some proposals on this topic .
fireball models of gamma - ray bursts assume the prompt burst emission in gamma rays to be produced at internal mildly relativistic shocks ( @xmath28 ) and the afterglow emission to originate at the blast wave propagating with initially ultrarelativistic velocity ( @xmath29 ) into the external medium ( see , e.g. , * ? ? ?
the evidence for ultra - high - energy cosmic ray production at grb sources is based on two claimed coincidences @xcite .
first , the constraints imposed on fireball model parameters , inferred from prompt gamma - ray and early afterglow observations , are similar to the constraints imposed on the shock in order to allow for proton acceleration to the energies above @xmath30 ev .
second , energy generation rate in gamma rays by gamma - ray bursts is similar to the rate required to account for the uhecr energy density .
the model of @xcite assumes cosmic ray acceleration to very high energies by the first - order fermi processes ( see also * ? ? ?
* ) at internal shocks .
the claimed agreement of model estimates with the observed uhecr flux and spectrum is however based on an oversimplified assumption about the spectrum of accelerated protons , which is taken as the one derived for non - relativistic shocks , where @xmath10 .
such assumption can not be justified for mildly relativistic shocks , where spectral indices are predicted to be much steeper in the most probably realized cases of superluminal shock configuration . moreover , the medium into which an internal shock propagates may be already modified , either by previous shell collisions , leading to a highly relativistic temperature of the upstream plasma , or by magnetic field amplification processes . in the latter case , the close to equipartition magnetic fields may occur near the shock , so that the magnetic field may play a non - negligible dynamical role .
both effects lead to the steeper particle spectra than those obtained at unmodified shocks @xcite .
note also , that in conditions of a highly turbulent close - to - equipartition fields near the relativistic shock the _ second - order _ acceleration processes may play a significant role in uhecr production @xcite , modifying the resulting particle spectrum .
the proposal of uhecrs production at grb external ultrarelativistic shocks , based on a claimed evidence of the first - order fermi acceleration seen in burst afterglow data ( see sec .
2 ) , was first advanced by @xcite .
analysis performed by @xcite shows however , that the fermi process is inefficient in accelerating particles when the blast wave propagates in the galactic - like interstellar medium .
the maximum energy a particle can gain in this case is @xmath31 ev .
therefore , in order to reach higher particle energies , the acceleration process must operate in the region with much stronger magnetic fields . note , however , that the generation of high - energy particles at ultrarelativistic shocks probably requires the presence of long - wave field perturbations ( see * ? ? ?
* ) , even if a highly turbulent medium exists near the shock .
ultra - high particle energies can be also reached at grb external shocks if there are energetic particles preexisting in the medium into which a blast wave propagates .
these particles , when reflected from the shock , examine the @xmath32 energy boost .
the particle reflection process is efficient if the energy density content in relativistic particles @xmath33 is sufficiently large , as the fraction of the blast wave energy that can go into boosting these particles to uhecr energies is proportional to @xmath34 .
cosmic rays of energy above @xmath35 ev ( sufficient to reach @xmath30 ev if @xmath36 ) in the galactic - like interstellar medium comprise only a very small fraction @xmath37 of the total ism energy density , and the reflection process can not account for the observed uhecrs , as indicated by @xcite .
more promising scenario is when the fireball blast wave expands into the pulsar wind bubble .
then , the bubble energy density is dominated by energetic ions , which can be boosted to uhecr energies with high efficiency . as shown in @xcite , spectra of the boosted ions
are determined by the blast wave deceleration , and the evaluated spectral index is @xmath10 in this case .
this leads to the uhecr spectrum compatible with observations .
note , however , that predictions of the presented model are based on a number of assumptions about physical conditions which lead to the most efficient acceleration . to be considered seriously as the source of uhecrs
, this mechanism should be modeled for definite realistic conditions , including the magnetic field structure in the pulsar wind bubble .
the possible sites for uhecr production are also the hot spots of frii radio sources .
these are believed to be the downstream regions of mildly relativistic shocks formed by interaction of relativistic jets with intergalactic medium .
the model of cosmic ray production at hot spots was presented by @xcite , who assumed that protons are accelerated in the first - order fermi process at non - relativistic parallel shock . from comparison of the acceleration and the radiative loss time scales , they estimated @xmath38 to lie in the uhecr range .
they show that in order to explain the observed uhecr flux , the particle spectral index at the shock must be on average @xmath39 or slightly flatter .
although this is in agreement with recent results obtained on cosmic ray acceleration at mildly relativistic parallel shocks @xcite ( results for subluminal shocks may also apply given the typical hot spot shock speeds @xmath40 ) , the assumption made by rachen & biermann of the kolmogorov magnetic field turbulence downstream , extending up to the scales corresponding to gyroradii of the most energetic protons , may not be valid ( see sec .
2 ) . moreover , it is not certain how much the finite size of the hot spots , and corresponding diffusive particle losses , influence the accelerated proton spectrum .
authors estimates of the critical energy above which these losses lead to the spectrum cutoff give ( for a hot spot size @xmath41 kpc and the magnetic field @xmath42 mg ) @xmath43 ev for diffusion parallel to the mean magnetic field and @xmath44 ev for cross - field diffusion .
these estimates depend on the acceleration time scale , which may be smaller for oblique shocks @xcite , but even if the perpendicular field configuration is assumed , the critical energy can not be much larger then @xmath45 ev ( see * ? ? ?
note also , that in that case the proton spectrum is probably very steep .
it is worthy to mention about an interesting non - standard fermi mechanism of particle acceleration proposed recently by @xcite .
this mechanism takes advantage of multiple particle conversions ( photon - induced or resulting from nucleon collisions ) from the charged state into neutral state and back . the charged particle ( proton ) downstream from the shock , when converted into neutral state ( neutron ) ,
can get into the upstream region with no influence from the magnetic field . there , conversion to the charge state may occur and the particle is magnetically confined to the plasma flow . at the moment of conversion , the particle can be far from the shock , so it may be deflected by a large angle before reaching the shock again .
the distribution of such particles is then isotropic upstream , what enables them to increase their energies in a whole cycle by a factor of @xmath46 . because of this
, the process can be very efficient in producing uhecr particles , even if the efficiency of the converter mechanism alone is low .
it should however be investigated in more details , also using numerical simulations .
the purpose of this work was to present the current understanding of the first - order fermi acceleration processes at relativistic shocks .
the present knowledge in that matter is confronted with various proposals of uhecr production at sources harboring relativistic shocks .
it was not the aim of the work to fully discuss these models , we comment only on the features related to the assumed details of acceleration processes .
there are however additional conditions that should be met by the models to account for the observed uhecr characteristics .
any successful model , apart from being able to fit the spectrum of uhecrs , has also to match their composition and arrival direction distribution , account for possible small - scale clustering of events etc .
such the issues are described in detail in , e.g. , @xcite .
i am grateful to micha ostrowski for enlightening discussions and critical remarks on the manuscript .
the work was supported by the polish state committee for scientific research through the grant pbz - kbn-054/p03/2001 . [ 1 ] achterberg , a. , et al . : 2001 , _ mnras _ * 328 * , 393 .
[ 2 ] ballard , k. r. , heavens , a. : 1992 , _ mnras _ * 259 * , 89 .
[ 3 ] bednarz , j. , ostrowski , m. : 1996 , _ mnras _ * 283 * , 447 . [ 4 ] bednarz , j. , ostrowski , m. : 1998 , _ phys .
lett . _ * 80 * , 3911 .
[ 5 ] begelman , m. c. , kirk , j. g. : 1990 , _ astrophys .
j. _ * 353 * , 66 .
[ 6 ] derishev , e. v. , et al . : 2003 , _ phys .
d _ * 68 * , 043003 .
[ 7 ] dermer , c. d. , humi , m. : 2001 , _ astrophys .
j. _ * 556 * , 479 .
[ 8 ] drury , l. oc .
, et al . : 2001 , in _ proc .
. cosmic ray conf .
_ , k .- h .
kampert , g. heinzelmann & c. spiering ( eds . ) , copernicus gessellschaft , p. 2096 .
[ 9 ] ellison , d. c. et al .
: 1990 , _ astrophys .
j. _ * 360 * , 702 .
[ 10 ] ellison , d. c. , double , g. p. : 2002
, _ astrop .
phys . _ * 18 * , 213 . [ 11 ] frederiksen , j. t. , et al . : 2004 , _ astrophys . j. _ * 608 * , l13 .
[ 12 ] gallant , y. a. , achterberg , a. : 1999 , _ mnras _ * 305 * , l6 .
[ 13 ] heavens , a. , drury , l. oc . : 1988 , _ mnras _ * 235 * , 997 .
[ 14 ] kirk , j. g. , et al . : 2000 , _ astrophys .
j. _ * 542 * , 235 .
[ 15 ] kirk , j. g. , heavens , a. : 1989 , _ mnras _ * 239 * , 995 .
[ 16 ] kirk , j. g. , schneider , p. : 1987
, _ astrophys .
j. _ * 315 * , 425 .
[ 17 ] medvedev , m. v. , loeb , a. : 1999 , _ astrophys .
j. _ * 526 * , 697 .
[ 18 ] niemiec , j. , ostrowski , m. : 2004 , _ astrophys .
j. _ * 610 * , 851 .
[ 19 ] nishikawa , k .-
i . , et al . : 2003 , _ astrophys .
j. _ * 595 * , 555 .
[ 20 ] olinto , a. v. : 2000 , _ phys .
_ * 333 - 334 * , 329 .
[ 21 ] ostrowski , m. : 1991 , _ mnras _ * 249 * , 551 . [ 22 ] ostrowski , m. : 1993 , _ mnras _ * 264 * , 248 . [ 23 ] ostrowski , m. : 2002 , _ j. phys
. studies _ * 6 * , 393 . [
24 ] ostrowski , m. , bednarz , j. : 2002 , _ astron .
astrophys . _
* 394 * , 1141 . [ 25 ] ostrowski , m. , siemieniec - ozibo , g. : 2002 , _ astron .
astrophys . _ * 386 * , 829 .
[ 26 ] rachen , j. p. , biermann , p. : 1993
, _ astron .
_ * 272 * , 161 .
[ 27 ] schmitz , h. , et al .
: 2002 , _ astrophys . j. _ * 570 * , 637 .
[ 28 ] silva , l. o. , et al .
: 2003 , _ astrophys .
j. _ * 596 * , l121 .
[ 29 ] vietri , m. : 1995 , _ astrophys .
j. _ * 453 * , 883 .
[ 30 ] waxman , e. : 2004 , _ astrophys .
j. _ * 606 * , 988 . [ 31 ] waxman , e. : 1995 , _ phys .
lett . _ * 75 * , 386 .
[ 32 ] zhang , b. , meszaros , p. : 2004
* 19 * , 2385 . | the current status of the theory of particle acceleration at relativistic shocks is presented , and a few possible scenarios for ultra - high - energy cosmic ray production at such shocks are briefly discussed . |
the increasing amount of interest in quantum information processing attracted many experimentalists and theoreticians to investigate the transport properties of low dimensional charge systems . beyond interest to build quantum computational units , i.e. q - bits , and the calculation algorithms which are supposed to be used by these units , the coherency of the information processing , or in other words information transport , is an essential challenge . since some of the proposed units are defined on the 2deg , the information carriers are naturally the electrons . in the absence of an external magnetic field ,
the well - known and well - defined landauer channels @xcite are the best candidates for coherent transport .
these channels are by definition ballistic and 1d .
they are formed due to the size quantization of the system under investigation .
a similar non - interacting single particle picture commonly used to describe the transport , also in the presence of an external magnetic field , is the so - called landauer - bttiker edge states ( lb - es ) @xcite .
however , its relevance in explaining some of the recent quantum hall , in particular the local probe , experiments is questionable , where the importance of electron - electron ( @xmath0 ) interaction is shown to be dominant @xcite .
the simplest way of including @xmath0 interactions is the so - called electrostatic approximation ( esa ) which is based on the landau quantization introduced by the external magnetic field @xcite . in this approach , the 2deg
is split into two regions depending on the value of the fermi energy with respect to landau level energy , @xmath1 , where @xmath2 is the cyclotron frequency and @xmath3 is the effective mass of the electron ( @xmath4 ) , with landau index @xmath5 .
if the fermi energy equals one of the landau energies , which has a @xmath6fold degeneracy , the electronic system is quasi - metallic so - called compressible and screening is nearly perfect . otherwise , when the fermi energy lies in between two consequent landau levels , screening is poor and system is called to be incompressible . beyond the differences in their screening properties ,
the transport properties of compressible and incompressible regions are different as black and white .
first of all we should mention that , within a compressible strip ( cs ) the screened potential is ( ideally ) flat and density is varying , as a result of high density of states ( dos ) at the fermi energy , whereas within the incompressible strip ( is ) the potential is varying ( since the external potential could not be screened perfectly ) and electron density is constant ( see fig .
[ fig : fig1 ] ) .
the discussion on `` where the current flows ? '' is a long lasting question . as early as , the formation of cs / iss was realized ,
it was conjectured that , the current should flow from the is , since the screened potential has a gradient only within these regions @xcite and back scattering is suppressed @xcite .
however , later it was argued that since there are no states available at the fermi energy the current should flow from the css , where it is possible to generate an excitation @xcite .
moreover , the non - interacting limit of the cs was claimed to reduce to the landauer - bttiker formalism @xcite .
the recent local potential @xcite and transparency @xcite experiments supported the findings of the first model , where it was shown that the positions of the hall potential drops coincide with the predicted positions of the ( innermost ) iss within esa . even before the esa , a self - consistent interaction model was developed by r.r .
gerhardts and his co - workers @xcite , which excellently agrees with the experimental results after including the influence of an imposed current @xcite . in this model ,
a hartree type mean - field approximation was considered , which provides the local distribution of the electron density and potential profiles as a function of magnetic field and temperature @xcite .
a local transport model was incorporated to describe the current distribution , where the entries of the conductivity tensor were calculated from a reasonable conductivity model , e.g. self - consistent born approximation ( scba ) @xcite , provided that local electron distribution is known .
moreover , the position dependent electrochemical potential , thereby the local current distribution , was obtained within a local version of the ohm s law under the condition of a _ local equilibrium _ @xcite .
this self - consistent theory was implemented to many interesting 2deg systems successfully explaining the relevant experiments @xcite .
the recent quantum interference experiments considering 2des under strong magnetic fields , provided unexpected information on the transport electrons . in the mach - zehnder interference setups
@xcite , it was observed that the contrast in the interference oscillations ( visibility ) is path - length independent , which is in strong contradiction with the optical counterpart of the experimental setup .
the basic idea is to split a monochromatic ( mono - energetic ) light ( electron ) beam from a half - transparent mirror ( qpc ) and measure the interference after these two split beams interfere . in these experiments the lb - es are assumed to simulate the light beam and qpcs to replace the mirrors .
however , the out come of the experiments were not easy to explain within a naive single particle model , without many assumptions @xcite . another very interesting experiment concerning qpcs is performed at the westervelt group ( harvard ) , where a local probe technique is used to observe electron flow near a qpc @xcite .
there it was clearly shown that the actual realization of the sample is dominant in determining the transport taking place ( see the paper by t. kramer within this issue ) .
the essential physics was explained comprehensively within the wave - packet formalism at relatively low magnetic fields ( @xmath7 t ) , where one should not expect the formation of the incompressible strips .
however , if one considers higher magnetic fields one should in principle include the effect of @xmath0 interactions in a self - consistent manner . beyond the interference experiments , the transmission measurements performed at
a 2deg using a single qpc has shown peculiar effects @xcite .
the suppression or enhancement of the transmission amplitude presents significant deviation from the chiral luttinger liquid model @xcite , meanwhile this effect could be explained by considering an electron - hole asymmetry @xcite . in this treatment
the formation of a large incompressible region is assumed , however , the @xmath0 interaction is not taken into account . here
we provide , such a calculation scheme within thomas - fermi approximation . in sec .
[ sec : sec1 ] we briefly describe the self - consistent model to obtain the electrostatic quantities and discuss the limitations of this approach .
we then proceed with the formulation of the current flow within a local ohm s law in sec .
[ sec : sec2 ] .
we finalize , our work by showing the interrelation between the quantized hall plateaus ( qhp ) and the expected distribution of the is , and thereby local current , near a qpc , also in the out of linear response regime . a sketch of the density profile as a function of lateral coordinate ( upper panel ) and the corresponding energy dispersion , i.e. 1@xmath8 and 2@xmath9 landau levels ( lower panel ) .
the vertical dashed lines indicate the region where the incompressible strip ( is ) resides , whereas the horizontal line in the lower panel is the fermi energy .
the dashed curve in the upper panel is the density distribution , when one considers finite extend of the wave function ( dashed ( green ) curve in the lower panel ) and the corresponding curve calculated within the tfa . ]
the essential physics of a 2deg subject to strong perpendicular magnetic field is governed by the single particle hamiltonian h = h_0+v(x , y ) [ eq : ham]with the kinetic term h_0=(*p*-*a*)^2 [ eq : ho]where @xmath10 stands for momentum and @xmath11 for the vector potential .
the potential energy is composed of the external @xmath12 and interaction @xmath13 termsv(x , y)=v_ext(x , y)+v_h(x , y ) [ eq : vext].using a relevant gauge ( here we use the landau gauge to exploit the symmetry of the system ) and assuming a translational invariance , one ends with the reduced 1d hamiltonian in @xmath14 direction , which is nothing but a harmonic oscillator centered at @xmath15 .
the solutions of the hamiltonian in @xmath16 direction are properly normalized plane waves with quasi - continuous momentum @xmath17 @xmath18 , where @xmath19 is the sample length ) , and @xmath20 is the square of the magnetic length . by translational invariance we mean that , the quantization effects arising from the finite length of the sample are negligible . the solution of the schrdinger equation in @xmath14 direction results in the well - known landau wave functions _ n_x(x)= h_n ( ) , where @xmath5 is the landau index and @xmath21 the @xmath5th order hermite polynomial with the argument @xmath22 , whereas the eigen energies are e_n , x=_c(n+1/2).the external potential is obtained by solving the 3d poisson equation @xcite for the structure shown in fig.[fig : fig2]a .
the calculation scheme and its details of implementing to a qpc is far beyond the scope of this paper and is discussed elsewhere , however , the for the present work a reasonable confinement potential profile is sufficient to initialize the self - consistent scheme described below .
the hartree potential is defined from the electron distribution via , v_h(x , y)=_ak(x , y , x,y)n_el(x,y)dxdy,[eq : tfapotential]where the kernel @xmath23 gives the solution of the poisson equation at @xmath24 for a point charge residing at the point @xmath25 , assuming periodic boundary conditions in 2d and @xmath26 is an average dielectric constant defined by the sample parameters .
once the hamiltonian given in eq .
[ eq : ham ] is solved one can obtain the electron density from , n_el(x , y)=_n , k_y|_n , k_y(x , y)|^2f(e_n , k_y-^ * ) , where f(e ) is the fermi function and @xmath27 is the ( position dependent , in the presence of an external current ) electrochemical potential .
we first make the assumption that , the total potential varies smoothly on the scale of the magnetic length , i.e. thomas - fermi approximation ( tfa ) .
this allows us to replace the wave functions with delta functions , the eigen energies are then given by : e_n(x , y)=e_n+v(x , y).the above approximation results in a simpler description of the electron density , n_el(x , y)= [ eq : tfaden]within the tfa , where @xmath28 stands for the local density of states ( ldos ) .
it is more common to use the filling factor @xmath29 instead of the electron density , which is defined as @xmath30 .
the name itself is explanatory , filling factor describes how many of the low - lying landau levels is filled .
if it is a integer , it means that all the states below the fermi level are occupied ( i.e. incompressible ) , otherwise show how much of the top most landau level has available states .
the calculation procedure is as follows , we start with a given external potential generated by the donors and gates , then calculate the electron density distribution in the absence of an external magnetic field .
the average electron density , @xmath31 , is kept constant all through the calculation , which enables to have a convergence criteria .
this electron density initializes the potential given in eq.[eq : tfapotential ] , which in turn determines the electron density via eq.[eq : tfaden ] .
the numerical procedure is a bit more complicated than described , however , the details can be found in ref .
@xcite .
schematic draw of the sample under investigation .
the gates defining the qpc are deposited on the surface of the sample along @xmath16 direction with a separation @xmath32 nm , under a potential bias @xmath33 v , such that the 2deg beneath is depleted .
the width of the gates are @xmath34 nm and assumed to be thinner than 10 nm .
a homogeneous donor layer with a number density @xmath35 @xmath36 is laying 43 nm below the surface , which donates almost 10 percent of its excess electrons to the 2deg .
the resulting electron density of the 2deg is @xmath37 @xmath36 without gates . in order to satisfy the open boundary conditions a larger lattice
is considered in actual calculations spanned with @xmath38 mesh points in @xmath39 , @xmath40 and @xmath41 directions , respectively . ] before proceeding with the discussion of the self - consistent results obtained for a qpc geometry , we would like to make some comments on the weak points of the tfa .
the first issue is that , if an is becomes narrower than the magnetic length our approximation fails @xcite .
such a potential profile is shown in fig .
[ fig : fig1 ] , therefore one should avoid these situations and perform a spatial averaging over the magnetic length both for the density and the potential profiles as a first order approximation @xcite .
this approach has been used previously for these systems and is shown to be sufficiently powerful to simulate the effect of the quantum mechanical wave functions @xcite .
another issue about narrow iss , is that the ldos ( @xmath42 ) become broader , when considering steep potential drops , i.e. large transverse electric fields @xcite . due to this local broadening of the ldos
the landau gap is smeared out and as a result is disappears @xcite .
these two observations together , i.e. overlap of the wave functions and local broadening , justifies the validity of performing a spatial average over magnetic length or even over the fermi wavelength @xmath43 , which we also will perform through out this work . the sample that we investigate is shown in fig .
[ fig : fig2 ] , a multi - layer gaas is grown on top of the substrate , which is followed by a thick layer of algaas .
this layer is ( @xmath44- ) doped by silicon , which provide electrons to the 2deg formed at the interface of the two different materials . on top of the sample ( metallic ) gates
are patterned to manipulate the electron density at the 2deg .
if a sufficiently large negative potential is applied to these gates ( here patterned as a qpc ) the electrons underneath the gates are depleted , however , the potential profile satisfies the conditions of the tfa @xcite
. we will consider a geometric pattern of the gates such that the opening of the qpc is 150 nm and is smooth again on the length scales of the magnetic length .
spatial distribution of the electron density ( a - d ) together with an illustrative sketch of the hall resistance ( broken red line ) and the expected visibility ( dotted green line ) as a function of magnetic field strength ( e ) .
the solid black regions indicate an integer filling factor equal to 2 , i.e. an is , whereas the degrading gray scale denotes the compressible regions with varying electron density , whereas the colored ( pink ) region is the electron depleted region .
the initial potential calculated at @xmath45 is obtained for a geometry shown in fig .
[ fig : fig2 ] .
the characteristic @xmath46 values are chosen depending on the existence and the width of the iss , such that the resulting magnetic field strengths are @xmath47 t ( a ) , @xmath48 t ( b ) , @xmath49 t ( c ) , @xmath50 t ( d ) .
all the calculations are done at temperatures satisfying @xmath51 for @xmath52 . for the dos broadening parameter see the related text . ]
[ fig : fig3 ] shows a sequence of density distributions calculated at selected magnetic field values , for which the 2deg system is out of the @xmath53 qhp at low @xmath46 fields ( a ) , at the lower edge of the qhp ( b ) at the higher edge of the qhp ( c ) out of the qhp at high @xmath46 .
pedagogically , it is preferable to start the discussion considering high magnetic fields since the density distributions at @xmath45 and at strong @xmath46 are similar . at zero magnetic field a 2deg
is known to be quasi - metallic , due to high ( but not infinite ) dos , screening is perfect and the external potential is almost flat in the interior of the sample .
the situation is similar at high magnetic fields , since only the lowest ( spin degenerate ) landau level is partially occupied , hence the system acts as a metal and external potential is almost perfectly screened .
however , decreasing magnetic field explicitly implies that the degeneracy @xmath54 is also reduced therefore the lowest landau level can not accommodate all the electrons .
this essentially means that the fermi energy of the system lies between two consequent landau levels all over the sample , this situation corresponds to fig .
[ fig : fig3]c . now , once the higher landau level is being occupied the @xmath0 interaction forces the system to establish an electrostatic equilibrium by minimizing the energy .
therefore the is become narrower and higher filling fractions are present , fig .
[ fig : fig3]b . even lowering the degeneracy ,
these is become more and more narrow until their widths become smaller or comparable with the quantum mechanical length scales as observed in fig [ fig : fig1 ] .
these results point out clearly that even in a single qhp the edge states exhibit different behavior by changing their widths .
moreover , at a very clean sample ( where no long - range potential fluctuations are present ) the coherence of the edge state is lost due to averaging of the phase . on the other hand ,
if an is is well separated and sufficiently wide , it will become `` more '' coherent .
this is true , if the current is confined to these regions , which we will demonstrate in the next section .
we should also remind that our classical approach in defining the current disables us to define a phase coherent transport in the sense of lb - es .
although there are attempts in defining a coherent transport including spin - orbit coupling , due to large zeeman splitting the arguments of this model @xcite are rather energetically irrelevant . however , our hand waving arguments coincides with the findings of most recent mach - zehnder experiments @xcite , where the amplitude of the visibility is measured as a function of applied magnetic field .
the reduction of this amplitude is observed at the higher and the lower edge of the qhp , meanwhile a local maxima occurred closer to the higher edge .
in this section we briefly summarize the previously developed local ohm s law applied to a 2deg under quantized hall conditions .
this model assumes that the ohm s law is valid on the scales larger than the fermi wavelength and a local equilibrium is established .
it is essentially based on the argument that if the local potential and electron distribution is known one can calculate the local current distribution with in the linear response regime , i.e. at very small currents .
gray scale plot of the density ( a ) and current distribution within the @xmath53 quantized hall plateau .
an electric field is applied in the @xmath16 direction resulting a fixed current ( @xmath55 ) , where the 1d current density is set to @xmath56 a / m .
all the other parameters are the same as fig .
[ fig : fig3 ] for @xmath57 t. ] the local ohm s law assumes that the relation * e*(*r*)=(*r*).*j*(*r * ) is valid locally , i.e on the scales of the fermi wavelength .
the next step is to satisfy the maxwell equation * * e*(*r*)=0 and the equation of continuity * * j*(*r*)=0simultaneously , for a stationary situation . from the above equations one can obtain the local current distribution ,
if the local resistivity tensor is known for an external fixed currenti=_a*j*(*r*)d*r*.here , we assume that @xmath58 is nothing but the inverse of the conductivity tensor which is related to the electron distribution via some reasonable conductivity model . for consistency reasons we assume a gaussian broadened spectral function a_n(e)= , where @xmath59 is the broadening parameter defined by the impurity potential , which we set @xmath60 in order to smear out the narrow iss . for the given spectral function the longitudinal conductivity is calculated as _ l=_-^de_n=0^ ( n+ ) [ e^(-[]^2 ) ] [ eq : sigmal]whereas the hall conductivity is simply the drude result_h=[eq : sigmah].one can already see the essence of the local model related with the existence of the iss : the longitudinal conductivity vanishes if the filling factor is an integer , which is the case when considering an is .
this is a result of the landau gap , if the fermi energy lies in between two landau levels there are no available states at the fermi level .
therefore , the conductivity and consequently the longitudinal resistivity vanishes : * * gray scale plot of the electron ( a ) and current density ( b ) , considering a sufficiently large current density , @xmath61 a / m , considering @xmath57 t. the asymmetry is induced by this current is significant on the righthand side of the sample . ]
_ l(x , y)= = = 0.in the linear response regime we assume that the relation v((x , y);i)-v((x , y);0)^*((x , y);i)-^*_eq , holds , i.e. the imposed external current does not change the density and potential profile , thereby the position dependent electrochemical potential can still be considered as constant . in fig.[fig : fig4 ] we plot the current distribution corresponding to a density distribution shown in fig.[fig : fig3 ] .
it is apparent that the current is confined to the regions where a well - developed is is present .
therefore , at least in the linear response regime , once the spatial distribution of the iss is obtained the current distribution can be calculated mutually . however
, when the imposed current is strong enough to influence the density distribution , one should calculate the position dependent electrochemical potential @xmath62 via e(*r*)=^*(*r*)/e .this relation brings a new set of self - consistent equations , which have to be solved numerically . in our ( numerical ) calculation scheme , we start with a small total current and in each iteration step increase the strength of @xmath63 slowly so that a convergence is guaranteed . the details of the calculation is given elsewhere @xcite .
the density and current distribution is shown in fig.[fig : fig5 ] for a situation where the external current is sufficiently large .
we observe that , the is on the right side of the qpc is relatively larger than the one of at the left side .
the physics leading to this effect is rather simple : when an external current is driven , a hall potential is built which tilts the landau levels , therefore the electrons have to compensate this potential change .
however , one can not add more electrons to the is since there are no states available . the only way to screen the landau tilting is to increase the widths of the iss on one side where the hall potential is larger @xcite .
this effect clearly demonstrates the importance of @xmath0 interactions . without the self - consistent treatment one would only get a higher chemical potential one one side , however
, the widths of the edge channels would remain unchanged . the asymmetry induced by the large external current on the widths of the is
is also observed in the local probe experiments , which is a strong evidence that our model of local ohm s law is relevant to explain the edge physics of the iqhe .
in conclusion we have demonstrated that the spatial distribution of the current carrying incompressible strips vary depending on the magnetic field strength considering a quantum point contact induced at a 2deg .
therefore , in interpreting the experimental results one should keep in mind that the so - called edge states may present different transport and coherency properties even within a single quantum hall plateau .
our electrostatic model relies on the self - consistent solution of the poisson and schrdinger equations using numerical techniques .
we have explicitly shown that the @xmath0 interactions result in the formation of incompressible and compressible regions .
based on the overlap of the quantum mechanical wave functions and the local broadening of the dos within the incompressible strips , which has an extend smaller or comparable to the fermi wavelength , we performed a spatial average over @xmath64 to cure the artifacts of the tfa .
the second set of self - consistent equations has been introduced to obtain the current distribution , namely solving the maxwell equation and equation of continuity simultaneously . in the small current regime , we have shown that the current is confined to the incompressible strips since the back scattering is suppressed and the longitudinal resistivity vanishes . meanwhile , the hall resistivity assumes its quantized value .
we have also shown that , considering sufficiently large external currents induces an asymmetry both in the density and in the current distribution , which we think is an interesting issue to check with the experiments .
the author would like to thank r.r .
gerhardts for his initiation , supervision and contribution in developing the model , s. arslan and a. weichselbaum in providing the data concerning the solution of the 3d poisson equation and d. eksi and s. aktas for all their contribution in the improvement of the numerical code .
the enlightening discussions with v. golovach is also highly appreciated .
this work was supported by the nim area a and sfb 631 . | we present the essential findings of the screening theory of the integer quantum hall effect ( iqhe ) considering a quantum point contact ( qpc ) .
our approach is to solve the poisson and the schrdinger equations self - consistently , taking into account electron interactions , within a hartree type approximation for a two dimensional electron gas ( 2deg ) subject to high perpendicular magnetic fields .
the coulomb interaction between the electrons separates 2deg into two co - existing regions , namely quasi - metallic compressible and quasi - insulating incompressible regions , which exhibit peculiar screening and transport properties . in the presence of an external current ,
we show that this current is confined into the incompressible regions where the drift velocity is finite .
in particular , we investigate the distribution of these incompressible strips and their relation with the quantum hall plateaus considering a quasi 1d constriction , i.e. a qpc . |
in general , quantum field theories have been built in the context of local models .
however , there exist physical situations that lead to non local interactions in a straightforward way .
let us mention , for instance , wheeler and feynman s description of charged particles @xcite , string theories with non local vertices @xcite , and non local kinetic terms that appear when bosonizing fermions in ( 2 + 1 ) dimensions @xcite @xcite .
as we shall see , several recently considered non local field theories are related to the study of electronic systems in one spatial dimension ( 1d ) @xcite @xcite .
indeed , in recent years the physics of 1d systems of strongly correlated particles has become a very interesting subject since one can take advantage of the simplicity of the models at hand and , at the same time , expect to make contact with experiments .
for instance , the recently discovered carbon nanotubes are perfect experimental realizations of 1d conductors @xcite . on the other hand , as the dimensionality of a system decreases , charge screening effects become less important and the long - range interaction between electrons is expected to play a central role in determining the properties of the system .
in fact , from a theoretical point of view the effects of long - range interactions have been recently discussed in connection to several problems such as the fermi - edge singularity @xcite , the insulator - metal transition @xcite , the role of the lattice through umklapp scattering and size dependent effects @xcite , etc . in the specific context of carbon nanotubes ,
a low - energy theory including coulomb interactions has been also recently derived and analyzed @xcite .
non local fermionic models have been also used in the study of fluctuation effects in low - dimensional spin - peierls systems @xcite . as shown in @xcite ,
starting from a non - local and non - covariant version of the thirring model @xcite , in which the fermionic densities and currents are coupled through bilocal , distance - dependent potentials , one can make direct contact with the `` g - ology '' model @xcite currently used to describe different scattering processes characterized by coupling functions @xmath3 , @xmath4 , @xmath5 and @xmath6 .
when one bosonizes this theory by either operational or functional methods , due to the contributions of @xmath3 ( backscattering ) @xcite and @xmath5 ( umklapp ) @xcite one finds an even more drastic departure from the local case .
indeed , instead of the well - known integrable sine - gordon model ( sg ) one gets a non local extension of it , which , as far as we know , is not exactly solvable .
recently , in ref.@xcite , the physical content of this model was explored by using the self - consistent harmonic approximation ( scha ) @xcite . as it is well - known
the scha is a non controlled approximation , i.e. there is no perturbative parameter involved .
it is then desirable to have an alternative analysis of this problem .
this is the main motivation of the present work .
we will apply the renormalization group ( rg ) technique @xcite , usually employed in local cases , to the non local sine - gordon model ( nlsg ) mentioned above . for simplicity
we shall assume that non locality plays a role only in umklapp interactions ( @xmath5 ) whereas all other potentials are local , i.e. proportional to delta functions . in section 2
we briefly show how the nlsg action is obtained from the non local thirring model . in section 3
we derive the rg equations and compute the gap in the charge - density spectrum .
this allows us to determine the effect of non contact @xmath5 couplings .
finally , in section 4 , we discuss our results .
let us sketch the derivation of the nlsg action .
we start from the fermionic ( 1 + 1)-dimensional quantum field theory with euclidean action given by @xmath7 where @xmath8 is the unperturbed action associated to a linearized free dispersion relation .
the contributions of the different scattering processes can be written as @xmath9 and @xmath10 where the @xmath11 are the usual two - dimensional dirac matrices and @xmath12 , @xmath13 .
the coupling potentials @xmath14 and @xmath15 are assumed to depend on the distance @xmath16 and can be expressed in terms of `` g - ology '' parameters as @xmath17 in the above equations @xmath18 and @xmath19 are just numerical constants that could be set equal to one .
we keep them to facilitate comparison of our results with those corresponding to the usual thirring model .
indeed , this case is obtained by choosing @xmath20 and @xmath21 . on the other hand , the non - covariant limit @xmath20
, @xmath22 gives one version ( @xmath23 ) of the tl model @xcite .
the terms in the action containing @xmath4 and @xmath6 represent forward scattering events , in which the associated momentum transfer is small . in the @xmath4
processes the two branches ( left and right - moving particles ) are coupled , whereas in the @xmath6 processes all four participating electrons belong to the same branch . on the other hand , @xmath3 and @xmath5
are related to scattering diagrams with larger momentum transfers of the order of @xmath24 ( bs ) and @xmath25 ( us ) respectively ( this last contribution is important only if the band is half - filled ) . for simplicity , throughout this paper we will consider spinless electrons .
the extension of our results to the spin-@xmath26 case with spin - flipping interactions , though not trivial , could be done by following the lines of ref .
@xcite . at this point
we consider the partition function @xmath27 expressed as a functional integral over fermionic variables .
the implementation of a generalized hubbard - stratonovich identity @xcite allows to write @xmath27 in terms of a fermionic determinant .
although this determinant is highly non trivial , one can combine a chiral change in the fermionic path - integral measure with a formal expansion in @xmath28 in order to obtain a bosonic representation ( see ref .
@xcite for details ) .
one thus establishes an equivalence between the original fermionic action and the following bosonic action depending on five scalars @xmath29 , @xmath30 , @xmath31 , @xmath32 and @xmath33 : @xmath34\\ + \frac{1}{2}{\int d^2x\,d^2y\,}c_{\mu } ( x ) u^{-1}_{(\mu)}(x , y ) c_{\mu } ( y ) + \frac{{g ' } \lambda c}{\pi } { \int d^2x\ , } [ c_{(0 ) } ( x)f_{0}(x ) + i c_{(1 ) } ( x)f_{1}(x ) ] \label{39}\end{gathered}\ ] ] where @xmath35 \\
b(p)&=\frac{1}{2}\left[p_0 ^2 \hat{v}_{(0)}^{-1 } ( p ) + p_1 ^2 ( \hat{v}_{(1)}^{-1 } ( p ) - \frac{g^2}{\pi})\right ] \\
c(p)&=p_0p_1\left(\hat{v}_{(0)}^{-1 } ( p)- \hat{v}_{(1)}^{-1 } ( p ) + \frac{g^2}{\pi}\right),\end{aligned}\ ] ] and @xmath36 since the integrals in @xmath31 and @xmath32 are quadratic these fields are easily integrated out and one gets @xmath37 } \label{41}\ ] ] with @xmath38 = s_{0 } + s_{int } \label{42}\ ] ] where @xmath39\ ] ] @xmath40\\ -\frac{(\lambda c)^2}{2\pi^2}{\int d^2x\,d^2y\,}g_{3 } ( x , y ) \cos\left[\sqrt{4 \pi}(\varphi(x ) + \varphi ( y ) ) -2i g ( \phi(x ) + \phi(y))\right].\end{gathered}\ ] ] it is now convenient to diagonalize the quadratic part of the effective action by introducing the fields @xmath41 , @xmath42 and @xmath43 : @xmath44 where we have defined @xmath45 and @xmath46
. we then obtain @xmath47 , \label{44}\ ] ] @xmath48\\-\frac{(\lambda c)^2}{2\pi^2}{\int d^2x\,d^2y\,}g_{3 } ( x , y ) \cos\sqrt{4\pi}[\phi(x ) + \phi ( y)].\label{45}\end{gathered}\ ] ] one can see that the @xmath41 and @xmath42 fields become completely decoupled from @xmath43 . moreover , it becomes apparent that the @xmath43-dependent piece of the action @xmath49 is the only one containing potentially relevant contributions ( i.e. gapped modes ) .
in this section we shall focus our attention on the non local action derived above . for simplicity , from now on we will consider the case in which @xmath50 and @xmath51 are constants ( local forward scattering ) and @xmath52 , i.e. a pure non local umklapp interaction .
thus , we start from the action @xmath53={\int\frac{d^2p}{(2\pi)^2}\,}\phi(p)\frac{f(p)}{2}\phi(-p ) -\frac{(\lambda c)^2}{2\pi^2}{\int d^2x\,d^2y\,}g_3(x , y)\cos\sqrt{4\pi}[\phi(x)+\phi(y)]\label{46}\end{gathered}\ ] ] with @xmath54 where we have now expressed all formulae in terms of @xmath18 coupling functions . in the local case ( @xmath55 ) the action ( [ 46 ] ) corresponds to the well - known sine - gordon model , which is an integrable , exactly solvable field theory .
in particular , a rg analysis shows that the stiffness constant " , @xmath56 has to be lower than @xmath57 in order to have a relevant cosine interaction , i.e. to have a gap in the spectrum .
recently , by reinterpreting bethe - ansatz results , zamolodchikov obtained the exact expression for this gap @xcite .
unfortunately , as far as we know , the present non local version of the theory is not exactly soluble and one is then forced to consider an approximation . in ref .
@xcite a scha was employed in order to estimate the energy gap .
but this is a non controlled , non perturbative approximation . besides this general disadvantage ,
the implementation of the scha technique led to a set of coupled algebraic equations that could be numerically solved only for very weak non locality .
it is then natural to try another method to attack the problem and eventually improve the approximation .
let us consider the wilsonian approach to the rg ( see for instance @xcite ) .
first of all we restrict our analysis to a non local interaction of the form : @xmath58 with @xmath59 where @xmath60 is the second derivative of the delta function with respect to @xmath61 . at this point
it is worth mentioning that rg calculations involving fermionic non local interactions already exist in the literature .
for instance , the authors of ref.@xcite used a jordan - wigner fermion representation for the 1d heisenberg - ising model which includes not only a non local fermion - fermion interaction but also a linear fermion - lattice coupling . due to the curvature of the band , the non trivial fermion - lattice coupling and the presence of both forward and umklapp scattering , comparison of their rg equations with ours
is not straightforward .
let us point out , however , that in @xcite fermions corresponding to nearest - neighbor lattice sites interact through a potential of the form @xmath62 , i.e. a unique constant @xmath63 is associated to @xmath64 . as shown in the above equation ( [ g3 ] ) , in the present work we are interested in a coupling which depends at least on two constants @xmath65 and @xmath66 . in fact , the derivation of rg equations for these coupling constants will be our next task . in condensed matter problems
one is usually interested in the physics at long distances , compared to a lattice spacing of the order of @xmath67 . since
, in momentum space , this corresponds to small @xmath68 , it is natural to consider correlations between fields with momenta @xmath69 , with s very large .
these are the so called `` slow modes '' @xmath70 . on the other hand ,
the `` fast modes '' @xmath71 are those carrying momenta that satisfy @xmath72 . in the present approach to rg these fast modes are integrated in the path integral framework , giving rise to an effective theory depending only on slow modes . from this action , in turn , one can read the flow equations for the couplings . indeed , writing the initial action as @xmath73 , to first order , after a suitable rescaling of coordinates and momenta and a redefinition of the fields ( see the appendix for details ) we obtain the following relationship between the original and rg transformed actions : @xmath74\delta^2(x - y)-\frac{\epsilon - 4 k \epsilon
t}{\lambda^2}\,\partial_1 ^ 2\delta^2(x - y ) \right)\times \\ \times \cos\sqrt{4\pi}(\phi(x ) + \phi ( y ) ) = \\
= \frac{(\lambda c)^2}{2\pi^2}\int d^{2}x\,d^{2}y\left(\lambda \delta^2 ( x - y)-\frac{\epsilon}{\lambda^2}\,\partial_1 ^ 2\delta^2(x - y)\right ) \cos\sqrt{4\pi}(\phi(x ) + \phi ( y ) ) = s_{int}^{'}.\end{gathered}\ ] ] as usual , imposing the invariance of the action under rg , we get the flow equations for the couplings @xmath65 and @xmath66 : @xmath75 @xmath76 where @xmath77 , with the initial conditions @xmath78 and @xmath79 .
the solution of this system is elementary , yielding : @xmath80 + k \epsilon_{0}\exp[-4 k t],\ ] ] @xmath81.\ ] ] from the last equation one clearly sees that the non local piece of the interaction is irrelevant , as expected .
concerning the local interaction one sees that it is relevant for @xmath82 , i.e. non locality does not modify this well - known condition already found for the local sg model .
therefore , for @xmath82 , @xmath65 will grow with increasing t and there will be a gap in the cd spectrum which can be estimated by determining the value @xmath83 for which @xmath84 . from now on we shall restrict our analysis to the case @xmath82 .
the gap is then given by @xmath85 , where @xmath86 .
it is also convenient to define the dimensionless gap @xmath87 . from equation ( [ rg3 ] )
one can thus derive an equation for @xmath88 which gives the behavior of the energy gap as function of the forward scattering potentials ( @xmath56 ) and the non local contribution of the umklapp scattering ( @xmath89 ) . before analyzing this non trivial equation it seems reasonable to check if it predicts sensible results for the local case . to this end
we set @xmath90 and @xmath91 in ( [ rg3 ] ) , obtaining : @xmath92 this result can be compared with the exact solution obtained by zamolodchikov @xcite and with the approximated result given by the scha method @xcite .
the corresponding expressions for the gap are respectively given by : @xmath93 and @xmath94 in order to compare @xmath95 and @xmath96 with @xmath97 in an efficient and easy to visualize way , we have computed the relative error @xmath98 as function of @xmath56 for both approximations .
the results are depicted in fig.1 where one sees that our rg computation gives values of the gap closer to the exact values for a wide range of the stiffness constant @xmath56 .
interestingly , the scha result works well when one approaches the end points of the interval .
going back to the case @xmath99 , by combining eqs.([rg1 ] ) and ( [ rg2 ] ) one readily gets a phase diagram in the @xmath100 plane ( see fig .
2 ) . there is a critical line given by @xmath101 .
if the initial parameters are tuned to lie on this line , the system will flow to the tomonaga - luttinger fixed point , at the origin . in this case , of course , the system remains gapless . on the other hand ,
for initial conditions outside the critical line , the system flows to strong coupling , giving rise to a gap @xmath88 , as mentioned above . for simplicity
let us consider the case @xmath102 and define the variable @xmath103 .
the gap equation can then be written as @xmath104 where @xmath105 .
this equation is one of our main results . for fixed values of @xmath56 and @xmath106
it gives the behaviour of the energy gap as function of the non local contribution to umklapp scattering , associated to non contact interactions .
the form of this formula suggests that it could be easier to handle the inverted equation : @xmath107 where @xmath108 . in fig.3
we show the numerical solution of this last equation for @xmath109 and @xmath110 .
we see that @xmath111 decreases for increasing @xmath89 , in qualitative agreement with the scha prediction @xcite .
a quantitative comparison between both approximations is given in fig.4 , for the same fixed values of @xmath106 and @xmath56 .
since the scha result obtained in @xcite is valid for small values of the coefficient associated to non locality , we have plotted the solutions in the interval @xmath112 .
we find that the gap decay predicted by rg , for increasing non locality , is much slower than the one obtained through the gaussian approximation . as a final comment
, we note that the rg treatment for non local interactions depicted in this work can be extended to a more general coupling function including an arbitrary number of even powers of @xmath113 . in coordinate space
such an interaction can be written as @xmath114 where @xmath115 .
the corresponding set of rg equations for the @xmath116 can be obtained from the expansion of the cosine integral function which appears when one integrates the fast modes after the mode separation ( see appendix ) .
the result is @xmath117 the general solution of this system can be expressed as a combination of exponentials and the computation of the gap can not be done , in principle , in an analytical way .
it is then illustrative to consider a particular case in which this calculation is simplified .
indeed , for small @xmath118 , it can be proved by induction that the solutions of this system of equations are of the form : @xmath119.\ ] ] in this limit one obtains the following expression for the gap : @xmath120 which is consistent with the conditions @xmath121 and @xmath122 .
in this paper we have considered a non local extension of the sine - gordon model .
this theory is obtained when one bosonizes a non local and non covariant version of the thirring model used to describe certain 1d many - body systems .
since the integrability of this non local sine - gordon model has not been proved , one needs to implement some approximation to study its physical content .
we have performed a rg calculation up to first order in the coupling function @xmath5 , which in a condensed matter context is associated to the so called umklapp scattering .
we obtained an expression for the energy gap as function of the non local piece of the interaction @xmath123 . for purely local interactions ( the exactly solvable sg ) our result seems to be a sensible approximation , improving the scha predictions for a wide range of forward interactions . in the non local case , in which no exact answer is known , we predict decreasing values for the gap for increasing values of @xmath123 , in qualitative agreement with a previous scha computation .
we were able to give a precise comparison between both approximations in the interval @xmath124 , showing that the gap decrease , for increasing non locality , is much weaker according to the rg computation .
since , as is well - known , the scha method is not a controlled approximation , the present results contribute to a better understanding of the physics of non local field theories .
we think that our results are also of interest in the context of 1d many - body systems ( luttinger liquids ) in which most of the previous investigations involving umklapp scattering do not consider non local effects associated to long range interactions @xcite .
this work was partially supported by universidad nacional de la plata and consejo nacional de investigaciones cientficas y tcnicas , conicet ( argentina ) .
in order to illustrate the computation leading to the flow equations we first define the free bosonic propagator : the next step is the analysis of @xmath49 , as given by the second term of equation ( [ 46 ] ) . for simplicity , in this appendix
we disregard the overall constant @xmath128 .
going to momentum space and performing the separation in slow and fast modes @xmath70 and @xmath71 , according to : we obtain @xmath132 \\ \cos[\frac{\sqrt{4\pi}}{(2\pi)^{2}}\int\,d^{2}p\,\phi_{>}(p)\,f(p , x , y_{1 } ) ] - \sin[\frac{\sqrt{4\pi}}{(2\pi)^{2}}\int\,d^{2}p\,\phi_{<}(p)\,f(p , x , y_{1 } ) ] \\
\sin[\frac{\sqrt{4\pi}}{(2\pi)^{2}}\int\,d^{2}p\,\phi_{>}(p)\,f(p , x , y_{1})])\end{gathered}\ ] ] where @xmath133 @xmath135>_{0}\ , =
\\ = \exp \left(-k \int\,dp_1 \,(\theta(p_1)-\theta(-p_1))\,\frac{1+\cos(p_1(x_1-y_1))}{p_1}\right)=\\ = \exp \left(-k \left[2\ln s + 2 ci\,(\lambda(x_1-y_1))-2 ci ( \frac{\lambda}{s}(x_1-y_1))\right]\right)\end{gathered}\ ] ] where @xmath136 is the cosine integral function and the free vacuum expectation values are , of course , taken with respect to fast modes .
rescaling momenta , coordinates and fields such that the free piece of the action @xmath137 remains invariant : @xmath138 and using the fact that @xmath139 , @xmath49 can be written as @xmath140\times \\ \times\ , ( 1 + [ ( 3 - 4 k ) + k ( \lambda ( x_{1}^{'}- y_{1}^{'}))^{2 } - \frac{k}{12}(\lambda ( x_{1}^{'}- y_{1}^ { ' } ) ) ^{4 } ] t ) , \end{gathered}\ ] ] where we have used the power expansion of the function @xmath136 . finally ,
using the explicit expression for @xmath5 in terms of @xmath141 and @xmath123 we obtain @xmath142 t - 2 k \epsilon_0 t ) \delta^2(x_{1}^{'}- y_{1}^{'})-\\-\frac{\epsilon_0 - 4 k \epsilon_0
t}{\lambda^2}\partial_1 ^ 2\delta^2(x_{1}^{'}- y_{1}^ { ' } ) ] \cos\sqrt{4\pi}[\phi(x^ { ' } ) + \phi ( y^ { ' } ) ] = \\
= \int d^{2}x^{'}\,d^{2}y^{'}[\lambda^ { ' } \delta^2(x_{1}^{'}- y_{1}^{'})-\frac{\epsilon^{'}}{\lambda^2}\partial_1 ^ 2\delta^2(x_{1}^{'}- y_{1}^ { ' } ) ] \cos\sqrt{4\pi}[\phi(x^ { ' } ) + \phi ( y^ { ' } ) ] = s_{int}^{'},\end{gathered}\ ] ] 99 r.p.feynman and j.a.wheeler , rev .
* 17 * ( 1945 ) 157 .
d.a.eliezer and r.p.woodard , nucl .
b * 325 * ( 1989 ) 389 .
e.c.marino , phys .
b * 263 * ( 1991 ) 63 .
nucl . phys .
b * 408 * ( 1993 ) 551 .
d.g.barci , c.d.fosco and l.oxman , phys .
b * 375 * ( 1996 ) 267 .
d.g.barci and l.oxman , mod .
a * 12 * ( 1997 ) 493 .
i. v. krive et al . , phys .
b * 52 * 10865 ( 1995 ) . c.nan , m.c.von.reichenbach and m.l .
trobo , nucl . phys .
* b 435 * [ fs ] 567 ( 1995);ibid . * b 485 * [ fs ] 665 ( 1997 ) .
+ m.manas , c.nan , m.l.trobo , nucl .
b 525 [ fs ] , * 721 * ( 1998 ) . + v. i. fernndez and
c. m. nan , theor .
( russia ) , * 118 * , 385 ( 1999 ) , hep - th/9810048 .
s.iijima , nature * 354 * , 56 , ( 1991 ) .
+ t.w.ebbesen , physics today * 49 ( 6 ) * , 26 , ( 1996 ) .
h.otani , t.ogawa , phys .
b * 54 * , 4540 , ( 1996 ) .
d. poilblanc et al . , phys .
b * 56 * , r1645 , ( 1997 ) .
g. fano et al . , phys .
b * 60 * , 15654 , ( 1999 ) .
+ s.capponi et al .
, ibid . * 61 * , 13410 , ( 2000 ) .
r.egger , a.o.gogolin , phys .
lett . * 79 * , 5082 , ( 1997 ) .
b.dumoulin et al . , phys .
lett . * 76 * , 1360 , ( 1996 ) .
v.i.fernndez , a.iucci and c.m.nan , nucl .
phys.[fs]*b 636 * , 514 ( 2002 ) .
hep - th/0112057 .
w.thirring , ann.of phys .
* 3 * 91 , ( 1958 ) .
+ b.klaiber , lectures in theoretical physics , boulder , 1967 , vol 10a , ed.a.barut and w.brittin , new york , gordon and breach , 141 .
+ v.e.korepin , teor . mat . fiz.(russia),*41
* 169 , ( 1979 ) .
j.solyom , adv . phys . * 28 * , 209 ( 1979 ) . a.luther and v.j.emery , phys .
lett . * 33 * , 589 ( 1974 ) .
+ r.heidenreich , b.schroer , r.seiler and d.a.uhlenbrock , phys .
a * 54 * , 119 ( 1975 ) .
v.j.emery , a.luther and i.peschel , phys .
b * 13 * , 1272 , ( 1976 ) .
y. saito , z. phys .
b * 32 * , 75 , ( 1978 ) . + m.p.a .
fisher and w. zwerger , phys .
b * 32 * , 6190 , ( 1985 ) .
+ m. stevenson , phys .
d * 32 * , 1389 , ( 1985 ) .
+ r. ingermanson , nucl .
b * 266 * , 620 ( 1986 ) .
+ a.iucci and c.m.nan , j. phys .
a * b 35 * , 8005 ( 2002 ) .
hep - th/0209019 .
r. shankar , rev .
* 66 * , 129 ( 1994 ) .
+ h. j. schulz , g. cuniberti and p. pieri , cond - mat/9807366 .
+ s. rao and d. sen , cond - mat/0005492 .
s.tomonaga , prog .
. phys . * 5 * , 544 ( 1950 ) .
+ j.luttinger , j. math . phys . * 4 * , 1154 ( 1963 ) .
+ e.lieb and d.mattis , j. math .
* 6 * , 304 ( 1965 ) .
a. iucci , k. li and c.m.nan , nucl .
phys.[fs]*b 601 * , 607 ( 2001).hep - th/0012269 .
b. zamolodchikov , int . j. mod
. phys . a * 10 * , 1125 , ( 1995 ) .
see for instance , u. zlicke , phys .
* 83 * , 5330 , ( 1999 ) ; a. rosch , cond - mat/0104181 , to appear in advances in solid state physics * 41*. * figure captions * + figure 1 : relative error @xmath143 as function of the stiffness constant , for the local case .
the dashed line corresponds to the scha whereas the solid line shows the rg result .
+ figure 2 : phase diagram in the @xmath144 plane , for @xmath110 .
the dashed line is the critical line @xmath145 .
+ figure 3 : numerical solution of the gap equation for @xmath109 and @xmath110 .
the gap @xmath88 decreases for increasing @xmath89 .
+ figure 4 : numerical solutions of the gap equations for @xmath109 , @xmath110 , and @xmath112 .
the dashed line corresponds to the scha whereas the solid line shows the rg result .
the unit function is included to allow comparison with the local case . | we study a non local version of the sine - gordon model connected to a many - body system with backward and umklapp scattering processes . using renormalization group methods
we derive the flow equations for the couplings and show how non locality affects the gap in the spectrum of charge - density excitations .
we compare our results with previous predictions obtained through the self - consistent harmonic approximation .
carlos m. nan@xmath0 and mariano j. salvay@xmath0 @xmath1 _ instituto de fsica la plata , departamento de fsica , facultad de ciencias exactas , universidad nacional de la plata .
cc 67 , 1900 la plata , argentina . _
@xmath2 _ consejo nacional de investigaciones cientficas y tcnicas , argentina . _ _ keywords : _ field theory , non local , renormalization group _ pacs : _
11.10.lm , 05.30.fk |
in cosmological models based on cold dark matter ( cdm ) , the first stars are believed to form within small protogalaxies , with virial temperatures @xmath2 @xcite .
cooling within these protogalaxies is dominated by molecular hydrogen , @xmath0 , which forms via the gas - phase reactions @xmath3 and @xmath4 even in the absence of dust . although the fractional abundance of @xmath0 that forms in this way is small , it is sufficient to allow for effective cooling and the formation of stars @xcite .
as soon as massive stars form , however , they immediately begin to photoionize and photodissociate this @xmath0 .
photoionization requires photons with energies greater than @xmath5 , which are strongly absorbed by neutral hydrogen , and is only of importance within regions .
photodissociation , by contrast , occurs through the absorption of photons in the lyman - werner band system @xcite , with energies in the range @xmath6 @xmath7 .
these photons are not strongly absorbed by neutral hydrogen and can readily escape into the intergalactic medium @xcite .
initially , many of these photons will be absorbed by intergalactic @xmath0 , but its abundance is small and it is rapidly photodissociated ( see section [ ther_chem ] below ) .
consequently , the onset of star formation is soon followed by the appearance of an ultraviolet background radiation field .
this ultraviolet background acts to suppress further star formation by photodissociating @xmath0 within newly - forming protogalaxies .
the effects of this background have been studied by a number of authors @xcite .
in particular , @xcite study the coupled problem of the evolution of the ultraviolet background and its feedback on the global star formation rate using a simple galaxy formation model based on the press - schechter formalism @xcite .
they find that cooling ( and hence star formation ) within small protogalaxies is completely suppressed prior to cosmological reionization .
taken at face value , their results suggest that star formation within small , @xmath0-cooled protogalaxies is a transient phenomenon , with little impact on later stages of galaxy formation . these conclusions , however , rest on the assumption that the only free electrons present in the protogalactic gas come from the small residual fraction remaining after cosmological recombination .
this is important , as free electrons ( and protons ) catalyze @xmath0 formation , as we can see from equations [ h2f1 ] to [ h2f4 ] .
if the free electron abundance were significantly higher , then the @xmath0 formation rate would also be higher , offsetting the effects of photodissociation . at the very least
, this would delay the suppression of star formation , and in principle could entirely negate it .
it is therefore important to determine whether there is any way in which an enhanced level of ionization could be produced .
one possible source is the residual ionization that would remain after the recombination of regions produced by an earlier generation of stars .
this has recently been studied by @xcite , who find that it can be an effective source of @xmath0 and can dramatically reduce the effectiveness of photodissociation feedback .
however , their conclusions are still somewhat uncertain , both because their simulations are under - resolved ( see figures 57 in * ? ? ?
* ) and because they neglect the effects of supernovae , which would act to disperse the gas and significantly lengthen the recombination timescale , thereby delaying the formation of @xmath0 .
an alternative possibility is ionization by a high redshift x - ray background . at x - ray energies ,
the optical depth of the intergalactic medium ( igm ) is small and any x - ray sources present will naturally generate an x - ray background .
moreover , x - rays can penetrate to large depths within newly - formed protogalaxies , allowing them to raise the fractional ionization throughout the gas .
the potential importance of such a background was first highlighted by @xcite , with high redshift quasars suggested as a possible source . using a very simple model for a quasar - produced x - ray background , they showed that if quasars contribute more than 10% of the uv background , then the ionization produced by the associated x - ray background is sufficient to negate the effects of uv photodissociation .
indeed , they found evidence that such a background could actually promote cooling within the dense gas in the centres of protogalaxies . this scenario has not yet been firmly ruled out , but observational evidence suggests that quasars are unlikely to be present in sufficient number at high redshift @xcite .
quasars , however , are not the only potential source of x - rays .
star formation also leads to the production of x - rays , primarily through the formation of massive x - ray binaries @xcite , but also through bremsstrahlung and inverse compton emission from supernova remnants .
these sources generate only a small fraction of the present - day x - ray background @xcite , but may become dominant at high redshifts . in this paper , we examine the effects of the x - rays produced by these sources , both on the cooling of gas within virialized protogalaxies and also on the chemical and thermal evolution of the igm .
the outline of the paper is as follows . in section [ back ] , we discuss the sources responsible for producing the uv and x - ray backgrounds , and show how the build - up of these backgrounds can be computed . in section [ meth ] ,
we outline the method used to study the effect of this radiation on the primordial gas , and in section [ res ] apply it for a number of different x - ray source models .
we present our conclusions in section [ conc ] .
for an observer at redshift @xmath8 , we can write the mean specific intensity of the radiation background at an observed frequency @xmath9 as @xcite @xmath10 where @xmath11 , @xmath12 is the proper space - averaged volume emissivity , @xmath13 is the optical depth at frequency @xmath9 due to material along the line of sight from redshift @xmath8 to @xmath14 and @xmath15 is the cosmological line element . to solve this equation
, we need to know how the emissivity and opacity evolve with redshift . for simplicity , we write the space - averaged emissivity in terms of the global star formation rate ( sfr ) as @xmath16 where @xmath17 is the luminosity density ( i.e. the luminosity per unit frequency ) per unit star formation rate ( in solar masses per year ) , and @xmath18 is the global star formation rate , with units @xmath19 . in principle , @xmath17 may be a complicated function of frequency and redshift . in practice , however , lyman - werner band emission is dominated by massive , short - lived ob stars and declines rapidly once star formation comes to an end ( see , for example , the instantaneous starburst models of * ? ? ?
* in which the lyman - werner flux declines by an order of magnitude within 4@xmath20 ) .
similarly , x - ray emission is dominated by short - lived sources such as massive x - ray binaries and supernova remnants which are end products of the same massive stars .
both kinds of emission are therefore strongly correlated with the star formation rate ; to a first approximation , we can assume that they are directly proportional to it , and that any redshift dependence of @xmath17 can be neglected . with this simplification , determining the emissivity breaks down into two independent problems : determining the global star formation rate as a function of redshift , and determining the luminosity density as a function of the star formation rate .
although we have observational constraints on the star formation rate up to @xmath21 , we have no direct constraints ( and few indirect ones ) at higher redshift .
consequently , any model of high redshift star formation must inevitably be highly theoretical .
moreover , this lack of constraints motivates us to choose as simple a model as possible ; more complicated ( and realistic ) models can always be considered once our observational knowledge improves .
a good example of this kind of simple model is the one used by @xcite ; we adopt the same model here .
we assume that star formation proceeds primarily through starbursts , of duration @xmath22 years , that are triggered when galaxies form . during the starburst
, the star formation rate is assumed to be constant .
the global star formation rate in this model is given by @xmath23 where @xmath24 is the star formation efficiency , @xmath25 is the baryon fraction ( ie the ratio of baryons to dark matter ) , and where @xmath26 is the cosmological density of matter in newly - formed galaxies ( with units of @xmath27 ) .
we assume that the value of @xmath25 in the protogalaxies is the same as in the igm , or in other words that @xmath28 .
for the cosmological model adopted in section [ res ] , this corresponds to @xmath29 .
we further assume that the rate of change of @xmath30 is approximately the same as the rate of change of @xmath31 , the total fraction of matter in halos with virial temperatures greater than a critical temperature @xmath32 , or in other words that @xmath33 where @xmath34 is the cosmological matter density .
here , @xmath32 represents the minimum virial temperature required for efficient cooling ; to a first approximation , halos with @xmath35 are unable to cool , while those with @xmath36 cool rapidly and can form stars .
various different definitions of @xmath32 are in use in the literature ( see , for example * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) ; we discuss our particular choice in a later section . making this approximation
is equivalent to assuming that the growth in @xmath31 is dominated by the formation of new halos with @xmath37 ( either by monolithic collapse or by the merger of smaller objects ) rather than by the accretion of matter by existing halos with @xmath36 .
this is justified at high redshift when such halos are rare and @xmath31 is dominated by objects near @xmath32 , but becomes less accurate at lower redshifts .
for this reason ( and others , to be discussed later ) we do not attempt to simulate the evolution of the background below @xmath38 . to solve equation [ diff_rho ] we need to know the value of @xmath32 .
in general , this will depend both on redshift ( see , eg * ? ? ?
* ) and on the intensities of the uv and x - ray backgrounds . indeed , understanding the evolution of @xmath32 with redshift is one of the main goals of this paper .
our procedure for determining @xmath32 is discussed at length in section [ meth ] ; for now , we assume that it is known . in this case
, we can calculate @xmath31 using the press - schechter formalism @xcite : @xmath39,\ ] ] where @xmath40 is the critical density threshold for collapse , @xmath41 is the square root of the variance of the cosmological density field , as smoothed on a mass scale @xmath42 , and where @xmath43 is the mass of a protogalaxy with virial temperature @xmath32 .
although both @xmath44 and @xmath41 depend upon the choice of cosmological model , their behaviour is well - known , and the problem of determining @xmath45 reduces to the relatively simple one of relating @xmath43 to @xmath32 . to do this , we need to know the protogalactic density profile .
hydrodynamical simulations suggest that it is approximately isothermal @xcite , but representing it as a singular isothermal sphere is physically unrealistic due to the latter s infinite central density .
accordingly , we follow @xcite and represent it as a truncated isothermal sphere @xcite . with this choice , we find that @xmath46 to determine @xmath47 , we must integrate equation [ diff_rho ] over a redshift interval @xmath48 , corresponding to the duration of the starburst ; hence , @xmath49 this simple model has a number of shortcomings . for instance , it assumes that the star formation efficiency @xmath24 and starburst duration @xmath22 are both constant , independent of redshift or galaxy mass .
it also assumes that each galaxy forms its stars in a single starburst and thus ignores the effects of continuous star formation and of subsequent , merger - triggered starbursts . .
] nevertheless , it has the virtue of simplicity , and is a good point from which to start our examination of the effects of the x - ray background .
the ultraviolet flux of a star - forming galaxy is dominated by emission from young , massive o and b - type stars .
these are short - lived , with the most massive having lifetimes of only a few myr , and thus the ultraviolet luminosity density is closely correlated with the star formation rate .
its value depends upon the spectral properties of the newly - formed stellar population , and thus on their initial mass function ( imf ) , metallicity and age . in a recent paper
, @xcite presents values for the photon flux in the lyman - werner bands calculated for a number of different metal - free stellar populations .
if we assume that the spectrum within the bands is flat ( a reasonable approximation ) , then we can convert this photon flux into a luminosity density . for a salpeter imf with minimum mass @xmath50 and maximum mass @xmath51 ( model a in
* ) , we find that @xmath52 reducing @xmath53 to the more conventional value of @xmath54 reduces this luminosity to @xmath55 as we form a greater number of low mass stars that do not contribute significantly to the lyman - werner flux .
both of these results assume that the dissociative flux has stabilized at its equilibrium value and is therefore proportional to the star formation rate .
this equilibrium is typically established after only 2 3 myr , so this is generally a good approximation , even for starbursts of relatively short duration .
the above figures are appropriate so long as we are dealing with stars formed out of _ entirely _ metal - free gas .
such stars are somewhat unusual , however , as the absence of carbon means that they are unable to generate energy via the cno cycle , which otherwise would dominate energy production in stars of mass @xmath56 . as a result ,
metal - free stars are hotter than their metal - enriched counterparts @xcite and have harder spectra .
a surprising consequence of this fact is that a low - metallicity stellar population will produce a larger dissociative flux than a metal - free population
the lower effective temperatures move the peak in the thermal emission from the most massive stars closer to the lyman - werner band , causing the ionizing flux to fall but the dissociative flux to rise .
we can use the data presented in @xcite for a stellar population with @xmath57 to examine the difference that this effect makes to the lyman - werner flux . using the same imf as in equation [ flux_dis ] , we find that @xmath58 thus , raising the metallicity increases the lyman - werner flux , but only by about 6070% ; as we will see in section [ res_nox ] , this has little effect on the evolution of @xmath32 .
clearly , there are many possible models other than those considered here .
indeed , there is growing evidence that the imf of population iii stars is strongly biased towards high masses @xcite .
however , this remains uncertain , and in this paper we have chosen to err on the side of caution and assume that the high - redshift imf is similar to that at the present day . in a recent study , @xcite collate data on a number of local starburst galaxies and compare the 2 10 @xmath59 x - ray fluxes measured by asca with the 8 1000 @xmath60 infrared fluxes measured by iras .
they find that a clear correlation exists , with the total x - ray and infrared fluxes related by @xmath61 similar correlations have previously been reported by @xcite and @xcite for x - rays in the 0.5 4.5 @xmath59 and 2 30 @xmath59 energy bands respectively .
theoretically , we would expect such a correlation , with both the x - ray and infrared emission tracing the underlying star formation rate . for the x - rays
, this occurs because the emission is dominated by massive x - ray binaries ( mxrbs ) : binary systems consisting of a massive ob star accreting onto a compact companion ( a neutron star or black hole ) .
x - ray emission from such systems generally switches on a few million years after the formation of the compact object , and the lifetime of the emitting phase is short , typically of the order of ( 2 5@xmath62 @xcite .
these short timescales tie the emission closely to the underlying star formation rate @xcite .
the far - infrared flux , on the other hand , tracks star formation far more directly , being dominated by emission from dust heated by short - lived , massive stars . to use this observed correlation to determine the x - ray luminosity of a star - forming galaxy as a function of its star formation rate , we use the result from the starburst models of @xcite that @xmath63 together with equation [ x_to_ir ] to write the x - ray luminosity as @xcite @xmath64 we then assume that this relationship between x - ray luminosity and star formation rate remains valid as we move to higher redshifts .
note that the same need not be true for the relationship between x - ray luminosity and infrared luminosity , or infrared luminosity and star formation rate : although they serve to establish the correlation between x - ray luminosity and star formation rate at @xmath65 , the assumption that this correlation remains valid at higher redshift does not imply that these other correlations also remain valid .
indeed , we would expect the infrared luminosity of dust - free protogalaxies to be very much lower than would be predicted by equation [ ir ] .
evidence that the correlation between x - ray luminosity and star formation rate does indeed remain valid at high redshift is provided by the recent stacking analysis of individually undetected lyman break galaxies in the _ chandra _ deep field - north @xcite .
this analysis finds that the average rest frame luminosity of the lyman break galaxies in the 2 8 @xmath66 energy band is @xmath67 . assuming a typical star formation rate of @xmath68 for these galaxies @xcite , this corresponds to an x - ray luminosity of @xmath69 consistent with the value derived above .
is measured in a slightly different energy band from that of equation [ lx ] , so the agreement between the two values is not quite as good as may at first appear . nevertheless , the necessary correction is small , and the values agree to within 50% . ]
although far from conclusive , this result suggests that we can extrapolate the locally observed correlation to at least as far as @xmath70 . comparing our determination of @xmath71 with a recent calculation by @xcite , we find a difference of a factor of ten in our results .
some of this disagreement is due to the difference in x - ray energy bands considered ( @xmath72 @xmath73 in @xcite , compared to @xmath74
@xmath75 here ) , but some must surely be due to intrinsic scatter in the observational data , suggesting that equation [ lx ] should properly be regarded as an order of magnitude estimate of the true x - ray luminosity . given equation [ lx ] for the x - ray luminosity as a function of the star formation rate , we calculate the x - ray luminosity density by assuming a template spectrum of power - law form @xmath76 where @xmath77 , and requiring that @xmath78 where @xmath79 and @xmath80 .
@xcite find that a weighted average of the galaxies in their sample gives a value for the spectral index of @xmath81 .
adopting this value and solving for @xmath82 , we find that @xmath83 altering @xmath84 changes @xmath82 , but never by more than 50% for values consistent with the @xcite measurement . clearly , individual galaxies may have spectra that differ markedly from this simple template , but it should be a reasonable approximation when averaging over a large number of galaxies .
the above model is simple , and empirically motivated , but does assume that the x - ray emission of high - redshift star - forming galaxies is very similar to that of starbursts observed locally .
this is a reasonable assumption in the absence of evidence to the contrary , and is probably valid as long as massive x - ray binaries continue to dominate the galactic x - ray emission .
however , it is quite possible that at high redshift some other type of source will come to dominate the emission , particularly if the number of binary systems is small , as is suggested by recent simulations @xcite .
it is therefore prudent to consider the effects of other potential sources of x - rays .
the obvious candidates are supernova remnants ( snr ) ; next to x - ray binaries , they are the most significant galactic sources @xcite
. they can emit x - rays through a variety of different emission mechanisms , but at high redshift the most significant will be thermal bremsstrahlung emission and non - thermal inverse compton emission .
thermal bremsstrahlung is produced by the hot gas within the snr .
detailed modeling properly requires a hydrodynamical treatment ( see , for example , * ? ? ?
* ) , but for our purposes a simple parameterization suffices .
if we assume that the hot gas has a single characteristic temperature @xmath85 , then we can write the luminosity density per unit star formation rate as @xmath86 where @xmath87 and where @xmath88 is constant . moreover , we can write the total x - ray luminosity as @xmath89 where @xmath90 is the typical supernova explosion energy ( in units of @xmath91 ) , @xmath92 is the fraction of this energy radiated as bremsstrahlung , and where @xmath93 is the number of supernovae that explode per solar mass of stars formed .
the value of @xmath93 depends on the imf ; for the standard salpeter imf adopted previously , @xmath94 . finally , since @xmath95 we can fix the value of @xmath88
; we find that @xmath96 for typical supernova parameters ( @xmath97 and an ambient density @xmath98 ) , @xcite find that a fraction @xmath99 of the explosion energy is radiated , at a characteristic temperature @xmath100 .
on the other hand , the higher mean density at high redshift , together with the comparative weakness of outflows from low metallicity stars @xcite suggest that the typical ambient density may be very much higher .
in particular , if it is as high as @xmath101 , then a supernova remnant will radiate its energy extremely rapidly , before the ejecta have time to thermalize @xcite . in this case
, the fraction of energy radiated as bremsstrahlung is very much higher ( @xmath102 ) , as is the characteristic temperature ( @xmath103 ) .
we examine both of these models in section [ res ] , with the understanding that the true picture lies somewhere in between .
in addition to this thermal emission , supernova remnants also produce non - thermal x - rays .
these are generated as the relativistic electrons produced by the snr gradually lose energy through synchrotron radiation , non - thermal bremsstrahlung and the inverse compton scattering of photons from the cosmic microwave background . at high redshift ,
the latter is likely to dominate @xcite .
the spectrum of the resulting emission depends upon the energy spectrum of the relativistic electrons , but at the energies of interest is well represented by a power law : @xmath104 .
the intensity of the emission depends upon the fraction of the supernova energy transferred to the electrons ; this is not well constrained , with estimates ranging from 0.1% to 10% .
accordingly , we model the emission as @xmath105 where @xmath87 , @xmath106 is the fraction of energy deposited in the electrons , and where we have assumed that @xmath97 and @xmath107 as in the thermal bremsstrahlung case .
this expression assumes a high - energy cutoff for the x - ray spectrum at @xmath73 but is only logarithmically dependent on the value of this cutoff . in section [ res ] , we examine results for models with @xmath108 and @xmath109 , which bracket the range of plausible values . much more detail on high redshift inverse compton emission , including a discussion of potential observational tests , is given in @xcite .
the opacity @xmath13 can be separated into two distinct components absorption by dust and gas within the emitting protogalaxy , which we term intrinsic absorption , and absorption by gas along the line of sight through the igm .
intrinsic absorption is difficult to model with any degree of accuracy as it depends upon a number of variables the size and shape of the galaxy , its ionization state , the position of the sources within it , the dust content etc . rather than attempt to model these in detail a significant undertaking in itself we instead adopt a highly approximate representation .
we assume that the emitted x - rays are attenuated by absorption by a neutral hydrogen column density of @xmath110 plus an associated neutral helium column density @xmath111 .
this absorption is assumed to be the same for all sources .
these values are chosen because they are representative of the column densities of the protogalaxies studied in this paper ( which presumably contain the bulk of the x - ray sources ) .
reducing @xmath112 ( as would be appropriate if much of the surrounding were photoionized and/or dispersed by the progenitors of the x - ray sources ) has little or no effect on @xmath32 , as gas in the igm and within the protogalaxy itself quickly come to dominate the total absorption .
increasing @xmath112 , on the other hand , has more significant effects : an order of magnitude increase in @xmath112 produces similar results to an order of magnitude decrease in the strength of the x - ray background , which , as we will see in section [ res ] , is generally sufficient to render x - ray feedback ineffective .
consequently , we will overestimate the effect of the background if the bulk of the x - ray sources reside within massive galaxies . turning to the lyman - werner bands
, we note that intrinsic absorption will generally be negligible within small protogalaxies , as their @xmath0 content is rapidly photodissociated @xcite .
moreover , we also assume that the effects of dust absorption are negligible . in galaxies of primordial composition
, this is obviously true ; in metal - poor galaxies , it should also be a good approximation , as very large column densities are required for significant dust obscuration ( for instance , @xmath113 for @xmath114 gas , if the dust - to - gas ratio is similar to that in the milky way ) .
again , these assumptions break down if the majority of sources are to be found in massive , metal - rich galaxies , but we expect such galaxies to be extremely rare at the redshifts of interest in this paper . compared to intrinsic absorption , the effects of absorption due to the igm are much simpler to treat , particularly if we can assume that the bulk of the gas remains at an approximately uniform density .
this assumption proves reasonable at high redshift for photons with mean free paths much greater than the typical clumping scale , as is the case for both lyman - werner band photons and x - rays .
our treatment of igm absorption is discussed in the following sections .
the continuum opacity of metal - free gas is very small @xcite and for our purposes can be neglected .
consequently , the only significant sources of opacity encountered by lyman - werner photons are absorption by the lyman series lines of neutral hydrogen , and by the lyman - werner lines of molecular hydrogen .
the lyman series lines have the effect of absorbing any lyman - werner photons of the same frequency , and reprocessing them to lyman-@xmath84 photons plus associated softer photons .
as lyman-@xmath84 lies outside of the lyman - werner band , the net effect is to block from view any sources at redshifts higher than some maximum , @xmath115 , given by @xmath116 where @xmath117 is the frequency of the appropriate lyman series line and @xmath9 and @xmath8 are the observed frequency and redshift . clearly , the size of @xmath115 depends upon the distance between @xmath9 and @xmath117 , and thus more sources are seen at frequencies that are a long way from a line . as a result ,
the spectrum develops a characteristic ` sawtooth ' shape ( see figure 1 in @xcite ) , with the effect becoming more pronounced as one nears the lyman limit .
absorption by molecular hydrogen is rather more complicated , due to the large number of lyman - werner lines that contribute to the opacity . if we approximate the lines as delta functions then an individual line produces an opacity @xmath118 where @xmath119 and @xmath120 are the oscillator strength and wavelength of the transition , @xmath121 is the number density of @xmath0 molecules in the level giving rise to the line at @xmath122 , the redshift of absorption , and @xmath123 is the hubble constant at @xmath122 .
the value of @xmath122 is given by @xmath124 where @xmath125 .
if we assume that all of the photons that are absorbed in the lines are permanently removed from the lyman - werner band , then the total opacity @xmath13 is simply given by the sum over all lines with @xmath126 : @xmath127 this sum potentially involves a very large number of lines , but can be greatly simplified by assuming that all of the @xmath0 is to be found in its ortho or para ground state ; at the redshifts of interest , the population of excited states will be negligible .
in deriving this expression , we have assumed that every absorption permanently removes a lyman - werner band photon .
this is not entirely correct . on average ,
only 15% of absorptions are followed by photodissociation of the @xmath0 molecule @xcite ; the rest of the time , the molecule decays back to a bound state , emitting a photon . in their treatment of this problem
, @xcite assumed that the excited @xmath0 molecule would always decay directly back into the original state , and would thus emit a photon of the same energy as the one initially absorbed .
in fact , this is not correct ( t. abel , private communication ) ; most decays occur initially to highly excited vibrational states , producing photons redwards of the lyman - werner bands .
only a small fraction of decays ( about 5% ) take place directly into the original state , while a slightly larger fraction ( about 15% ) produce photons that lie elsewhere in the lyman - werner band system @xcite .
we do not include the effect of these photons , however ; an accurate treatment would be quite complicated and is almost certainly unnecessary as we shall see in section [ res ] , @xmath0 in the igm is rapidly destroyed by the growing lyman - werner background and is completely negligible by the time that negative feedback begins . at x - ray energies ,
the opacity of the intergalactic gas is dominated by the ionization of neutral hydrogen and helium ; prior to recombination , the @xmath128 abundance is small and can be neglected .
the x - ray opacity can thus be written as @xmath129 \frac{{\,{\rm d}l}}{{\,{\rm d}z } } { \,{\rm d}z},\ ] ] where @xmath130 and where @xmath131 and @xmath132 are the absorption cross - sections of neutral hydrogen and helium respectively , with @xmath133 and @xmath134 being the corresponding number densities .
as long as the fractional ionization of the igm remains small ( i.e. a few percent or less ) , the ratio between @xmath135 to @xmath136 can be accurately approximated by its primordial value @xmath137 where @xmath138 is the helium mass fraction , and we can write equation [ tauigm ] purely in terms of @xmath136 as @xmath139 n_{{\rm{h } } } \frac{{\,{\rm d}l}}{{\,{\rm d}z } } { \,{\rm d}z}.\ ] ] this integral is readily computable by means of numerical integration .
in the previous section , we showed that , given a simple star formation model , it is relatively easy to calculate the evolution of the lyman - werner and x - ray backgrounds .
two of the parameters of our star formation model the star formation efficiency @xmath24 and starburst lifetime @xmath22
we treat as free parameters ( although they can be constrained to some extent see , e.g. @xcite ) .
the remaining parameter , @xmath32 , the temperature at which efficient cooling becomes possible , is determined by the strength of the backgrounds themselves . this clearly presents us with a problem : the evolution of @xmath32 is coupled to that of the backgrounds , and to know one we must first know the other . fortunately , this difficulty is easily avoided .
we know that at high redshift the number of protogalaxies , and hence the star formation rate , must be very small .
consequently , there must be some redshift above which the external radiation field will become too weak to affect galaxy formation .
the precise redshift at which this occurs is model dependent , but for the models examined in this paper we typically find that radiative feedback is negligible above @xmath140 . by choosing an initial redshift @xmath141 ,
therefore , we can be sure that in our initial simulation the background radiation will have no effect .
given this starting point , we next proceed incrementally to lower redshifts via the following procedure : 1 .
given @xmath142 , we calculate the background radiation field at @xmath143 , assuming that @xmath144 .
2 . using the computed background
, we simulate the chemical and thermal evolution of a protogalaxy with @xmath145 ; the details of this simulation are outlined in sections [ comp ] to [ halt ] below .
the main aim of this simulation is to determine whether the protogalactic gas can cool efficiently .
if the protogalactic gas cools , then our assumed value of @xmath32 is correct ; we store this result , and return to step one to proceed to the next redshift . if the gas fails to cool , we continue to step four .
we increment our assumed value of @xmath146 by a small amount @xmath147 , and recalculate the background radiation field .
we assume that @xmath32 varies linearly over @xmath148 .
given the new background , we return to step two .
provided that @xmath148 and @xmath147 are both small , the error in @xmath146 will also be small ; this is particularly the case once emission from larger protogalaxies ( which cool via lyman-@xmath84 radiation ) begins to dominate the background .
this approach reduces the coupled problem to the simpler one of determining whether a protogalaxy with virial temperature @xmath149 and formation redshift @xmath150 will cool when exposed to a particular background radiation field . to answer this question
, we need to be able to model the thermal and chemical evolution of the protogalaxy .
our approach to this problem is outlined in the following sections .
ideally , we would like to use a high - resolution hydrodynamical simulation to follow the thermal and chemical evolution of the protogalactic gas ( see , e.g. * ? ? ?
unfortunately , including the effects of radiative transfer , particularly of photons in the lyman - werner bands , into such a simulation is not currently feasible .
we are thus forced to approximate . in choosing an appropriate approximation ,
we are also motivated by the desire to minimize the computational requirements of our simulations , so that we can explore the effects of a variety of different source models .
we make three main approximations : 1 .
we assume spherical symmetry .
this is a reasonable approximation for the first generation of protogalaxies , but clearly is incorrect in detail ( see , e.g. figure 2 of * ? ? ?
we assume that the protogalactic gas is _
static _ , at least on the timescale of the simulation .
this allows us to ignore the hydrodynamical evolution of the gas , and also substantially simplifies the treatment of radiative transfer .
this assumption clearly breaks down once the gas begins to cool strongly and loses its pressure support , but as we are only interested in the evolution up to this point , this is not a significant problem .
we assume that all of the @xmath0 molecules remain in the rotational and vibrational ground state , in either ortho or para form .
this simplification allows us to evolve the chemistry and radiative transfer on the timescale on which the total @xmath0 abundance changes ( typically @xmath151
@xmath152 ) rather than that on which the individual level populations change ( @xmath153 @xmath154 ) .
it also simplifies our treatment of the radiative transfer .
we discuss this approximation in more detail in section [ pd ] .
together , these approximations allow us to solve for the chemical and thermal evolution of a model protogalaxy in a matter of minutes on a fast desktop computer .
this allows us to study the redshift evolution of @xmath32 at high resolution in both temperature and redshift and for a number of different x - ray source models .
however , this approach has an obvious drawback we can not be sure that our approximations give a fair representation of the real protogalaxy . of particular concern
is the neglect of the hydrodynamic evolution of the gas , and the consequent error in the density profile .
this is potentially significant because the @xmath0 cooling rate , along with many of the chemical reaction rates , scales as the square of the density .
small errors in the density can thus lead to larger ones in the computed temperature .
these concerns are mitigated to some extent , however , by the close agreement between the results of detailed numerical simulations and previous semi - analytic treatments .
for instance , the values of @xmath32 obtained from the smoothed - particle hydrodynamics simulations of @xcite agree well with the results of @xcite , despite the highly approximate uniform density profile adopted by the latter group .
similarly , the results of @xcite , obtained with a three - dimensional adaptive mesh hydrodynamical code broadly agree with those of @xcite , who use a static model similar to that presented here .
together , these results suggest that @xmath32 is insensitive to the precise details of the density profile , but clearly this remains an area of concern .
our computational method can be broken down into three main stages initialization of the density profile and the chemical abundances , computation of the thermal and chemical evolution of the gas , and termination of the simulation at a suitable point .
these are described below .
the protogalactic density profile is modeled as a truncated isothermal sphere , with central overdensity @xmath155 and truncation radius @xmath156 the virial temperature and redshift of formation of the protogalaxy completely specify its density profile .
we subdivide this profile into @xmath157 spherical shells of uniform thickness and compute the mean density within each shell .
we have run a number of test simulations with different values of @xmath157 , and find that setting @xmath158 provides sufficient spatial resolution to accurately determine @xmath32 .
we assume that the initial chemical composition of the protogalactic gas is the same as that of the intergalactic medium . at our initial redshift @xmath159
, we take this from @xcite . at lower redshifts ,
the chemical evolution of the igm is influenced by the lyman - werner and x - ray backgrounds .
we therefore calculate the intergalactic abundances explicitly , using the chemical model outlined in section [ chem ] , by solving the chemical rate equations : @xmath160 where @xmath161 and @xmath162 are source and sink terms for @xmath163 .
at the same time we also solve for the temperature of the intergalactic gas @xcite @xmath164 where @xmath165 is the hubble constant , @xmath166 is the total particle number density and where @xmath167 and @xmath168 are the net heating and cooling rates ( see section [ cool ] ) .
we solve this set of ordinary differential equations with the stifbs integrator of @xcite .
as the size of the required timestep is generally much smaller than the redshift interval @xmath148 that separates our individual determinations of @xmath32 , we compute intermediate values by interpolation and from these determine the strength of the radiation background and hence the photochemical rates .
although our main aim in following this chemistry is to determine the correct initial abundances for our simulations of protogalactic evolution , the results are of interest in their own right and are presented and discussed in section [ ther_chem ] .
our treatment of the chemistry of the igm does not include the effects of the ionizing photons from stars ( and/or quasars ) that are ultimately responsible for the reionization of the intergalactic gas .
this is justified at early epochs , as these photons are confined within small regions surrounding the luminous sources , but this simplification restricts the validity of our results to the period prior to cosmological reionization .
the post - reionization epoch , and the effect of reionization on galaxy formation , have received extensive study elsewhere ( see * ? ? ?
* and references therein ) . to simulate the chemical evolution of the protogalactic gas , we adopt a chemical model consisting of thirty reactions between nine species : @xmath169 , @xmath170 , @xmath171 , @xmath172 , @xmath128 , @xmath173 , @xmath0 , @xmath174 and free electrons .
the reactions included in the model are summarized in table [ chemtab ] , together with the source(s ) of the data used .
this model is based in large part on that of @xcite , but has been modified to improve its accuracy when applied to optically thick gas . aside from a number of updates to the reaction coefficients in the light of new data , the main differences are as follows : 1 .
we include the contribution to the hydrogen ionization rate arising from the ionizing photons produced by @xmath128 recombination , in line with the discussion in chapter 2 of @xcite .
although commonly a small correction to the total rate , this can become significant when x - ray photoionization dominates .
2 . to enable us to accurately determine the @xmath128 abundance
, we find that we need to include the effects of charge transfer between @xmath128 and @xmath169 ( reaction 20 ) , as this can be comparable to the recombination rate when the fractional ionization is small . for completeness
we also include the inverse reaction ( no .
21 ) , although this is unimportant at @xmath175 .
we include the contribution to the ionization rates of hydrogen and helium arising from secondary ionization by energetic photoelectrons , based on the recent calculations of @xcite .
the contribution of secondary ionization to the other photoionization rates is small and can be neglected .
4 . we do not include the photodissociation of @xmath0 by photons above the lyman limit ( reaction 28 in @xcite ) , as in optically thick gas this will be negligible compared to the effects of @xmath0 photoionization ( reaction 26 ) . on the other hand , we do include the effects of dissociative photoionization ( reaction 27 ) , which becomes significant for photon energies greater than 30@xmath176 .
more information about all of these points , and the chemical model generally , can be found in @xcite .
.a list of the reactions included in our chemical model of protogalactic gas .
values for the rate coefficients ( or radiative cross - sections where appropriate ) are given in @xcite .
references are to the primary source(s ) of the data whenever possible ; in many cases , we have also used analytical fits to this data from @xcite , @xcite or @xcite .
[ cols= " < , < , < " , ] given the initial temperature and chemical abundances , plus the set of chemical reaction rates , actually solving for the thermal and chemical evolution is relatively easy .
as in the igm case , we simply solve the coupled set of chemical rate equations @xmath160 @xmath177 using the stifbs integrator of @xcite . at the start of each timestep ( hereafter time @xmath178 ) , we compute the photochemical rates as outlined in sections [ rtmethod ]
. we then use stifbs to solve for the new chemical abundances and new temperature at the end of the timestep ( time @xmath179 ) , repeating this for each shell in turn .
we next store these values , return to time @xmath178 , and recalculate them using the same procedure , but with _ two _ timesteps of length @xmath180 .
we recalculate the photochemical rates at the intermediate point .
we next test for convergence by comparing our two sets of results .
if any of the chemical abundances or temperatures of any of the shells differ by more than 0.1% , then we reject the results and begin again from time @xmath178 with a smaller timestep .
otherwise , we check to see whether we need to halt the simulation , using the criteria discussed below , and , if we do not , we begin the computations for a new timestep starting from @xmath179 .
one final approximation that we find useful in practice is to fix the @xmath170 and @xmath174 abundances at their equilibrium values .
this allows the integrator to take much larger timesteps than would otherwise be possible , but introduces very little error into the computed @xmath0 abundances . using the method outlined in the preceding sections , we compute the chemical and thermal evolution of the protogalaxy until one of two conditions is met : either the protogalactic gas begins to cool strongly , or we exceed a preset time limit , @xmath181 . to assess whether gas cooling is ` strong '
enough requires an objective cooling criterion .
a number of different possibilities have been suggested in the literature @xcite . in our simulations
, we adopt the criterion used by @xcite : we require that the elapsed time exceeds the cooling time , as computed at the edge of the protogalactic core , at a distance @xmath182 from the centre of the protogalaxy .
the advantage of this choice is that it avoids giving us a false positive result in cases where @xmath183 drops briefly below @xmath184 at early times , but remains so for a time @xmath185 . as an additional sanity check
, we also require that the final temperature be smaller than the initial temperature .
if the protogalactic gas does not cool strongly , then the simulation will terminate when it reaches @xmath181 .
this pre - set time limit is required on purely practical grounds , to prevent simulations in which the gas does not cool from running for excessive amounts of time , but also has a physical justification . in our simulations ,
we treat protogalaxies as isolated objects , uninfluenced by external events . in reality , they are part of a dynamically evolving mass distribution , and the majority will only survive for a limited time before merging with other protogalaxies of a similar or larger size .
it is possible to use the press - schechter formalism to calculate the distribution of survival times as a function of mass @xcite , but for rare objects the mean survival time is typically of the order of the hubble time and thus for simplicity we set @xmath186 .
in the following sections , we present results from a number of simulations that examine the effects of the x - ray backgrounds produced by the various source models discussed in sections [ def_xray ] and [ alt_xray ] . unless otherwise noted ,
all of these simulations assume the same parameters for the star formation model : a standard salpeter imf , with @xmath187 and @xmath188 , a star formation efficiency @xmath189 and a starburst lifetime @xmath190 .
additionally , all of the simulations use the same cosmological model , the @xmath168cdm concordance model of @xcite , which has parameters @xmath191 . before examining the effect of an x - ray background on protogalactic evolution , we first briefly study the evolution of @xmath32 in its absence . as well as allowing us to determine the sensitivity of our results to variations in the uv source model , this also provides us with a necessary baseline against which to compare our other results . in figure [ baseline ] , we plot the evolution of @xmath32 with redshift for the three different star formation models discussed in section [ uvdensity ] .
all three models assume a salpeter imf , with maximum stellar mass @xmath51 , as well as our standard star formation efficiency , starburst lifetime and cosmological model , described previously .
our basic model assumes a metal - free stellar population , with a minimum stellar mass of @xmath187 ; the corresponding results are given by the dotted line in figure [ baseline ] .
the dashed line illustrates the effect of increasing the minimum mass to @xmath50 ; the solid line assumes the same @xmath53 , together with a metallicity @xmath57 .
figure [ baseline ] demonstrates that although the strength of the lyman - werner background increases by almost a factor of five as we move from our basic model to the metal - enriched model , this has less effect on @xmath32 than we might expect : the difference in @xmath32 between the three models is never more than 50% , and the qualitative details of its evolution are very similar in all three models .
this suggests that the uncertainty introduced by our lack of knowledge of the properties of the primordial stellar population need not be unduly limiting .
however , it is also clear that some uncertainty remains , and this will place a lower limit on the magnitude of any effect that we can reliably claim to detect , as small variations in @xmath32 due to the x - ray background will be swamped the by the error resulting from the uncertainty in the lyman - werner background . in the work that follows ,
we take as our baseline the results of our basic , metal - free , low @xmath53 model ; as figure [ baseline ] demonstrates , this minimizes the strength of the lyman - werner background , and thus will tend to maximize the effectiveness of the x - ray background . in figure [ base_xray ] , we plot the evolution of @xmath32 in the presence of the x - ray background generated by the massive x - ray binary model described in section [ def_xray ] . for the purposes of comparison
, we also plot the results of our basic x - ray free model .
initially , the evolution of @xmath32 is the same in both models , implying that the x - ray background has little or no effect on the gas . at a redshift @xmath192 and critical temperature @xmath193
, however , the models begin to diverge significantly . in the x - ray free model , the critical temperature continues to increase rapidly until it reaches its maximum value of @xmath194 . in the x - ray binary model , by contrast ,
the rate of increase of @xmath32 is significantly slower , and it fails to reach its maximum value by the end of the simulation at @xmath195 . because at lower redshift we expect the effects of ionizing radiation from both stellar sources and quasars to become increasingly important and thus our results to become unreliable .
clearly if reionization occurs at @xmath196 , the same is true for some portion of the results plotted here . ]
it is clear from figure [ base_xray ] that the presence of the x - ray background significantly affects the evolution of gas in the larger of the @xmath0-cooled protogalaxies . in small protogalaxies , on the other hand ,
the negative feedback caused by the lyman - werner background remains as strong as ever . a simple way to judge
the importance of this effect is to examine the difference it makes to the fraction of gas in the universe that can collapse and cool . as we saw in section [ sfr ] , we can use the press - schechter formalism to express the cool gas fraction as @xmath39.\ ] ] using the relationship between @xmath43 and @xmath32 derived in that section , it is straightforward to calculate the evolution of @xmath45 in both the x - ray binary and x - ray free models .
to better highlight the difference between the two models , we plot in figure [ fcoll ] the ratio of the cooled gas fraction in the x - ray binary model to that in the x - ray free model . at high redshift ,
the evolution of @xmath32 is the same in both models , and thus the ratio is one .
below @xmath197 , the behaviour of the models begins to diverge , but this has little effect on the ratio until the models begin to diverge sharply at @xmath198 .
thereafter , the ratio rises sharply to a peak at @xmath199 , and subsequently declines as the growth in the cooled mass fraction becomes dominated by the formation of larger protogalaxies that cool via lyman-@xmath84 emission .
figure [ fcoll ] shows us that by ignoring the effect of the x - ray background we underestimate @xmath45 by at most a factor of two .
in fact , the difference is likely to be even smaller : we have assumed that all of the gas in a protogalaxy with @xmath200 will cool , but this need not be the case if an x - ray background is present , as x - ray heating will prevent gas from cooling in the low - density outer layers of the protogalaxy . whether this will affect the star formation rate within individual protogalaxies is not clear , but in any case the difference in the globally - averaged star formation rate is unlikely to be greater than a factor of two .
note , however , that this extra star formation occurs entirely in low - mass systems , from which ionizing photons @xcite and supernova ejecta @xcite can more readily escape .
doubling the star formation rate may thus significantly increase the feedback of primordial star formation on the igm . finally , given the uncertainties in the data underlying our simple massive x - ray binary model ( see section [ def_xray ] ) , it is of interest to investigate the sensitivity of our results to changes in the strength of the x - ray background .
accordingly , we have run additional simulations in which the strength of the x - ray sources was increased or decreased by a factor of ten .
the results are plotted in figure [ alt_norm ] , together with the results of our basic x - ray binary model and of the x - ray free model . unsurprisingly ,
increasing or decreasing the strength of the x - ray background alters the evolution of @xmath32 . decreasing it by an order of magnitude increases @xmath32 to the point where its evolution is little different from that in the x - ray free model .
increasing it by an order of magnitude , on the other hand , systematically lowers @xmath32 , although never by more than a factor of two .
as we discussed in section [ alt_xray ] , it is possible that massive x - ray binaries are much less abundant at high redshift than at the present day .
if so , then the x - ray emission of star - forming galaxies will be dominated by supernova remnants ( snr ) .
these will generate x - rays through two main emission mechanisms : thermal bremsstrahlung from hot gas , and inverse compton scattering of the cmb by relativistic electrons .
we consider the effects of these mechanisms separately .
we examined two possible variants of the thermal bremsstrahlung model .
in one , we assumed that the characteristics of the emission are broadly the same as those observed locally , with a fraction @xmath99 of the supernova energy being radiated as x - rays with a characteristic temperature @xmath201 @xcite .
in the other model , we assumed that all high - redshift supernovae explode in extremely dense surroundings , producing x - ray bright , ultra - compact remnants with @xmath202 and @xmath203 @xcite .
realistic models should lie somewhere between these two extremes .
however , we found that in neither of these cases does the x - ray background affect the evolution of @xmath32 : at our level of temperature resolution , the results are identical to those obtained for the x - ray free model .
the obvious conclusion is that the background produced by bremsstrahlung is simply too weak to be effective .
our inverse compton model fares somewhat better . in this model ,
the strength of the x - ray background is proportional to the mean fraction @xmath106 of the supernova explosion energy that is transferred to relativistic electrons within the remnant .
this value is not known accurately and so we considered two possible cases , one with @xmath108 and another with @xmath109 , these being conservative lower and upper bounds on the true value . in the former case , we again saw no significant effect on the evolution of @xmath32 . in the latter case ,
on the other hand , we saw results very similar to those obtained for the x - ray binary model , as illustrated in figure [ sn_ic ] . in principle , therefore , inverse compton emission from supernova remnants could be as important an x - ray source as emission from x - ray binaries .
ultimately , however , its importance depends upon the value of @xmath106 , and no firm conclusions are possible until this value is better constrained . in this context , the possible observational tests suggested by @xcite could prove extremely valuable .
as well as determining the evolution of @xmath32 , our simulations also allow us to study the thermal and chemical evolution of the diffuse intergalactic medium , as outlined in section [ model_pg ] . as a simple example , we plot in figure [ h2_in_igm ] the evolution of the fractional @xmath0 abundance in the igm for the x - ray free model ( solid line ) and the x - ray binary model ( dashed line ) .
in both cases , the fractional abundance rapidly decreases from its primordial value due to photodissociation by the ultraviolet background , reaching @xmath204 by @xmath205 .
subsequently , its decline slows , in part because the rate of increase in the strength of the ultraviolet background also slows .
below @xmath206 , the behaviour of the two models diverges . in the x - ray free model
, @xmath207 continues to decline until the end of the simulation . in the x - ray binary model ,
on the other hand , the increasing ionization of the igm boosts the @xmath0 formation rate to the point where it overtakes the photodissociation rate and the @xmath0 abundance , after reaching a minimum at @xmath208 , begins to climb .
nevertheless , it remains extremely small at the end of the simulation , readily justifying our assertion in section [ lwabs ] that absorption by intergalactic @xmath0 does not play a significant role in determining the strength of the lyman - werner background . in figure [ igm_ion ]
, we plot the evolution of the fractional ionization of the igm . in the x - ray free model ,
this remains approximately constant over the lifetime of the simulation , as the recombination timescale is significantly longer than the hubble time . in the x - ray binary model , on the other hand , photoionization by the growing x - ray background
eventually overcomes the very small recombination rate and drives the fractional ionization upwards , increasing it by just over an order of magnitude by the end of the simulation .
even so , it remains small , demonstrating that the x - ray background does not contribute significantly to cosmological reionization .
the effect on the fractional ionization of helium ( the ratio of @xmath128 to @xmath172 ) is rather more striking .
the post - recombination @xmath128 abundance is extremely small @xcite , and in the x - ray free model remains at this low level throughout the simulation . in the x - ray binary model , on the other hand ,
it increases dramatically over the course of the simulation , reaching @xmath209 by @xmath38 .
this is illustrated in figure [ igm_heion ] .
finally , in figure [ igm_heat ] we plot the evolution with redshift of the temperature of the igm . in the x - ray free model ,
adiabatic cooling dominates the thermal evolution , and the temperature falls off approximately as @xmath210 . in the x - ray binary model , on the other hand , photo - electric heating begins to heat the igm strongly at @xmath211 , driving the temperature up to @xmath212 by the end of the simulation .
thus , although the x - ray background does not contribute significantly to the reionization of the igm , it does produce substantial reheating prior to reionization .
moreover , given the large mean free path of the x - ray photons , this reheating occurs almost uniformly throughout the igm , rather than being localized to the vicinity of star - forming galaxies .
one consequence of this reheating is that the formation of very small - scale structure will be suppressed , as the increased temperature of the igm leads to an increased jeans mass .
this is unlikely to affect the global star formation rate , however , as star formation within these small structures would in any case be strongly suppressed by the ultraviolet background .
nevertheless , it will reduce the mean clumping factor of the igm below the level that we would otherwise predict , which may in turn speed up reionization @xcite .
reheating also affects the visibility of the igm in the redshifted @xmath213 line of neutral hydrogen .
@xcite show that scattered lyman-@xmath84 emission from high redshift galaxies efficiently couples the spin temperature of the hyperfine levels to the kinetic temperature of the gas . if the gas temperature is smaller than the cmb temperature , this results in @xmath213 line absorption ; if it is greater , then it results in emission .
absorption is easier to detect than emission @xcite , but the heating produced by the x - ray background implies that absorption occurs only at @xmath214 , and that concentrating on detecting @xmath213 emission may be the more viable strategy .
the results of the previous section allow us to assess the impact of the high redshift x - ray background that is produced by star - forming galaxies .
if we assume that the x - ray emission of these galaxies is similar to that observed locally , and that the same correlation between x - ray luminosity and star formation rate applies , then we find that the background produced is strong enough to partially offset the effects of uv photodissociation in large ( @xmath215 ) , @xmath0-cooled protogalaxies . however , local emission is dominated by massive x - ray binaries , which may not form in large numbers at high redshift .
therefore , we have also explored the effect of an x - ray background produced by emission from supernova remnants .
if this emission is dominated by inverse compton scattering and if the fraction of the supernova energy transferred to the relativistic electrons powering this emission is large , then the resulting background has very similar effects to one produced by x - ray binaries . on the other hand , if the fraction of energy transferred is small , then the background has little or no effect .
in addition to inverse compton emission , we have also examined the effect of thermal bremsstrahlung emission from hot gas in the remnants , and find that even if all supernovae were to form x - ray bright , ultra - compact remnants , the resulting x - ray background would still be too weak to significantly affect protogalactic evolution . finally , none of these models produces an x - ray background that is strong enough to balance uv photodissociation in small protogalaxies , with virial temperatures @xmath216 . in these protogalaxies , negative feedback always dominates .
how significant are these results ?
one simple way to assess this is to study the evolution of the mass fraction of cooled gas , which represents the total amount of matter available to form stars . comparing its evolution in the x - ray binary model to that in the absence of an x - ray background
, we find that it is increased by approximately a factor of two .
given our star formation model , this corresponds to an increase in the global star formation rate by the same amount .
however , this is small compared to the order of magnitude increase that would result if we were simply to ignore the effect of the uv background ( see figure 7 of * ? ? ? * ) . in reality
, the difference between the two models may be greater than this because the additional star formation takes place entirely in low - mass systems , from which ionizing photons and supernova - produced metals can readily escape @xcite .
doubling the star formation rate may therefore have more than double the impact on the intergalactic medium . however , to properly assess the ultimate importance of this effect requires more detailed modelling , beyond the scope of this paper .
ultimately , understanding the history of star formation in the small protogalaxies studied in this paper remains important even if they do not contribute to the reionization or enrichment of the igm to any great degree .
this is simply because , in a hierarchical universe , these protogalaxies are the building blocks from which larger galaxies form and therefore set the initial conditions for later stages of galaxy formation . in particular , very little metal enrichment of the primordial gas is required in order to allow the cno cycle to operate and population ii ( rather than population iii ) stars to form , and yet this can have a profound effect on the predicted spectral energy distribution of an early stellar population .
although the main purpose of our study was to examine the effects of the x - ray background on the thermal and chemical evolution of gas within protogalaxies , our approach also allows us to examine the effects of the background on the diffuse igm .
our main results are three - fold : 1 .
we confirm the rapid destruction of @xmath0 in the intergalactic medium noted by @xcite , but also show that when an x - ray background is present the @xmath0 abundance does not continue to decline indefinitely , but eventually stabilizes and may even begin to increase .
however , it never becomes large enough to significantly affect the lyman - werner background .
we show that although photoionization by the x - ray background significantly increases the fractional ionization of the igm ( and in particular the fractional ionization of helium ) , the bulk of the gas remains mostly neutral , demonstrating that the contribution of the x - ray background to cosmological reionization is small .
we find that the x - ray background will also heat the intergalactic gas , raising its temperature to @xmath217 by @xmath38 ( compared to @xmath218 in the x - ray free model ) .
this will suppress the formation of structure on the smallest scales by increasing the jeans mass .
it is unlikely to affect the global star formation rate , since @xmath219 , but may speed up the process of reionization by reducing the mean clumping factor of the igm .
a number of other authors have studied the effects of radiative feedback on the formation of @xmath0-cooled protogalaxies @xcite .
most of these studies have concentrated solely on the effects of the lyman - werner background , generally finding that it suppresses cooling ( and hence star formation ) by @xmath19830 , prior to cosmological reionization .
our x - ray free simulations support this conclusion .
in particular , comparison with the results of @xcite , who use a very similar method but with a different implementation of radiative transfer , gas chemistry and cooling , shows good agreement .
rather less work has been done on the effects of the high redshift x - ray background .
@xcite examined the formation of molecular hydrogen in a constant density primordial cloud illuminated by a power - law uv spectrum extending into the hard x - rays , and found that in some cases , the background could enhance @xmath0 formation .
this lead @xcite to examine the effects of a quasar - produced x - ray background , using a very simple model in which the x - ray flux is a fixed fraction of the lyman - werner band flux , modulated by absorption by a fixed column density of @xmath169 and @xmath172 .
they found that an x - ray to uv flux ratio of 10% was enough to overcome negative feedback , and that a higher flux ratio could potentially produce positive feedback .
recently , @xcite have studied the effects of similar model backgrounds using an adaptive - mesh hydrodynamics code .
they also find that x - rays reduce the effectiveness of negative feedback , but that the latter still dominates .
they do not find evidence for the positive feedback predicted by @xcite .
however , their simulations do not include the effects of @xmath0 self - shielding , and thus potentially underestimate the amount of @xmath0 that forms ( although see * ? ? ? * for a different view ) .
there are several significant differences between our work and these previous investigations .
firstly , we do not assume a fixed spectrum or intensity for the x - ray background ; rather , we specify the properties of the x - ray _ sources _ and subsequently compute the build - up of the background in a self - consistent fashion .
moreover , we consider source models where the x - ray emission is proportional to the star formation rate , as is observed to be the case for star - forming galaxies at low redshift ; the relationship between star formation rate and x - ray emission in quasar - based models is far less clear . as a result , we generally consider x - ray backgrounds signficantly fainter than those studied in the papers cited above .
this makes direct comparison of our results difficult .
however , we note that @xcite find x - ray feedback to be ineffective below @xmath220 , regardless of the strength of the x - ray background ; at the redshifts they consider , this corresponds to a virial temperature @xmath221 , and thus agrees well with the similar result obtained in this paper . in closing
, we note that a number of uncertainties still remain in our treatment of this problem .
some of these the high redshift star formation rate or the appropriate population iii initial mass function , for instance we simply do not know at the present time .
fortunately , changes to our assumed values can be readily incorporated in the framework laid out in this paper .
other uncertainties arise from our method of simulation ; in particular , from our assumption of a static density profile for the protogalactic gas .
we hope to address these issues in future work .
the authors would like to acknowledge detailed comments by the anonymous referee which helped to significantly improve this paper .
scog would also like to acknowledge useful discussions on various aspects of this work with tom abel , omar almaini , rennan barkana , marie machacek and si peng oh .
financial support for much of this work was provided by a pparc studentship .
additional support was provided by nsf grant ast99 - 85392 . | the first generation of stars ( commonly known as population iii ) are expected to form in low - mass protogalaxies in which molecular hydrogen is the dominant coolant .
radiation from these stars will rapidly build up an extragalactic ultraviolet background capable of photodissociating @xmath0 , and it is widely believed that this background will suppress further star formation in low - mass systems .
however , star formation will also produce an extragalactic x - ray background .
this x - ray background , by increasing the fractional ionization of protogalactic gas , promotes @xmath0 formation and reduces the effectiveness of ultraviolet feedback . in this paper , we examine which of these backgrounds has the dominant effect . using a simple model for the growth of the uv and x - ray backgrounds , together with a detailed one - dimensional model of protogalactic chemical evolution , we examine the effects of the x - ray backgrounds produced by a number of likely source models .
we show that in several cases , the resulting x - ray background is strong enough to offset uv photodissociation in large @xmath0-cooled protogalaxies . on the other hand , small protogalaxies ( those with virial temperatures @xmath1 ) remain dominated by the uv background in all of the models we examine .
we also briefly investigate the effects of the x - ray background upon the thermal and chemical evolution of the diffuse igm .
cosmology : theory galaxies : formation molecular processes radiative transfer |
the observation of large anisotropic flow or in particular elliptic flow of different hadronic species at the relativistic heavy ion collider ( rhic ) at brookhaven national lab , new york , provides a strong confirmation of the formation of qgp .
photons are known to probe the properties of qgp in a relatively clean manner compared to hadrons . unlike hadrons , photons are emitted from every stage of the expanding system and also carry undistorted information from the production point to the detector .
elliptic flow is known to provide information about early thermalization and collectivity in the hot and dense matter .
the flow parameter @xmath2 is quantified as the 2nd fourier co - efficient of the particle distribution in the transverse momentum plane as ( at mid - rapidity and for collisions of same type of nuclei only even cosine terms survive in the series below ) : @xmath3\end{aligned}\ ] ] we have shown that elliptic flow of thermal photons for @xmath4 gev au+au collisions at rhic ( using ideal hydrodynamic model ) reflects the anisotropies of the early partonic phase at large values of transverse momentum @xmath0 @xcite . in a recent calculation of photon @xmath2 ( considering prompt and jet photons along with thermal )
, it is shown that the thermal @xmath2 dominates over others for @xmath5 gev @xcite . for thermal photons along with contributions from qm and hm .
right panel : @xmath0 integrated @xmath2 results from different phases as a function of @xmath6.,title="fig:",height=139 ] for thermal photons along with contributions from qm and hm .
right panel : @xmath0 integrated @xmath2 results from different phases as a function of @xmath6.,title="fig:",height=139 ]
the time evolution of photon spectra and @xmath2 is studied for impact parameter @xmath7 fm and @xmath8 fm considering the same initial conditions as given in ref .
rates of photon production from quark matter and hot hadronic gas are taken from arnold _
et al . _
@xcite and turbide _ et al . _
@xcite respectively , which provide good description of the single photon data at rhic as well as at sps energies .
the @xmath0 dependent elliptic flow parameter for thermal photons along with contributions to @xmath2 from quark matter ( qm ) and hadronic matter ( hm ) phases are shown separately in left panel of fig.[fig:1 ] . in the right panel of same figure ,
the @xmath0 integrated @xmath2 from different phases as a function of proper time @xmath6 and normalized with @xmath9 from respective phases are shown , where @xmath10 ( @xmath11 fm ) is the time when freeze - out is completed @xcite .
as shown in the figure , the @xmath12 saturates early within a time period of @xmath13 fm .
the @xmath14 is small at the beginning and saturates much later compared to @xmath12 .
contours of @xmath0 dependent spectra and @xmath2 at various values of @xmath6 from qm , hm and the sum of the two phases are shown explicitly in fig .
[ fig:2 ] . at large @xmath0
( @xmath15 gev ) , most of the photons from qm are emitted very early within a time period of @xmath16 fm . however , the @xmath12 is not that strong at early times and it gradually builds - up as shown in the left panel of fig .
[ fig:2 ] . photons from hm and their @xmath2 , both are very small at the beginning and become significant only after 4 - 5 fm .
the sum spectra and @xmath2 show clear qm dominance at early times or large @xmath0 and the hm contribution is substantial only for lower @xmath0 or later times .
spectra [ upper panel ] and elliptic flow [ lower panel ] of thermal photons for different values of @xmath6 from qm , hm and the sum of the two phases.,title="fig:",height=105 ] spectra [ upper panel ] and elliptic flow [ lower panel ] of thermal photons for different values of @xmath6 from qm , hm and the sum of the two phases.,title="fig:",height=105 ] spectra [ upper panel ] and elliptic flow [ lower panel ] of thermal photons for different values of @xmath6 from qm , hm and the sum of the two phases.,title="fig:",height=105 ] spectra [ upper panel ] and elliptic flow [ lower panel ] of thermal photons for different values of @xmath6 from qm , hm and the sum of the two phases.,title="fig:",height=105 ] spectra [ upper panel ] and elliptic flow [ lower panel ] of thermal photons for different values of @xmath6 from qm , hm and the sum of the two phases.,title="fig:",height=105 ] spectra [ upper panel ] and elliptic flow [ lower panel ] of thermal photons for different values of @xmath6 from qm , hm and the sum of the two phases.,title="fig:",height=105 ]
the final @xmath17 for photons is quite sensitive to the initial thermalization time @xmath1 for a fixed value of particle multiplicity , and the value of @xmath1 can thus be estimated with the help of experimental results @xcite . the differential elliptic flow @xmath17 for different @xmath1 ranging from 0.2 to 1.0 fm at @xmath18 fm , are shown in left panel of fig . [ fig:3 ] .
for smaller @xmath1 , the number of photons from qm increases compared to larger @xmath1 at high @xmath0 . as @xmath12 is smaller at earlier times and @xmath14 does not change significantly with changing @xmath1 , the total @xmath2 decreases with smaller @xmath1 . as hadrons are emitted from freeze - out surface , their @xmath2 is affected only marginally with changing values of @xmath1 ( right panel of fig .
[ fig:3 ] ) . in conclusion ,
the time evolution results for photon spectra and elliptic flow show gradual build - up of the flow parameter along with relative contributions from different phases with time and at different @xmath0 .
photon flow is quite sensitive to the value of @xmath1 , which can be estimated by experimental determination of the flow parameter . | elliptic flow of thermal photons has a great potential to explore the early time dynamics and evolution of quark gluon plasma . @xmath0
dependent temporal contours for photon spectra and elliptic flow from quark matter and hadronic phases , as well as the @xmath0 integrated results show gradual build - up of flow with time and relative contributions from different phases to that very clearly . unlike hadrons , photon flow is quite sensitive to the initial thermalization time @xmath1 , and its value can be estimated with the experimental determination of @xmath2 .
photons , elliptic flow , time evolution , @xmath1 etc . |
let us consider a gaas / algaas heterostructure .
the point symmetry group of the constituent semiconductors is @xmath8 , but the presence of the interface reduces the symmetry of the heterostructure to @xmath7 . lowering the symmetry has important consequences on optical properties , leading to an anisotropy in the ( 001 ) plane .
this anisotropy is of the current interest and is studied both experimentally and theoretically .
the landmark experiments were carried out by jusserand _
@xcite , kwok _ et al . _
@xcite , krebs and voisin @xcite , and kudelski _ et al .
these measurements were accompanied by theoretical considerations concentrating on the anisotropic spin splitting of the conduction band @xcite , the anisotropy of the raman scattering @xcite and the anisotropy due to the heavy hole - light hole mixing by the interface potential @xcite .
a comprehensive review of experimental data and theoretical models developed to investigate the anisotropy of heterostructures and quantum wells can be found in @xcite . in the present paper
we consider an influence of the @xmath9 symmetry lowering on the degree of circular polarization of the luminescence originating from a single gaas / algaas heterostructure .
the investigated heterstructure was grown in the [ 001 ] direction .
a high mobility quasi two dimensional electron gas ( 2deg ) is located at the interface due to si donor doping of the barrier .
a diluted sheet of be acceptors is introduced into the gaas layer at @xmath10 = 30 nm from the interface .
the photoluminescence analyzed originated from transitions between the 2deg and photo excited holes bound to be acceptors .
application of the magnetic field ( @xmath3 ) perpendicular to the ( 001 ) plane splits the electron and hole levels which leads to a nonzero degree of circular polarization at @xmath11 and at sufficiently low temperatures .
a near - band luminescence observed in 2d structures grown from zinc blende materials is often interpreted as a @xmath12 transition , where @xmath13 ( conduction band ) and @xmath14 ( valence band ) are representations of the @xmath8 symmetry group of the bulk crystal at zero wavevector @xmath15 .
the corresponding optical selection rules are shown in figure [ jl1]a .
in such a case , the degree of circular polarization of the luminescence , @xmath4 , is determined only by the magnetic field and the temperature ( @xmath16 ) : the magnetic field determines the value of the zeeman splitting of the electron and hole levels while the temperature determines their occupation .
the selection rules shown in figure [ jl1]a are valid for the perfect @xmath8 symmetry of the electron and hole states . a single heterojunction is a structure of the @xmath7 symmetry and the selection rules of figure [ jl1]a are relaxed .
the resulting scheme ( derived further on ) is shown in figure [ jl1]b .
such a relaxation of the selection rules leads to a decrease of the degree of circular polarization .
electron with a @xmath14 hole in the @xmath8 symmetry .
solid arrows : @xmath17 , dashed arrows : @xmath18transitions .
b ) lowering the symmetry of the system leads to a relaxation of the selection rules and mixing of polarization of transitions between given pair of levels .
intensity of transitions is mimicked schematically by lines thickness .
hole levels are not eigenstates of @xmath19 orbital momentum operator but no possible perturbation - related energy shifts are indicated . ]
the origin of the present paper was an observation of a dependence of @xmath4 on the concentration of the 2deg ( @xmath5 ) which can not be explained on the basis of the selection rules shown in figure [ jl1]a . to explain the effect observed , we discuss two mechanisms which are related to the heterostructure electric field and lead to selection rules shown in figure [ jl1]b .
the first one is the stark effect on a photo excited hole bound to be acceptor .
the second mechanism is an in - plane anisotropy of optical transitions induced by a low symmetry of the heterostructure potential .
the paper is organized as the following . to introduce the idea of the present experiment , we start , in section 2 , with a description of basic facts concerning polarization of the luminescence in acceptor @xmath0-doped heterostructures .
we follow , in section 3 and section 4 , with the experimental part describing the sample investigated , the experimental set - up and results .
next , in section 5 , we introduce a model of the stark effect on the @xmath14 hole .
application of the model to the experimental data leads to the conclusion that the stark effect can not be responsible for the @xmath6 dependence observed .
then , in section 6 , we propose that it results from a low symmetry of the heterojunction electrostatic potential .
we argue that the anisotropy appears due to conduction band electron wavefunctions , and not wavefunctions of holes taking part in the optical transition . within a phenomenological model we derive appropriate selection rules shown in figure [ jl1]b .
finally , we summarize and conclude the paper .
heterostructures similar to that investigated in the present work were already studied by other groups @xcite .
we recall here some basic facts that are relevant to understand the present experiments and results . in acceptor
@xmath0-doped heterostructures , acceptors are introduced into the gaas channel some tens of nanometers from the gaas / algaas interface . a 2deg
is created in a heterostructure quantum well due to doping of the algaas barrier with shallow si donors .
the photoluminescence ( pl ) is excited by a laser beam and some of photo excited holes are captured by acceptors in the @xmath0-layer .
a typical luminescence spectrum shows a number of lines corresponding to different transitions , like donor - acceptor or exciton ones .
here we focus only on the radiative recombination of 2d electrons with holes bound to acceptors in the @xmath0-layer .
since this is the only optical transition analyzed in this paper , the notion `` hole '' will always have the meaning of a hole bound to an ionized acceptor in the @xmath0-layer . if the magnetic field is applied perpendicular to the 2deg plane , the electron density of states is quantized into landau levels ( lls ) and the degeneracy of the hole ground state is totally removed .
in such a case , the pl arising from the recombination of 2d electrons with holes is polarized . the degree of circular polarization @xmath20 , where @xmath21 and @xmath22 are intensities of the @xmath18 and @xmath17 components of the luminescence , respectively .
there are two factors that contribute to @xmath4 : polarization of the 2deg and polarization of the holes .
let us consider an even electron filling factor when the electron gas is composed of an equal number of electrons with the spin up and spin down .
then , @xmath4 is determined by the polarization of holes only , since the electron gas is unpolarized .
this is the case considered in the present paper : the polarization of the luminescence observed is caused by a thermal distribution of holes on acceptor levels split by the magnetic field .
tuning the 2deg concentration can be achieved by the electrical polarization of a semi - transparent electrode prepared on the sample surface .
we use this possibility to perform luminescence measurements as a function of @xmath5 and @xmath3 , and we show that @xmath4 depends on @xmath5 when @xmath3 and @xmath16 are constant .
the sample under investigation was a high quality gaas / al@xmath1ga@xmath2as heterostructure grown on semi - insulating gaas substrate .
the gaas channel of about 1 @xmath23 m above a 50 periods of 5 nm/5 nm gaas / alas superlattice contains unintentional acceptors at a concentration less than 10@xmath24 @xmath25 .
the algaas barrier comprises an undoped 45 nm thick algaas spacer and a uniformly si - doped 35 nm thick algaas layer ; the doping level amounts to 10@xmath26 @xmath25 .
the @xmath0-layer of be atoms with the concentration of 10@xmath27 @xmath28 was introduced into the gaas channel at the distance @xmath10 = 30 nm away from the gaas / algaas interface .
the barrier was covered with a 15 nm thick gaas cap layer .
the measurements were carried out in an optical helium cryostat supplied with a 5 t split - coil .
all measurements were carried out at 1.6 k and the temperature was stabilized within 0.02 k by pumping the helium gas through a manostat .
the luminescence was excited by a he - ne laser .
all data presented in this paper were obtained at the same laser excitation power .
both cw and time - resolved measurements were performed .
the power of excitation was a few mw/@xmath29 and was a few orders of magnitude smaller than that corresponding to a saturation of the luminescence signal .
in the case of time - resolved measurements , laser pulses were generated by passing the laser beam through an acoustooptical modulator driven by a generator of rectangular voltage pulses .
the time resolution was 5 ns .
the luminescence passed through a @xmath30 plate followed by a quartz linear polarizer .
the circular polarizations were separated by turning the @xmath30 plate .
the luminescence was analyzed by a spectrometer supplied with a ccd camera ( for cw measurements ) or with a photomultiplier ( for time - resolved measurements ) . in the latter case ,
the photomultiplier signal was amplified and directed to a photon counter .
a semi - transparent au gate electrode and an ohmic contact were fabricated on the sample surface and the concentration of the 2deg was tuned by polarizing the gate in the backward direction .
@xmath5 was estimated by determination the magnetic field , @xmath31 , at which the luminescence from the @xmath32=1 lls disappears ( two landau levels correspond to each @xmath32 ) .
then @xmath33 , where @xmath34 is the electron charge and @xmath35 is the planck constant .
an example of the magnetic field evolution of the luminescence spectrum is shown in figure [ jl2]a . with an increase of @xmath3 , the number of populated lls of the first electrical subband decreases as their degeneracy grows .
( indicated ) .
lls of the ground electrical subband and the peak corresponding to the first excited ( @xmath36 ) subband are visible .
lls are labeled with their number , @xmath32 . in this case
, the @xmath37 peak disappears at @xmath38 t. the spectra are vertically shifted for better presentation .
b ) an example of deconvolution of a luminescence spectrum into lorentzians . ] the analysis of the polarization of the luminescence starts with a deconvolution of each spectrum into separate lorentzian peaks corresponding to pairs of lls ( figure [ jl2]b ) .
we subtract from the total spectrum lorentzians corresponding to the second electrical subband and the highest in energy pair of lls of the first electrical subband ( @xmath36 and @xmath39 peaks in figure [ jl2]b ) .
this leaves that part of the spectrum which corresponds to an equal number of lls occupied with spin - up and spin - down electrons , i.e. , to the totally unpolarized electron gas .
the area of that part of the spectrum is proportional to the intensity of the luminescence , and used to calculate @xmath4 .
this procedure allows to determine @xmath4 at given @xmath3 for different @xmath5 .
the results are shown in figure [ jl3 ] . clearly , @xmath4 depends on @xmath5 , and to interpret this result is the purpose of the present paper .
dependence for 1 t ( squares ) , 2 t ( circles ) , 3 t ( triangles ) and 4 t ( diamonds ) .
solid lines are linear extrapolations . estimated error of @xmath4 is 0.03 . ]
( solid squares ) and @xmath18 ( open squares ) components of the luminescence from @xmath40 lls for a few values of @xmath3 and applied gate polarizations ( indicated ) .
each @xmath17 signal was scaled to coincide with the corresponding @xmath18 signal .
the exciting laser pulse ends at about 270 ns . ]
figure [ jl4 ] shows examples of a temporal evolution of the luminescence signals measured in @xmath18 ( open squares ) and @xmath17 ( solid squares ) polarizations .
the @xmath17 data was multiplied by a constant factor to coincide with the @xmath18 data .
corresponding @xmath18 and @xmath17 signals are proportional in the whole time domain , both within the laser pulse , and after the pulse end .
the degree of polarization does not depend on time ( within the time resolution of the present experiment ) which indicates that the system is stationary and time - resolved polarization measurements give the same value of @xmath4 as cw measurements .
we used this fact to analyze polarization data obtained from cw measurements which essentially improved the signal to noise ratio .
the fact that the system investigated is stationary does not mean that the distribution of holes on acceptor levels correspond to the temperature of the helium bath surrounding the sample .
we refer here to time - resolved polarization studies on similar structures @xcite which show that at 1.6 k the relaxation time of photoexcited holes on the acceptor levels is of the order of @xmath41 s. this is a few orders of magnitude shorter than the luminescence decay time which is of the order of @xmath42 s , as can be concluded from figure [ jl4 ] . for this reason
, we can assume that photoexcited holes are distributed on the acceptor levels according to an equilibrium thermal distribution corresponding to the helium bath temperature of 1.6 k. the problem of distribution of electrons is avoided in the present considerations because we take into account the luminescence originating from fully occupied landau levels only .
to interpret @xmath43 dependence shown in figure [ jl3 ] , we begin with an analysis of the stark effect on the acceptor bound hole .
this problem was recently considered by smit _
@xcite who calculated the energy of acceptor levels in the electric field .
our approach is complementary : we take into account the magnetic field , not considered in @xcite , and we discuss an influence of mixing of hole levels by the electric field on @xmath4 . a group theory based hamiltonian for the @xmath14 hole in the electric ( @xmath44 ) and magnetic field ( @xmath3)was introduced by bir , butikov and pikus @xcite . in the case of both @xmath44 and @xmath3 in the @xmath45 direction , one gets the following perturbation hamiltonian of the linear zeeman and linear stark effects : @xmath46 where @xmath47 is the bohr magneton , @xmath48 and @xmath49 are isotropic and anisotropic @xmath50-factors , respectively , @xmath51 are the @xmath52 = 3/2 orbital momentum matrices , and @xmath53 is the dipole moment of be acceptor .
let us denote the unperturbed acceptor wavefunctions as @xmath54 , @xmath55 , where @xmath56 is an envelope function .
we choose the following basis of the wavefunctions @xmath57 : @xmath58 ( corresponding to the magnetic quantum number @xmath59= 3/2 , 1/2 , - 1/2 and -3/2 , respectively ) . @xmath60 and @xmath61 are spin @xmath62 up and down spinors , and @xmath63 , @xmath64 and @xmath65 are functions transforming like @xmath66 , @xmath67 and @xmath45 under operations of the @xmath8 symmetry group , respectively @xcite .
the above hamiltonian takes the form : @xmath68 \label{hamiltonmatrix}\ ] ] where @xmath69 , @xmath70 describe the linear zeeman effect and @xmath71 describes the linear stark effect . the hamiltonian given by equation [ hamiltonmatrix ]
can be diagonalized analytically and explicit formulae for mixed wave functions ( @xmath72 ) are : @xmath73 and @xmath74 , where @xmath75 is a normalizing constant and @xmath76 describes mixing of the wave functions ; @xmath77 , where @xmath78 .
the corresponding energies are : @xmath79 and @xmath80 .
the conduction band @xmath13 states are not affected by the electric field @xcite .
@xmath28 ( thin line ) and @xmath81 @xmath28 ( thick line ) .
a ) the electric field at @xmath10 = 30 nm as a function of the electron concentration @xmath5 . dashed line is guide for eye .
b ) @xmath82 dependence calculated on the basis of energies and wave functions obtained by diagonalization of the hamiltonian given by eq .
[ hamiltonmatrix ] at 1.6 k and for @xmath83 d. data for @xmath44= 0 and @xmath84 v / m can not be distinguished in the scale of the figure . ] in the presence of the electric field , the selection rules shown in figure [ jl1]a , valid for @xmath85 , are replaced by the scheme shown in figure [ jl1]b which is a result of mixing of @xmath86 with @xmath87 and @xmath88 with @xmath89 states described by the above hamiltonian . in consequence , a decrease of @xmath4 in comparison with the @xmath90 case can be expected . to determine the magnitude of this effect we use the functions @xmath72 to calculate the intensity of the @xmath18 and @xmath17 transitions in the following way . according to the time - dependent perturbation theory , the intensity of an optical transition between the states @xmath91 and @xmath92 is proportional to @xmath93 , where @xmath94 is the polarization vector of the electromagnetic wave and @xmath95 is the electron momentum . in the case of a transition between electrons occupying landau levels and holes occupying acceptor levels ,
the intensity observed in the polarization @xmath94 is equal to @xmath96 @xmath97 is a polarization independent factor , @xmath98 is the wave function of a conduction band electron with the spin projection @xmath99 equal to @xmath60 or @xmath61 .
statistical weights @xmath100 describe the probability of occupation of each of four acceptor zeeman @xmath101 : @xmath102/\sum_i\exp[(e_1-e_i)/k_bt]$ ] , where @xmath103 is the energy of the lowest lying level , corresponding to @xmath104 .
statistical weights for electron spin levels , @xmath105 , are equal to 1 because we consider equal number of fully occupied landau levels with spin up and down .
let @xmath106 and @xmath107 . then @xmath108 in considering the stark effect we assume in this section that both electron wavefunctions @xmath47 and hole wavefunctions @xmath109 transform according to representations of the @xmath8 symmetry group .
matrix elements in eq .
[ gammac ] can be expressed by @xmath110 .
this equality results from the assumed symmetry of wavefunctions and the fact that the @xmath66 and @xmath67 directions are equivalent in the @xmath8 symmetry . we get
@xmath111 where @xmath112 . to calculate @xmath4 as a function of @xmath44 using ( [ sigmapm ] ) we estimate the heterostructure electric field in the plane where be acceptors
are placed , @xmath113 .
the distribution of charges and the electrostatic potential in the heterostructure were calculated by a self - consistent solution of the schrdinger and poisson equations .
the overall charge neutrality was guarantied by taking into account interface and surface charges , which simulate also the effect of a gate electrode and which are used to control the concentration of the 2deg in the heterostructure .
the results of the calculations are the subband energies and wavefunctions as a function of the total 2deg concentration @xmath5 .
the calculations show that at @xmath114 = 30 nm , where be acceptors are placed , the electric field is almost insensitive to @xmath5 : it increases from @xmath115 v / m to @xmath116 v / m when @xmath5 increases from @xmath117 @xmath28 to @xmath118 @xmath28 ( see figure [ jl5 ] ) .
the @xmath50-factors of holes bound to be acceptors in bulk gaas were obtained in a far infrared experiment by lewis , wang and henini @xcite : @xmath119 and @xmath120 . to calculate @xmath76 , the value of the be acceptor dipole moment , @xmath53 , should be known . since , apparently , this data is not available in the literature , let us assume that @xmath53 = 1 d ( 1 d = 3.3 @xmath121 10@xmath122 cm ) .
this value is higher than the dipole moment of several different acceptors in bulk si investigated by kpf and lassman @xcite , which was found to fall in the range 0.26 d - 0.9 d. taking @xmath123 d , @xmath124 k and @xmath125 v / m we can estimate the upper limit of the stark effect energy corresponding to the present experimental conditions .
we get @xmath126 5 @xmath23ev , which is negligible in comparison with the thermal energy at 1.6 k. it means that the distribution of holes over the zeeman levels is not changed by the heterostructure electric field .
an influence of mixing of hole levels by the electric field on @xmath4 can be investigated by calculation of @xmath82 for different values of @xmath44 .
results for @xmath85 and @xmath127 v / m are shown in the inset to figure [ jl5 ] .
the two curves practically coincide : for the magnetic field of interest , @xmath4 for these two values of @xmath44 differs by less than @xmath128 .
an analysis presented in this section shows that it is not possible to explain the @xmath6 dependence by the stark effect : neither energy shifts of hole levels nor mixing of levels is strong enough to influence @xmath4 significantly .
let us analyze now the influence of the low symmetry of the heterostructure potential on @xmath4 . in structures of the @xmath7 symmetry ,
the directions @xmath129 ( [ 1@xmath1300 ] ) and @xmath131 ( [ 110 ] ) are not equivalent which leads to an anisotropy of optical transitions in the ( 001 ) plane .
this leads to experimentally observed nonzero degree of the linear polarization : @xmath132 } - i_{[1\overline{1}0]})/(i_{[110 ] } + i_{[1\overline{1}0 ] } ) \neq 0 $ ] .
it will be shown below that in systems of the @xmath7 symmetry , @xmath4 and @xmath133 are related to each other:@xmath134 , where @xmath135 is the degree of circular polarization for the @xmath8 symmetry .
the increase of the strength of perturbation causing @xmath136 symmetry lowering results in an increase of @xmath133 and a corresponding decrease of @xmath4 .
for this reason , effects of the symmetry lowering can be investigated by an analysis of @xmath4 .
a basis for modeling the relation between a low symmetry of a quantum system and its optical anisotropy is a theory of ivchenko _ et al .
_ @xcite . according to this theory
, the anisotropy originates from mixing of light and heavy valence band holes while it is explicitly assumed that the @xmath8 symmetry of conduction band electron wavefunctions is preserved .
rsller and kainz , starting from the theory of ivchenko _ et al .
_ , showed that the @xmath137 coupling of the valence and conduction bands leads to [ 110 ] vs. [ 1@xmath1300 ] anisotropy of the conduction band wavefunctions @xcite .
these calculations were carried out for the @xmath138 case only .
a special feature makes the heterostructure investigated in this work different from other systems of the @xmath7 symmetry studied by other groups .
the point is that holes participating in optical transitions are localized as far as 30 nm from the interface ( i.e. , about 10 bohr radii ) and the influence of the localized interface potential on the holes can be neglected .
we showed that the long - range heterostructure potential is also too weak to mix hole levels significantly .
other mixing mechanisms , like the internal stress or stress - related piezoelectric effects , even if present , do not depend on the electron concentration .
for this reason , we propose that the anisotropy results from a low symmetry of conduction band electron wavefunctions . we propose that the source of the anisotropy is a perturbation of the valence band induced both by the localized interface potential and the long - range heterostructure electric field .
these perturbations do not influence holes localized on be acceptors , as we argued above , but only the valence band states in the region of the gaas channel where the heterostructure electric field is strong and the density of conduction band electron wavefunction is large .
next , the symmetry of conduction band electron wavefunctions is lowered due to the @xmath137 coupling of bands described in @xcite .
the @xmath137 perturbation increases with the electron wavevector which means that it becomes stronger when @xmath5 increases .
we propose that the @xmath6 dependence observed results from changes of the heterostructure electric field correlated to changes in the 2deg concentration .
as in the theory of @xcite , we point at two factors contributing to the symmetry lowering : the long - range heterostructure electric field ( leading to the pockels effect ) and the localized interface potential .
in fact , figure [ jl5 ] shows that for @xmath139 nm a large increase , of the order of 10@xmath140 v / m , of the heterostructure electric field is observed with the increase of @xmath5 .
this means that a perturbation of the valence band , and in consequence , of the conduction band , increases with @xmath5 . also , when @xmath5 increases , the heterostructure well becomes narrower , electron wavefunctions shift towards the interface and the influence of the localized interface potential on the electron states increases .
the latter mechanism should lead to much stronger effects in the case of the single interface in the heterostructure investigated in comparison to quantum wells .
the reason is that in quantum wells of the @xmath7 symmetry , contributions of the two interfaces enter with opposite signs and partially cancel each other .
also , the fact that be - bound holes are unperturbed , makes effects of lowering the symmetry of the conduction band states more evident .
carrying out calculations similar to these presented in @xcite with taking into account the magnetic field is beyond the scope of the present paper .
we propose , however , a phenomenological description of the polarization in the system investigated which allows us to derive selection rules for optical transitions between an electron of the @xmath7 symmetry and a bulk - like hole of the @xmath8 symmetry .
> from eq .
[ intensity ] one can calculate the intensity of the @xmath141 transition for the [ 110 ] ( @xmath142 ) and [ 1@xmath1300 ] ( @xmath143 ) polarizations to get @xmath144 } - i_{[1\overline{1}0 ] } = 2\xi\mbox{re}\sum_{\sigma i}w_{i}\langle\mu\sigma|p_x|a\phi_i\rangle\langle a\phi_i|p_y|\mu\sigma\rangle . \label{realis}\ ] ] in the @xmath8 symmetry , the directions [ 110 ] and [ 1@xmath1300 ] are equivalent ( @xmath133 = 0 in this case ) and the above expression is equal to zero .
this is possible only when @xmath145 to get a non zero value of @xmath133 with perturbed electron wavefunctions @xmath23 of the @xmath7 symmetry , it is necessary that @xmath146 where the first equation reflects the equivalence of the [ 100 ] and [ 010 ] directions in the @xmath7 symmetry
. calculation of matrix elements @xmath147 with taking into account ( [ inequality ] ) , gives the selection rules of optical transitions observed in the circular polarizations in the @xmath7 symmetry .
non zero intensities are shown by arrows in figure [ jl1]b .
lowering the symmetry of conduction band wavefunctions leads to relaxation of selection rules and a decrease of @xmath4 caused by mixing of @xmath17 and @xmath18 transitions . to quantify the model
, we introduce a real ( see below ) parameter @xmath148 and we assume that @xmath149 @xmath148 describes an evolution of matrix elements when the symmetry is lowered .
we get @xmath150 where @xmath151 the degree of circular polarization is @xmath152 while @xmath153 .
( assuming that @xmath148 is complex , one can calculate the degree of linear polarization between the [ 100 ] and [ 010 ] directions : @xmath154}-i_{[010]})/(i_{[100]}+i_{[010]})=-\frac{2 \mbox{\tiny{im } } \zeta}{1+\zeta^2}\frac{r - s}{r+s}$ ] which is equal to zero because these directions are equivalent .
this implies @xmath155 . ) assuming that @xmath156 , where @xmath47 transforms into @xmath23 when the symmetry lowers from @xmath8 to @xmath7 , @xmath157 does not depend on @xmath148 and is equal to @xmath135 - the degree of circular polarization that would be measured in the case when both the electron and hole states were of the @xmath8 symmetry .
these allow us to derive the relation between @xmath133 and @xmath4 mentioned above . to estimate the range of @xmath148 corresponding to @xmath4 measured , we use eq . [ gammaccalc ] which gives @xmath158 , where @xmath159 . with data of figure [ jl3 ] and @xmath160 , we get results presented in figure [ jl6 ] showing @xmath148 increasing from about 0.2 to 0.5 with @xmath5 increasing from 10@xmath161 @xmath28 to @xmath162 @xmath28 . calculated from data in figure [ jl3 ] .
squares : 1 t , circles : 2 t , triangles : 3 t , diamonds : 4 t. as an example , solid line correspond to the linear fit for @xmath3 = 3 t in figure [ jl3 ] . for clarity , error bars for @xmath3
= 3 t data only are shown .
the error of @xmath148 corresponds to the error of @xmath4 equal to 0.03 . ]
the results presented in this section allow us to suggest that a widely used approximation of conduction band electron wavefunctions as unperturbed functions of the @xmath8 symmetry is not generally valid .
the symmetry lowering of conduction band states results from the perturbation of the valence band by a long - range heterostructure electric field and the localized interface potential .
measurements on heterostructures with different barrier height could help to separate these two contributions .
this is because the role of the localized interface potential should strongly increase with lowering the barrier height due to an exponential increase of penetration of the barrier by electron wavefunctions .
low temperature polarized magnetoluminescence experiments on a high quality gaas / algaas heterostructure with be acceptor @xmath0-layer incorporated into the gaas channel were carried out . a metallic gate on the sample surface allowed to tune the 2deg concentration in the heterostructure .
the degree of circular polarization of the luminescence originating from transitions between 2deg and photoexcited holes bound to be acceptors was analyzed .
we observed that @xmath4 decreases with the increase of @xmath5 . in principle , this effect can be explained by the stark effect on be - localized hole and/or by an in - plane anisotropy of optical transitions , reflecting the @xmath7 symmetry of the heterstructure potential .
an analysis of the stark effect was based on a group theory hamiltonian , with taking into account results of self - consistent calculations of the heterostructure electric field .
it was shown that the stark effect can be neglected in the heterstructure investigated , which led to the conclusion that the effect responsible for the @xmath6 dependence is a low symmetry of the heterostructure potential .
we argued that be - localized holes are of the @xmath8 symmetry which led us to the conclusion that the optical anisotropy is related to a low symmetry of the conduction band electron wavefunctions .
we propose that this symmetry lowering results from a perturbation of the valence band which induces perturbation of the conduction band due to the @xmath137 coupling .
we used symmetry arguments to derive selection rules of optical transitions in the case when the conduction band electrons and holes are of the @xmath7 and @xmath8 symmetry , respectively , and we provided a phenomenological description of the experimental data . in conclusion , it is proposed that due to a large separation of holes from the interface , the in - plane anisotropy observed is due to lowering of the symmetry of the conduction band electron wavefunctions only .
the mechanism responsible for the symmetry lowering is a perturbation of the valence band states which influences the conduction band states through the @xmath137 coupling .
the increase of the perturbation with @xmath5 is caused by an increase of the heterostructure electric field with @xmath5 .
measurements on heterostructures with different barrier heights could help to separate a contribution from the localized interface potential from that of the long - range electric field .
j. . is thankful to m. grynberg and a. usakowski for discussions . | photoluminescence measurements were carried out on be @xmath0-doped gaas / al@xmath1ga@xmath2as heterostructure at 1.6 k in magnetic fields ( @xmath3 ) up to 5 t. luminescence originating from recombination of a two - dimensional electron gas ( 2deg ) and photo excited holes localized on be acceptors was analyzed .
the degree of circular polarization ( @xmath4 ) of the luminescence from fully occupied landau levels was determined as a function of @xmath3 and the 2deg concentration , @xmath5 . at @xmath3 constant , @xmath4 decreased with the increase of @xmath5
. two mechanisms of the @xmath6 dependence are discussed : a ) the stark effect on a photo excited hole bound to be acceptor and b ) the in - plane anisotropy of the intensity of optical transitions .
a quantitative analysis shows that the influence of the stark effect on @xmath4 is negligible in the present experiment .
we propose that the @xmath6 dependence results from the @xmath7 symmetry of conduction band electron wavefunctions and we give qualitative arguments supporting this interpretation . |
error fields ( efs ) have been and are the subject of intense study in tokamaks @xcite where relative errors as small as @xmath0 or even @xmath1 are known to affect stability , cause disruptions , and degrade confinement and plasma rotation .
error fields are obviously also very important in modern transport - optimized stellarators @xcite , whose performances rely on carefully optimized 3d magnetic fields .
the specially shaped coils that generate such fields are numerically optimized and , typically , are built and positioned with very high precision .
in other words , errors are minimized at the construction stage .
vacuum fields are then experimentally characterized by a standard technique involving an electron beam and a fluorescent rod @xcite .
incidentally , it should be noted that , since this technique requires flux surfaces to exist in a vacuum , it is not applicable to devices such as tokamaks that require plasma current to generate flux surfaces .
the measured surfaces are usually confirmed to be in good agreement with the desired , optimal configuration . however ,
if not in agreement , a comparison of computed and measured flux surfaces can shed light on possible sources of errors .
at that point one can either ( 1 ) correct the error `` at the source '' ( reposition one or more coils ) @xcite or ( 2 ) apply ef corrections by means of a separate set of dedicated coils .
this latter approach is quite common in tokamaks @xcite . as for stellarators ,
ef corrections were applied in lhd by means of resonant magnetic perturbation ( rmp ) coils , also deployed in other mhd studies @xcite .
w7-as used `` special '' and `` control '' coils to vary the toroidal mirror term and boundary island geometry and study their effect on plasma properties @xcite , and similar uses are envisioned for the `` trim coils '' in w7-x @xcite .
earlier possibilities for ef correction in w7-x were discussed in ref .
@xcite .
stellarators are considered complicated to build .
it has been suggested that their attractiveness as reactors could increase if their construction is simplified , or construction tolerance relaxed , without significantly degrading the plasma properties @xcite . at this point
it is highly hypothetical , but a possible route to simplification could consist in : i ) slightly relaxing the tolerance and ii ) correcting the errors _ a posteriori _ , either by approach ( 1 ) or ( 2 ) listed above . in either approach , it is useful to experimentally quantify the displacements , tilts and , possibly , deformations of the actual coils , compared with design values .
this is useful anyway , even if stricter tolerance is adopted , to confirm that the construction imperfections are indeed smaller . as a first step in exploring such route ,
we have complemented the well - established experimental technique mentioned above with metrology measurements and with a numerical method that `` inverts '' the flux surface errors into error sources such as imperfections in coils positions and tilts .
the cnt stellarator , constructed in 2004 @xcite , is notable for its simple coil configuration and for having the lowest aspect ratio of any stellarator ever built @xcite .
the magnetic field is generated by a system of four planar , circular coils : two interlocked ( il ) coils and two poloidal field ( pf ) coils , the latter of which form a helmholtz pair ( fig .
[ fig : coil_schematic ] ) .
this configuration , as designed , generates a set of toroidal nested closed flux surfaces with two field - periods .
the radial profile of rotational transform [ line - through]*@xmath2 * , as well as the general shape of the flux surfaces , can be controlled by varying two main parameters : the tilt angle between the il coils and the ratio of the current in the il coils to the current in the pf coils , @xmath3 .
cnt was constructed to permit three different il coil tilt angles @xmath4 : @xmath5 , @xmath6 , and @xmath7 .
these angles were chosen for their distinct [ line - through]*@xmath2 * profiles as well as for their resiliency to efs @xcite .
at which poincar cross - sections are measured as described in section [ sect : flux_surf_meas].,scaledwidth=42.5% ] error field resiliency was an especially important consideration for cnt s coil configurations due to the permissive tolerances used in the construction of the coils and the vacuum vessel . whereas present - day stellarators are typically built to tolerances on the order of @xmath8 to @xmath0 @xcite , cnt s tolerances were of order @xmath9 to @xmath8 to minimize the cost and complexity of construction . from 2005 - 2010 ,
the interlocked coils were kept in the @xmath10 configuration .
the flux surfaces for this configuration were measured experimentally and found to agree very well with numerical predictions @xcite .
more recently , cnt s coils were switched to the @xmath11 configuration .
this configuration is predicted to have less magnetic shear than the previous one . as a result ,
efs that resonate with rational surfaces within the [ line - through]*@xmath2 * profile are expected to have more significant effects on the magnetic geometry , leading to larger islands and equilibrium deformation .
a detailed understanding of the magnetic geometry in cnt is a central objective of the cnt research program .
it will allow for more precise alignment of diagnostics and improve the accuracy of equilibrium modeling and reconstruction .
it will also assist in the design of equipment whose shaping and placement are heavily dependent on the field , such as island divertors .
in this paper we present the first detailed measurements of the flux surface geometry of the @xmath11 configuration , which exhibits noticeable disagreements with predictions ( sec .
[ sect : flux_surf_meas ] ) .
we then describe efforts to diagnose the sources of the efs under the assumption that the main sources of error are displacements of the coils from their design positions .
these efforts include ( 1 ) photogrammetric measurements of the positions of the pf coils ( sec .
[ sect : photogrammetry ] ) , ( 2 ) studies of the effects of different classes of displacements on the rotational transform [ line - through]*@xmath2 * ( sec .
[ sect : numerical_studies ] ) , and ( 3 ) design of an optimization algorithm to find the most likely displacements of the il coils that lead to the observed flux surface geometry ( sec .
[ sect : optimization ] ) .
we then discuss the potential broader applicability of the optimization algorithm to more complex coil configurations ( sec .
[ sect : discussion ] ) and briefly describe future work ( sec .
[ sect : summary ] ) .
the flux surface geometry in cnt is measured with a standard technique involving an electron beam and a fluorescent rod @xcite .
electrons are emitted from an electron gun at energies of roughly 80 ev .
the electrons travel along a field line until they strike an obstacle , either a part of the vessel ( if an open field line ) or an aluminum rod that extends into the confining region .
the aluminum rod is coated with zno : zn fluorescent powder that emits blue - green light ( 505 nm ) when struck by electrons . in cnt ,
two fluorescent rods are positioned at the mid - plane of the vacuum vessel , which coincides with the toroidal cross - section at @xmath12 ( fig .
[ fig : coil_schematic ] ) .
the rods can be rotated across this surface with an external actuation mechanism described in ref .
background pressures are maintained at the base pressure of the cnt vacuum vessel ( @xmath13 torr ) to minimize collisions with neutral atoms and molecules .
if a long - exposure photograph is taken of the fluorescent rods as they are scanned across the @xmath12 plane during the emission of an electron beam , the resulting image will show a cross - section of the flux surface on which the beam was emitted .
certain parts of the cross - section may not appear in the image due to shadowing by the electron gun or the limited extent of the fluorescent rods .
for the measurements described in this paper , a digital camera was positioned outside the vacuum vessel to face the plane of the cross - section through a fused - silica viewport .
images were acquired using ten seconds of exposure time .
in addition to the fluorescent glow from the rods , the images contained some regions of stray light .
this was due to blackbody emission from the electron gun filament reflecting off components in the vacuum vessel .
regions of stray light were eliminated manually from each image before processing .
the relative positions of each pixel were determined based on the length of the rod in the image .
the absolute positions of each pixel were then determined by comparing the position of the rod to the position of a fixed landmark in the chamber .
the lens axis of the camera was confirmed to be within one degree of perpendicular to the @xmath12 plane through the use of a leveling tool .
this uncertainty could contribute to up to 3 mm of displacement of the resulting image , as the camera was placed roughly one meter away from the cross - section . ; ( b ) @xmath14,scaledwidth=50.0% ] two examples of experimentally measured poincar cross - sections are shown in fig .
[ fig : ideal_poinc_comp ] , overlaid with numerically computed poincar data for the corresponding current - ratios .
the numerical data , determined using a biot - savart field line tracer described in ref .
@xcite , were generated assuming that the coils were perfectly aligned .
note that the qualitative agreement is poor : the outboard side of the surfaces at @xmath15 ( fig .
[ fig : ideal_poinc_comp]a ) are flatter and have more vertical elongation than the experimental surfaces .
in addition , the experimental data for @xmath14 ( fig . [ fig : ideal_poinc_comp]b ) have prominent islands that are not predicted numerically .
in addition to the geometric disagreements in the poincar sections at toroidal angle @xmath12 , the measured rotational transform [ line - through]*@xmath2 * has been observed to differ from numerical predictions . while detailed profiles of [ line - through]*@xmath2 * have not been measured in cnt ,
it is possible to identify low - order rational surfaces through field line visualizations @xcite . in this technique
, the electron gun is operated while the vacuum vessel is back - filled to a neutral pressure between @xmath1 and @xmath0 torr . at these pressures ,
electron - neutral collisions are frequent enough that the path of the electron beam emits a glow that is visible to the naked eye . in the vicinity of a low - order rational surface
, the beam can be seen to strike the back side of the electron gun after a finite number of toroidal transits .
for example , for @xmath16 , a region exists in which the electron beam is observed to strike the electron gun after three toroidal transits , indicating the presence of a surface ( or island chain ) with [ line - through]*@xmath2 * = 1/3 . on the other hand ,
numerically calculated [ line - through]*@xmath2 * profiles ( fig .
[ fig : iota_unp ] ) only contain [ line - through]*@xmath2 * = 1/3 for @xmath17 . evidently , an ef is causing a systematic offset of the rotational transform . * @xmath2 * profiles for selected current - ratios in cnt in the @xmath18 il coil configuration . in the red curves , which are labeled with respective values of @xmath3 , each dot represents a closed flux surface and dashed segments indicate island chains .
some low - order rational numbers are shown as horizontal dashed lines .
note that current - ratios of 3.18 and above are not predicted to contain [ line - through]*@xmath2 * = 1/3 , contrary to measurements.,scaledwidth=50.0% ] the observed discrepancies between experiment and calculations motivated a more detailed study of the possible sources of efs .
displacements of the pf coils from their design positions were measured directly using photogrammetry . in this procedure , reflective markers were affixed to all parts of the pf coils as well as the exterior of the vacuum vessel . the vessel and coils
were then photographed from several different angles .
the photographs were analyzed by v - stars software ( produced by geodetic systems , inc . ) , which generated a point cloud giving the relative positions of the markers in 3d space .
the point cloud was then fitted to the nominal surfaces of the machine as provided by a cad file .
the result was a set of displacement vectors ( fig .
[ fig : photogrammetry ] ) for each marker corresponding to the displacement of its respective part of the coil from the design location .
displacement vectors that deviated significantly from those of neighboring markers were viewed as erroneous and were ignored in subsequent analysis .
an attempt was also made to measure the positions of the il coils using a romer six - axis measuring arm . however , limited access to the il coils restricted their measurements to a small region of each coil .
the measurements obtained were sufficient to estimate the location of the center of the device to within less than 1 mm ( which was then used as a reference for the pf coil marker offsets ) , but did not yield accurate estimates of the misalignments of the il coils relative to one another .
we will note that it is possible to configure the chamber to permit full access for the romer arm , but this could not be done on the time scale of the arm s availability for the measurements described in this paper and will be an objective for future work .
blue ( red ) arrows represent deviations in the same ( opposite ) direction of the vector normal to the surface where they originate.,scaledwidth=45.0% ] according to the marker displacements determined by the software , certain regions of the pf coils are separated by as much as @xmath19 mm from their intended locations , which is well outside of the 10 mm tolerance specified for construction .
it is noteworthy that cnt s coils still produce good flux surfaces despite having displacements of this magnitude .
the observed misalignments of the pf coils are believed to have arisen during a procedure to change the il coil configuration from @xmath20 to @xmath11 .
the pf coils must be tilted out of the way to permit access to the il coils , and they sit on hinged support structures for this purpose . by the end of the procedure , the structures had visibly deformed due to internal stresses arising from the tilting , likely affecting the positions of the pf coils themselves . to determine the effects of the measured pf coil displacements on the flux surfaces ,
new field line traces were conducted with the pf coils offset according to the photogrammetry data . for these calculations , the pf coils were still assumed to be circular and planar but were tilted and translated so as to best fit the displacement vectors . according to these fits , the northern ( @xmath21 ) and southern ( @xmath22 ) pf coils were translated 5 mm and 22 mm respectively , and their axes were tilted by @xmath23 and @xmath24 respectively . cross - section with @xmath14 from a configuration incorporating the measured pf coil displacements ( blue ) and from the design configuration ( black ) .
( b ) calculated profiles of [ line - through]*@xmath2 * for this configuration.,scaledwidth=50.0% ] a comparison of the numerical poincar plots with and without the displacements is shown in fig .
[ fig : pfc_disp_compare ] using @xmath14 .
although the plots are not identical ( in particular , the magnetic axis has shifted slightly outboard and field lines initiated at the same locations do not overlap perfectly ) , the effects of the pf coil displacements do not appear to be sufficient to explain the observed experimental discrepancies .
in particular , the slight increase in the [ line - through]*@xmath2 * profile is actually the opposite effect of what would be needed to explain the [ line - through]*@xmath2 * = 1/3 island chain seen at an intermediate minor radius in fig .
[ fig : ideal_poinc_comp]b . since the pf coil displacements were not enough to explain the observed differences in poincar cross - sections , additional sources of field error were sought
in the search for additional sources of field error , the most likely candidate was thought to be displacements to the il coils .
previous work @xcite has shown that a translational displacement of either of the il coils has a significantly greater effect on the magnetic geometry than the same displacement applied to either of the pf coils .
( for this reason , the tolerance for the il coils was set at 2 mm , one - fifth of the pf coil tolerance . ) to identify the possible causes of the systematic offset in [ line - through]*@xmath2 * discussed in section [ sect : flux_surf_meas ] , a series of numerical field line traces was computed for different classes of il coil displacements .
field lines were also traced for similar classes of pf coil displacements for comparison .
the classes of displacements were sorted into symmetric ( equal coil movement in opposite directions ) and antisymmetric ( equal coil movement in the same direction ) . for each class of displacement ,
[ line - through]*@xmath2 * profiles were calculated for a series of magnitudes of the displacement . to represent the calculated trends in [ line - through]*@xmath2 * , fig .
[ fig : diota_compare ] shows derivatives of [ line - through]*@xmath2*@xmath25 ( _ i.e. _ , the limit of [ line - through]*@xmath2 * as the minor radius approaches zero ) with respect to each type of displacement .
derivatives were estimated by second - order finite differences about the nominal positions of the coils at @xmath15 .
* @xmath2*@xmath25 with respect to the magnitude of different classes of coil displacements .
( a ) translations of the coils in the cartesian directions illustrated in fig .
[ fig : coil_schematic ] .
( b ) tilts of the coils in the directions of their respective @xmath26 and @xmath27 axes .
the @xmath26 and @xmath27 axes for the il coils are illustrated in fig .
[ fig : coil_axes ] ; for both pf coils , @xmath28 and @xmath29 . a more detailed description of each displacement class is given in the text.,scaledwidth=50.0% ] the displacements plotted in fig .
[ fig : diota_compare ] collectively represent the twenty degrees of freedom that the four coils have for rigid motion ( five for each coil ; rotation of a coil about its axis is ignored due to symmetry ) .
the translational displacements shown in fig .
[ fig : diota_compare]a have fairly intuitive interpretations .
a positive `` ilc co-@xmath30 '' displacement , for example , involves motion of both the il coils in the positive @xmath30 direction as indicated in fig .
[ fig : coil_schematic ] .
an `` ilc counter-@xmath30 '' displacement , on the other hand , involves motion in opposite ways in the @xmath30 direction , with positive displacement indicating the il coils moving apart and negative displacement indicating the il coils moving toward one another .
the angular displacements shown in fig .
[ fig : diota_compare]b represent tilts of the coil in two orthogonal directions denoted by @xmath26 and @xmath27 .
the @xmath26 and @xmath27 vectors for the il coils are illustrated in fig .
[ fig : coil_axes ] .
`` ilc co - b - tilt '' displacement , for example , refers to both il coils tilting by some angle @xmath31 such that their axes rotate toward their respective @xmath27 unit vectors .
a change in ilc co - b - tilt is equivalent to an adjustment of @xmath4 ( nominally @xmath6 ) .
`` ilc counter - a - tilt '' displacement , as another example , refers to the il coils tilting such that il1 tilts its axis in the positive @xmath32 direction and il2 tilts its axis in the negative @xmath33 direction . for both pf coils , @xmath28 and @xmath29 .
vectors ( green arrows ) , and @xmath27 vectors ( blue arrows ) as described in the text .
the @xmath34 , @xmath35 , and @xmath36 directions as defined in fig .
[ fig : coil_schematic ] are shown in yellow , magenta , and cyan respectively.,scaledwidth=30.0% ] from fig .
[ fig : diota_compare ] , it is evident that the two classes of displacements that have the greatest influence on [ line - through]*@xmath2 * are the separation of the il coils ( along the @xmath30 axis ) and the tilt angle between the il coils . another less - prominent influence arises from the separation of the pf coils .
this is roughly equivalent to changing the pf coil current , as it strengthens or weakens the helmholtz field created by the coils , and illustrates how adjusting the pf current relative to the il current serves as a fine adjustment to the iota profile . the relative sensitivity of [ line - through]*@xmath2 * to the different classes of coil displacements in fig .
[ fig : diota_compare ] suggests that , if the offset of the [ line - through]*@xmath2 * profile is indeed caused by coil displacements , it is very likely that displacements of the il coils play a significant role .
this observation motivated a more detailed study of il coil displacements to be described in the following section .
the optimization method used in this study tweaks the coil positions in small increments until the calculated poincar data resulting from the perturbation matches with the experimentally obtained poincar cross - sections .
the procedure is conceptually similar to methods used in some plasma equilibrium reconstruction codes such as efit and v3fit @xcite , which determine plasma equilibrium parameters that fit to diagnostic signals . in this procedure ,
poincar cross - sections take the role of the diagnostic signals , whereas coil displacements have the role of the equilibrium parameters . for the optimizations described in this paper ,
the observed poincar cross - sections are mapped into a vector @xmath37 of discrete geometric parameters , which may be thought of equivalently as physics parameters derived from flux surface topology .
the definitions of these parameters are provided in [ sect : parametrization ] , and the procedure for determining them from experimental and numerical poincar data is given in [ sect : coeff_determination ] .
the coil misalignments were encapsulated in a vector @xmath38 of displacement parameters ( _ i.e. _ , engineering parameters ) .
the work in this paper used two variants of @xmath38 , both of which will be described later in this section .
an optimization begins with target geometric parameters @xmath39 for the desired poincar cross - section geometry and an initial guess @xmath40 of coil displacement parameters .
the geometric parameters associated with @xmath40 ( or any set of dispacement paramameters @xmath38 ) are then @xmath41 .
the discrepancy @xmath42 between @xmath39 and @xmath41 is defined as @xmath43 the optimization algorithm attempts to find @xmath44 such that @xmath45 using the newton - raphson method @xcite .
expanding @xmath46 about a coil configuration @xmath38 that differs from @xmath44 by @xmath47 , i.e. , @xmath48 the iterative newton step @xmath47 is given by a solution to the equation @xmath49 here , @xmath50 is the jacobian , @xmath51 in general , @xmath38 is not of the same dimension as @xmath46 .
we thus find the optimal @xmath47 through linear least - squares .
the best linear unbiased estimator ( blue ) for @xmath47 is that which minimizes @xmath52 @xcite : @xmath53 where @xmath54 is the covariance matrix for the target parameters @xmath39 .
this estimator for @xmath47 is given by @xmath55 for experimental poincar data , the covariance matrix @xmath56 is estimated by evaluating @xmath37 for multiple samples of the pixels obtained from composite flux surface images . for each sample , pixels are selected randomly with replacement ( bootstrapping ) @xcite . the number of pixels available to sample from a particular flux surface in the images used for this analysis ranged from 300 to more than 10,000 .
each sample consists of 400 pixels per flux surface . for the verification studies conducted in sec .
[ subsect : verification ] , the optimizer was programmed to fit the coils to numerically determined poincar data ( _ i.e. _ , a manufactured solution ) . to obtain a covariance matrix for numerical poincar data , multiple poincar plots are obtained by tracing field lines from randomized initialization points .
each sample consists of a trace of 200 field lines followed for 200 toroidal revolutions .
( a certain percentage of the field lines will terminate if they are initialized outside of the last closed flux surface ; these are ignored . ) as indicated in eq .
[ eqn : blue ] , @xmath56 must be inverted for the analysis
. this can introduce significant numerical error if @xmath56 is ill - conditioned . to mitigate this risk , rather than directly inverting @xmath56 , its inverse is approximated through singular value decomposition :
@xmath57 here , the orthogonal matrices @xmath58 to the left and right of @xmath59 are the same due to the inherent symmetry of @xmath56 .
@xmath59 is a diagonal matrix of the singular values .
if @xmath56 is ill - conditioned , the lowest singular value(s ) will be much less than the greatest singular value . as a workaround
, a cutoff is imposed such that all singular values of @xmath56 less than a factor @xmath60 of the greatest singular value are set to zero .
the pseudoinverse of this approximation is given by @xmath61 where @xmath62 is a diagonal matrix in which the @xmath63 nonzero diagonal element is equal to the inverse of the @xmath63 nonzero diagonal element of @xmath59 . for the work described in this paper , @xmath64 was chosen empirically to be @xmath65 for the calculations in this paper as the lowest value for which the newton direction was not noticeably compromised .
the elements of the jacobian in this implementation are computed by second - order finite differences : @xmath66 note that , due to its dependence on @xmath39 via @xmath46 , @xmath67 contains random error and may be correlated with other matrix elements . because of this , eq .
[ eqn : blue ] does not , strictly speaking , give the blue for @xmath47 , since one of the underlying assumptions is that @xmath50 is non - random .
nevertheless , we have found that this estimator is sufficient in many cases as long as the finite differencing interval @xmath68 is sufficiently large .
if @xmath46 depends nonlinearly on @xmath38 , there is a risk that the newton step @xmath47 will overshoot a local minimum in @xmath52 . to rectify such occurrences
, a line - search algorithm checks the newton step after each iteration to ensure that the average rate of decrease in @xmath52 over the interval @xmath47 is at least as great as the gradient of @xmath52 evaluated at the starting point @xmath38 of the iteration .
if this condition is not met , the algorithm samples @xmath52 at shorter distances along the direction of the newton step to identify the local minimum .
more details on this algorithm are given in ref .
@xcite . to verify the performance of the algorithm , some tests were conducted in which the code was used to solve for a known coil displacement @xmath44 using target parameters @xmath69 generated from a field - line trace that used @xmath44 as its input .
one early test was meant to determine whether the algorithm could distinguish two classes of displacements that produced qualitatively similar outcomes .
one such pair of displacement classes is ( 1 ) separation of the coils along the @xmath30 axis and ( 2 ) adjustment of @xmath4 .
incidentally , these two classes were also found to have the greatest impact on rotational transform ( fig .
[ fig : diota_compare ] ) , a strong determiner of the cross - section geometry .
a @xmath38 vector was thus defined with just two components : @xmath70\ ] ] here , @xmath71 and @xmath72 are the displacements along the machine @xmath30 axis of the first and second il coil , respectively , from their nominal positions .
@xmath73 is the displacement of the coil angle from its nominal value of @xmath6 . @xmath74 and @xmath75 are scaling constants equal to 1 m and 1 radian , respectively .
the target parameters @xmath76 were generated using @xmath77 and @xmath78 ; _ i.e. _ , no discrepancy in coil separation and a decrease of @xmath79 in @xmath4 .
the initial guess @xmath40 was ( 0 , 0 ) and the finite differencing interval for both components was @xmath8 .
the results of this test are shown in fig .
[ fig : contours_tilt_dx ] .
note that the contours of @xmath52 exhibit a shallow valley along a line consisting of a family of displacements that would result in similar poincar geometry .
nevertheless , the algorithm succeeded in moving the components of @xmath38 to within one finite differencing interval of the target values on the fourth iteration . and the path taken by the optimizer in a verification test with two free parameters .
the initial guess @xmath40 is ( 0 , 0 ) and the target parameters are denoted by the red @xmath80 .
contours are on a logarithmic scale with solid contours representing powers of 10.,scaledwidth=50.0% ] subsequent verifications tested the ability of the code to identify similar coil displacements when more coil parameters were free . in these tests ,
the il coils were permitted to undergo rigid rotations and transformations ; hence , each coil was allowed five degrees of freedom : three translational and two angular .
the translational parameters , @xmath30 , @xmath81 , and @xmath82 , are simply cartesian displacements along the respective unit vectors indicated in fig .
[ fig : coil_schematic ] . the angular parameters
are illustrated in fig .
[ fig : coil_disp ] . as shown in the diagram ,
@xmath83 and @xmath84 are orthogonal projections of a unit vector representing the _ perturbed _ coil axis onto the plane of the _ unperturbed _ coil .
the @xmath85 and @xmath86 components correspond to the directions @xmath32 and @xmath87 illustrated in fig .
[ fig : coil_axes ] .
the polar displacement angle of the perturbed axis of the first il coil can thus be computed as @xmath88 . and @xmath89 components of the angular displacement parameters of the coils .
in this image , the il1 coil is used as an example .
the blue surface represents the plane of the unperturbed coil , and the red arrow represents a unit vector along the axis of the perturbed coil.,scaledwidth=50.0% ] mm and the rest of the parameters zero .
( b ) parameter evolution toward targets of @xmath90 , @xmath91 , and the rest zero .
( c ) descent of @xmath52 for both optimizations.,scaledwidth=50.0% ] fig .
[ fig : verification ] shows the outcomes of two calculations that optimized the ten il coil parameters as described above to a target @xmath39 vector generated numerically from a chosen set @xmath44 of displacements . in fig .
[ fig : verification]a , the only nonzero component of @xmath44 was @xmath92 , which was set to 5 mm , a simple translation of the first il coil . in fig .
[ fig : verification]b , there were two nonzero components of @xmath44 : @xmath86 = 0.002 and @xmath93 = 0.004 ; _
i.e. _ , both coils were tilted toward their respective @xmath27 vectors , but by different amounts .
both optimizations descended by multiple orders of magnitude in @xmath52 in the first three steps ( fig .
[ fig : verification]c ) , but afterward , @xmath52 flattened out : the optimizations did not converge any further toward @xmath44 .
one possible explanation is that a relatively long finite differencing interval @xmath68 was used to determine the jacobian in each step : all translational ( @xmath30 , @xmath81 , and @xmath82 ) components used an interval of 1 mm , and the angular components ( @xmath94 and @xmath89 ) used an interval of 0.001 .
all of the final values @xmath95 of the components are well within @xmath96 of the target values @xmath97 ( fig .
[ fig : verification]a - b ) .
another contributing factor may be the numerical uncertainty of @xmath46 .
after the verification tests , the algorithm was applied to experimental poincar data .
the target geometric parameters @xmath98 were determined from the cross - section data for the current - ratio 3.68 ( fig .
[ fig : data2params ] ) .
this dataset was chosen over datasets from other current - ratios for its abundance of fully - characterized flux surfaces and for its lack of magnetic islands , two attributes that facilitate accurate calculations of @xmath98 ( sec .
[ subsect : islands ] ) .
the pf coils were held fixed with displacements determined by a best - fit to the photogrammetry data discussed in sec .
[ sect : photogrammetry ] .
all ten parameters associated with the il coils were free .
the initial guess @xmath40 was the nominal il coil configuration . as in the 10-parameter verifications conducted in sec .
[ subsect : verification ] , the finite differencing interval was 1 mm for all translational motion and 0.001 for all angular displacements .
the set of displacements obtained in the final iteration will be referred to hereafter as @xmath99 .
the shifts in il coil positions during the fifteen steps of the optimization are shown in fig .
[ fig : params_exp_10free]a , with the accompanying descent of @xmath52 shown in fig .
[ fig : params_exp_10free]b . the largest translational moves from the starting positions occurred in the negative @xmath35 direction for both coils , averaging to -22 mm .
the @xmath36 displacement was nearly identical at around 6 mm for both coils , and the @xmath34 displacement averaged to about 3 mm with a 3 mm `` counter- '' component ( cf . the displacement classes in fig .
[ fig : diota_compare ] ) .
it should be noted that , while the `` co- '' translational motion of the coils may indicate that the il coils are indeed off of their nominal positions , it may also reflect misalignment of the fluorescent rod .
the largest angular shift was of the first il coil , which tilted about @xmath100 away from its nominal axis along @xmath32 , effectively decreasing @xmath4 to @xmath101 . with the results of fig .
[ fig : diota_compare ] in mind , this shift , combined with the 3 mm of counter - motion in the @xmath34 direction , is likely responsible for most of the downward offset in [ line - through]*@xmath2 * observed experimentally . during the optimization , the value of @xmath52 ( fig .
[ fig : params_exp_10free]b ) for the poincar data associated with the displacements decreased by a factor of more than 100 .
most of this decrease occurred during the first five iterations , after which improvement was insignificant .
the final value of @xmath52 is lower for this optimization than in the verifications ( fig .
[ fig : verification]c ) by about an order of magnitude ; however , this is primarily due to the fact that the target parameters @xmath98 determined here from the experimental data have greater uncertainty than those of the manufactured solutions .
associated with the poincar data resulting from the displacements.,scaledwidth=50.0% ] the qualitative improvement in the numerically predicted poincar plots generated using @xmath99 versus @xmath40 is shown in fig .
[ fig : opt_poinc_comp ] . fig .
[ fig : opt_poinc_comp]a , identical to fig .
[ fig : ideal_poinc_comp]a , refers to @xmath15 and compares the experimental data to a numerical prediction generated from @xmath40 ( _ i.e. _ , coil displacements were neglected ) .
note , again , the disagreements in the vertical elongation on the outboard side , the concavity on the inboard side , and in the position of the magnetic axis .
most of the numerical cross - sections intersect multiple experimental ones .
[ fig : opt_poinc_comp]b shows the same experimental data overlaid on numerical data computed from @xmath99 .
the incorporation of the displacements in the comparison has dramatically improved the qualitative agreement .
the discrepancies in magnetic axis , as well as the shaping of the inboard and outboard sides , has essentially vanished , and there far fewer cases of numerical cross - sections intersecting multiple experimental ones .
[ fig : opt_poinc_otherratios]a - b show comparisons at other current - ratios ( @xmath102 for [ fig : opt_poinc_otherratios]a and @xmath14 for [ fig : opt_poinc_otherratios]b ) in which the numerical poincar data were generated from @xmath99 .
the agreement is not as good as in fig .
[ fig : opt_poinc_comp]b , which is to be expected because the error vector @xmath46 used in the optimization was based exclusively on the geometry of the @xmath103 cross - section .
it is noteworthy , however , that for @xmath104 the optimization nonetheless predicts three islands near the edge of size and position comparable to what is observed .
note also the improvement in agreement for @xmath104 using @xmath99 in fig .
[ fig : opt_poinc_otherratios]b over fig .
[ fig : ideal_poinc_comp]b in which the numerical data were calculated from the nominal coil positions . and ( b ) @xmath104 , the coil displacements inferred for @xmath103 ( _ i.e. _ , @xmath99 ) .
measurements for the respective current ratios are shown for comparison.,scaledwidth=50.0% ] finally , calculated values of [ line - through]*@xmath2 * at various current - ratios @xmath3 using @xmath99 are shown in fig .
[ fig : iota_opt]b . note that , for @xmath16 , the profiles of [ line - through]*@xmath2 * contain the value 1/3 , which is consistent with the observations described in sec .
[ subsect : exp_results ] .
recall that the [ line - through]*@xmath2 * profiles that had been predicted for the nominal coil positions ( fig .
[ fig : iota_unp ] ) failed to account for the presence of [ line - through]*@xmath2 * = 1/3 for @xmath105 and that , furthermore , the pf coil displacements alone were insufficient to explain this disagreement ( fig .
[ fig : pfc_disp_compare ] ) .
hence , the optimization of the il coil displacements has resolved the discrepancy between the observations and predictions .
for the optimization conducted for the cnt il coils in this paper , the target parameters in @xmath39 came exclusively from a single current - ratio topology ( @xmath3 = 3.68 ) that had no significant magnetic islands . in principle , a cross - section containing magnetic islands should also be useable for an optimization if adequate experimental surface data are available , in the sense that many puncture points are distributed evenly around the cross - section of the surface .
such data were not available from cnt cross - sections featuring prominent islands , largely as a result of two factors .
the first was that cross - sections containing the large [ line - through]*@xmath2 * = 1/3 island chain were cut off at the inboard side due to the finite extent of the fluorescent rod .
on the other hand , had the rod extended further inboard , it would have collided with the il coils during rotation . thus , as seen in fig .
[ fig : opt_poinc_otherratios]b , there are no complete surfaces recorded outside the island chain .
this factor may be unique to cnt as it relates to the particular coil configuration .
the second factor , more likely to arise in other stellarators , is the shadowing of near - rational surfaces in the vicinity of island chains .
this is a result of the electron beam striking the back side of the gun before reaching the fluorescent rod .
this effect is visible in surfaces near the magnetic axis in fig .
[ fig : opt_poinc_comp]a - b , in which only three dots appear , corresponding to the first three toroidal transits of the electron beam . in this case , the shadowing was not a major concern because only a small portion of the cross - section was affected .
but in low - shear profiles with one or more rational surfaces at intermediate minor radii , large portions of the cross - section may be shadowed , thereby obscuring a large amount of geometric information and preventing the determination of accurate geometric parameters .
however , if the above factors are absent or limited in extent , we expect that cross - sections with island chains should be useable for optimization .
poincar data from field lines within a chain of large islands , identifiable by large empty intervals in the poloidal angle @xmath31 , are ignored in the parametrization procedure ( [ subsect : island - param ] ) .
thus , only the closed flux surfaces on either side of the island chain will determine @xmath37 , and , in turn , the netwon steps @xmath47 in an optimization .
although the information about the magnetic geometry within the islands themselves is lost , the kinking of nearby closed flux surfaces adjacent to the island chain is retained .
it is hypothesized that the closed flux surfaces with island - induced kinking should be sufficient for the optimization .
although the vector @xmath38 of coil parameters used for the optimization in this paper had only ten components , this vector can in principle be expanded to contain arbitrarily many parameters characterizing non - rigid displacements for multiple coils .
one logical extension for cnt would be to allow for elliptical compression of the il coils , which would add two more parameters for each coil ( one for ellipticity and one for phase ) .
more general deformations may be treated , for example , as fourier series in which each additional @xmath106 number would require the addition of two parameters ( coefficients ) if the coil is assumed to stay planar . with the addition of even more parameters , the coil could be allowed to deform off of its nominal plane .
it follows that , for an arbitrary stellarator , the size of @xmath38 would scale as the number of coils to be optimized times the number of degrees of freedom for rotation , translation and deformation afforded to each coil . of course , as seen in the foregoing study ,
each coil need not be given the same number of degrees of freedom .
assuming @xmath37 consists of geometric data from only one current - ratio ( which need not be the case ) , each column of the jacobian can be computed with a single call to the field line tracer .
different columns , however , corresponding to perturbations to different components of @xmath38 , require separate field line traces .
since the field line traces are the most demanding part of the optimization procedure , the computation will scale linearly with the size of @xmath38
. it should be noted , however , that since the columns of the jacobian may be determined in parallel with one another , the time need not scale linearly if multiple processors are available .
in summary , significant vacuum field errors have been observed in the cnt stellarator with the coils configured to have @xmath107 .
photogrammetric measurements of the pf coils showed misalignments , although those misalignments were not sufficient to fully explain the disagreements , in particular the observed offset in the [ line - through]*@xmath2 * profile . a numerical analysis of the influences of coil displacements on [ line - through]*@xmath2
* indicate that [ line - through]*@xmath2 * is most sensitive to displacements of the il coils .
this motivated the development of a numerical optimization method to calculate displacements for the il coils that , in combination with the measured pf coil displacements , fit to the observed poincar cross - sections .
the application of this algorithm to cnt at @xmath103 led to a set of il coil displacements that exhibited significantly improved quantitative and qualitative agreement with observations .
future work will focus on improving the optimization algorithm to achieve even better fits to the observed data .
these improvements will include ( 1 ) simultaneously optimizing to data from multiple current - ratios using @xmath46 vectors that include geometric parameters for multiple poincar cross - sections , ( 2 ) operating the field line tracer at greater numerical precision to reduce the uncertainty in numerically generated fourier coefficients and allow for smaller finite difference intervals for the computation of the jacobian , ( 3 ) expanding the @xmath38 vector to include other sources of error including coil deformations , uncompensated coil leads , and displacements of the pf coils , and ( 4 ) investigating ways of generalizing the @xmath37 vector to include information about poincar data inside islands , possibly analogous to generalizations of 3d toroidal equilibria found in codes like pies @xcite , siesta @xcite , and spec @xcite .
many of these improvements will be more demanding computationally , although much of the algorithm can be parallelized ( in particular , the calculation of the covariance and jacobian matrices ) .
while the misalignments in the cnt coils could in principle be corrected by , for example , repairing or replacing the support structures , this is not a high priority for the cnt program .
one reason for this is that , in spite of the field errors , we still have access to configurations with good sets of nested , closed flux surfaces such as the one attained for @xmath3 = 3.68 .
in addition , the large island chains that the field errors have given rise to at lower current - ratios present an opportunity for research in island divertor physics .
the authors would like to thank s. lazerson for assistance with the field line tracing code employed in this study , as well as r. diaz - pacheco and y. wei for their assistance with data collection .
the authors would also like to acknowledge the financial support of the department of energy and the national science foundation of the united states , grant no .
nsf - phy-04 - 49813 .
29 url # 1#1urlprefix[2][]#2 reimerdes h , buttery r j , garofalo a m , in y , haye r j l , lanctot m j , okabayashi m , park j k , schaffer m j , strait e j and volpe f a 2011 _ fusion science and technology _ * 59 * 572585 hender t c , wesley j c , bialek j , bondeson a , boozer a h , buttery r j , garofalo a , goodman t p , granetz r s , gribov y , gruber o , gryaznevich m , giruzzi g , gnther s , hayashi n , helander p , hegna c c , howell d f , humphreys d a , huysmans g t a , hyatt a w , isayama a , jardin s c , kawano y , kellman a , kessel c , koslowski h r , haye r j l , lazzaro e , liu y q , lukash v , manickam j , medvedev s , mertens v , mirnov s v , nakamura y , navratil g , okabayashi m , ozeki t , paccagnella r , pautasso g , porcelli f , pustovitov v d , riccardo v , sato m , sauter o , schaffer m j , shimada m , sonato p , strait e j , sugihara m , takechi m , turnbull a d , westerhof e , whyte d g , yoshino r , zohm h , the itpa mhd disruption and group m c t 2007 _ nuclear fusion _ * 47 * s128s202 mynick h e 2006 _ physics of plasmas _ * 13 * 058102 xanthopoulos p , mynick h e , helander p , turkin y , plunk g g , jenko f , grler t , told d , bird t and proll j h e 2014 _ physical review letters _ * 113 * 155001 sinclair r m , hosea j c and sheffield g v 1970 _ applied physics letters _ * 17 * 92 colchin r j , anderson f s b , england a c , gandy r f , harris h h , henderson m a , hillis d l , kindsfather r r , lee d k ,
million d l , murakami m , neilson g h , saltmarsh m j and simpson c m 1989 _ review of scientific instruments _ * 60 * 26802689 jaenicke r , ascasibar e , grigull p , lakicevic i , weller a , zippe m , hailer h and schwrer k 1993 _ nucl .
fusion _ * 33 * 687704 pedersen t s , kremer j p , lefrancois r g , marksteiner q , sarasola x and ahmad n 2006 _ phys
. plasmas _ * 13 * 012502 colchin r j , harris j h , anderson f s b , england a c , gandy r f , hanson j d , henderson m a , hillis d l , jernigan t c , lee d k , lynch v e , murakami m , neilson g h , rome j a , saltmarsh m j and simpson c m 1989 correction of field errors in the atf torsatron _ proceedings of the 16th european conference on controlled fusion and plasma physics , venice _ ( _ europhysics conference abstracts _
vol 13b ) ( iaea ) pp
615618 otte m , lingertat j and wagner f 2003 magnetic flux surface measurements with vertical field and compensation coils at the wega stellarator _ proceedings of the 14th international stellarator workshop , greifswald _ sakakibara s , narushima y , okamoto m , watanabe k y , suzuki y , ohdachi s , ida k , yoshinuma m , tanaka k , tokuzawa t , narihara k , yamada i , yamada h and the lhd experiment group 2013 _ nuclear fusion _ * 53 * 043010 hirsch m , baldzuhn j , beidler c , brakel r , burhenn r , dinklage a , ehmler h , endler m , erckmann v , feng y , geiger j , giannone l , geiger g , grigull p , hartfuss h j , hartmann d , jaenicke r , knig r , laqua h p , maassberg h , mccormick k , sardei f , speth e , stroth u , wagner f , weller a , werner a , wobig h , zoletnik s and the w7-as team 2008 _ plasma physics and controlled fusion _ * 50 * 053001 rummel t , risse k , kisslinger j , kppen m , fllenback f , neilson h , brown t and ramakrishnan s 2012 _ ieee transactions on applied superconductivity _ * 22 * 4201704 kisslinger j and andreeva t 2005 _ fusion engineering and design _ * 74 * 623626 landreman m and boozer a h 2016 _ phys . plasmas _ * 23 * 032506 pedersen t s , boozer a h , kremer j p , lefrancois r g , reiersen w t , dahlgren f d and pomphrey n 2004 _ fusion sci .
technol . _ * 46 * 200 imagawa s , masuzaki s , yanagi n , yamaguichi s , satow t , yamamoto j , motojima o and the lhd group 1998 _ fusion engineering and design _ * 41 * 253 wanner m and the w7-x team 2000 _ plasma physics and controlled fusion _ * 42 * 1179 lazerson s , otte m , bozhnekov s , biedermann c , pedersen t s and the w7-x team 2016 _ nuclear fusion ( submitted ) _
brenner p w , pedersen t s , berkery j w , marksteiner q r and hahn m s 2008 _ ieee transactions on plasma science _ * 36 * 1108 kremer j p 2006 _ the creation and first studies of electron plasmas in the columbia non - neutral torus _ ph.d .
thesis columbia university new york , ny 10027
lao l l , john h s , stambaugh r d , kellman a g and pfeiffer w 1985 _ nuclear fusion _ * 25 * 1611 hanson j d , hirshman s p , knowlton s f , lao l l , lazarus e a and shields j m 2009 _ nuclear fusion _ * 49 * 075031 press w h , teukolsky s a , vetterling w t and flannery b p 1992 _ numerical recipes in c _
2nd ed ( cambridge university press ) jones c s and finn j m 2006 _ nuclear fusion _ * 46 * 335 james g , witten d , hastie t and tibshirani r 2013 _ an introduction to statistical learning with applications in r _ ( springer ) rieman a and greenside h 1986 _ computer physics communications _ * 43 * 157 hirshman s p , sanchez r and cook c r 2011 _ physics of plasmas _ * 18 * 062504 hudson s r , dewar r l , dennis g , hole m j , mcgann m , von nessi g and lazerson s 2012 _ physics of plasmas _ * 19 * 112502
the level of agreement between two sets of poincar data is determined by fitting the nested flux surface cross - sections to a discrete set of geometric parameters @xmath37 .
the components of @xmath37 are the coefficients of a linear combination of orthogonal functions consisting of a fourier series in the poloidal angle @xmath31 and a polynomial series in the normalized minor radius @xmath108 . for a flux surface characterized by some particular @xmath108 , the @xmath109 and @xmath110 coordinates of the point with poloidal angle @xmath31
are expressed as @xmath111 @xmath112 @xmath113 @xmath114 as indicated in the above representation , the coefficients of each fourier mode are themselves functions of @xmath108 .
this dependence is represented as a linear combination of polynomials @xmath115 ; for example : @xmath116 the polynomials @xmath115 are chosen to be orthonormal in the inner product defined by @xmath117 and the first few polynomials in this set are listed in table [ tab : polynomials ] .
.the first few polynomials @xmath115 as discussed in the text .
[ cols="<,<",options="header " , ] the vector @xmath37 of geometric parameters , then , is just a list of these coefficients : @xmath118 @xmath119 to ensure that the set of coefficients is unique for a particular set of poincar data , it is necessary to use a precise definition for @xmath108 and @xmath31 . in this work ,
@xmath108 for a particular cross - section is defined as @xmath120 , where @xmath121 is the enclosed area and @xmath122 is a reference area not to exceed the area enclosed by the last closed flux surface . in this way , @xmath123 at the magnetic axis and @xmath124 on the edge of the region of the poincar cross - section to be parametrized . the poloidal angle @xmath31 is defined to advance in direct proportion to the arc length along a flux surface cross - section and is set to zero on the outboard side where @xmath110 is equal to @xmath125 .
an example of a fit of coefficients to experimental data is shown in fig . [
fig : data2params ] .
as discussed in section [ sect : flux_surf_meas ] , the experimental data were separated according to flux surface . for each flux surface ,
a curve was fit iteratively to the poincar points in such a way as to minimize the disagreement with the data points .
an example of a set of fit curves is shown in fig . [
fig : data2params]a .
when the fitting iterations are complete , the curve was parametrized in @xmath31 as described in [ sect : parametrization ] and the @xmath126 measured surface was assigned a @xmath127 value based on the enclosed area .
each fit curve @xmath128 was then projected onto a fourier series , yielding one set of coefficients @xmath129 for each value @xmath127 .
sets of corresponding fourier coefficients ( e.g. , @xmath130 ) were then fit by linear least - squares onto polynomials in @xmath108 of degree @xmath131 , from which the polynomial coefficients as shown in eq .
[ eqn : polycoeffs ] were obtained .
the calculations described in this paper used @xmath132 and @xmath133 .
because experimental data points are often limited near the magnetic axis , certain constraints are enforced in the least - squares fitting in order to avoid spurious oscillations in the @xmath108 polynomials . in particular , all polynomials for fourier coefficients of @xmath134 are constrained to be zero at the axis .
in addition , all coefficients of @xmath135 are constrained to have a first derivative of zero at the axis .
furthermore , coefficients with @xmath136 are only expanded to second order in @xmath108 ( _ i.e. _ , requiring @xmath137 ) .
once coefficients are calculated for a set of poincar data , they are arranged into a vector @xmath37 containing @xmath138 elements .
because of the constraints described in the above paragraph , @xmath139 of those components are redundant , and are therefore removed for subsequent analysis .
ideally , the coefficients @xmath37 should be unique for a given set of poincar data . however , in practice the coefficients will vary slightly depending on which poincar points are used for this fit .
specifically , for experimental data , this depends on which surfaces are measured and which of the subsequent pixels are used for the analysis . for numerical data , the dependence is on the locations where the field line traces are initialized .
hence , each component of @xmath37 will have an associated uncertainty that may be correlated with that of other components .
[ subsect : opt_proc ] and [ subsect : numerical ] describe how these correlations are accounted for in the optimization .
if a chain of large islands exists within a cross - section , it will be identified during the curve - fitting process by poloidal gaps in the poincar data .
since the parametrization described in [ sect : parametrization ] only permits closed flux surfaces , the islands are simply ignored during in the fits of fourier coefficients to the polynomials @xmath115 decribed above . since some fourier coeffients may change abruptly from the axis - facing side of the island to the edge - facing side , it may be necessary to incorporate higher - order polynomials @xmath115 in the fits , leading to more components in the @xmath37 vector . if one were to plot the curves of constant @xmath108 specified from the geometric parameters @xmath37 ( as in fig .
[ fig : data2params]b ) , one would observe a continuous deformation from the shape of the last closed flux surface on the axis - facing side of the island chain to the first closed surface on the edge - facing side .
these deformed curves in the island region clearly do not reflect the actual island geometry and are simply a result of interpolation between the data for the core - facing side and the edge - facing side .
in other words , the laminar flux surfaces occurring in place of the islands reflect the fact that the @xmath37 vector contains information only about the flux surfaces on either side of the island chain . | sources of error fields were indirectly inferred in a stellarator by reconciling computed and numerical flux surfaces .
sources considered so far include the displacements and tilts of the four circular coils featured in the simple cnt stellarator .
the flux surfaces were measured by means of an electron beam and fluorescent rod , and were computed by means of a biot - savart field - line tracing code . if the ideal coil locations and orientations are used in the computation , agreement with measurements is poor .
discrepancies are ascribed to errors in the positioning and orientation of the in - vessel interlocked coils . to that end , an iterative numerical method was developed .
a newton - raphson algorithm searches for the coils displacements and tilts that minimize the discrepancy between the measured and computed flux surfaces .
this method was verified by misplacing and tilting the coils in a numerical model of cnt , calculating the flux surfaces that they generated , and testing the algorithm s ability to deduce the coils displacements and tilts .
subsequently , the numerical method was applied to the experimental data , arriving at a set of coil displacements whose resulting field errors exhibited significantly improved agreement with the experimental results . |
dielectric spectroscopy has provided many informations to the investigation of complex materials , and especially to the field of glass transition phenomenon ( glass - forming systems ) during the last decades .
this technique gives direct access to the polarization of molecular dipoles allowing the study of relaxation processes .
it can probe the molecular structure and dynamics of several materials like liquids , polymer composites , colloids , porous materials , or ferroelectric crystals @xcite .
it covers a wide range of frequency , typically from @xmath0 to @xmath2 hz , going up to @xmath3 hz with coaxial techniques @xcite .
dielectric liquids of small molecules like the glass - forming glycerol @xcite or glass - forming polymers @xcite have been extensively studied .
dielectric spectroscopy gives direct access to relaxation times@xcite like rheology , but it is a rotational relaxation one , not a translational .
nevertheless , it has been showed that on molecular liquids at equilibrium , these two relaxation times have the same temperature dependence@xcite , at least above the glass transition .
recent progress in the glass transition were provided by direct measurement of a dynamical growing length scale@xcite .
this technique has always been improved , e.g. , the important work done regarding the electrodes in order to measure charged liquids even at low frequency @xcite .
dielectric spectroscopy has also been coupled to scattering techniques such as x - rays @xcite or neutrons @xcite .
furthermore , recent theories on aging and out of equilibrium physics predict interesting low frequency properties of the response and fluctuations , either during the relaxation of the system towards equilibrium after a quench of the control parameter @xcite , or when the system is driven by external fields @xcite . in order to test these models
it is important to measure the response of the system as a function of time after the beginning of the time dependent phenomenon@xcite .
single frequency measurements will force to repeat the experiment many times for each selected frequency making the experiment extremely long and difficult . in those cases multi - frequency measurements are absolutely necessary .
the standard technique to perform multi - frequency response functions is to drive the system with a white noise signal .
the main disadvantage of this approach is to require long averages before reaching reliable response measurements ( see section [ procedure ] ) .
indeed , the lower frequencies need long time sampling . to overcome this problem we present an innovative setup that allows simultaneous impedance measurements at several frequencies over a typical range of @xmath4 hz with low noise and high sensibility .
this has been obtained with a careful choice of the input signal , of the current amplifier , and an appropriate correction , which allow us to measure high capacitances with very small losses ( > @xmath5 is the typical order of magnitude ) .
this means that our system is particularly useful for the study of relaxation processes in polymer films which have very high resistivity .
the length of the measurement time window is imposed only by the smallest frequency that one needs to resolve because all other frequencies are measured simultaneously .
the paper is organized as follows . in section [ bkg ]
we give some general , basic background on the dielectric spectroscopy .
section [ principle ] describes the experimental principle , [ setup ] the experimental setup , and [ procedure ] the measurement procedure .
application of the technique to the study of some polymer films is presented in section [ samples ] .
we conclude in section [ sumcon ] .
dielectric spectroscopy investigates the frequency dependent dielectric properties of a material @xcite .
it is based on the interaction of an external field with the electric dipole moment of the sample , often expressed by the complex dielectric permittivity or dielectric constant : @xmath6 where @xmath7 is the dielectric displacement and @xmath8 the electric field , both dependent on the angular frequency @xmath9 .
@xmath10 can be expressed as @xmath11 being @xmath12 the stored permittivity , which is related to the stored energy by the sample , and @xmath13 the loss permittivity , which indicates the loss energy .
their loss tangent is @xmath14 where @xmath15 is the phase shift between @xmath7 and @xmath8 .
the dielectric constant @xmath16 depends on the dielectric material , the frequency , the temperature , age .
the orientation polarization of dipoles due to applied disturbances allows the investigation of the dielectric relaxation of the material in the frequency range below @xmath17hz .
moreover , if those dipoles are of molecular origin , the dielectric spectroscopy allows to investigate molecular motions .
the experimental techniques and procedures for dielectric spectroscopy depend on the frequency range of interest . at low and intermediate frequencies ( @xmath18 hz )
the sample can be modeled as a parallel plate capacitor where the sample is the dielectric insulator in between the two electrodes @xcite . assuming no edge capacitances
, the capacitance of an ideal parallel plate capacitor reads @xmath19 where @xmath20 is the distance between electrodes , @xmath21 the surface occupied by the sample and @xmath22 the vacuum permittivity . using eq .
[ eq - permittivity2 ] in eq .
[ eq - permittivity1 ] , we see that the impedance @xmath23 of the sample is modeled by a resistor @xmath24 in parallel with a capacitor @xmath25 and it can be expressed as : @xmath26 with @xmath27 and @xmath28
. the complex impedance @xmath23 can be measured by imposing the voltage @xmath29 and measuring the current @xmath30 flowing through : @xmath31 as @xmath32 is low on pure dielectric systems this means that the resistive part of the impedance is quite high and the in - phase current becomes hardly detectable .
so far we described the ideal capacitor , but in the real experiments there are other parameters to take into account .
on the one hand , corrections due to the measurement system like the influence of the extra capacitance in the cables . on the other hand , sample preparation must be very careful to avoid difficulties , such as sample geometry , bad electric contacts between electrodes and sample , edge capacitances , or electrode polarization .
sample geometry has to be well known .
we do not discuss these standard experimental details but we focus on the description of the features of the electronics , the driving signal , and the data analysis that we use in our experimental setup . our goal is to measure @xmath10 simultaneously in a frequency range covering 4 orders of magnitude in @xmath9 . as dielectric losses are in general small in pure dielectric ( such as polymers ) , any spurious effect must be taken into account in order to have a good accuracy on the wide frequency range . to define notation and to clarify the principle of our technique
let us recall basic electronic concepts .
in order to measure @xmath33 we use a current amplifier whose entrance impedance is just the sample impedance @xmath23 as sketched in fig .
[ fig - circuit ] . in order to analyze the drawbacks of this scheme
let scheme of the working principle of the main amplifier of our device . ] us assume first that the amplifier u in fig .
[ fig - circuit ] is an ideal one .
thus , we suppose that the bias currents at both entrances are zero ( @xmath34 ) , and that the open - loop amplification of u is infinite for all frequencies .
this implies that @xmath35 because the positive entrance is at the ground potential .
therefore , the impedance @xmath36 ( given by the parallel of the capacitances of the cables and of the amplifier entrance ) does not play any role in this case . from the standard circuit analysis ,
the current balance between the current flowing in the feedback impedance @xmath37 ( @xmath38 ) and the entrance current we obtain @xmath39 where @xmath40 is the reference and @xmath41 the output potentials ( see fig .
[ fig - circuit ] ) .
we have omitted the obvious dependence on @xmath9 .
thus , the impedance @xmath42 can be calculated as @xmath43 nevertheless , considering a real amplifier , ( i.e. with a finite open loop gain ) one has to take into account in the current balance ( fig .
[ fig - circuit ] ) that @xmath44 .
thus , @xmath45 can not be neglected and the current balance is : @xmath46 where we have considered that the random output current @xmath47 takes generically into account all the sources of noise produced by the entrance bias currents , the voltage noise , and the component thermal noise . in order to get @xmath42 we first multiply eq .
[ eq - real ] by the conjugate complex @xmath48 of @xmath49 and we take an average over several realizations : @xmath50 where we took into account that @xmath47 is uncorrelated with @xmath40 thus @xmath51 . we consider the transfer functions of both the signal and the correction : @xmath52 @xmath53 using eq .
[ eq - real_2 ] , and the definitions of @xmath54 and @xmath55 , the impedance of the sample can be calculated as @xmath56 the knowledge of @xmath57 is very important in order to have reliable estimations of the dielectric constant at relatively high frequencies when the amplifier open loop gain begin to decrease and the phase shifts are not negligible with respect to the one that we want to measure . in our experimental setup
we actually measure @xmath57 as we describe in the next section .
the real scheme of our device is presented in fig .
[ circuitall ] .
the characteristics of each component are presented in table [ table1 ] .
the source signal @xmath58 is sent to a first operational amplifier @xmath59 .
this amplifier is a voltage follower or a non - inverting buffer .
this leads to decouple the impedance of the source and the sample with a negligible additional noise .
for that , we choose the lme 49990 amplifier , which has very low voltage noise .
the output of @xmath59 is sent to the sample via @xmath60 which is the equivalent to @xmath61 in fig .
[ fig - circuit ] .
note that @xmath62 is very useful in order to avoid self oscillations of the operational amplifier @xmath59 when large @xmath63 are used ( see appendix[app ] ) .
the signal sent to the sample is @xmath40 as defined in the previous section . .[table1 ] technical details of the components of the circuit of fig .
[ circuitall ] .
u are the different amplifiers of types lme 49990 ( low voltage noise , 1.4 nv/@xmath64 at 10 hz ) or ad549 ( low current noise , 0.5@xmath65a@xmath66 , @xmath67 to 10 hz , bias current 30 fa ) ; r are resistances . [ cols="^,^,^,^ " , ] scheme of our real circuit .
@xmath59 , @xmath68 , and @xmath69 are low noise voltage amplifiers . @xmath70 and @xmath71 are low noise current amplifiers .
details of the components are on table [ table1 ] . ]
@xmath40 is also sent to a secondary buffer amplifier @xmath68 ( lme 49990 ) , and the output of this amplifier is used as the reference signal @xmath72 ( assuming @xmath73 because the phase - shift of @xmath68 is negligible in our frequency range ) .
the current flowing through the sample can be very small if the sample impedance is very high so this part of the circuit uses amplifiers with a low current noise and bias current ( ad549 ) .
there are two acquisition paths : in the main one , the voltage from the sample goes through a first amplifier @xmath70 with @xmath74 .
this amplifier is the main one of the circuit and is equivalent to the one modeled in fig .
[ fig - circuit ] and by eq .
[ eq - real ] .
it is followed by @xmath69 ( lme 49990 ) which amplifies the voltage by a factor 4 taken into account in data treatment .
the second path is the measurement of the negative entrance potential of amplifier @xmath75 , that is to say @xmath57 in eq .
[ eq - real ] .
this signal also goes through a buffer amplifier @xmath71 and is amplified 4 times by @xmath76 , getting @xmath77 ( i.e. @xmath78 ) . in order to accurately measure impedances
, we have to take into account all the contributions from the different components in the electronic device .
the impedance of the cables @xmath45 between the sample and the amplifiers has to be considered @xmath79 where @xmath80 f is the capacitance of 0.5 m long cables we used .
the entrance capacitance @xmath81 f of @xmath75 can be neglected .
we made careful choice of good quality cables between the sample and the amplifier other cables have less sensitive contribution to the noise in the circuit in order to reduce noise in the signal .
we also have to take into account the feedback impedance of the amplifiers @xmath82 where @xmath83 is the feedback resistance and @xmath84 f its parallel capacitance ( fig .
[ fig - circuit ] ) .
this capacitance is negligible compared to the impedance of @xmath85 in our frequency range .
the signals @xmath72 , @xmath77 , and @xmath86 are sent to a national instrument pxi-4472 acquisition card ( 24 bits resolution ) .
data acquisition is performed via a labview home - made program and data treatment using matlab to compute the transfer functions @xmath87 ( eq . [ eq - transferinput ] ) and @xmath88 ( eq . [ eq - transfercorr ] ) using standard fft algorithm .
the transfer functions of the three amplifiers composed by @xmath68 , @xmath69 , and ( @xmath68+@xmath76 ) are measured in order to correct the phase of the three measured signals .
in order to perform a simultaneous and precise measurement of @xmath42 in a wide frequency range using the device described in the previous section we have optimized the frequency dependence of the signal @xmath40 . indeed the standard technique employed to perform a response measurement on a wide band is to excite the system with an almost white noise in the band of interest and to measure simultaneously the input and output signals .
then , by taking fft of these two signals input and output , one can compute the transfer function @xmath54 by a very long average of the response function on many configurations of the noise .
this system although very practical and adaptable on almost all cases presents several drawbacks in the case of a circuit like the one presented in the previous sections .
the first is that in the circuit of fig .
[ circuitall ] the high frequencies are amplified much more than the low ones . therefore , one has to compensate for this effect .
the second drawback is using a random noise at the input , its phase is rapidly changing and many averages are required in order to smooth this phase noise and to have a reliable measurement . in order to avoid these difficulties and reduce the number of averages in our system we apply the following method .
we first fix a measurement interval @xmath89 and a sampling frequency @xmath90 , where @xmath91 is an integer .
then , we generate a signal composed by a sum of sinusoids whose pulsations @xmath92 are harmonics of the slowest frequency @xmath93 , specifically @xmath94 where @xmath95 are the amplitudes in volts .
the amplitudes are chosen to be @xmath96 because favoring low frequencies allows us to compensate the increasing gain of the amplifier at high frequencies .
these kinds of signal are numerically generated and recorded in the computer memory which is then sampled by the same clock at the sampling frequency @xmath97 , which is also used for sampling the acquired signals @xmath40 , @xmath41 , and @xmath57
. in such a way all phase noises are suppressed as each frequency contains a fix integer number of acquired samples .
we use two different devices to generate the signal either a national instrument pxi 5411-card generator or an agilent 33500b series waveform generator . as an example we plot in fig .
[ fig - psdfreq ] the spectra of @xmath72 , @xmath86 , and @xmath77 signals obtained using @xmath98 f. in this specific example we select twenty - five frequencies @xmath99 in a logarithmic scale between a minimum value of @xmath100 hz and @xmath101 hz .
the sampling rate is @xmath102 sample / s .
the peak of the noise at about 3 khz is explained in the appendix[app ] .
notice that the amplitude of the correction signal is not negligible above @xmath103 hz ( fig .
[ fig - psdfreq]c ) and it is important to keep it into account in the calculation of @xmath42 . to summarize , this electronics combined with the described data analysis can measure with a good accuracy @xmath104 and @xmath105 f. , ( b ) @xmath86 , and ( c ) @xmath77 signals .
circles point to the chosen frequencies . ]
pf(@xmath106 ) , @xmath107 nf(@xmath108 ) and @xmath109 nf ( @xmath110 ) are plotted .
the losses of the @xmath111 pf(@xmath106 with error bars ) and @xmath107nf@xmath112 capacitances are plotted in ( b ) , whereas those of the @xmath109 nf in fig.[fig - capatest]b .
the capacitance of @xmath111 pf has very small losses and this allows us to fix the limits and the losses of our apparatus ( blue line with error bars in b ) . in ( b )
we also plot the value of two pure resistances of @xmath113 m@xmath114 ( @xmath115)and @xmath116 g@xmath117 .
these two resistances have of course a small parallel capacitance of about @xmath118 pf shown in ( a ) @xmath119 and @xmath120 , which also fixes the accuracy of our system . in ( b )
we also plot the measured resistance ( @xmath121 ) when @xmath42 is the parallel of the @xmath107nf capacitance with the @xmath113 m@xmath114 resistance .
notice that for each @xmath42 the whole frequency interval is covered by only two simultaneous measurements in two overlapping four order of magnitude intervals : [ @xmath122 hz ] and [ @xmath123 hz ] ] nf capacitance of fig.[fig_all_z]a .
its losses are plotted in ( b@xmath110 ) together with the pure resistance of @xmath116 g@xmath117 .
we also plot the measured total resistance @xmath106 of @xmath42 composed by the parallel of the @xmath109 nf with the @xmath116 g@xmath114 resistance .
the dashed line represent the results of the measurement without using the correction signal @xmath57 in the calculation .
the whole measurement in the range 0.1 hz-1000 hz takes less than 60 s ( i.e. 6 periods of the smallest frequency ) . ] from the measured impedance ( eq . [ eq - impedance ] ) we can calculate the resistance and the capacitance of the dielectric as @xmath124^{-1 } \label{eq - resistance}\ ] ] @xmath125 in order to check the efficiency of our device in the chosen frequency window , we use several impedances @xmath42 whose values have been checked by standard apparatuses @xcite . in fig.[fig_all_z]a
we plot the measured values of the capacitances and in fig.[fig_all_z ] b ) those of the measured resistances .
this plot illustrates the whole dynamics of our system ranging from @xmath107pf ( at @xmath118pf the error is large but the measurement can be still used ) to @xmath109 nf and larger if a smaller @xmath126 is used @xcite .
the @xmath111 pf capacitance is an air capacitance with extremely small losses and this allows us to determine the losses of our circuit as a function of frequency which are plotted in fig.[fig_all_z]b . in order to appreciate more precisely the accuracy we plot in fig.[fig - capatest]a ) the measured values of the @xmath109 nf capacitance and fig.[fig - capatest]b its losses in the frequency range @xmath127 .
we see that the fluctuations of the measurement are less than @xmath128 in all the frequency range . in fig.[fig - capatest]b
we compare the losses of the @xmath109 nf capacitance with a measured resistance @xmath129 g@xmath114 .
the capacitance @xmath63 has an almost constant value , whereas its losses increase , which correspond to a decrease of @xmath24 as a function of @xmath9 ( see eq.[eq - losses ] and fig.[fig - capatest]b ) .
when @xmath63 is mounted in parallel to @xmath130 then the capacitance @xmath63 is unchanged whereas the measured resistance is now the parallel of @xmath131 which is constant at low frequencies up to around 10 hz , and then , decreases as a function of the increasing frequency due to the capacitive losses @xcite .
it is important to stress that , using our method , we need to average less than @xmath132 s , in order to measure simultaneously the values of @xmath133 and @xmath63 on 4 orders of magnitude in frequency , plotted in fig .
[ fig - capatest ] .
as already mentioned , standard techniques use a white noise in the bandwidth of interest . by applying such a signal to our circuit ,
the result is much more noisy ( even impossible to find correct values ) than that obtained using our specifically chosen input signal described in the previous section ( see eq.[eq - sin ] ) . in fig .
[ fig - capatest ] we can also see the importance of the correction signal @xmath57 ( see fig . [ fig - psdfreq]c ) in order to properly calculate the values of @xmath133 and @xmath63 in the whole chosen bandwidth .
finally we stress that because of the limited memory of our voltage generator we always measure simultaneously at most 4 order of magnitude in frequency in the range @xmath134 .
furthermore the accuracy is not constant as a function of frequency .
it is @xmath135 below @xmath118 hz and above @xmath107 khz about @xmath136 in the range @xmath127
in the case of a pure capacitance , from the values of @xmath133 and @xmath63 we can calculate both @xmath137 and @xmath32 : @xmath138 @xmath139 with @xmath140 .
therefore , @xmath141 ( eq . [ eq - tandelta ] ) can be calculated as @xmath142 which is independent of the geometry of the sample .
we check our experimental device with well - known polymer films .
as already mentioned , our experimental device allows to perform simultaneous measurement of impedance at several frequencies in a range from @xmath0 to @xmath101 hz with low noise and high sensitivity .
first of all , we investigate an extrusion film makrofol^^ de 1 - 1 000000 ( from bayer ) based on makrolon^^ polycarbonate ( pc ) of 125 @xmath143 m of thickness and with a glass transition temperature ( @xmath144 ) of about 150 @xmath145c . the sample consist of a pc film of 18 cm of diameter .
we measure the impedance and compute @xmath63 and @xmath133 from 22 to 167 @xmath145c . fig .
[ fig - pc ] shows @xmath133 and @xmath63 at 22 @xmath145c .
@xmath133 presents huge values at low frequencies decreasing with frequency .
@xmath63 presents a small dependence in frequency . from @xmath63
we calculate the dielectric constant @xmath16 ( eq . [ eq - permittivity1 ] ) to be similar to the value found in the literature@xcite ( @xmath146 ) .
@xmath147 ( eq . [ eq - tandelta2 ] ) as a function of temperature is presented in fig .
[ fig - pctandelta ] .
a peak corresponding to the segmental relaxation of pc ( glass transition ) appears from low to high frequencies when increasing temperature ( well - known in polymers , see e.g. ref . ) .
the advantage here is that one can do easily multi - frequency dynamical measurements changing the rate of the temperature ramp .
m of thickness and @xmath148 cm of diameter as a function of frequency at room temperature . ] of a pc film of 125 @xmath143 m of thickness and 18 cm of diameter as a function of frequency at different temperatures . ]
c at 1 , 11 , and 91 hz .
both magnitudes are normalized to their values just after the quench ( time @xmath149 ) : @xmath150 ( 1 hz ) , @xmath151 ( 11 hz ) , and @xmath152 ( 91 hz ) for the resistance ; @xmath153 f ( 1 hz ) , @xmath154 f ( 11 hz ) , and @xmath155 f ( 91 hz ) for the capacitance . ]
then we investigate a film of poly(vinyl - acetate ) ( pvac ) in order to study the aging phenomenon .
the pvac ( @xmath156c ) is cycled between 19 and 53@xmath145c , with a very fast thermal quench ( around 30@xmath145c / min ) .
the electrodes are made out of thin films of aluminum ( 150 nm ) deposited by evaporation over a sapphire disc .
such crystal plate is chosen for its high thermal conductivity because heating and cooling are done through the substrate .
the polymer is dissolved in chloroform , then poured over the lower electrode .
after careful drying of the solvent , the second aluminum thin film is deposited over the dielectric .
connexions are fixed on the electrodes thanks to conductive glue .
the thickness of the dielectric is around @xmath157 , such that the overall capacitance is a few nf in order to match the amplifier s impedance .
[ fig - pvacr ] shows resistance and capacitance of pvac above and below @xmath144 as a function of frequency .
the @xmath158-relaxation appears in the measurement range around 53@xmath145c but not at lower temperature ( 19@xmath145c ) .
[ fig - pvacr2 ] shows the time evolution of @xmath133 and @xmath63 after a fast thermal quench from 53@xmath145c to 19@xmath145c at 1 , 11 , and 91 hz .
the relaxation of @xmath133 and @xmath63 are due to the aging of the polymer below the glass transition going toward equilibrium .
this technique allows to work on a time range limited at small times by the speed of the quench ( typically one minute ) .
our device is absolutely necessary in this kind of dynamical studies .
we present here a technique for impedance measurements , specially designed for highly resistive samples and time dependent phenomena .
indeed it allows us to measure the complex dielectric constant simultaneously in a wide frequency range .
the electronics combined with the described data analysis allow to measure with a good accuracy @xmath104 and @xmath105 f. this technique is based on two main elements .
first of all , a home - made amplifier specially built to measure small phase shifts in the whole frequency range of interest .
the proposed method based on a correction signal is very efficient for obtaining accurate values of both on the real and the imaginary part of the impedance , even in the cases of very small phase shift .
secondly , the reference signal is carefully chosen to give optimal accuracy over the whole frequency range , which is typically @xmath4 hz .
we use a sum of sinus with decreasing intensity at high frequency in order to keep a constant signal - to - noise ratio with frequency .
the chosen frequencies of the input signal are exact subharmonic of the sampling frequency , which is used to generate the input signal and to sample the outputs . compared to the use of a white noise
, this signal has well - defined phases , ensuring the good phase accuracy we need .
furthermore , measuring known frequencies allows shorter acquisition times than using white noise in order to get high precision on the impedances .
then , this device is particularly useful for time evolving impedances studies , especially in the field of aging of glass - forming polymers .
four orders of magnitude in frequency , chosen in the range of @xmath4 hz , can be simultaneously investigated over typical time range from one minute to arbitrary long time .
we also present here few results on two polymer samples : a pc film evolving to the glass transition with temperature , and a pvac sample evolving with time after a fast quench .
this last result demonstrates the rich field of non - equilibrium physics that can be probed using this device .
r. prez - aparicio acknowledges the funding by solvay .
caroline crauste - thibierge and sergio ciliberto thank the founding by erc grant 267687 outeflucop .
we also thank debjani bagchi who made use of the setup presented here .
in this appendix we give an explanation of the noise peak in the output spectrum of fig .
[ fig - psdfreq ] .
the amplification of the ad549 is very well approximated at low frequencies by @xmath159 with @xmath160 and @xmath161 .
the analysis of the circuit of fig .
[ circuitall ] shows that the spectrum of the output signal @xmath162 is related to that at the input @xmath163 : @xmath164 where @xmath165 ^ 2 \omega^2 \label{eq_denominator } \end{split}\ ] ] here we have used the approximation @xmath166 and @xmath167 ( see fig . [ fig - circuit ] ) .
these are very good for the purpose of understanding the maximum of the noise .
this spectrum has a resonance at @xmath168}$ ] which is rather broad because of the strong damping term @xmath169 . inserting the numerical value
we see that the resonance is at about @xmath170 and the broadening a few hz . these values are in agreement with the enhancement of the noise in the power spectrum of fig .
[ fig - psdfreq]b , if we consider that the noise of the input signal , fig .
[ fig - psdfreq]a , has a wide bandwidth with a slow decrease as function of frequency . in fig .
[ fig - simul ] we plot an output spectrum computed from eq .
[ eq_power_spectrum ] using a realistic input spectrum , which is plotted in the same figure .
we see that the output spectrum is very close to the one measured with @xmath171f .
22ifxundefined [ 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 10.1080/001075100181259 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.77.318 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.1470235 [ * * , ( ) ] link:\doibase 10.1021/ma992039y [ * * , ( ) ] \doibase doi : 10.1016/0032 - 3861(67)90021 - 3 [ * * , ( ) ] link:\doibase 10.1021/jp9640989 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.104.165703 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.3678324 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.4746992 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.1150528 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.1876992 [ * * , ( ) ] \doibase http://dx.doi.org/10.1103/physreve.55.3898 [ * * , ( ) ] http://iopscience.iop.org/0295-5075/46/5/637 [ * * , ( ) ] \doibase http://dx.doi.org/10.1103/physreve.63.012503 [ * * , ( ) ] \doibase http://dx.doi.org/10.1103/physrevlett.103.040601 [ * * , ( ) ] http://iopscience.iop.org/0295-5075/53/4/511 [ * * , ( ) ] \doibase http://dx.doi.org/10.1103/physrevlett.88.257202 [ * * , ( ) ] http://iopscience.iop.org/0295-5075/63/4/603 [ * * , ( ) ] @noop _ _ ( , , ) | we present an innovative technique which allows the simultaneous measurement of the dielectric constant of a material at many frequencies , spanning a four orders of magnitude range chosen between @xmath0 hz and @xmath1 hz .
the sensitivity and accuracy are comparable to those obtained using standard single frequency techniques .
the technique is based on three new and simple features : a ) the precise real time correction of the amplification of a current amplifier ; b ) the specific shape of the excitation signal and its frequency spectrum ; and c ) the precise synchronization between the generation of the excitation signal and the acquisition of the dielectric response signal .
this technique is useful in the case of relatively fast dynamical measurements when the knowledge of the time evolution of the dielectric constant is needed . |
with one exception ( konacki et al .
2003 ) , all confirmed extrasolar planets have been discovered by the doppler velocity technique . the selection effects inherent to radial velocity surveys ( cumming ,
marcy & butler 1999 ) favor the detection of planets at small orbital radii . to date , about half of the known planets have semi - major axis @xmath1 au , while the most distant - 55 cnc d - lies at @xmath2 au from its parent star .. indirect evidence , however , suggests that there could be a sizable population of massive planets at much greater radii .
recent observations of dusty debris around vega have been interpreted as suggesting the presence of a planet of a few jupiter masses with @xmath3 ( wilner et al .
further , simulations modeling circumstellar dust discs suggest a planet lies at a distance of @xmath4 au from epsilon eridani ( ozernoy et al .
2000 ) .
forming planets in these outer locations is difficult .
gas giants must form before the disc dissipates , at timescales no greater than @xmath5 myr ( haisch , lada , & lada 2001 ) . in standard core accretion models ( safranov 1969 ) , the timescale for building the core of a giant planet increases rapidly with radius , with a @xmath6 scaling approximately as @xmath7 ( pollack et al .
although such models are undoubtedly oversimplified ( pollack et al .
1996 ; bryden , lin & ida 2000 ) , it is hard to avoid the conclusion that forming massive planets at radii of several tens of au within @xmath8 myr is difficult .
indeed , this has led to the suggestion that uranus and neptune may have formed at smaller radii in our own solar system ( thommes , duncan & levison 1999 , 2002 ) .
motivated by these issues , we investigate the possibility of forming massive planets at small @xmath9 , followed by outward migration . in sections 2 and 3 ,
we consider sequentially the two mechanisms that have been extensively studied in the context of inward migration : planet - disc interactions ( goldreich & tremaine 1980 ; lin & papaloizou 1986 ; lin , bodenheimer & richardson 1996 ; trilling et al . 1998 ) and
gravitational scattering after disc dissipation ( rasio & ford 1996 ; weidenschilling & marzari 1996 ; lin & ida 1997 ; ford , havlickova & rasio 2001 ; terquem & papaloizou 2002 ) .
our conclusions are briefly summarized in section 4 .
we calculate the orbital evolution of massive planets embedded within an evolving protoplanetary disc using a variant of the approach described by armitage et al .
we use a simple , one - dimensional ( i.e. vertically averaged ) treatment to model the evolution of a protoplanetary disc evolving under the action of both internal viscous torques and external torques from one or more embedded planets ( goldreich & tremaine 1980 ; lin & papaloizou 1986 ; trilling et al . 1998 ; trilling , lunine & benz 2002 ) . for a disc with surface density @xmath10 ,
the governing equation is , @xmath11 + \dot{\sigma}_{w}. \label{eqsigma}\ ] ] here , @xmath12 is the kinematic viscosity which models angular momentum transport within the disc gas , and @xmath13 is a term which allows for mass to be lost from the disc for example as a consequence of photoevaporation . the second term within the brackets describes how the disc responds to the planetary torque , @xmath14 , where this function is the rate of angular momentum transfer per unit mass from the planet to the disc . for a planet in a circular orbit at radius @xmath9 , we take , @xmath15 where @xmath16 , the mass ratio between the planet and the star , @xmath17 and @xmath18 is the scale height of the disc . guided by detailed protoplanetary disc models ( bell et al .
1997 ) , we adopt @xmath19 .
the transfer of angular momentum leads to orbital migration of the planet at a rate , @xmath20 if the only torque on the planet comes from the gravitational interaction with the disc . in the core accretion model for giant planet formation ( e.g. pollack et al .
1996 ) , the accretion of the gaseous envelope is predicted to take longer than any other phase of the formation process ( several myr in the baseline model for jupiter presented by pollack et al .
in particular , the time scale for accretion is much longer than the time scale on which a sufficiently massive planet can establish a gap in the protoplanetary disc , since numerical simulations show that an approximately steady - state gap can be set up by a massive planet within @xmath21 orbital periods ( e.g. lubow , seibert & artymowicz 1999 ) .
a consequence of the inequality of these time scales is that massive planets those of several jupiter masses probably accrete most of their envelopes _ subsequent _ to the development of a gap in the protoplanetary disc .
numerical simulations show how this accretion may occur .
gas from the outer disc penetrates the leaky tidal barrier created by the planet , and flows inward to form a small circumplanetary disc around the growing planet ( lubow , seibert & artymowicz 1999 ; dangelo , henning & kley 2002 ; bate et al .
2003 ) .
the existence of mass flow across gaps onto planets is intrinsically a two ( or three ) dimensional phenomenon ( e.g. the discussion in artymowicz & lubow 1996 ) .
the torque function ( eq . [ eqtorque ] ) used in our one dimensional code establishes a clean gap for all planet masses above about @xmath22 , and this gap precludes any mass flow across the gap , or onto the planet . to allow for the mass growth of planets , we have therefore modified the one dimensional treatment to explicitly include mass flow from the outer disc on to the planet .
we begin by making an approximate fit to the results of two - dimensional numerical simulations ( lubow et al . 1999 ; dangelo et al . 2002 ) .
we define the _ efficiency _ of mass accretion across the gap via a parameter @xmath23 , which is the planetary accretion rate as a fraction of the disc accretion rate at large radii ( away from the location of the planet ) .
the results of the aforementioned numerical simulations can then be approximated by the formula , @xmath24 where @xmath25 is the mass of jupiter and @xmath26 is an adjustable parameter which can be used to test how the results depend upon the overall efficiency of planetary accretion .
we use the above equation to calculate at each timestep the appropriate planetary accretion rate .
we then remove the required amount of mass from the first zone on the outer edge of the gap , and add it to the mass of the planet .
note that we assume that all the mass flow onto the planet originates from the outer disc , and do not permit any material to ` bypass ' the planet and flow directly from the outer disc to the inner disc .
mass accretion across the gap onto a planet may also be expected to lead to accretion of angular momentum , given that the specific angular momentum of gas at the outer gap edge exceeds that of the planet .
this effect which it is easy to show can have a significant influence on the migration rate can not be straightforwardly measured from existing numerical simulations . for this paper ,
we adopt the simplest approach , and assume that the accreted gas has the same specific angular momentum as gas at a radius @xmath27 .
this fixed radius approximates the location of the outer edge of the gap throughout most of the calculation .
( [ eqsigma ] ) is solved on a fixed , non - uniform mesh using standard explicit numerical methods ( e.g. pringle , verbunt & wade 1986 ) .
the mesh is uniform in a scaled variable @xmath28 .
typically , 300 grid points are used , with an inner boundary at 0.1 au and an outer boundary at 200 au . a zero - torque ( @xmath29 ) boundary condition is applied at @xmath30 .
for the protoplanetary disc model adopted , the outer boundary is at sufficiently large radius that the choice of boundary condition there has no influence on the results .
we model the protoplanetary disc as a viscous accretion disc ( lynden - bell & pringle 1974 ) suffering mass loss at large radius as a consequence of photoevaporation ( e.g. shu , johnstone & hollenbach 1993 ) . motivation for considering models of this form is provided first and foremost by observations of photoevaporative flows in orion ( johnstone , hollenbach & bally 1998 ) , and is discussed further by clarke , gendrin & sotomayor ( 2001 ) , matsuyama , johnstone & hartmann ( 2003 ) , and armitage , clarke & palla ( 2003 ) .
briefly , we write the viscosity as a fixed ( in time ) power - law in radius . for most of the calculations
we have adopted the form , @xmath31 this yields a steady - state surface density profile @xmath32 , which is similar to that derived from more detailed protoplanetary disc models over the radii of interest ( bell et al . 1997 ) . to test
how sensitive the conclusions are to this assumption , we have also run one model with a different form for the viscosity , @xmath33 mass loss from the disc scales with radius as , @xmath34 with @xmath35 .
we express the normalization of the mass loss via a parameter @xmath36 , which is defined as the total mass loss from the disc for a disc with an outer edge at 25 au .
the instantaneous rate of mass loss will therefore differ from this value depending upon the extent of the disc .
the initial surface density profile for the @xmath37 disc model is , @xmath38 while the @xmath39 model is identical except for the replacement of @xmath40 by @xmath41 . here
, @xmath42 is a constant used to define the initial accretion rate , while @xmath43 is a truncation radius which sets a smooth exponential cut off to the surface density at large radius . for @xmath44 , a disc described by this initial condition
has a constant accretion rate , so we specify the initial surface density of our models via this inner accretion rate @xmath45 .
[ fig_gas1 ] shows how the evolution of the accretion rate and disc mass varies with the strength of photoevaporative mass loss .
we have computed models with our standard viscosity ( @xmath37 ) that have @xmath46 , @xmath47 , and @xmath48 .
for consistency with observational determinations of protoplanetary disc parameters in nearby star - forming regions ( e.g. gullbring et al .
1998 ) , we adopt for all of these models an initial accretion rate of @xmath49 , and a truncation radius of @xmath50 .
this yields an initial disc mass of @xmath51 .
an additional model ( shown as the dashed curve in fig .
[ fig_gas1 ] ) was calculated with the @xmath39 viscosity law and a mass loss of @xmath48 . with identical choices of @xmath45 and @xmath43 ,
the initial disc mass for this model was @xmath52 . as expected from previous calculations ( clarke et al .
2001 ; matsuyama et al .
2003 ) , all four models show qualitatively similar evolution .
there is an initial phase in which the disc mass and accretion rate decline slowly , due primarily to mass accretion onto the star .
subsequently , the mass and accretion rate drop more rapidly as the evolution becomes dominated by mass loss via the wind ( clarke , gendrin & sotomayor 2001 ) .
higher rates of mass loss reduce the disc lifetime , but all models have observationally acceptable lifetimes in the range between 4 myr and 7 myr .
most importantly for our purposes , in all four models mass loss is at least reasonably important ( relative to accretion ) for the overall evolution of the disc .
the fraction of the initial disc mass that is lost in the wind varies between 40 percent and 52 percent for the @xmath37 models , and is 39 percent for the @xmath39 model .
we note that the fraction lost in the wind is only a weak function of @xmath36 , because the interval of time over which the wind acts is reduced for higher instantaneous mass loss rates .
-1.8truein to investigate how planets migrate within the evolving disc , we run the protoplanetary disc models repeatedly . in each run , we add an initially low mass planet ( @xmath53 ) into the disc at a specified time and radius .
we then allow the planet to grow and migrate within the evolving disc , and record the final planet mass and orbital radius once the disc has been either accreted or lost in the wind . by varying the formation time @xmath54 , and the formation radius @xmath55 ,
we study how the final outcome depends upon when and where in the disc massive planets form .
[ fig_gas2 ] shows the results for the disc with the standard viscosity law and varying rates of mass loss .
we considered planet formation radii of 5 au and 10 au , safely outside any estimate of the snow line ( sasselov & lecar 2000 ) , and took @xmath56 .
the final planet mass and orbital radius are plotted as a function of the formation time .
for the low and intermediate rates of mass loss , the sense of orbital migration is predominantly _
inward_. planets formed near the end of the disc lifetime end up in orbits close to where they formed , accrete relatively little disc gas , and remain as low mass objects .
planets formed earlier migrate inwards under the action of gravitational torques , and have time to grow to several jupiter masses .
these results are consistent with previous studies of migration ( trilling et al .
1998 ; armitage et al . 2002
; trilling , lunine & benz 2002 ) .
for the highest rate of mass loss from the disc , however , qualitatively different evolutionary tracks , shown in the right - hand panels of fig .
[ fig_gas2 ] , are obtained .
the enhanced mass loss means that there is a larger range of disc radii across which the radial velocity of the gas is outward
. this can drive significant outward migration . for our choices of parameters
, we find that inward migration persists for all planets formed at 5 au , while outward migration is the rule for planets formed at 10 au
. for formation times between about 1 myr and 2 myr , migration approximately stalls ( similar to the behaviour reported by matsuyama , johnstone & murray 2003 , though for slightly different reasons ) , while for earlier formation times the gas can drive these outer planets to radii of around 20 au or greater .
significant accretion onto the planet occurs throughout this time , so the planets stranded at larger radii are all predicted to be massive objects .
any attempt to distil the inherently multi - dimensional physics of planetary migration into a fast one dimensional scheme is bound to be approximate , and there are particularly obvious uncertainties in our models for planet growth and disc evolution .
we have already demonstrated , as our main result , that for planets forming at radii of around 10 au a switch between inward and outward migration occurs when @xmath36 is varied by a factor of a few .
the mass loss rate via photoevaporation is clearly a vital control parameter . to check how important some of the other parameters are ,
we have recalculated the migration of planets formed at 10 au in two different models . in one
, we altered the assumed disc viscosity ( to @xmath39 rather than @xmath37 ) , with the new viscosity chosen to produce a steady - state surface density profile @xmath57 .
this scaling is one used often in studies of the solar nebula ( weidenschilling 1977 ) .
a second model was computed with the standard viscosity law , but with an accretion efficiency parameter @xmath58 .
this change halves the rate of growth of planets via accretion .
both models used the high rate of mass loss previously found to be conducive to outward migration .
[ fig_gas3 ] shows the final radii attained by planets in the three models .
outward migration occured in all three models , reflecting the primary importance of the assumed mass loss rate in determining the fates of the model planets . changing the efficiency of accretion on to the planet made negligible difference to the final planetary radii ( though it reduced the final masses of the model planets by approximately a factor of two ) .
substantially greater migration , however , was obtained in the calculation with the different viscosity law , despite the fact that a _ smaller _ fraction of the initial disc mass was actually lost to the wind in this case .
we interpret this as being a side effect of the different surface density profile .
outward migration occurs when the torque from the inner disc exceeds that from the disc at radii beyond the planet s orbit .
the steeper surface density profile of the @xmath57 means that there is less mass initially exterior to the planet . as this mass
is lost in the wind , the now unbalanced torque from the inner disc is more effective in driving the planet outward .
# 1 # 1 # 1 = 0.5 mm l|c|c|c|c|c|c|c|c|c|c|c|c|c|c|c| + @xmath59 & ' '' '' 1(-5 ) & 2.5(-5 ) & 5(-5 ) & 7.5(-5 ) & 1(-4 ) & 2.5(-4 ) & 5(-4 ) & 7.5(-4 ) & 1(-3 ) & 2.5(-3 ) & 5(-3 ) & 7.5(-3 ) & 1(-2 ) & 2.5(-2 ) & 5(-2 ) + & * inner & * inner & * inner & * both & * both & * outer & * & * & * & & & & & & + & & & * & * & * none & * both & * both & * & * & * none & * none & & & & + & & & & & * none & * & * outer & * outer & * none & * none & * none & * & * none & & + & & & & & & & * none & * none & * none & * & * none & * none & * none & * outer & * outer + * * * * * * * * * * * * * * * * * * * * * * * * * * what are the implications of our gas disc migration calculations for the origin of planets at large orbital radii ?
we believe that three general conclusions can be drawn .
first , mass loss from the outer disc can drive substantial outward migration , even when the mass loss is modest enough that the disc can survive for several myr .
higher rates of mass loss would lead to more dramatic migration , but the resulting short disc lifetimes might preclude planet formation . in the central regions of the orion nebula , for example , johnstone et al .
( 1998 ) infer mass loss rates between @xmath60 and @xmath61 , with correspondingly small estimated disc lifetimes .
our results suggest that photoevaporation can have a major impact on planetary migration even in substantially more benign environments .
second , outward migration driven by the gas disc is favoured in systems where photoevaporative mass loss is stronger .
the predicted rate of mass loss due to photoevaporation for a solar mass star is rather small ( shu , johnstone & hollenbach 1993 ) , so for relatively isolated stars we would expect outward migration only for stars significantly more massive than the sun .
an alternative possibility is that the mass loss is driven by external irradiation , as in the case of orion ( johnstone , hollenbach & bally 1998 ) .
finally , outward migration via this mechanism is a relatively slow process , which occurs on the viscous time of the protoplanetary disc .
there is ample time for initially low mass planets to accrete substantial gaseous envelopes , so we would expect planets at large radius to be massive objects .
if the initial outcome of the planet formation process is a system of several massive planets , subsequent gravitational interactions can lead to many possible outcomes .
planets may collide , be ejected from the system , or settle into quasi - periodic or periodic orbits .
theoretical investigations of the orbital evolution of two - planet systems ( ford , havlickova , & rasio 2001 ) and few - planet systems ( chambers , wetherill , & boss 1996 ) have been conducted with detailed descriptions of special cases , such as equal mass planets on coplanar orbits .
gladman ( 1993 ) established that systems with two close planets exhibit chaotic but quasi - periodic behavior given the appropriate initial conditions .
ford , havlickova , and rasio ( 2001 ) explored the evolution of such systems when the mass of each of two planets revolving around a solar - type star in nearly circular , coplanar orbits is equal to @xmath62 .
we will expand on this study by considering different mass ratios for the two planets revolving around a central star .
in doing so , we will show that outward migration _ is _ possible , for _ either _ planet .
the setup for our set of simulations is motivated primarily by the gravitational scattering experiments performed by ford , havlickova , & rasio ( 2001 ) .
that study considered the interaction between two planets of mass @xmath62 ( on the order of a jupiter mass ) and in an initially close configuration .
the resulting branching ratios of system outcomes was explored .
systems became unstable by ejecting a planet or by a collision , one between planets or one between a planet and a star .
stable systems remained or settled into quasi - periodic orbits over @xmath63 myr .
the initial configuration of both planets in ford , havlickova , & rasio s ( 2001 ) simulations is motivated by the `` hill stability criterion '' , an analytic result first applied to planetary systems by gladman ( 1993 ) . a system is said to be hill stable if the planets can not approach each other closely for all time .
we adopt gladman s ( 1993 ) notation : @xmath64 represents the least separation for which both planets , _ for sure _ , will be hill stable .
nothing can be said about the hill stability of a system for a separation less than @xmath64 .
@xmath65 represents the greatest separation at which global chaos " will occur .
a system that is `` globally chaotic '' might produce collisions or ejections . for a separation greater than @xmath65 , the planets _ might _ exhibit stable quasi - periodic orbits .
the geometry is illustrated in fig .
( 1993 ) analytic second order expansion for small values of @xmath66 and @xmath67 yields an approximate expression for @xmath64 , @xmath68 unlike @xmath69 an approximation for @xmath65 can only be found empirically . by developing overlap resonance criteria for two - planet systems , wisdom ( 1980 ) derived the following approximation when @xmath70 : @xmath71 ford , havlickova , & rasio ( 2001 ) ran simulations where the initial semimajor axis separation of the planets lay between @xmath64 and @xmath65 , so that the planets would neither be in a hill stable configuration nor become unstable immediately . in this work
, we sample the entire initial separation range spanned by @xmath64 and @xmath65 in order to best detect planets that may migrate outwards and remain on quasi - periodic orbits .
we denote the semimajor axis of the initially inner planet @xmath72 , the initially outer planet @xmath73 , and the respective planet mass / central mass ratios as @xmath66 and @xmath67 .
other orbital elements will distinguish the planets with a subscript of `` 1 '' or `` 2 '' .
all runs were performed with @xmath74 au , so that the results may be scaled easily to any ratio of semimajor axes , such as for @xmath75 au .
all numerical runs of the three - body system were performed with a bulirsch - stoer routine from the hnbody integration package ( rauch & hamilton 2002 ) .
the routine ran for 2 myr with an accuracy parameter of @xmath76 , and was rerun with uniformly smaller initial timesteps when necessary until this accuracy was achieved . for each run ,
the bulirsch - stoer routine began with an initial timestep of @xmath77 yr , and the orbital elements of each body were output every @xmath78 myr .
in most cases , energy and angular momentum errors , expressed by @xmath79 and @xmath80 , where @xmath81 is the initial total system energy and @xmath82 is the initial total angular momentum in the direction perpendicular to the orbital plane , did not exceed @xmath83 . in the most pathological cases , energy and z - angular momentum errors
were conserved to within @xmath84 .
angular momenta in the other two directions were typically conserved to two order of magnitudes better than the z - angular momentum .
we are most interested in the evolution of stable systems in which both planets remain bound .
we define stable systems as systems where both planets have semimajor axes which never exceed @xmath85 au and eccentricities which never exceed @xmath86 .
further , we deemed a system unstable if at any time the hnbody code outputted a negative value for the semimajor axis or eccentricity of either planet .
a preliminary exploration of phase space revealed that of the systems that become unstable , most did so within 0.05 myr .
thus , each system underwent several checks at 0.05 myr ; unstable systems were terminated , and stable systems were evolved for a total of 2 myr . in order to reduce computer time , we imposed additional conditions on systems after 0.05 myr .
systems not satisfying these conditions were terminated .
given that @xmath74 au for each run , the conditions are \1 ) 1.2 au @xmath87 3.0 au , \2 ) @xmath88 au , \3 ) if @xmath89 , then @xmath90 , \4 ) if @xmath91 , then @xmath92 .
these conditions were chosen based on a preliminary exploration of the properties of unstable systems .
for example , we found that after 0.05 myr , if the semimajor axis of the outer planet is within 50% of @xmath72 , then a high ( @xmath93 ) eccentricity of the outer planet implies that the system will become unstable .
the inner planet the outer planet by definition , long - term chaotic behavior may differ drastically due to an infinitesimal change in initial conditions .
therefore , because computers use finite - precision arithmetic , individual runs are largely irreproducible from machine to machine . in this context , one can only make statements about the probability of a system behaving in a certain manner ( quinlan & tremaine 1992 ) .
thus , for each particular dimension of phase space explored , we performed 300 runs , each with randomly chosen initial orbital parameters that lay within specific ranges .
we expanded on the results of ford , havlickova , and rasio ( 2001 ) by investigating the effects of altering the planet / star mass ratio for each planet . for each pair of mass ratios ,
300 runs were performed , each run with both planets having a randomly chosen initial eccentricity between 0 and 0.01 and initial inclination between @xmath94 and @xmath95 , along with an argument of perihelion , longitude of ascending node , and mean anomaly each randomly chosen between @xmath94 and @xmath96 .
we performed 9 sets of 300 runs for each of the following four outer planet mass ratios @xmath67 : @xmath97 @xmath98 @xmath99 and @xmath100 each set of runs corresponds to a different inner planet mass ratio .
we varied the mass of the inner planet within one order of magnitude of the mass of the outer planet .
as mentioned , the initial semimajor axis of the inner planet for every run was set at 1.0 au .
the initial semimajor axis of the outer planet was chosen in a regime that exhibited unpredictable behavior but allowed for quasi - periodic orbits .
explanation of this regime follows .
the inner planet the outer planet the inner planet the outer planet the inner planet the outer planet as mentioned previously , the region bounded by @xmath64 and @xmath65 describe systems that @xmath101 exhibit hill instability and @xmath101 exhibit quasi - periodic motion .
when both are exhibited , @xmath102 planets exhibit outward migration , which is the topic of this section .
so as to include all regions from which migrating behavior might arise , we took @xmath103}^{\frac{2}{7}}.$ ] because @xmath104 au for each run , values of @xmath105 were randomly chosen between @xmath106 and @xmath107 for the simulations .
our initial conditions for the scattering experiments are similar to those adopted by most previous authors ( e.g. ford , havlickova & rasio 2001 ; marzari & weidenschilling 2002 ; chambers , wetherill & boss 1996 ) , in that we start with fully - formed planets in relatively close proximity to each other , and consider only their mutual gravitational interactions .
such initial conditions are chosen primarily for simplicity and compatibility with previous work , and can only partially be justified on physical grounds . in particular , since many of the trial systems prove to be unstable over time scales comparable to the lifetime of protoplanetary discs ( haisch , lada & lada 2001 ) , there is likely to be an earlier epoch during which _ both _ planet - planet gravitational interactions and planet - disc interactions are important .
this earlier stage of evolution could lead to an unstable multiple planet system in at least two ways .
first , two planets might form in well - separated orbits , but then subsequently migrate into an unstable configuration as a consequence of disc - driven migration ( goldreich & tremaine 1980 ; lin & papaloizou 1986 ; lin , bodenheimer & richardson 1996 ) . for this to happen
, the planets would have to avoid becoming trapped into resonance during the migration process ( snellgrove , papaloizou & nelson 2001 ; lee & peale 2002 ; murray , paskowitz & holman 2002 ) .
alternatively , the planets might form close together , and be stabilized against immediate violent instability by the presence of a surrounding gas disc ( lin & ida 1997 ; nagasawa , lin & ida 2003 ) .
further , unstable planetary systems may form after the disc has dissipated .
planet - planet interactions alone in crowded systems of jovian - mass planets most often leaves only two survivors in close , quasiperiodic but ultimately unstable orbits ( adams & laughlin 2003 ) .
although any of these pathways could plausibly lead to initial conditions that are similar to those which we ( and other authors ) have assumed , a full treatment will obviously need to model the more complex interactions that are possible during the phase when the gas disc is being dispersed . table 1 and figs .
[ fig1]-[fig4 ] displays the data we took .
table 1 summarizes the data in terms of mass ratio and extent of migration , while the figures provide a more detailed look at migration behavior .
[ fig5 ] provides a representative sample of the final semimajor axes and eccentricities of both planets when significant outward migration occurs , and provides a perspective on how data is plotted in figs .
[ fig1]-[fig4 ] .
for example , the 43 stable systems displayed in fig .
[ fig5 ] are plotted as the solid lines in fig .
[ fig5 ] illustrates that planets that migrate both inward and outward tend to increase their eccentricity .
although the extent of migration in both directions is similar , those planets that migrate inward tend to increase their eccentricities at a higher rate . figs .
[ fig1]-[fig4 ] each represent a set of six cumulative probability distributions that display the fraction of _ stable _ systems vs. semimajor axis ratios , eccentricities , and inclinations of the outer and inner planet after exactly 2 myr .
the upper panel in each figure represents the initially outer planet , and the lower panel represent the initially inner planet .
each figure keeps the mass of the outer planet fixed , but varies the mass of the inner planet .
although we obtained nine curves for each graph on each figure , we show only the curves that exhibit significant radially outward migration , plus some that do not , and we do not show any curves where less than 10% of the 300 runs were stable . detecting migration for the outer planets
is difficult because their semimajor axes were randomly chosen for each run . to aid in determining the net radial movement of these planets
, we define @xmath108 and @xmath109 .
each panel in figs .
[ fig1]-[fig4 ] contain distributions of semimajor axis ratios , rather than absolute semimajor axes , because those are scalable in this study .
table 1 and figs .
[ fig1]-[fig4 ] illustrate that outward migration of both the inner and outer planets occurs more frequently with smaller values of @xmath67 .
[ fig1 ] displays extensive migration . in the lowest mass case ,
almost 60% of the stable inner planets migrate at outward to at least 150% of their original semimajor axis . as @xmath66
is increased , migration occurs less frequently ; for @xmath110 no inner planets migrated out to 150% of their original semimajor axis .
the correlation of mass ratio to final state behavior is less apparent but still present in the final eccentricity and inclination curves . in both graphs ,
the least - mass case prompts the highest final eccentricity and inclination values for the orbit of the inner planet . the final state of the outer planet , represented by the lower panel of fig .
[ fig1 ] , reflects the exchange of angular momentum in this system .
one sees that when an inner planet migrates outward , it leaves the initially outward planet back so that @xmath111 .
the one instance where the outer planet experiences significant migration is the highest mass case - the same case where @xmath112 .
the eccentricity and inclination distributions for the outer planet in this massive case differ drastically from the behavior seen for other inner planet masses . in fig .
[ fig2 ] , @xmath113 and the result is significantly less migration than seen in fig .
[ fig1 ] . similarly to fig .
[ fig1 ] , the highest mass case exhibits the greatest migration of the outer planet . in this case , @xmath11420% of the outer planets and @xmath11415% of the inner planets migrate beyond 150% of their original semimajor axis . halving the inner mass value reduces the probability for @xmath112 or @xmath115 to @xmath1145% .
in contrast to fig . [ fig1 ] , the least mass case fails to show significant outward migration of the inner planet , despite the inner planet exhibiting the highest final inclinations . in fig .
[ fig3 ] , where the outer mass is a full order of magnitude more massive than in fig .
[ fig1 ] , we see even less migration .
inner bodies for the smallest mass case exhibit high eccentricities and inclinations , but only @xmath1145% satisfy @xmath116 . the case where @xmath117 is the same case studied by ford , havlickova , & rasio ( 2001 ) . their figs .
11 - 12 can be compared to the semimajor axis ratio distributions , eccentricity distributions , and inclination distributions in fig .
[ fig3 ] .
no migration of the inner planet occurs for any case in fig .
however , the outer planet shows significant migration .
further , as shown by the sequence of four curves in the lower panel , the higher the mass of the outer planet , the greater the extent of the migration . for the higher mass cases ,
about 25% of the initally outer planets satisfy @xmath118 .
remarkably , these planets all retain final eccentricities @xmath119 , in stark contrast to the large final eccentricities of the migrating planets of figs .
[ fig1]-[fig3 ] . because of the chaotic nature of the three body problem , one encounters difficulty when predicting the appropriate timescale over which to integrate .
we adopted a similar timescale to the one used by ford , havlickova , & rasio ( 2001 ) , but recognize that systems that appear to be stable at 2 myr might become unstable at some future time .
we extended the running time of one system to sample the consequences .
after 2 myr , 55 out of the 300 systems for the case @xmath120 remain stable . by running this system for 10 myr , we found only 3 out of the 55 systems became unstable , and did so quickly .
further , these three systems became unstable before 3 myr .
for any set of 300 runs of a chaotic system , the slightest change in any input parameter of the bulirsch - stoer algorithm might drastically alter the results of any individual runs , but keep the same global behavior .
the extent of the invariance of this global behavior is a function of the code used and the number of systems sampled . to explore this measure of invariance for the runs presented here ,
we reran two systems with four different initial tolerances ( @xmath121 , @xmath122 , @xmath123 , @xmath124 ) . one of these systems ( @xmath125 ) exhibited negligible migration , and the other ( @xmath126 ) exhibited significant migration . in order to achieve the largest feasible sample size for this error analysis ,
we tripled the number of runs performed to 900 for the system exhibiting negligible migration . as fig .
[ fig6 ] illustrates , runs with different tolerances are practically indistinguishable for the case of no migration , but vary up to 20% for the case of significant migration .
the gravitational interaction of a pair of unequal mass planets that lie in close initial configurations allow _ either _ planet to migrate outward to at least twice it s initial semimajor axis .
planets that migrate outward lie in quasi - stable , slightly inclined eccentric orbits with an eccentricity that spans the entire permissible range and an inclination up to about @xmath127 .
although either planet in any given system may drift outward , the smaller the mass of the planets , the higher the tendency for the initially inner planet to migrate outward .
further , less massive giant planets such as uranus or neptune are more likely to migrate outward than jupiter - mass planets .
this result , derived from our gravitational scattering simulations alone , is consistent with thommes , duncan & levison s ( 1999 , 2002 ) conclusion that uranus and neptune s current location is a result of outward migration amongst jupiter and saturn .
as a first step in explaining the presence of planets at large orbital radii , we have shown that outward migration of protoplanets is possible both by planet - disc interactions and by planet - planet gravitational scattering without the presence of a disc .
strong mass loss in discs coupled with planetary cores that are formed at about @xmath128 au allow planets to migrate outward in discs to radii that are as much as a factor of several in excess of their initial semimajor axes .
planets that migrate in such a manner are likely to be massive .
we predict that gas - driven outward migration should be most likely to occur around more massive stars , whose strong uv flux can drive a powerful photoevaporative outflow .
subsequently , when in the appropriate chaotic regime , planets within a multiple planet system may migrate outward due to gravitational scattering alone .
planets that migrate in this manner may be massive or not , however low mass objects tend to exhibit the most extensive outward migration .
orbital migration is typically accompanied by an increase in eccentricity that spans the allowable range for elliptic orbits .
we thank an anonymous referree for helpful comments , kevin rauch and doug hamilton for use of their integrator and assistance in it s operation , as well as the colorado rings group for useful discussions . | observations of structure in circumstellar debris discs provide circumstantial evidence for the presence of massive planets at large ( several tens of au ) orbital radii , where the timescale for planet formation via core accretion is prohibitively long . here
, we investigate whether a population of distant planets can be produced via outward migration subsequent to formation in the inner disc .
two possibilities for significant outward migration are identified .
first , cores that form early at radii @xmath0 can be carried to larger radii via gravitational interaction with the gaseous disc .
this process is efficient if there is strong mass loss from the disc either within a cluster or due to photoevaporation from a star more massive than the sun but does not require the extremely destructive environment found , for example , in the core of the orion nebula .
we find that , depending upon the disc model , gas disc migration can yield massive planets ( several jupiter masses ) at radii of around 20 - 50 au .
second , interactions within multiple planet systems can drive the outer planet into a large , normally highly eccentric orbit .
a series of scattering experiments suggests that this process is most efficient for lower mass planets within systems of unequal mass ratio .
this mechanism is a good candidate for explaining the origin of relatively low mass giant planets in eccentric orbits at large radii .
accretion , accretion discs stars : formation stars : pre - main - sequence planetary systems : protoplanetary discs planets and satellites : formation |
recently , studies in numerical general relativity have shown that merged binary black holes can have a large recoil velocity through anisotropic emission of gravitational waves ( e.g. * ? ? ?
* ; * ? ? ?
the maximum velocity would reach @xmath3 , although the actual distribution of kick velocities is very uncertain .
the discovery of such recoiled black holes is important for studies about the growth of black holes as well as the general relativity .
supermassive black holes ( @xmath4 ) at the centers of disk galaxies would be kicked through this mechanism .
although they would be bright just after the kick because of the emission from the accretion disk carried by the black holes , they would soon get dim as the disk is consumed by the black holes @xcite .
however , if the kick velocity of a black hole is not large enough to escape from the host galaxy , it will eventually fall back onto the galactic disk .
if the time - scale of dynamical friction is large enough , it will revolve around the galactic center many times before it finally spirals into the galactic center .
when the black hole passes the galactic disk , it will accrete the interstellar medium ( ism ) in the disk @xcite .
the accretion of the surrounding gas onto an isolated black hole has been studied by several authors ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and references therein ) .
most of the previous studies focused on stellar mass ( @xmath5 ) or intermediate mass black holes ( imbhs ; @xmath6 ) .
the accretion rate and thus the luminosity of a black hole depend on the mass of the black hole and the density of the gas surrounding it ( see equation [ [ eq : dotm ] ] ) .
since the mass of a stellar - mass black hole is small , its luminosity becomes large enough to be observed only when it plunges into a high density region such as a molecular cloud . on the other hand , in this letter
, we show that a recoiled supermassive black hole can shine even in the ordinary region of a galactic disk because of its huge mass .
we calculate the orbit of a recoiled black hole in a fixed galaxy potential .
the galaxy potential consists of three components , which are @xcite disk , hernquist spheroid , and a logarithmic halo : @xmath7 @xmath8 @xmath9\:,\ ] ] where @xmath10 ( @xmath11 ) and @xmath12 are cylindrical coordinates aligned with the galactic disk , and @xmath13 .
we adopt the parameters for the galaxy .
we take @xmath14 , @xmath15 , @xmath16 kpc , @xmath17 kpc , @xmath18 kpc , @xmath19 kpc , and @xmath20 ; @xmath21 is determined so that the circular velocity for the total potential is @xmath22 at @xmath23 kpc ( see * ? ? ?
we solve the equation of motion for the supermassive black hole : @xmath24 where @xmath25 is the velocity of the black hole , and @xmath26 .
the density of the disk is given by @xmath27^{5/2}(z^2+b^2)^{3/2}}\ ] ] @xcite .
we assume that part of the disk consists of the ism ; its density is represented by @xmath28 and @xmath29 .
the circulation velocity of the disk is given by @xmath30 the accretion rate of the ism onto the supermassive black hole is given by the bondi - hoyle accretion @xcite : @xmath31 where @xmath32 is the mass of the black hole , @xmath33 ( @xmath34 ) is the sound velocity of the ism , and @xmath35 is the relative velocity between the black hole and the surrounding ism .
we assume that the orbit of the black hole is confined on the @xmath36-@xmath12 plane ( @xmath37 ) .
thus , the relative velocity is simply given by @xmath38 .
the x - ray luminosity of the black hole is given by @xmath39 where @xmath40 is the efficiency .
since the accretion rate is relatively small for the mass of the black hole , the accretion flow would be a radiatively inefficient accretion flow ( riaf ; * ? ? ?
* ; * ? ? ?
* ) . in this case
, the efficiency follows @xmath41 for @xmath42 , where @xmath43 is the eddington luminosity ( e.g. * ? ? ?
therefore , we assume that @xmath44 for @xmath45 and @xmath46 for @xmath47 , where @xmath48 @xcite .
we assume that @xmath49 .
we solved equation ( [ eq : motion ] ) by mathematica 6.0 using a command ndsolve .
the black hole is ejected on the @xmath36-@xmath12 plane at @xmath50 .
the direction of the ejection changes from @xmath51 to @xmath52 , where @xmath51 corresponds to the @xmath12-axis .
we calculate the orbit until @xmath53 , which is chosen to be much larger than the period of revolution and to be smaller than the time - scale of dynamical friction .
the latter is estimated to be @xmath54 @xcite , where @xmath55 is the total density ( disk@xmath56sphere@xmath56halo ) .
the halo component does not much affect the dynamical friction . since @xmath57-body simulations for a spherically symmetric potential showed that the coulomb logarithm is @xmath583 @xcite
, we take @xmath59 .
the effects of dynamical friction on orbits in a complex potential like the one we adopted would be complicated and ideally should be studied with high - resolution @xmath57-body simulations .
thus , equation ( [ eq : dynfrc ] ) should be regarded as a rough estimate of the time - scale of the dynamical - friction .
the black hole is placed at the center of the galaxy at @xmath50 . since we do not know the distributions of mass and initial velocity ( @xmath60 ) of the black hole , we consider situations in which the emission from it would be observed easily . that is , the luminosity of the black hole would be large , and the observable time would be long .
we consider five combinations of @xmath32 and @xmath60 shown in table [ tab : par ] .
if we take larger @xmath32 and/or smaller @xmath60 , the dynamical friction becomes more effective and the black hole quickly falls into the galaxy center . on the other hand ,
if we take smaller @xmath32 and/or larger @xmath60 , the luminosity of the black hole becomes too small to be observed ( equation [ [ eq : dotm ] ] ) .
moreover , the black hole is not bound to the galaxy , if @xmath60 is too large .
the dynamical friction is most effective when @xmath61 . in table
[ tab : par ] , we show the time - average of the time - scale , @xmath62 , for @xmath63 and @xmath61 . fig .
[ fig : orbit ] shows the orbit of the black hole when @xmath64 and @xmath65 .
[ fig : lum ] shows the luminosity of the same black hole ( @xmath66 ) . in table
[ tab : par ] , we present the distance of the apastrons from the center of the galaxy ( @xmath67 ) when @xmath61 .
it is to be noted that @xmath67 is not much dependent on @xmath68 for a given @xmath60 .
we also present the maximum x - ray luminosity of the black hole ( @xmath69 ) when @xmath61 in table [ tab : par ] . for a given @xmath32 and @xmath60 ,
the x - ray luminosity is larger when @xmath68 is closer to @xmath52 , because the orbit is included in the galactic disk , where @xmath70 is large .
we found that for @xmath71 , the luminosity reaches its maximum when the black hole passes apastrons and when the apastrons reside in the disk of the galaxy .
this is because @xmath72 decreases , @xmath70 increases , and thus @xmath73 increases there ( equation [ [ eq : dotm ] ] ) . on the other hand , for @xmath74 ,
the distance of the apastrons from the galactic center ( @xmath67 ) is always large ( table [ tab : par ] ) .
thus , even if apastrons reside in the disk , @xmath70 is small there .
therefore , the luminosity of the black hole reaches its maximum between the apastron and periastron , and @xmath69 is smaller compared with the models of @xmath71 ( table [ tab : par ] ) . assuming that black holes are ejected in random directions at the centers of galaxies
, we estimate the probability of observing black holes with luminosities larger than a threshold luminosity @xmath75 . for given @xmath32 and @xmath60
, we calculate 91 evolutions of the luminosity by changing @xmath68 from @xmath76 to @xmath52 by one degree .
then , we obtain the period during which the relation @xmath77 is satisfied for each @xmath68 , and divide the period by @xmath78 .
this is the fraction of the period during which the black hole luminosity becomes larger than @xmath79 .
we refer to this fraction as @xmath80 and show it in fig .
[ fig : ftheta ] when @xmath81 , @xmath64 , and @xmath82 .
we average @xmath80 by @xmath68 , weighting with @xmath83 , and obtain the probability of observing black holes with @xmath84 . in table
[ tab : par ] , we present the probability @xmath85 when @xmath86 ; for the parameters we chose , @xmath870.56 .
we have found that a supermassive back hole that had been recoiled at the center of a disk galaxy could be observed in the galactic disk with an x - ray luminosity of @xmath88 .
one of the candidates of such objects is ultraluminous x - ray sources ( ulxs ) observed in disk galaxies @xcite .
they are found in off - nuclear regions of nearby galaxies and their x - ray luminosities exceed @xmath89 , which are larger than the eddington luminosity of a black hole with a mass of @xmath90 .
if ulxs are stellar mass black holes , they might be explained by anisotropic emission @xcite , slim - disks @xcite or thin , super - eddington accretion disks @xcite . on the other hand
, there is some evidence that they are imbhs at least for some of them @xcite .
considering their x - ray luminosities and off - center positions , some of the ulxs might be the recoiled supermassive black holes .
however , the fraction of supermassive black holes in the ulxs would not be large .
@xcite estimated that for comparable mass binaries with dimensionless spin values of 0.9 , only @xmath91% of all mergers are expected to result in an ejection speed of @xmath92@xmath93 .
since the ejection speed is smaller for mergers with large mass ratios and smaller spin values , the actual fraction would be smaller .
moreover , in our model , the time - corrected probability of observing black holes with @xmath94 is @xmath95 , where @xmath96 is obtained by averaging @xmath97 f(\theta)/t_{\rm age}$ ] by @xmath68 , weighting with @xmath83 , and @xmath98 ( @xmath91 gyr ) is the age of a galaxy ( table [ tab : par ] ) . here
, we note that @xmath99 should be regarded as the upper - limit of the actual time - scale , because @xmath100 should decrease through the dynamical friction every time the black hole passes the dense region of the galaxy .
furthermore , our model indicates that a traveling supermassive black hole needs to have a mass comparable to the one currently observed at the galactic center in order to have large @xmath101 .
it is unlikely that a galaxy would have undergone many mergers of black holes with such masses .
the number of such mergers that a galaxy has undergone would be @xmath102 .
thus , the probability that a galaxy has a traveling supermassive black hole with a luminosity comparable to that of ulxs is @xmath103 .
in fact , current radio observations seem to show that ulxs observed so far are not supermassive black holes .
our model predicts that the x - ray luminosity of a supermassive black hole traveling through the galaxy is comparable to the typical x - ray luminosity of a liner ( @xmath104@xmath105 ; * ? ? ?
liners seem to show core radio emission and many even have detectable jets @xcite . on the other hand ,
radio observations have shown that no ulx has been detected with a unresolved radio core @xcite .
moreover , it has been shown that the optical luminosities of ulxs tend to be smaller than their x - ray luminosities ( e.g. * ? ? ?
* ) , which is inconsistent with typical riaf spectra ( e.g. * ? ? ?
thus , it is unlikely that most of the ulxs are the supermassive black holes traveling through the galaxies .
however , the recoiled supermassive black holes could be found through future extensive surveys .
our model predicts that the x - ray luminous black holes should not be observed far from the centers of the host galaxies ( say @xmath106 kpc ) , because @xmath70 should be small there ( [ sec : results ] ) .
our model also predicts that the relative velocity between the x - ray source and the surrounding ism and stars is @xmath107 .
if atomic line emission associated with the x - ray source is observed , the velocity could be estimated through the doppler shift . instead of x - rays , @xcite argued that radio detections may be best to search for isolated accreting black holes .
the detailed analysis of the spectra and the time variability would be useful to determine the masses of the black holes @xcite . in the future , statistical studies could observationally constrain the probability of the mergers of black holes and the recoil .
we have shown that a supermassive black hole ejected from the center of the host disk galaxy will return to the galactic disk , if the initial velocity is smaller than the escape velocity of the galaxy .
the black hole accretes the surrounding ism and the resultant x - ray luminosity can reach @xmath108 , when it passes the apastrons in the disk .
although the luminosity of a recoiled supermassive black hole is comparable to that of ultra - luminous x - ray sources ( ulxs ) , it is unlikely that many of the observed ulxs are the supermassive black holes .
i would like to thank the anonymous referee for useful comments .
i am grateful to h. tagoshi and t. tsuribe for useful discussion .
yf was supported in part by grants - in - aid from the ministry of education , culture , sports , science and technology of japan ( 20540269 ) .
cccccccc @xmath109&500 & 0.3 & 4.7 & 1 & @xmath110 & 0 & 0 + @xmath111&500 & 0.3 & 1.4 & 1 & @xmath112 & 0.064 & 0.010 + @xmath111&600 & 1 & 4.5 & 3 & @xmath112 & 0.016 & 0.012 + @xmath111&700 & 2 & 27 & 13 & @xmath113 & 0 & 0 + @xmath114&500 & 0.3 & 0.47 & 1 & @xmath115 & 0.56 & 0.031 + @xmath114&600 & 1 & 1.5 & 3 & @xmath115 & 0.15 & 0.11 + @xmath114&700 & 2 & 9.0 & 13 & @xmath116 & 0 & 0 + | recent studies have indicated that the emission of gravitational waves at the merger of two black holes gives a kick to the final black hole .
if the supermassive black hole at the center of a disk galaxy is kicked but the velocity is not large enough to escape from the host galaxy , it will fall back onto the the disk and accrete the interstellar medium in the disk .
we study the x - ray emission from the black holes with masses of @xmath0 recoiled from the galactic center with velocities of @xmath1 .
we find that their luminosities can reach @xmath2 , when they pass the apastrons in the disk . while the x - ray luminosities are comparable to those of ultra - luminous x - ray sources ( ulxs ) observed in disk galaxies , ulxs observed so far do not seem to be such supermassive black holes .
statical studies could constrain the probability of merger and recoil of supermassive black holes . |
for a polytope @xmath6 of dimension @xmath4 , the _ face vector _ of @xmath6 is given by @xmath7 where @xmath8 is the number of @xmath9-dimensional faces of @xmath6 .
it is an ongoing investigation to describe the face vectors for @xmath4-dimensional convex polytopes .
the case @xmath10 are the only cases where this description is complete . a nice theorem by steinitz @xcite states that an integer vector @xmath11 is a face vector for a three dimensional polytope @xmath6 if and only if ( i ) we have the euler relation @xmath12 , ( ii ) @xmath13 and ( iii ) @xmath14 .
moreover , the first inequality is tight if and only if @xmath6 is simplicial , the second one is tight if and only if @xmath6 is simple .
the only linear equation satisfied by the entries of the face vector is the euler - poincar relation @xmath15 , that holds for all polytopes of dimension @xmath4 . for higher dimensions
@xmath16 no such nice description of face vectors has been found yet . in the hope of obtaining a description of a polytope in terms of parameters / entries of a vector
, there is the _ flag vector _ @xmath17 , containing @xmath18 entries where for each @xmath19 the entry @xmath20 denotes the number of strict chains ( or _ flags _ ) @xmath21 of faces of @xmath6 where @xmath22 .
the entries @xmath20 do satisfy the generalized dehn - sommerville equations which , together with the existence of matching polytopes , implies that the linear span of flag vectors of @xmath4-dimensional polytopes is precisely @xmath23 , the @xmath4-th fibonacci number , as shown in the seminal paper @xcite .
one way to encode the information contained in the flag vector of a @xmath4-polytope @xmath6 is to define the @xmath24-index , a linear combination of non - commuting monomials in @xmath25 and @xmath26 of degree @xmath4 in the following way : for @xmath27 let @xmath28 , where @xmath29 if @xmath30 and @xmath31 if @xmath32 .
the _ @xmath24-index _
@xmath33 for @xmath6 is then defined as @xmath34 , where the sum is over all @xmath27 and @xmath35 .
an elegant consequence of the generalized dehn - sommerville equations for the entries @xmath20 is that each @xmath24-index can be written as a non - commuting polynomial in @xmath36 and @xmath37 .
the resulting polynomial called the _
@xmath38-index _ , also denoted @xmath33 and first appeared in @xcite . within the free non - commuting @xmath39-algebra @xmath40 in two variables @xmath25 and @xmath26 , the degree of @xmath41
is set to be 1 and the degree of @xmath42 is set to be two . in this way
the number of all homogeneous monomials of degree @xmath4 in @xmath41 and @xmath42 is precisely @xmath23 .
hence , the @xmath38-index is the most compact way to encode all the entries of the flag vector of a polytope @xmath6 w.r.t .
linear relations of the coefficients of the monomial in @xmath41 and @xmath42 .
a lot of current research is therefore directed toward understanding the @xmath38-index for polytopes .
although the @xmath38-index is an elegant and compact way to encode the flag vector for a polytope @xmath6 , it can be at times difficult to handle in many explicit computations since algebraic manipulation in @xmath43 can become particularly cumbersome .
the purpose of this article is to do the opposite of the above : namely , instead of attempting to compactify the encoding of the flag vector further , then we shall instead relax the constrictions and allow some redundancies among coefficients in order to facilitate direct computations .
this will in certain cases allow us to write down explicit ( albeit involved ) formulae from which the entries of the flag vector can be obtained .
we will define the _ flag polynomial _ for a polytope , a polynomial in commuting variables that will behave reasonably well when constructing polytopes by known methods i.e. prisms , pyramids , products , free joins and minkowski sums .
the article is organized as follows : in section [ sec : basic ] we present our basic definitions and write some direct consequences . in section [ sec : flag - poly ] we state ( but do not prove ) theorem [ thm : flag - main ] , the main result of the section ; a formula for the flag polynomial of a minkowski sum of @xmath0 standard simplices .
we then parse through an explicit application of theorem [ thm : flag - main ] and note some consequences .
section [ sec : join ] is devoted to the derivation of theorem [ thm : p - acts - flag ] ; a formula for the flag polynomial of a polytope @xmath44 , a polytope obtained when the geometric polytope @xmath6 in @xmath45 acts on the polytopes @xmath46 .
this action generalizes the notion of the free join of two polytopes .
finally , we note that theorem [ thm : flag - main ] is a corollary of theorem [ thm : p - acts - flag ] . in the last section [ sec : f - poly-2 ] we present two examples on the application of the flag polynomial . in the first example we compute the maximum number of faces of an arbitrary minkowski sum of two simplices . in the latter example
we compute the maximum discrepancy between the number of @xmath5-chains of faces of a minkowski sum of two simplices and the number of @xmath5-chains of faces of a simple polytope of the same dimension and on the same number of vertices as the minkowski sum .
this article can be viewed as a self contained continuation of @xcite . for related references and history of the significance of minkowski sums of simplices
we refer to the introduction of @xcite . also , for self - containment we will present definition [ def : master ] and some terminology from @xcite . for @xmath47
we let @xmath48 = \{1,2,\ldots , r\}$ ] .
the _ standard simplex _ @xmath49}$ ] of dimension @xmath50 is given by @xmath51 } = \ { \tilde{x } = ( x_1,\ldots , x_r)\in { { \mbox{$\mathbb r$}}}^r : x_i\geq 0 \mbox { for all $ i$ } , x_1+\cdots+x_r = 1\}. $ ] each subset @xmath52 $ ] yields a _ face _ @xmath53 of @xmath54}$ ] given by @xmath55 } : x_i = 0 \mbox { for } i\not\in f\}. $ ] clearly @xmath53 is itself a simplex embedded in @xmath56 . if @xmath57 is a family of subsets of @xmath48 $ ] , then we can form the _ minkowski sum _ of simplices @xmath58 note that a face of any polytope @xmath59 ( in particular of @xmath54}$ ] ) is given by the set of points that maximize a linear functional @xmath60 for @xmath61 . recall the @xmath3-polynomial of a polytope : let @xmath6 be a @xmath4-dimensional polytope . for each @xmath62 let @xmath63 denote the set of all @xmath9-dimensional faces of @xmath6 and so @xmath64 is the number of @xmath9-dimensional faces of @xmath6 .
then @xmath65 is the set of all the faces of @xmath6 except the empty face .
@xmath3-polynomial _ of @xmath6 is then given by @xmath66 .
[ obs : f - poly - standard ] the @xmath3-polynomial for the standard @xmath67-dimensional simplex @xmath49}$ ] is given by @xmath68
we now define our main object for this paper , a generalization of the @xmath3-polynomial of a polytope , and derive some basic results .
conventions : ( i ) for an @xmath69-tuple @xmath70 of variables and an @xmath69-tuple of numbers @xmath71 , let @xmath72 .
( ii ) for @xmath71 let @xmath73 .
[ def : flag - poly ] let @xmath6 be a polytope with @xmath74 and @xmath75 .
for an @xmath69-tuple of variables @xmath70 the _
@xmath69-flag polynomial _ is defined by @xmath76 where the sum is taken over all chains @xmath77 with @xmath78 and @xmath79 denotes the number of chains of faces @xmath80 of @xmath6 with @xmath22 for each @xmath81 .
note that each @xmath69-tuple @xmath82 of non - negative integers yields an @xmath69-tuple @xmath83 of increasing chain of integers @xmath84 where @xmath85 for each @xmath9 and so @xmath86 the reasons why it might be worth working with the flag polynomial of a polytope is ( i ) its expression is simple for many basic polytopes like the simplex and hypercube and ( ii ) it behaves well under various natural constructions , the first of which is the cartesian product .
[ prp : prod - good ] for polytopes @xmath6 and @xmath87 , a positive integer @xmath88 and @xmath70 we have @xmath89 .
any pair of chains @xmath90 and @xmath91 yields a chain @xmath92 of faces in the product @xmath93 .
vice versa , any such chain of faces in @xmath93 of length @xmath69 yields a pair of chains of faces of length @xmath69 in @xmath6 and @xmath87 respectively .
since the mapping @xmath94 is linear the proposition follows .
consider now the standard @xmath67-dimensional simplex @xmath95 .
let @xmath96 be an @xmath69-chain of nonnegative integers . since any choice of @xmath97 vertices from the @xmath98 vertices of @xmath95 is an @xmath99-face of @xmath95 , the number of chains @xmath100 of faces of @xmath95 with @xmath22 for each @xmath9 is , by the product rule , given by @xmath101 since @xmath102 we have the following .
[ obs : flag - poly - simp ] for @xmath103 the @xmath69-flag polynomial of the standard simplex @xmath95 is given by @xmath104 by proposition [ prp : prod - good ] and observation [ obs : flag - poly - simp ] we have obtain the following .
[ cor : flag - poly - prism ] for @xmath105 and a polytope @xmath6 the @xmath69-flag polynomial of the prism @xmath106 of @xmath6 is given by @xmath107 since the @xmath4-dimensional hypercube @xmath108 is given by @xmath109 ( @xmath4-times ) , then we also obtain the following .
[ cor : flag - poly - hcube ] for @xmath110 the @xmath69-flag polynomial of the hypercube @xmath108 is given by @xmath111 from this we see that the number of @xmath69-chains of faces of @xmath108 is @xmath112 . in particular , for @xmath113 we see that the total number of faces of @xmath108 is @xmath114 , the conjectured maximum number of faces of a centrally symmetric @xmath4-polytope @xcite .
should one want to see the chains from the monomials more explicitly we have the following directly by definition [ def : flag - poly ] of the flag polynomial .
[ obs : part->seq ] let @xmath6 be a polytope and @xmath105 .
if @xmath115 and @xmath116 are two tuples of variables such that @xmath117 for each @xmath118 , then the coefficient of @xmath119 in @xmath120 is the number of chains of faces @xmath80 of @xmath6 where @xmath22 for each @xmath9 .
remarks : ( i ) if for some reason one prefers to consider a polynomial directly corresponding to the flag vector ( i.e. terms of the polynomial corresponding to strict chains of faces ) , then that can be obtained by the inclusion / exclusion principle : for each fixed @xmath69 , the polynomial corresponding to the number of _ strict _ chains @xmath121 of faces of a polynomial @xmath6 with @xmath122 is given by @xmath123 where @xmath124 where @xmath125 is the @xmath126-flag polynomial on the @xmath126-tuple obtained from @xmath127 by removing the variables @xmath128 for @xmath129 .
\(ii ) the number of terms in @xmath130 is @xmath131 whereas the number of terms in @xmath132 equals the number of solutions @xmath133 to @xmath134 in nonnegative integers @xmath135 , namely @xmath136 .
comparing the number of entries of the flag vector to the combined number of terms of all flag polynomials @xmath132 for @xmath137 , the combined number of terms equals @xmath138 not surprisingly an exponential expression which is considerably less than @xmath139 , the square of the number of entries in the flag vector , but also more than @xmath23 , the @xmath4-th fibonacci number , the exact number of linearly independent terms @xcite . in the next section we state theorem [ thm : flag - main ] , our main result , then parse through an elaborate example , before proving the main result in section [ sec : join ] .
if @xmath57 consists of @xmath0 subsets of @xmath48 $ ] and is presented ( merely for convenience ) as an ordered tuple @xmath140 , then for each @xmath141 $ ] we let @xmath142:i \in f_j\}$ ] .
the following definition is from @xcite : [ def : master ] for @xmath2 let @xmath143 be the family of @xmath0 ordered subsets of @xmath144 $ ] so that for @xmath145 then @xmath146 is the @xmath9-th nonempty subset of @xmath147 $ ] in the lexicographic order .
then the minkowski sum of these @xmath0 simplices @xmath148 is called the @xmath149 _ master polytope_. example : for @xmath150 we have @xmath144 = [ 3 ] = \{1,2,3\}$ ] and the lexicographic ordering of the nonempty subsets of @xmath151 = \{1,2\}$ ] is given by @xmath152 . by definition
we must have @xmath153 where @xmath154 and @xmath155 . in this case
@xmath156 is the 2-dimensional rhombus lying in the plane @xmath157 in @xmath158 that is the convex hull of the points @xmath159 , @xmath160 , @xmath161 and @xmath162 . in a similar fashion , for @xmath163 we have @xmath144 = [ 7]$ ] and the lexicographic ordering of the nonempty subsets of @xmath164 $ ] is given by @xmath165 . by definition @xmath166 where @xmath167 , @xmath168 and @xmath169 . in this case
@xmath170 is a 6-dimensional polytope in @xmath171 with 41 vertices as verified by the computer program polymake @xcite ( see ( * ? ? ?
* observations 2.13 and 2.16 ) ) .
note that for a general @xmath2 then @xmath172 where @xmath173 for each @xmath9 .
the master polytopes @xmath1 for @xmath174 are fundamental in understanding the overall structure of a general minkowski sum of standard simplices , as we will see in theorem [ thm : flag - main ] , the main result in this section , where we express the flag polynomial of a general minkowski sum of simplices .
example : since @xmath175 , the 2-dimensional rhombus , has the combinatorial structure of @xmath176 we get by proposition [ prp : prod - good ] and observation [ obs : flag - poly - simp ] the following .
@xmath177 ^ 2 \\ & = & \left(\frac{(x_1+x_2+x_3 + 1)^2 - ( x_2+x_3 + 1)^2}{x_1}\right)^2 \\ & = & ( x_1 + 2x_2 + 2x_3 + 2)^2,\end{aligned}\ ] ] and hence by observation [ obs : part->seq ] @xmath178 something which is evident by inspection . from now on , we will assume @xmath2 fixed and we will always present our collection of subsets as an ordered @xmath0-tuple @xmath140 of @xmath48 $ ] , and the minkowski sum as @xmath179 we will assume that @xmath57 covers all of @xmath48 $ ] , that is @xmath180 $ ] , since otherwise we can simply restrict to this union . for a general minkowski sum @xmath181 of simplices as in ( [ eqn : minkssum ] ) , then for each @xmath182 $ ] we let @xmath183 : n_{\cal{f}}(j ) = n_{{\cal{h}}(k)}(i)\}$ ] and @xmath184 , so @xmath185 is the number of elements in the set @xmath186 .
since @xmath57 covers @xmath48 $ ] we have @xmath187 .
note that for any polytope @xmath6 we have that @xmath188 only has terms from @xmath189 where the coefficient for @xmath190 is zero .
hence , @xmath188 is the generating function for the number of chains @xmath191 of faces of @xmath6 , which is equivalent to the generating function for the number of chains @xmath192 of faces of @xmath6 .
[ obs : x_ell=0 ] for a polytope @xmath6 we have @xmath193 .
example : considering @xmath175 again , we have from the above example that @xmath194 note that the total number of 2-chains is the sum of the coefficients of @xmath195 , 25 altogether .
conventions : ( i ) for a set of point @xmath196 let @xmath197 : x_i\neq 0\mbox { for some } \tilde{x } = ( x_1,\ldots , x_r ) \in s\}$ ] be the _ support _ of @xmath198 .
( ii ) for a face @xmath199 of @xmath6 let @xmath200 denote the dimension of the face @xmath199 .
( iii ) for a polytope @xmath6 we let @xmath201 denote the set of all @xmath69-chains @xmath202 of faces of @xmath6 with @xmath80 .
( iv ) for each such @xmath203 let @xmath204 denote the chain of their corresponding dimensions . with the above conventions in mind we have @xmath205 we can now state the following main result in this section , whose proof is in section [ sec : join ] .
[ thm : flag - main ] for @xmath105 and @xmath70 the flag @xmath69-polynomial for a general minkowski sum of simplices @xmath206 is given by @xmath207 in the formula above we interpret @xmath208 , since there is no @xmath209
. for each @xmath210 let @xmath211 for a concrete example we will parse through the very special case of theorem [ thm : flag - main ] when @xmath212 and @xmath213 . in this case @xmath214 and @xmath215 , @xmath216 and @xmath217 .
we will label the vertices of @xmath175 as follows : @xmath218 , @xmath219 , @xmath220 and @xmath221 .
a 2-chain of faces of @xmath175 will be given by @xmath222 where each face is presented as a string of the vertices whose convex combination forms the face .
as noted in the last example there are 25 2-chains of @xmath175 and they can be grouped together by what @xmath223 is as follows : @xmath224 for @xmath225 and @xmath226 we have @xmath227 @xmath228 , @xmath229 , and hence @xmath230 from this we obtain @xmath231 for @xmath232 , and @xmath233 we have @xmath234 @xmath235 , @xmath229 , and hence @xmath236 from this we obtain @xmath237 for @xmath238 we have @xmath239 @xmath240 , @xmath229 , and hence @xmath241 from this we obtain @xmath242 for @xmath232 , and @xmath243 we have @xmath244 @xmath228 , @xmath245 , and hence @xmath246 from this we obtain @xmath247 for @xmath248 , and @xmath249 we have @xmath250 @xmath235 , @xmath251 and hence @xmath252 from this we obtain @xmath253 finally , for @xmath225 , and @xmath254 we have @xmath255 @xmath228 , @xmath256 , and hence @xmath257 from this we obtain @xmath258 hence , collecting all the data for @xmath214 with @xmath215 , @xmath216 and @xmath217 we have by theorem [ thm : flag - main ] that @xmath259 where the sum is taken over the 25 different 2-chains @xmath203 given in ( [ eqn:2-chain ] ) .
substituting we obtain the following .
[ cor:2-chain ] if @xmath260 $ ] and @xmath215 , @xmath216 and @xmath217 then for @xmath214 we have @xmath261 where @xmath262 and @xmath263 for @xmath264 are as in observation [ obs : flag - poly - simp ] . example : for @xmath265 and @xmath266 we have @xmath267 and @xmath268 and we get in this case that @xmath269 something that can be seen by inspection , since the minkowski sum @xmath270 in this case has the same combinatorial structure as a cube where one side / edge has been contracted to a vertex .
the polytope @xmath270 has been drawn in figure [ fig : contracted - cube ] . where @xmath265 and @xmath266 .
] example : using the setup from corollary [ cor:2-chain ] above , we note that the total number of vertices of @xmath214 is the number of @xmath271 2-chains of faces , which is precisely the constant term in @xmath272 given by @xmath273 .
since @xmath274 and @xmath275 by observation [ obs : flag - poly - simp ] , we get @xmath276 for each @xmath9 , and hence by corollary [ cor:2-chain ] the number of vertices is in this case @xmath277 we will revisit this example in section [ sec : f - poly-2 ] .
in this section we describe how viewing the free join of two polytopes from a different angle , naturally generalizes to an action of a polytope on a set of polytopes .
we then present the @xmath69-flag polynomial for the resulting polytope and finally apply that to the minkowski sum of simplices . for polytopes @xmath6 and @xmath87 recall that the _ free join _ ( or just the _ join _ as it is sometimes called when there is no danger of ambiguity @xcite ) @xmath278 is a polytope that is obtained by positioning @xmath6 and @xmath87 into skew affine subspaces and then take the convex hull @xmath279 .
the faces of @xmath278 are given by ( i ) type @xmath6 : the faces of @xmath6 , ( ii ) type @xmath87 : the faces of @xmath87 and ( iii ) type @xmath280 : the disjoint union @xmath281 where @xmath199 is a face of @xmath6 and @xmath282 a face of @xmath87 .
the dimension of a face of type @xmath6 or @xmath87 is equal to its dimension as a face in @xmath6 or @xmath87 , but the dimension of a face @xmath281 of type @xmath280 is equal to @xmath283 because of the convex combination of points of @xmath199 and points of @xmath282 .
the three types of faces of @xmath278 is a partition of its faces .
the free join of a polytope @xmath6 and a singleton @xmath284 is called the _ pyramid over @xmath6 _ and denoted @xmath285 @xcite . for a chain @xmath286 of faces of @xmath278
there are five types of chains depending on the type of the starting and terminal face : types @xmath287 , @xmath288 , @xmath289 , @xmath290 and @xmath291 .
the @xmath69-flag polynomial for type @xmath287 and @xmath291 chains is @xmath132 and @xmath292 respectively .
the @xmath69-flag polynomial for a type @xmath289 chain is the same as for @xmath93 except for the fact that the dimension of each face @xmath281 is one more that that of @xmath293 .
hence , the @xmath69-flag polynomial must be shifted upward by one at each level of the chain . by definition of the flag polynomial , that is obtained by multiplying by the the first variable @xmath294 and we obtain @xmath295 . for a type @xmath288 chain , there are @xmath296 possible places in the chain where we encounter the first type @xmath280 face , say at step @xmath297 .
each such chain is determined by an @xmath69-chain of faces in @xmath6 ( the @xmath298s ) and an additional @xmath299-chain in @xmath87 ( the @xmath300s ) and has the form @xmath301 again , since the dimension of each @xmath302 is one more than that of @xmath303 , we must shift the dimension upward by one from and including at level @xmath9 , which is obtained by multiplying by @xmath304 .
hence , the @xmath69-flag polynomial for all chains of the form ( [ eqn:(p , pq ) ] ) is given by @xmath305 .
since the case for type @xmath290 chains is treated in the same way we obtain the following .
[ prp : p*q ] for polytopes @xmath6 and @xmath87 and a positive integer @xmath88 , the @xmath69-flag polynomial for the free join @xmath278 is given by @xmath306 note that for the empty polytope @xmath307 we have @xmath308 and @xmath309 .
also , for each @xmath69 we have @xmath310 . when @xmath87 is a singleton @xmath311 , then @xmath312 for each @xmath69 , and hence we obtain the following .
[ cor : flag - poly - pyr ] for a polytope @xmath6 and @xmath105 , the @xmath69-flag polynomial for the pyramid @xmath285 over @xmath6 is given by @xmath313 remark : for a @xmath4-polytope the @xmath4-flag - polynomial contains all the information of the flag vector and the @xmath38-index of @xmath6 with many redundancies ( i.e. the coefficient of different monomials of the flag polynomial will count the same number of strict chains ) . working with the most compact encoding of the flag vector , elegant formulae for computing the @xmath38-indices of the prism @xmath106 and the pyramid @xmath314 in terms of the @xmath38-index @xmath33 of @xmath6 and some linear derivations @xmath315
are given in @xcite .
but even with these nice recursions , it is hard to obtain explicit formulae for the @xmath38-index of specific polytopes , like the simplex @xmath316 , largely since derivations on non - commuting monomials in @xmath317 expand exponentially in repeated steps and do not simplify easily .
these recursions are generalized to a recursion to compute the @xmath38-indices for the free join and the cartesian product in @xcite .
we now generalize the idea of the free join and investigate the effect on the corresponding flag polynomial from proposition [ prp : p*q ] .
first note that a concrete geometric realization of the free join @xmath318 can be obtained by an action @xmath319 of the 1-simplex @xmath320 on the polytopes @xmath321 and @xmath322 .
here @xmath320 is the standard geometric 1-dimensional simplex @xmath323 } = \{(x_1,x_2 ) : x_1 , x_2 \geq 0 , x_1 + x_2 = 1\}$ ] , and we attach @xmath321 to one endvertex @xmath324 and @xmath322 to the other @xmath325 .
here we also choose an embedding of each @xmath326 in @xmath327 where @xmath328 , and the action is given by @xmath329 note that for a nonempty polytope @xmath87 we have @xmath330 .
however , if both @xmath321 and @xmath322 are nonempty , then every face of @xmath331 has the form @xmath332 where each @xmath333 is a face of @xmath326 or @xmath326 itself , and @xmath199 is a face of @xmath320 or @xmath320 itself .
each @xmath69-chain of faces of @xmath331 yields , by projection , an @xmath69-chain of faces @xmath334 of @xmath320 , and hence we have a partition of the @xmath69-chains of faces of @xmath331 where each part consists of the chains that yield the same chain of @xmath320 .
note that each proper face @xmath199 of @xmath320 has @xmath335 and hence each @xmath69-chain @xmath203 of faces of @xmath320 has @xmath336 .
writing @xmath321 for @xmath6 and @xmath322 for @xmath87 in proposition [ prp : p*q ] , the @xmath69-flag polynomial in proposition [ prp : p*q ] for @xmath318 equals that of is geometric realization @xmath331 , and we obtain @xmath337 conventions : ( i ) call an embedding of a polytope @xmath6 in @xmath338 _ simplicial _ if @xmath339}$ ] .
( ii ) by a _ full face _ of a polytope we will mean either a face or the entire polytope itself .
the repeated free @xmath340-join @xmath341 of polytopes @xmath326s is well defined since the operation @xmath342 is associative and commutative , and it has the empty polytope @xmath307 as the identity element .
this operation @xmath342 makes the collection @xmath343 of all polytopes @xmath87 a commutative semigroup @xmath344 .
[ obs : m - fold ] let @xmath345 be polytopes ( the empty polytope @xmath307 is allowed ) .
the @xmath340-fold free join @xmath346 is a polytope of dimension @xmath347 in which each full face has the the form @xmath348 where each @xmath333 is a full face of @xmath326 or the empty face . as @xmath349 , the dimension of each face
is given by @xmath350 .
one nice advantage of ( [ eqn : form2 ] ) over the formula in proposition [ prp : p*q ] is that it generalizes naturally in light of observation [ obs : m - fold ] .
this is since a geometric realization of the @xmath340-fold free join @xmath346 can , as above , be obtained by an action @xmath351 of the standard @xmath352-simplex @xmath353 on @xmath345 .
we attach each @xmath326 to an endvertex @xmath354 of @xmath353 and so @xmath355 here we do assume each @xmath326 to be a nonempty polytope , and so each full face of @xmath356 has the form @xmath357 where @xmath199 is a full face of @xmath353 and each @xmath333 is a full face of @xmath326 .
as @xmath356 is the geometric realization of @xmath346 , their flag-@xmath69 polynomials are identical , and we have the following corollary . [ cor : flag - m - fold ] for @xmath105 and simplicially embedded polytopes @xmath46 , the flag-@xmath69 polynomial for the @xmath340-fold free join @xmath346 is given by @xmath358 we now consider a generalization @xmath44 of @xmath356 where @xmath6 is now an arbitrary geometric polytope of dimension @xmath359 with @xmath360 . here
we assume @xmath361 to be simplicially embedded in @xmath45 and @xmath362 for each @xmath9 .
the polytope @xmath6 acts on simplicially embedded @xmath46 as @xmath363 and the action is given by the following .
[ def : p - acts ] for a simplicially embedded polytope @xmath361 with @xmath360 and simplicially embedded polytopes @xmath46 the action of @xmath6 is given by @xmath364 similar to observation [ obs : m - fold ] we have the following generalization .
[ prp : p - acts ] the dimension of the polytope @xmath44 from definition [ def : p - acts ] is given by @xmath365 and it is simplicially embedded .
each full face has the form @xmath357 where @xmath199 is a full face of @xmath6 and each @xmath333 is a full face of @xmath326 .
also @xmath366 for each @xmath9 let @xmath328 and @xmath367 a tuple of variables and @xmath368 .
each full face of @xmath369 is determined as the maximum of a functional @xmath370 . for a given @xmath371 and @xmath9 , let @xmath333 be the face of @xmath326 that maximizes @xmath372 and assume this maximum is @xmath373 .
let @xmath199 be the face of @xmath6 that is maximizes @xmath374 .
then the face that maximizes @xmath370 is @xmath357 .
remark : note that if @xmath375 then @xmath376 regardless of the face @xmath333 .
we now briefly discuss how we interpret @xmath369 when @xmath377 is allowed : let @xmath378 and assume that @xmath379 is the empty polytope .
as @xmath361 is simplicially embedded in @xmath45 , we can assume that @xmath380 and @xmath381 .
in this case let @xmath382 where @xmath383 is the projection obtained by omitting the @xmath9-th coordinate .
in this case we make the following definition .
@xmath384 by ( [ eqn : q = empty ] ) we can therefore assume that each @xmath326 in @xmath369 is a proper nonempty polytope .
we now state and prove our main result in this section , that the @xmath69-flag polynomial for @xmath44 has the same form as displayed in corollary [ cor : flag - m - fold ] .
[ thm : p - acts - flag ] for @xmath105 and simplicially embedded polytopes @xmath46 , the flag-@xmath69 polynomial of the polytope @xmath44 from definition [ def : p - acts ] is given by @xmath385 we will prove theorem [ thm : p - acts - flag ] in a few steps and use that to prove theorem [ thm : flag - main ] .
for simplicity we will write @xmath386 . assuming that @xmath387 for each @xmath9 we have linear maps @xmath388 and @xmath389 given by @xmath390 and @xmath391 respectively .
[ lmm : aa ] for faces of the polytope @xmath392 we have @xmath393 if and only if @xmath394 are subfaces in @xmath6 and @xmath395 are subfaces in @xmath326 for each @xmath396 .
we will prove the more involved direction .
assume that @xmath393 .
since each of the @xmath326 are simplicially embedded , we have @xmath397 . for @xmath396 , then by the definition of @xmath398 , we have @xmath399f_i$ ] where @xmath400 $ ] and @xmath401 .
( here @xmath402 is the projection onto the @xmath9-th coordinate ) .
since we have that @xmath394 , then for each @xmath9 we have @xmath403\subseteq [ a_i',b_i']$ ] or @xmath404 , or @xmath405f_i = \pi_{\tilde{\imath}}(a(\mathbf{f } ) ) \subseteq \pi_{\tilde{\imath}}(a'(\mathbf{f } ' ) ) = [ a_i',b_i']f_i'.\ ] ] for @xmath406 $ ] we have @xmath407f_i\subseteq [ a_i',b_i']f_i'$ ] . since @xmath408f_i : \sum_jx_j = c_i\}$ ] , then @xmath409f_i'\cap\{\tilde{x } : \sum_jx_j = c_i\ } = c_if_i'$ ] ( here @xmath326 is simplicially embedded and so are all of its faces ) .
therefore @xmath410 which completes the proof of the lemma .
by proposition [ prp : p - acts ] each @xmath69-chain of faces of @xmath392 has the form @xmath411 which by lemma [ lmm : aa ] firstly yields a unique @xmath69-chain @xmath412 of faces of @xmath6 , and hence @xmath413 . by a suitable permutation of the @xmath414
we may assume that @xmath415 ( where @xmath416 means that @xmath417 for all @xmath418 and @xmath419 ) .
secondly , by this mentioned assumption on the order on the @xmath420 , then for each fixed @xmath69-chain @xmath421 ( [ eqn : l - chain ] ) yields a young tableau ( or ferres diagram ) of boxes in @xmath422\times[m]$ ] such that for each @xmath423 $ ] the boxes numbered @xmath424 , where @xmath425 , contain the faces @xmath426 from @xmath427 that are indexed by numbers of @xmath420 .
let @xmath428 be the complete list of numbers from @xmath422 $ ] where @xmath429 for each @xmath430 .
hence , for each such @xmath431 we obtain @xmath432 chains of faces @xmath433 in @xmath434 , one for each @xmath435 .
conversely , every @xmath69-chain as in ( [ eqn : l - chain ] ) is presented by such a young tableau in which the horizontal sequence of boxes at hight @xmath436 yields a chain of faces in @xmath434 . this together with proposition [ prp : p - acts ] yields the following lemma .
[ lmm : key - chain ] for @xmath105 let @xmath421 be a fixed @xmath69-chain of faces of @xmath6 .
then the coefficient of @xmath437 in the polynomial given by @xmath438 is the number of @xmath69-chains @xmath439 of faces of @xmath44 with @xmath440 and @xmath441 . with the convention that a product with an empty index set equals one , ( [ eqn : key - poly ] ) can be written as @xmath442 theorem [ thm : p - acts - flag ] now follows directly from lemma [ lmm : key - chain ] .
now let @xmath443 and consider a minkowski sum @xmath181 of simplices as in ( [ eqn : minkssum ] ) . for @xmath182 $ ]
let @xmath185 be as before , the number of elements of @xmath183 : n_{\cal{f}}(j ) = n_{{\cal{h}}(k)}(i)\}$ ] and so @xmath187 . if @xmath444 and @xmath445 is a convex set , then @xmath446 .
hence , from this and by definitions [ def : master ] of the master polytope @xmath1 and [ def : p - acts ] we have that @xmath447},\delta_{[r_2]},\ldots,\delta_{[r_{2^k-1 } ] } ) \nonumber \\ & = & p(k)(\delta_{r_1 - 1},\delta_{r_2 - 1},\ldots,\delta_{r_{2^k-1}-1 } ) .
\label{eqn : minks = p(k)act}\end{aligned}\ ] ] theorem [ thm : flag - main ] now follows from ( [ eqn : minks = p(k)act ] ) and theorem [ thm : p - acts - flag ] .
in this section we consider two extremal examples of the minkowski sum of two simplices , where we apply the flag polynomial . in our first example
we obtain the @xmath3-polynomial of a general minkowski sum of @xmath0 standard simplices in terms of the master polytope and use that to compute the maximum number of @xmath4-faces of the minkowski sum of two simplices when @xmath212 . in our second example we compute the maximum discrepancy between the number of @xmath5-chains of faces of the minkowski sum and the minimum number of such @xmath5-chains of a simple polytope of the same dimension and on the same number of vertices . as before , @xmath181 will be the minkowski sum of @xmath0 simplices as in ( [ eqn : minkssum ] ) .
note that each face @xmath87 of @xmath181 can be viewed as a @xmath448-chain of faces .
hence , letting @xmath113 in theorem [ thm : flag - main ] we get @xmath449 , the set of faces of @xmath1 , and we obtain a generating function in terms of @xmath294 alone , which we denote by just @xmath450 in the following corollary .
we will now use corollary [ cor : main ] to investigate in more detail the minkowski sum of two simplices , the @xmath3-polynomial and then the maximum number of @xmath4-dimensional faces such a polytope can have . by corollary [ cor : main ]
we have in particular for @xmath212 and @xmath452 ( and therefore @xmath215 , @xmath216 and @xmath217 , ) that the @xmath3-polynomial of @xmath214 is given by @xmath453 hence , together with observation [ obs : f - poly - standard ] we have the following . [ cor : k=2 ]
if @xmath214 and @xmath454 , where @xmath215 , @xmath216 and @xmath217 , then the @xmath3-polynomial of @xmath181 is given by @xmath455 and hence for each @xmath456 the number @xmath457 of @xmath4-dimensional faces of @xmath181 is given by @xmath458 for given @xmath47 and @xmath456 we want to find the the maximum number @xmath462 of @xmath4-dimensional faces a minkowski sum @xmath214 of two simplices with @xmath463 can have .
that is , we want to determine @xmath464 to do this , it is understood that for a nonzero integer @xmath465 the map @xmath466 defined as @xmath467 is given by @xmath468 for @xmath469 and @xmath470 for @xmath471 . * if @xmath483 , then @xmath484 will take it minimum value if , and only if , @xmath485 for each @xmath486 . * if @xmath487 , then any @xmath488 summing up to @xmath489 will yield the minimum value @xmath490 of @xmath484 .
in particular those with @xmath485 for each @xmath486 .
clearly @xmath457 from corollary [ cor : k=2 ] is maximized iff @xmath491 is minimized .
note that @xmath492 .
considering each of the cases of @xmath493 we obtain the following theorem , in which @xmath494 $ ] denotes the _ nearest _ integer to the real number @xmath450 .
note that for a fixed @xmath495 , where @xmath496 are integers and @xmath497 is odd , then @xmath498 $ ] is the unique integer that minimizes @xmath499 .
[ thm : max-2d - faces ] if @xmath500 is the maximum number of @xmath4-faces in a minkowski sum @xmath270 of two simplices where @xmath501 as given in ( [ eqn : max-2d - faces ] ) , then @xmath502}{d+2 } - \binom{{\lceil ( 2r+1)/3 \rceil}}{d+2}.\ ] ] remarks : ( i ) in the case when @xmath503 we note that @xmath504 .
moreover , any minkowski sum @xmath214 such that @xmath505 , @xmath506 and @xmath507 will have @xmath508 in this case .
( ii ) otherwise , if @xmath509 , then the minkowski sum @xmath214 with @xmath508 is more `` unique '' in the following sense : consider the minkowski sum @xmath214 of two sets @xmath452 , where @xmath454 , @xmath215 , @xmath216 and @xmath217 as before . directly by corollary [ cor : k=2 ] we have that @xmath519 , the number of vertices of @xmath181 , is given by @xmath520 , something which we saw alternatively in the last example of section [ sec : flag - poly ] .
if @xmath511 and @xmath512 are disjoint , then @xmath521 ( see proposition 1.2 in @xcite for a more general result ) which is a simple polytope by the simplicity of any simplex . hence we will assume that @xmath511 and @xmath512 have a nonempty intersection so @xmath522 , in which case the dimension of @xmath181 is precisely @xmath50 ( see @xcite for more detailed discussion ) .
for a polynomial @xmath523 , denote by @xmath524f(\tilde{x})$ ] the coefficient for @xmath525 in @xmath523 . note that the number of @xmath5-chains in @xmath214 is then given by @xmath526{\tilde{f}}^2_{p_{\cal{f}}}(x_1,x_2)$ ] .
note that when @xmath511 and @xmath512 are not disjoint , then each vertex of @xmath181 is contained in a minimum of @xmath527 @xmath4-faces since the degree of each vertex is at least @xmath50 and @xmath4 of these yield a face .
[ obs:(0,r-2)simple ] for @xmath214 where @xmath511 and @xmath512 have a nonempty intersection , @xmath181 is simple iff @xmath528{\tilde{f}}^2_{p_{\cal{f}}}(x_1,x_2 ) = \binom{r-1}{d}f_0 = \binom{r-1}{d}(r_1r_2 + r_1r_3 + r_2r_3 + r_1).\ ] ] since @xmath526{\tilde{f}}^2_{p_{\cal{f}}}(x_1,x_2 ) = [ x_2^d]{\tilde{f}}^2_{p_{\cal{f}}}(0,x_2)$ ] , and in addition @xmath529 and @xmath530 , we obtain by corollary [ cor:2-chain ] that @xmath531{\tilde{f}}^2_{p_{\cal{f}}}(0,x ) \\ & = & [ x^d]\left ( ( r_1r_2 + r_1r_3 + r_2r_3 + r_1)(x+1)^{r-1 } + r_2r_3(x+1)^{r-2 } - r_2r_3(x+1)^{r_2+r_3 - 1}\right),\end{aligned}\ ] ] and hence we have the following . the other question _ `` when is @xmath214 farthest from being simple ? '' _ is more interesting . to make this question precise and address it properly ,
we define the _ discrepancy _ @xmath537 as the difference between @xmath532 and the minimum number @xmath538 of @xmath5-chains in simple polytope of the same dimension and on the same number @xmath520 of vertices , @xmath539 the discrepancy @xmath537 can be thought of as a measure ( in terms of @xmath4 ) on how far from simple the polytope @xmath181 is .
hence , for a fixed @xmath47 we like to determine what kind of minkowski sum @xmath181 of two simplices will yield the largest discrepancy .
this is only of interest if @xmath540 and @xmath522 , @xmath541 since @xmath522 then @xmath542 , so to determine @xmath543 it suffices to compute @xmath544 where @xmath545 is given by @xmath546 considering @xmath545 in integers @xmath450 and @xmath547 on each line @xmath548 where @xmath465 is a constant , then clearly the maximum of @xmath545 is reached on each of these lines when @xmath549 . in particular , when @xmath550 the maximum of @xmath545 is @xmath551 . for the other range @xmath552 , the more interesting one ( since we like to consider @xmath4 fixed and @xmath98 much larger than @xmath4 ) , we can by symmetry assume that @xmath553 and write @xmath554 where @xmath555 .
then by expressing @xmath556 in terms of @xmath557 , we have that @xmath558 where @xmath559 and @xmath560 . for the moment we consider real @xmath561 in the mentioned interval ,
so by the intermediate value theorem for real continuous functions , @xmath562 is obtained at @xmath563 where @xmath564 and the difference @xmath565 , or @xmath566 we will first estimate asymptotically the solution in @xmath561 to ( [ eqn : dg=0 ] ) , and hence also @xmath567 $ ] , in terms of @xmath4 and @xmath98 and interpret the corresponding @xmath558 . note
that ( [ eqn : dg=0 ] ) can be rewritten as @xmath568 considering solely the asymptotic values as @xmath98 tends to infinity , the solution @xmath561 of ( [ eqn : prodh ] ) must also tend to infinity .
hence , as the right hand side ( rhs ) of ( [ eqn : prodh ] ) tends to @xmath569 as @xmath98 ( and hence @xmath561 as well ) goes to infinity , and the left hand side ( lhs ) of ( [ eqn : prodh ] ) is asymptotically @xmath570 as @xmath98 tends to infinity , we have for large @xmath98 that @xmath571{2/(d+2)}r$ ] .
hence , @xmath562 is obtained at @xmath563 where @xmath572{2/(d+2)}r$ ] as @xmath98 tends to infinity . from this
we deduce the following asymptotic result .
[ prp : discrep - asymp ] if a minkowski sum @xmath214 where @xmath501 and @xmath522 has the maximum discrepancy @xmath573 between the number of @xmath5-chains of faces of @xmath181 and the minimum number @xmath538 of @xmath5-chains in a simple polytope of the same dimension and with the same number @xmath574 of vertices , then for all fixed @xmath4 and @xmath575 we have @xmath576{\frac{2}{d+2}}\right)r , \\ r_i & \sim & \frac{1}{2}\sqrt[d]{\frac{2}{d+2}}r , \end{aligned}\ ] ] as @xmath98 tends to infinity , and hence @xmath577 we will now determine exactly which @xmath578 , @xmath579 and @xmath580 will yield the maximum discrepancy . in order to do that we determine concrete and tight upper and lower bounds for the solution @xmath561 of ( [ eqn : prodh ] ) to pinpoint the corresponding integer @xmath581 at which the maximum discrepancy @xmath582 for the minkowski sum @xmath214 is attained .
the remainder of this section will be spent computing this mentioned @xmath581 and interpreting the corresponding discrepancy for the minkowski sum .
first we centralize the lhs of ( [ eqn : prodh ] ) by letting @xmath583 and @xmath584 . in this way
the lhs of ( [ eqn : prodh ] ) becomes @xmath585 since @xmath586 are fixed then @xmath587 is increasing and the rhs of ( [ eqn : prodh ] ) is at most @xmath569 , we have that @xmath588 and therefore we obtain a concrete lower bound @xmath589 for @xmath590 , and hence also for @xmath561 , given by @xmath591{\frac{2}{d+2}}(r-2 ) + \left(1-\sqrt[d]{\frac{2}{d+2}}\right)\frac{d-1}{2}.\ ] ] incidentally , we will use this lower bound @xmath589 from ( [ eqn : zlower ] ) to obtain an upper bound for @xmath561 . by ( [ eqn : prodh ] ) ,
its lhs expressed in ( [ eqn : lhs ] ) is at least @xmath592 .
since @xmath593 is increasing , the rhs of ( [ eqn : prodh ] ) and hence also ( [ eqn : lhs ] ) is greater than @xmath594 .
regardless of the parity of @xmath4 , the expression in ( [ eqn : lhs ] ) is less than @xmath595 and hence we obtain an upper bound for @xmath590 , and hence also @xmath561 , determined by @xmath596 or @xmath597 where @xmath598{q(l(d;r))}$ ] . by the mean value theorem for the function @xmath599{t}$ ] , there is an @xmath600 = \left[\frac{2}{d+2 } , c^d \right]$ ] such that @xmath601{\frac{2}{d+2 } } = \frac{1}{{\alpha}^{1 - 1/d}(d+2)((d+2)l(d;r)+1 ) } < \frac{1}{2(d+2)(r-2)},\ ] ] and since @xmath602 for positive @xmath603 , ( [ eqn : zupper ] ) implies @xmath604 by the bounds of @xmath465 and definition of @xmath589 we have further that @xmath605{\frac{2}{d+2 } } + \frac{1}{2(d+2)(r-2)}\right)r + \frac{d-1}{2 } \\ & = & l(d;r ) + \frac{r}{2(d+2)(r-2 ) } \\ & < & l(d;r ) + \frac{1}{2(d+2)},\end{aligned}\ ] ] and hence ( [ eqn : zzupper ] ) implies @xmath606 since @xmath607 is decreasing we have further @xmath608{\frac{d+2}{2 } } + \sqrt[d]{\frac{2}{d+2 } } < \frac{2\log(d+2)}{d},\ ] ] and so ( [ eqn : z3upper ] ) implies the following . note : ( i ) by proposition [ prp : tight ] we have for any @xmath611 there is an integer @xmath612 and a function @xmath613 such that for any @xmath614 and @xmath615 then @xmath616 .
( ii ) also , as @xmath617{2/(d+2)})(d-1)/2 = \log(d/2)/2 + o(\log^2d / d)$ ] we immediately obtain that the real number solution @xmath561 to ( [ eqn : dg=0 ] ) satisfies @xmath618{\frac{2}{d+2}}(r-2 ) + \frac{\log(d/2)}{2 } + o\left(\frac{\log^2(d)}{d}\right ) + o\left(\frac{d\log(d)}{r}\right),\ ] ] for large @xmath4 and @xmath98 with @xmath610 .
[ thm : discrep ] if a minkowski sum @xmath214 where @xmath501 and @xmath522 has the maximum discrepancy @xmath573 between the number of @xmath5-chains of faces of @xmath181 and the minimum number @xmath538 of @xmath5-chains in a simple polytope of the same dimension and with the same number @xmath574 of vertices , then we have @xmath619 and @xmath620 where @xmath589 is as defined in ( [ eqn : zlower ] ) .
consider the upper bound of @xmath626 in theorem [ thm : discrep ] .
for every fixed @xmath627 we see that if @xmath628 then the real interval in which the integer @xmath626 is contained in is of length strictly less than 2 and hence there are at most two possibilities for the integer @xmath626 .
observation [ obs : dense ] now implies there are infinitely many @xmath628 for which the integer @xmath626 is uniquely determined by theorem [ thm : discrep ] . in this case we know exactly when the maximum discrepancy occurs .
[ cor : exact ] with the setup from theorem [ thm : discrep ] , if @xmath573 , then for all fixed @xmath627 , @xmath628 and @xmath575 , we have @xmath629 to hold , in which case we have further there are infinitely many @xmath98 such that the maximum discrepancy @xmath543 occurs exactly at @xmath630 remark : although @xmath626 is for all @xmath631 uniquely determined up to parity , we have by the uniform distribution property of @xmath632 in observation [ obs : dense ] that for every fixed @xmath4 , _ `` most '' _
@xmath631 will determined @xmath578 , @xmath579 and @xmath580 uniquely as stated in the latter display in corollary [ cor : exact ] .
that is to say , the probability of picking @xmath98 that uniquely determines @xmath578 , @xmath579 and @xmath580 can be made arbitrarily high by choosing large enough @xmath4 and large enough @xmath631 .
the author would like to thank james f. lawrence for helpful discussions regarding the theory of polytopes in general .
also , thanks to bernd sturmfels for introducing this investigation of minkowski sum of standard simplices .
last but not least , sincere thanks to walter morris for his encouragement , interest and helpful discussions .
figure [ fig : contracted - cube ] is the same that appears in @xcite and was drawn by walter morris . | for a polytope we define the _ flag polynomial _ , a polynomial in commuting variables related to the well - known flag vector and describe how to express the the flag polynomial of the minkowski sum of @xmath0 standard simplices in a direct and canonical way in terms of the _ @xmath0-th master polytope _
@xmath1 where @xmath2 .
the flag polynomial facilitates many direct computations . to demonstrate this we provide two examples
; we first derive a formula for the @xmath3-polynomial and the maximum number of @xmath4-dimensional faces of the minkowski sum of two simplices .
we then compute the maximum discrepancy between the number of @xmath5-chains of faces of a minkowski sum of two simplices and the number of such chains of faces of a simple polytope of the same dimension and on the same number of vertices . *
2010 msc : * 05a15 , 05a16 , 52b05 , 52b11 .
* keywords : * polytope , minkowski sum , master polytope , flag polynomial , @xmath3-polynomial . |
it is desired that most events to be safe , stable or rather predictable . identifying an unusual event
is a typical and vital topic in different fields , such as intrusion detection in cyber security , fraud detection for credit cards , insurance or health care and fault detection in safety critical systems @xcite .
we may call this unusual event as outlier or abnormality which is different from a usual event .
outlier is usually minor in a group of events / data while inlier is the majority .
od refers to detect any abnormal element in data which is not consistent with an expected behavior @xcite .
a good od method should be accurate to detect outlier and less erroneous judgment on inlier data .
many od methods @xcite have been developed in recent years . in @xcite,@xcite ,
nearest neighbor approach was proposed to consider the distance or the similarity between two data instances .
the assumption behind the approach is that inliers in data should be dense and outlier(s ) is far from these dense inliers .
nb classifiers were developed in the 1950s .
the nb classifiers are constructed based on bayes theorem , and are widely used for text categorization with superior performance @xcite .
another popular application is spam filtering among numerical emails @xcite .
the kernel smooth nb method in this paper supposes the traffic data could be modeled by kernel distributions so that the bayes theorem is applied .
in contrast , the gmm method presumed a mixture of gaussian distributions and confidence region is employed .
gmm , firstly introduced by aitkin and wilson ( 1980 ) @xcite , is an appropriate way of handle data with multiple outliers @xcite .
the gmm method is useful in dealing with speech recognition @xcite . for traffic data , detecting anomaly traffic event would be better to deal with the traffic problems
this research aims at detecting outliers for a large - scale traffic database from hong kong .
the original video was taken at a four - arm junction as shown in fig .
[ fig1:a ] .
the 4-arm junction can be expressed as an ideal map like fig .
[ fig1:b ] . in total , the traffic data for 31 days was recorded with two sessions per day : am ( 07:00 - 10:00 ) and pm ( 17:00 - 20:00 ) . for each session , the original video data are dividing into 19 spatial temporal ( st ) signals .
[ fig1:c ] demonstrates an sample of 4 normal st signals .
these st signals have different number of traffic cycles because smoother traffic flow would lead to a shorter traffic cycle and result in more traffic cycles in one session .
also , the st signals in each direction suffer a high degree of similarities ( fig .
[ fig1:d ] ) among each other or across signals in different traffic direction .
therefore , it requires a truncation process ( to standardize the length of a signal s cycles ) and a signal domain transformation ( to remedy the signal similarities ) .
herein , each st signal was processed by principal component analysis ( pca ) @xcite to reduce the dimension of signal representation .
then , an od method was carried out based on the pca - processed ( x , y)-coordinates plane . 0.5 0.5 0.5 0.5 in this paper , a detail investigation of nb and gmm classifiers in large - scale traffic data
is carried out , for these two methods are newly applied in traffic od .
experimental results demonstrate that the nb and gmm methods can achieve 93.78% and 94.50% detection success rates ( dsrs ) , respectively . these performance is comparable to previous evaluation on other od methods @xcite , @xcite including gaussian mixture model ( 80.86% ) , one - class svm ( 59.27% ) , s - estimator ( 76.20% ) and kernel density estimation ( 95.20% ) .
the organization of this paper is as follows : section ii gives a review of related work about od .
sections iii and iv present details of two proposed od methods and their experimental results , and section v concludes this paper .
the popular od methods include statistical , nearest neighborhood , spectral approaches and learning based approaches .
details about these methods are given as follows .
statistical approach is one of the earliest approaches in od .
it assumes that all inliers occur in a high probability region of this distribution model when outliers deviate strongly from the inliers .
statistical od methods can perform parametric techniques like gmm or regression , or non - parametric techniques like histogram based or kernel function based .
this approach would perform well if data distribution is assumed well . however , if the data distribution assumption is false , the result would be far away from the correct situation @xcite , @xcite .
nearest neighbor approach considers the distance or the similarity between two data instances .
the assumption behind the approach is that inliers in data should be dense and outlier(s ) is far from these dense inliers @xcite , @xcite .
this approach can be easily realized in an unsupervised way that could require high computational complexity . as a result , it is hard to deal with very complex data @xcite .
there are two major variant methods related to this approach , density measurement based methods , such as local outlier factor ( lof ) @xcite , influenced outlierness ( inflo ) @xcite and local outlier correlation integral ( loci ) @xcite , or distance measurement based methods , like mahalanobis distance @xcite , db(@xmath0)-outliers , outlier scoring based on k - nn distances @xcite , resolution - based outlier factor ( rof ) @xcite .
spectral approach suggests that inliers and outliers could appear significantly different in a spectral domain .
the approach detects outliers by embedding data into a lower dimensional subspace .
the approach is useless if inliers and outliers in data are not separable in the lower dimensional subspace .
in addition , it requires high computational complexity @xcite .
learning approach is using various training methodologies @xcite ( or so - called trained machines ) to train the input data .
the related od methods are neural networks , nave bayesian network , support vector machines ( svm ) .
the assumption is that inliers and outliers in data can be easily distinguished by the trained machines .
therefore , the machines can test and classify a test instance into either an inlier group or an outlier group in od @xcite .
there are many algorithms that can be used in multi - learning class approach , and they have a faster testing phase than other approach .
however , accurate labeling for various normal classes is often impossible @xcite . in this paper , an investigation of nb classifier and gmm classifier to model the pca - processed ( x , y)-data and their performance for od in traffic data
would be carried out .
nb classifier was developed in 1950s for text retrieval @xcite .
it is also widely used for spam filtering among numerical emails @xcite .
nb classifier @xcite is developed based on bayes theorem which takes the form of @xmath1 based on the probability p(e),p(h ) , and the conditional probability p(eh ) , the posteriori probability p(he ) , which denotes the possibility of event h conditioned on an occurred event e , can be obtained .
naive bayes classifiers are based on the information of the training data , and then determined the highest possible class of testing data from their information .
kernel distribution is a non - parametric distribution to estimate each training point as some independent distributions into the whole distribution @xcite .
@xmath2 kernel distribution is estimated by sample data points with a kernel density .
the commonly used kernels are listed as follows : + box kernel : @xmath3 triangle kernel : @xmath4 epanechnikov kernel : @xmath5 normal kernel : @xmath6 where _ x _ is replaced by _
y _ for y - axis kernel .
kernel smoothing nb ( ksnb ) classifier is a trained kernel distribution based classifier .
different kernels with different widths for each predictor or class are available in ksnb classifier .
we use the trained ksnb classifiers to detect outliers , where the classifiers would automatically set a bandwidth value for each feature and class which is optimal for a gaussian model .
then , a corresponding region from contours surrounding various kernel distributions is formed an inlier region e , therefore any data out of that region is classified as an outlier . in this method , box ,
triangle , epanechnikov and normal kernels are used for od .
the od procedures based on ksnb classifier are given as follows : + fig .
[ fignbflowchart ] illustrates the flowchart of the proposed nb based method , which includes a training stage and a testing stage .
the details are as below .
[ [ a .- training - stage ] ] a. training stage + + + + + + + + + + + + + + + + + step 1 : feed training data ( all inliers ) into nb classifiers .
+ step 2 : fit training data into the specified kernel distributions defined in eqns .
( 3 - 6 ) .
+ step 3 : combine each kernel distribution into a whole distribution .
+ step 4 : set a bandwidth ( bib ) of the inlier region automatically from the kernel distribution .
[ [ b .- testing - stage ] ] b. testing stage + + + + + + + + + + + + + + + + step 1 : input testing data into the trained nb classifiers .
+ step 2 : data label as outlier if the data is out of the inlier region .
+ step 3 : output result .
+ in addition , * algorithm [ nbalg ] * is provided below .
the inliers data set @xmath7 and testing data set @xmath8 @xmath9 @xmath10 every data point @xmath11 @xmath12 @xmath13 @xmath14 @xmath15 @xmath16 @xmath17 in @xmath18 @xmath19 @xmath20 @xmath13 experimental results based on the kernel smoothing nb method are listed in table [ diffnbkernelres ] .
accuracy ( acc . ) , positive predictive value ( ppv ) , negative predictive value ( npv ) , sensitivity ( sen . ) , false positive rate ( fpr ) are employed as measurement metrics and their definitions can be referred to @xcite . from the results , we can see that overall accuracy among all kernels is almost higher than * 90%*. overall npv among all kernels is higher than * 95% .
* .performance of the nb method among different kernels . [ cols="^,^,^,^,^,^",options="header " , ] [ diffgmm ]
in this paper , we present two od methods , kernel smoothing nb and gmm , to detect outliers in large - scale traffic data .
the kernel smoothing nb method utilizes the trained kernel smoothing nb classifiers to detect outliers , in which the classifiers set the best bandwidth value with the inlier region .
any data points out of that region are classified as outliers . in the gmm method ,
the rectangular confidence region is formed in bonferroni adjustment way .
true @xmath21 significant level region of the gmm can be constructed as the inlier region for the more accuracy .
experimental results show that the two algorithms can achieve pleasing detection accuracies compared with the od methods in our previous studies @xcite , @xcite , including gaussian mixture model , one - class svm , s - estimator and kernel density estimation .
this research is supported by hong kong rgc grf : 12201814 and hkbu frg2/14 - 15/075 .
g.h . john and p.
langley , estimating continuous distributions in bayesian classifiers , " _ in : hanks , p. b. eds . ,
11th conf . uncertainty in artificial intelligence _ , morgan kaufmann , san francisco , ca , pp .
338 - 345 , 1995 .
ngan , n.h.c .
yung and a.g.o .
yeh , a comparative study of outlier detection for large - scale traffic data by one - class svm and kernel density estimation , " _
is&t / spie electronic imaging _
, 94050i-94050i-10 , 2015 . | it is meaningful to detect outliers in traffic data for traffic management .
however , this is a massive task for people from large - scale database to distinguish outliers . in this paper
, we present two methods : kernel smoothing nave bayes ( nb ) method and gaussian mixture model ( gmm ) method to automatically detect any hardware errors as well as abnormal traffic events in traffic data collected at a four - arm junction in hong kong .
traffic data was recorded in a video format , and converted to spatial - temporal ( st ) traffic signals by statistics .
the st signals are then projected to a two - dimensional ( 2d ) ( x , y)-coordinate plane by principal component analysis ( pca ) for dimension reduction .
we assume that inlier data are normal distributed .
as such , the nb and gmm methods are successfully applied in od ( outlier detection ) for traffic data .
the kernel smooth nb method assumes the existence of kernel distributions in traffic data and uses bayes theorem to perform od .
in contrast , the gmm method believes the traffic data is formed by the mixture of gaussian distributions and exploits confidence region for od .
this paper would address the modeling of each method and evaluate their respective performances .
experimental results show that the nb algorithm with triangle kernel and gmm method achieve up to * 93.78% * and * 94.50% *
accuracies , respectively . |
quasar absorption line systems are an extremely useful means of statistically constraining the various scenarios of metal enrichment , inflow and outflow , ionization conditions , kinematics , and gas structure within galaxies and the igm .
various absorption line studies have concluded that these systems are cosmologically distributed @xcite , and numerous subsequent studies have identified specific galaxies associated with absorption ( see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . for the following discussion
, we adopt the terms `` weak '' , `` intermediate '' , and `` strong '' to refer to absorbers having @xmath2 @xmath3 , @xmath16 @xmath17 , and @xmath10 , respectively , where @xmath18 is the rest frame equivalent width of the @xmath192796 transition . -selected gas probes a wide range of column density environments .
weak absorption in particular , which samples optically thin gas over a large span of cosmic time , has been proposed to sample dwarf or lsb galaxies as well as the igm @xcite .
however , there are instances in which weak is identified in the circumgalactic medium of `` normal '' bright galaxies @xcite .
@xcite , using an ionization model , concluded that filamentary and sheetlike igm structures host at least a portion of weak absorption .
moreover , @xcite argue that @xmath20 absorbers likely reside in the igm .
photoionization modeling of weak systems with associated led @xcite to argue for a scenario of a shell geometry as might be expected for supernova remnants or high velocity clouds moving in a hot corona .
better insight into the nature of the structures selected by weak absorbers remains elusive and partially motivates this study .
measurements of the redshift path density @xmath4 of weak absorbers place important constraints on @xmath21 , the product of their number density and cross section @xcite . in the same vein ,
probing discrepancies in redshift path densities of absorption in various equivalent width ranges based on differences in the background sources may reveal information about these absorbing structures . despite the evidence indicating that absorption line systems are primarily associated with intervening gas as opposed to the background source itself ,
several studies have nevertheless revealed major discrepancies , depending on the nature of the background source , in the incidence of these absorbers per unit redshift .
@xcite , in a study of strong absorbers in bl lac objects , observed a redshift path density of 45 times greater than that expected from quasar surveys .
@xcite compared quasar and grb sightlines and found the latter to have a factor of @xmath22 excess in the redshift path density of strong absorbers .
a similar study by @xcite found a factor of @xmath23 excess of strong absorbers , but no excess of intermediate absorbers , in the grb sample .
@xcite reported a factor of @xmath23 excess of both strong and intermediate absorbers in blazar versus quasar sightlines .
it should be noted , however , that in that study the authors did find marginal statistical evidence in the strong sample of an excess occurring nearer the blazar ( even after excluding absorbers having velocity separations of less than 5,000 from the emission redshift ) .
finally , in a study analyzing absorbers down to @xmath24 in grb sightlines , @xcite found a factor of @xmath25 overabundance of strong absorbers and a @xmath26 _ reduction _ for @xmath27 compared to studies of quasar sightlines .
the latter discrepancy , however , was deemed insignificant since the results were consistent at the @xmath28 confidence level .
@xcite have so far presented the only study to compare redshift path density differences in weak absorbers based on different types of background sources . in this paper
we present the largest study of weak absorption to date , and for the first time provide a parameterized fit to the redshift path density evolution .
we also present our findings of differential absorber redshift path density , based on the absolute magnitude of the background quasar , for weak , intermediate , and strong systems . in [ sec : data ] we present our data , and in [ sec : results ] our results .
we discuss plausible interpretations in [ sec : discuss ] , and conclude in [ sec : conclude ] .
the cosmological parameters @xmath29 mpc@xmath30 , @xmath31 , and @xmath32 are adopted throughout .
we have searched 252 keck / hires and uves / vlt quasar spectra for doublet absorption .
all systems were objectively identified using the methods of @xcite and @xcite .
further details are provided in @xcite and @xcite .
the search space omitted redshifts within 5,000 of the quasar emission redshift , @xmath33 , and blueward of the emission of the quasar .
a total of 422 absorbers comprise our sample .
we divided the absorbers into three subsamples using historically motivated weak @xcite , intermediate @xcite , and strong @xcite equivalent width ranges . in order to determine whether our absorber subsamples are consistent with being cosmologically distributed along the lines of sight to the quasars
, we performed the second test of @xcite . performing the kolmogorov - smirnov ( ks ) test for each of our equivalent
width ranges , we could not rule out that their distributions are consistent with being cosmological parameter of each system , where @xmath34 , with @xmath35 representing a doublet at the minimum observed redshift included in the search , and @xmath36 representing a doublet at the maximum observed redshift .
following the formalism of @xcite , the sensitivity function @xmath37 of the survey was then calculated .
this is a measure of the number of lines of sight in the survey in which a system of a given minimum @xmath18 could have been detected at each value of @xmath38 .
the @xmath38 distributions of each absorber sample were then statistically compared to @xmath37 . ] .
this result is in agreement with earlier studies @xcite .
since none of the quasars in our sample were observed with _ a priori _ knowledge of weak absorption , our survey is unbiased for this population .
however , since the quasar sample is drawn from a broad range of targeted science programs , there is the possibility of bias in the intermediate and strong absorbers , which are known to sometimes be associated with dlas @xcite , or which in some cases were already known due to previous lower resolution surveys . to examine whether our intermediate and strong absorber samples are consistent with an unbiased population , we compared our measured equivalent width distributions , @xmath39 , where @xmath40 is the characteristic @xmath18 , to the distribution measured by @xcite . using the ks test for the three redshift bins measured by @xcite
, we obtained @xmath41 ( @xmath42 ) , @xmath43 ( @xmath44 ) and @xmath45 ( @xmath46 ) , respectively .
for the full redshift range encompassing all three bins , we obtained @xmath47 . even in the case of the lowest value of @xmath48 ,
corresponding to the lowest redshift range , the two populations are not inconsistent with each other to even a @xmath49 level .
we thus proceed under the assumption that our sample of absorbers is a fair sample . for our analysis in [ sec : results ] , we obtained absolute @xmath50-band and apparent magnitudes ( primarily @xmath50 , @xmath51 , and @xmath52 ) of the quasars .
the majority were obtained from @xcite ; 16 were obtained from the nasa / ipac extragalactic database ; and for two quasars , the magnitudes could not be determined so these lines of sight and their absorbers were omitted from analysis for which these quantities were required .
following the formalism of @xcite , modified to account for the doublet ratio @xcite , we calculated the number of absorbers per unit redshift , @xmath4 . for our full redshift range , @xmath53 , the cumulative redshift path is @xmath54 and is @xmath55% complete to a 5 @xmath12 equivalent width sensitivity of @xmath56 and @xmath57% complete to a 5 @xmath12 equivalent width sensitivity of @xmath58 . over the redshift range @xmath59 , the extent of a study by @xcite ,
we have @xmath60 , compared to their 70 ; and for @xmath61 , the extent of a study by @xcite , we have @xmath62 , compared to their 17 .
these represent the largest two previous weak surveys .
our larger cumulative redshift path reflects the larger number of lines of sight included in our survey . @xcite
surveyed 26 hires quasar spectra and found 30 weak systems , while @xcite surveyed 81 uves quasar spectra and found 112 weak systems . in our survey
we identified 188 weak systems .
we calculated @xmath4 for weak systems in four redshift bins in order to facilitate comparison with the works of @xcite and @xcite ; the bins and results are shown in table [ tab : dndzcomp ] .
ccccc crcv99 & @xmath63 & @xmath64 & @xmath65 & @xmath66 + nmck07 & @xmath67 & @xmath68 & @xmath69 & @xmath70 + this survey & @xmath71 & @xmath72 & @xmath73 & @xmath74 + all three studies obtained different results , with the trend being that the @xmath4 values have decreased with larger survey size .
all three used the identical code ( search , * ? ? ?
* ) for line and candidate identification , and all three ostensibly used the same algorithms in determining the redshift paths for each system . to test for possible differences in the calculations , we ran our code used for this survey on the identical spectra and set of systems used by @xcite and compared their @xmath75 and @xmath4 results with those of the reproduced study .
the result was that the @xmath75 values from the @xcite study were @xmath76 larger than our reproduced values from their data , and the @xmath4 values from the original study were thereby lowered compared to our reproduced study .
this suggests that our redshift path calculations are more conservative than those of @xcite .
if our redshift paths had been calculated exactly as theirs , the @xmath4 result of this survey would presumably have been lower , further widening the discrepancy with previous works .
a similar duplication of the survey of @xcite could not be performed because a significant number of the authors quasar spectra were unavailable to us . in an attempt to find differences in the quasar samples that could possibly lead to the discrepant @xmath4 results among the three works , we investigated the quasar apparent magnitude distributions .
the ks test was performed between all possible pairs of the three surveys .
our survey and that of @xcite exhibit remarkably similar distributions ; it could not be ruled out to greater than a @xmath77 confidence level that their apparent magnitudes had been drawn from the same population , and their median values were both 17.5 .
in contrast , the @xcite survey differed from each of the other two to a confidence level of @xmath78 and had a median apparent magnitude of 16.3 .
a uniform set of apparent magnitudes in the same band was not available , however , making these comparisons uncertain .
the smaller @xcite survey was undoubtedly overall biased toward brighter quasars .
we likewise investigated the quasar _ absolute _
@xmath50-band magnitude distributions , but when each survey was tested against the other two , it could not be ruled out to even a @xmath77 confidence level that their distributions had been drawn from the same population . similarly , no substantial differences were found in the overall median quasar absolute magnitude of the three surveys : @xmath79 , @xmath80 , and @xmath81 for @xcite , @xcite , and this study , respectively .
the reasons behind the different @xmath4 results among the three works remain unclear ; future studies will hopefully resolve the weak puzzle .
it may be that the inclusion or exclusion of weak systems very close to the limiting equivalent width may play a role ( anand narayanan , private communication ) , as well as differences in the codes used , since the calculation of @xmath75 is extremely sensitive for the weakest systems .
the number of absorbers per unit redshift is the product of the proper number density of absorbers and their proper geometric cross section . in the standard cosmological model , the no - evolution expectation ( nee ) for the redshift number density
can be parameterized as : @xmath82_{\hbox{\tiny nee } } = { c\over h_{0}}n_0\sigma_0{(1+z)^{2}\over\sqrt{\omega_{m}(1+z)^{3}+\omega_{\lambda}}}\ ] ] where @xmath83 is the mean comoving number density of absorbers and @xmath84 is the mean comoving geometric absorber cross section .
as shown in figure [ fig : residual]@xmath85 , our observed @xmath4 departs strikingly from the nee ( dashed curve ) , which was normalized to the mean @xmath86 , 1.098 , and to the overall @xmath4 , 0.83 , of our weak sample over the redshift range @xmath53 .
the general behavior of @xmath4 is in agreement with @xcite in that it peaks between @xmath87 and then decreases toward higher redshift .
if the product @xmath88 varies as a function of redshift , @xmath4 may depart from the no - evolution expectation .
the quantity @xmath89 can be written as @xmath90 where @xmath91 is a nonnegative function that parameterizes the evolution of @xmath4 .
figure [ fig : residual]@xmath92 depicts our weak @xmath4 result divided by the nee .
the data clearly motivate a linear fit ; this was achieved using a function of the form @xmath93 where @xmath94 is the slope and @xmath95 is the function normalization .
the result , @xmath96 and @xmath97 , is shown as a solid line in figure [ fig : residual]@xmath92 .
finally , the fit to @xmath91 was multiplied by the nee , yielding a fit to the @xmath4 result in figure [ fig : residual]@xmath85 ( solid curve ) that can be expressed as @xmath98\over \sqrt{\omega_{m}(1+z)^{3}+\omega_{\lambda } } } \qquad ( z<2.74 ) \ , .\ ] ] weak absorbers seem to vanish at high redshift ; this result is discussed in [ sec : discuss ] . motivated by studies that found differences in @xmath4 results based on background object as discussed in [ sec : intro ] , as well as our finding of differing apparent magnitude distributions among the weak studies as discussed in [ sec : weakdndz ]
, we attempted to discern some intrinsic difference that might affect the observed @xmath4 . using the absolute @xmath50-band quasar magnitudes of our survey , which had a range of @xmath99
, we divided our quasars into `` bright '' and `` faint '' subsamples according to the median , @xmath100 . the bright subsample has a median of @xmath101 , and the faint subsample has a median of @xmath102 .
lccccccccc & @xmath103&@xmath104 & @xmath105&@xmath106 { \tiny bright } & { \tiny $ 1.040 \pm 0.016$}&\ ] ] & @xmath107&@xmath108 + faint & @xmath109&@xmath110 & @xmath111&@xmath106 { \tiny bright / all}&{\tiny $ 1.117 \pm 0.019$}&{\tiny $ 6.2~\sigma$}&{\tiny $ 0.967 \pm 0.016$}&{\tiny $ 2.1~\sigma$}&{\tiny $ 0.905 \pm 0.015$}&{\tiny $ 6.3~\sigma$}\\[0pt ] { \tiny faint / all } & { \tiny $ 0.897 \pm 0.013$}&{\tiny $ 7.9~\sigma$}&{\tiny $ 1.021 \pm 0.014$}&{\tiny $ 1.5~\sigma$}&{\tiny $ 1.072 \pm 0.015$}&{\tiny $ 4.8~\sigma$}\\[0pt ] { \tiny
faint / bright}&{\tiny $ 0.803 \pm 0.020$}&{\tiny $ 9.9~\sigma$}&{\tiny $ 1.056 \pm 0.019$}&{\tiny $ 2.9~\sigma$}&{\tiny $ 1.185 \pm 0.019$}&{\tiny $ 9.7~\sigma$}\\[-5pt ] \cutinhead{\tiny $ 1.4 < z_{abs } \leq 2.34 $ } { \tiny all } & { \tiny $ 0.686 \pm 0.011$}&\ ] ] & @xmath112&@xmath113 + bright & @xmath114&@xmath115 & @xmath116&@xmath106 { \tiny faint } & { \tiny $ 0.579 \pm 0.028$}&\ ] ] & @xmath117&@xmath118 + bright / all&@xmath119&@xmath120&@xmath121&@xmath122&@xmath123&@xmath124 + faint / all & @xmath125&@xmath126&@xmath127&@xmath128&@xmath129&@xmath130 + faint / bright&@xmath131&@xmath132&@xmath133&@xmath134&@xmath135&@xmath136 + figure [ fig : dndzweakmags ] shows our weak @xmath4 results for the bright and faint samples binned as in figure [ fig : residual ] .
the redshift path density of the bright sample is clearly higher than that of the faint sample at all redshifts .
the lowest redshift bin of the weak bright sample contained only five systems and we believe this caused the @xmath4 result in that bin to be less reliable . figures [ fig : dndz]@xmath85[fig : dndz]@xmath137 plot our @xmath4 results for weak , intermediate and strong absorption .
the bins ( @xmath138 and @xmath139 , `` low '' and `` high '' redshift ) were selected based on previous studies @xcite . similarly , figures [ fig : dndz]@xmath140[fig : dndz]@xmath141 show @xmath4 calculated for the faint and bright quasar samples .
figures [ fig : dndz]@xmath142[fig : dndz]@xmath143 show the ratio of the @xmath4 results of the faint quasar sample to that of the bright .
these values are listed in table [ tab : dndz ] for each @xmath18 and redshift range . though the weak @xmath4 results of @xcite , @xcite , and this study all differed as discussed in
[ sec : weakdndz ] , the _ relative _ values among our own faint , bright and all quasar subsamples are robust , having been calculated in a consistent manner . similarly , though our intermediate and strong samples have @xmath18 distributions consistent with @xcite as mentioned in
[ sec : data ] , our @xmath4 results for these samples are higher than those of @xcite and @xcite .
this is expected since some of the quasars in our survey were targeted for their known @xmath144 absorption ; however , the @xmath4 ratios between our magnitude bins are robust . for weak systems ,
the @xmath4 values toward bright quasars are @xmath145 higher than for the faint in both redshift ranges , while for strong systems , @xmath4 is @xmath15 higher toward faint quasars than toward bright in both redshift ranges .
the faint to bright @xmath4 ratio , @xmath146 , departs for the weak systems from unity at the @xmath147 level for low redshift and at the @xmath148 level for high redshift .
the strong absorber ratios are similarly significant ( see table [ tab : dndz ] ) , though in that case the faint quasar @xmath4 values are higher , rather than lower , than those of the bright sample . in the case of the intermediate absorbers
the ratio is consistent with unity within @xmath149 . in the calculation of @xmath4 the sensitivity of each spectrum
is accounted for ( see * ? ? ?
* ; * ? ? ?
* ) , eliminating the possibility that the higher values of the weak bright sample compared to the weak faint sample might result from higher signal - to - noise ratios .
based on our weak @xmath4 result ( see figure [ fig : residual ] ) , the cosmic number density , geometric cross section , or both , of weak absorbers appear to be evolving .
the apparent dropoff in our fit toward @xmath150 may be overly steep ; it is possible that the weak @xmath4 peaks at @xmath151 , declines slightly and then levels off toward the present .
however , it is a first attempt to characterize weak absorber evolution using a functional form .
our result predicts that no such absorbers exist above @xmath6 .
@xcite speculated that the apparent paucity of weak above @xmath152 might be due to the high redshift analogs of low redshift weak absorption being associated with strong . in this scenario ,
weak absorption at high redshift would be in the kinematic vicinity of strong and thus would not be recognized as isolated weak absorption .
however , we have compared the high velocity weak kinematic subsystems of strong subsystems to isolated weak absorption @xcite .
morphologically these two types of profiles often appear very similar , but a ks test of their rest equivalent width distributions revealed that they are actually two distinct populations to a 99.98% confidence level , or greater than 3 @xmath12 . a ks test of the distributions of flux decrement - weighted velocity spreads , @xmath153 @xcite indicated to a greater than 6 @xmath12 confidence level that the two populations are unique .
the @xmath4 evolution we detect in weak may be due to changes in gas structure or ionization conditions ; neither we nor other studies find the same falloff in intermediate and strong absorption @xcite up to our maximum redshift of @xmath154 .
@xcite do report a decline in the strong population above @xmath155 , and note that this peak corresponds to that of the sfr .
our weak @xmath4 result , which exhibits no such peak , may provide indirect evidence that a substantial fraction of these absorbers resides in the igm , since their evolution appears not to correspond to star formation .
using cloudy 08.00 photoionization modeling @xcite , we investigated the evolution of weak absorber sizes @xmath156 and cosmic number densities @xmath157 ( for additional details see * ? ? ? * ) . in this scenario
selects relatively dense cloudlets embedded within plane parallel slabs of gas .
we modeled optically thin clouds having a range of hydrogen number densities based on past photoionization modeling results @xcite .
we assumed an ultraviolet background model that varies as a function of @xmath158 , following the work of @xcite , which includes the contribution of galaxies . though we examined a grid of clouds with a range of column densities and metallicities ,
we discuss here clouds having @xmath159 of @xmath160 @xmath161 and a metallicity of 0.1 solar . for weak
, @xmath162 is constrained to the range 15.517.0 @xmath161 @xcite .
the resulting cloud thicknesses , which we interpreted as absorber sizes @xmath156 and which are governed by the ionizing background , peak at @xmath163 and then decline toward the present , as shown in figure [ fig : cloudy]@xmath85 .
assuming spherical clouds , the corresponding absorber cross sections @xmath164 , combined with our weak @xmath4 constraint using the fit of equation [ equationfit ] , translate into cosmic absorber number densities @xmath157 that increase monotonically toward the present ( figure [ fig : cloudy]@xmath92 ) .
the absorber sizes produced by this model are on the order of a parsec , and yield absorber number densities on the order of @xmath7 mpc@xmath165 , for the middle range of @xmath166 values .
this corresponds to @xmath167@xmath168 absorbers per @xmath169 galaxy for @xmath170 @xcite as well as for @xmath171 @xcite . if the clouds are not spherical , but
instead the transverse extent @xmath172 scales with cloud thickness according to a factor @xmath173 such that @xmath174 , then @xmath157 would scale as @xmath175 . for @xmath176 , @xmath157
would then be reduced by a factor of @xmath177 , which yields @xmath9 weak absorbers per mpc@xmath8 .
it should be noted that changing the model s @xmath178 would change @xmath156 in direct proportion , while @xmath157 , using our @xmath4 constraints , would vary as @xmath179 .
our cloudy model is suggestive of a condensation mechanism into sheet or filament structures characteristic of the igm .
sheetlike geometries require far fewer weak absorbers than do spherical geometries per @xmath169 galaxy , and therefore we consider these to be a more realistic scenario .
although we do not fully explain the nature of weak absorption , this exercise does provide a limiting case in which to couch the phenomenon .
our absorber size estimate is not far from that of @xcite , who concluded that a sheetlike structure containing many embedded @xmath180 pc absorbers was required to account for the observed @xmath4 .
recent findings by @xcite and @xcite indicate that @xmath181 , but not @xmath182 , absorbers are found in the circumgalactic medium of normal galaxies at impact parameters of less than 200 kpc .
they suggest that the weaker of these two populations may then reside primarily in the igm . our @xmath4 results toward faint versus bright absolute magnitude quasars reveal that the weak absorbers have a higher redshift path density in the bright quasar sample than in the faint at both low and high redshifts . for the strong absorbers ,
the opposite is true .
the intermediate absorbers appear to follow no clear trend , and may represent an equivalent width range where the effects leading to the weak and strong differentials mostly cancel .
following the discovery by @xcite of the quasar
grb strong discrepancy , various researchers have attempted to explain the phenomenon .
@xcite modeled the effects of absorber size and impact parameter on observed equivalent width and concluded that the @xmath4 discrepancy may be due to the larger beam sizes of quasars versus grbs , the latter of which they state are on the order of the sizes of cloud cores .
the authors also predicted that different luminosity populations of quasars should contain different incidences and strengths of intervening absorbers .
@xcite countered this differential beam size argument , noting that no unsaturated doublets have been observed having a doublet ratio of one , as would be expected in the case of partial covering of a quasar beam by an absorber .
they stated that magnification bias could explain the discrepancy , and that dust obscuration bias and association of absorbers with the circumburst environment could also partially account for it .
a statistical study was also conducted by @xcite to look for systematically lowered equivalent widths over quasar broad line emission regions , which are substantially larger than quasar continuum regions ; no significant difference was found .
@xcite cite an intrinsic origin as a possible explanation for the grb excess .
absorbing gas could be ejected at relativistic velocities and masquerade as an intervening absorber ; however , the authors note that the presence of absorption in these systems , as well as the lack of fine structure transitions that are expected in the vicinity of a grb , cast doubt on this theory . @xcite
concur that the excess could be intrinsic , and estimate required ejection velocities of @xmath183 .
they also consider gravitational lensing to be a viable mechanism to account for the discrepancy . in two studies of absorbers toward grbs ,
@xcite and @xcite reported no excess incidence over quasar sightlines .
@xcite speculated that the difference in the cases of versus absorbers arose partially because the former introduced more dust extinction than the latter .
however , dust extinction in absorbers was subsequently modeled @xcite , and the authors concluded that the effect could only account for @xmath184 of the quasar
grb discrepancy .
@xcite rejected an intrinsic origin for excess grb absorbers due to the lack of both excess absorption and excess weak and intermediate absorption , and instead favored gravitational lensing as the relevant mechanism .
@xcite modeled gravitational lensing in quasar and grb sightlines and concluded that it was a feasible explanation for the excess , but that further grb data were necessary to support or refute their findings .
they noted that afterglows in which strong systems are found are brighter than average , implying a greater lensing rate . through their modeling of extinction curves toward quasars and grbs ,
@xcite determined that @xmath4 toward quasars would be significantly higher if corrected for dust , and that the correction varies with redshift .
the discrepancy compared to grbs arises , the authors state , because their absorber redshift distribution is shifted higher than that of quasars , resulting in less loss of detected absorbers .
they calculated that this effect could account for a factor of two excess in the grb @xmath4 . keeping these previous studies of the quasar
grb discrepancy in mind , and in an attempt to understand the possibly related phenomenon we have uncovered , we attempted to find some other metric within our faint and bright quasar samples that would shed light on these issues .
the redshift path density depends on the integrated equivalent width distribution , @xmath185 we therefore compared equivalent width distributions for our faint and bright quasar samples for the various @xmath18 ranges as well as for low , high , and all redshifts .
we also studied equivalent width distributions binned by relative beam sizes using @xmath186 as a proxy , i.e. assuming that the square of the source radius @xmath187 is proportional to the @xmath50-band luminosity of the quasar .
the ratio of the source radius for quasar @xmath143 relative to the median radius for the full quasar sample can then be written @xcite as @xmath188 we then studied the effect of changing beam size with redshift due to cosmology .
we calculated the ratio of the relative beam size of quasar @xmath143 at the redshift of absorber @xmath189 to the cross section of the source : @xmath190 ^ 2\ ] ] where @xmath191 , @xmath192 is the angular diameter distance at the absorption redshift of system @xmath189 , and @xmath193 is the angular diameter distance at the source redshift of quasar @xmath143 .
finally , we examined the combined effect of source size and cosmology , by using equation [ equationradii ] to scale @xmath194 . using the ks test ,
none of these equivalent width distributions yielded significant differences ( of at least 3 @xmath12 ) between the faint and bright quasar samples . since the @xmath4 discrepancy is in this case of a smaller magnitude than in the case of the quasar
grb phenomenon , it may require a larger data set to discern the reasons behind the observations .
the findings of @xcite do offer an intriguing possibility by relating absorption redshift distributions to @xmath4 .
our faint absorber samples do have lower median @xmath86 values than our bright samples across all three equivalent width ranges , probably a result of the correlation whereby intrinsically more luminous quasars tend to be selected at higher redshifts .
this result only supports the authors dust argument in the case of our weak absorbers , the only sample in which the absorption incidence is significantly higher toward bright quasars than faint . for our weak sample ,
the median absorption redshift is @xmath195 in the faint subsample and @xmath196 in the bright .
ks testing , however , revealed that it could not be ruled out to a greater than 98.32% confidence level that the faint and bright subsamples are drawn from the same underlying @xmath86 distribution .
though several authors have argued for an intrinsic origin for excess strong absorption toward grbs versus quasars , this does not appear to explain the @xmath4 discrepancy in the case of our bright and faint quasar populations .
our @xcite testing revealed absorber distributions consistent with cosmological within all equivalent width and quasar absolute magnitude subsamples , as well as in the aggregate populations .
the velocities of -selected gas ejected from a quasar would have to reach large fractions of the speed of light in order to pass for intervening systems .
it therefore seems highly unlikely that significant intrinsic absorption could be present in our sample .
we have found in a survey of 252 quasar spectra that the incidence of weak absorption evolves markedly , that it peaks at @xmath197 , and that it is fit by a function that is a product of the no - evolution expectation with a linear function .
our linear fit to the ratio of our @xmath4 data to the nee resulted in a slope of @xmath96 and a normalization of @xmath97 for the function @xmath198 .
our @xmath4 result predicts that no weak absorbers exist above @xmath6 .
we find that when our quasar survey is segregated by absolute magnitude , weak @xmath4 is significantly lower in the faint subsample than in the bright , with faint to bright @xmath4 ratios of @xmath199 at low redshift and @xmath200 at high redshift .
in contrast , strong @xmath4 is significantly _ higher _ in the faint subsample than in the bright , with faint to bright @xmath4 ratios of @xmath201 at low redshift and @xmath202 at high redshift .
intermediate equivalent width absorbers exhibited @xmath4 ratios consistent with unity within 3 @xmath12 . at this time it is uncertain whether these results stem from some intrinsic property of the quasars , from some difference in the intervening sightlines , or from some combination of these factors .
we thank wallace sargent , michael rauch , jason prochaska , and charles steidel for their contribution of spectra , and anand narayanan for helpful communications regarding @xcite .
this research 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 .
we are grateful for nsf grant ast 0708210 , the primary funding for this work ; jle was also supported by a three - year aerospace cluster fellowship administered by the vice provost of research at new mexico state university and by a two - year new mexico space grant graduate research fellowship .
mtm thanks the australian research council for a qeii research fellowship ( dp0877998 ) .
bahcall , j. n. , & peebles , p. j. e. 1969 , , 156 , l7 barton , e. j. , & cooke , j. 2009 , , 138 , 1817 bergeron , j. , aracil , b. , petitjean , p. , & pichon , c. 2002 , , 419 , 811 bergeron , j. , & boiss , p. 1991
, , 243 , 334 bergeron , j. , boiss , p. , & mnard , b. 2011 , , 525 , 51 budzynski , j. m. , & hewett , p. c. 2011 , , 416 , 1871 chen , h .- w . , helsby , j. e. , gauthier , j .-
r . , shectman , s. a. , thompson , i. b. , & tinker , j. l. 2010 , , 714 , 1521 chen , h .- w .
, & tinker , j. l. 2008 , , 687 , 745 churchill , c. w. , kacprzak , g. g. , nielsen , n. m. , steidel , c. c. , & murphy , m. t. 2012 , , submitted churchill , c. w. , kacprzak , g. g. , & steidel , c. c. 2005 , iau colloq . 199 : probing galaxies through quasar absorption lines , 24 churchill , c. w. , rigby , j. r. , charlton , j. c. , & vogt , s. s. 1999 , , 120 , 51 churchill , c. w. , mellon , r. r. , charlton , j. c. , jannuzi , b. t. , kirhakos , s. , steidel , c. c. , & schneider , d. p. 2000 , , 130 ,
91 churchill , c. w. , & vogt , s. s. 2001 , , 122 , 679 cucchiara , a. , jones , t. , charlton , j. c. , fox , d. b. , einsig , d. , & narayanan , a. 2009 , , 697 , 345 evans , j. l. 2011 , ph.d .
thesis , new mexico state university evans , j. l. , churchill , c. w. , & murphy , m. t. 2012 , , in preparation faber , s. m. 2007 , , 665 , 265 ferland , g. j. , korista , k. t. , verner , d. a. , ferguson , j. w. , kingdon , j. b. , & verner , e. m. 1998 , , 110 , 761 frank , s. , bentz , m. c. , stanek , k. z. , mathur , s. , dietrich , m. , peterson , b. m. , & atlee , d. w. 2007 , , 312 , 325 guillemin , p. , & bergeron , j. 1997 , , 328 , 499 haardt , f. , & madau , p. 1996
, , 461 , 20 kacprzak , g. g. , churchill , c. w. , evans , j. l. , murphy , m. t. , & steidel , c. c. 2011 , , 416 , 3118 kacprzak , g. g. , churchill , c. w. , steidel , c. c. , & murphy , m. t. 2008 , , 135 , 922 lanzetta , k. m. , turnshek , d. a. , & wolfe , a. m. 1987 , , 332 , 739 lundgren , b. f. , brunner , r. j. , york , d. g. , ross , a. j. , quashnock , j. m. , myers , a , d. , schneider , d. p. , al sayyad , y. , & bahcall , n. 2009 , , 698 , 819 lynch , r. s. , & charlton , j. c. 2007 , , 666 , 64 matejek , m. s. , & simcoe , r. a. 2012 , , submitted milutinovi , n. , rigby , j. r. , masiero , j. r. , lynch , r. r. , palma , c. , charlton , j. c. 2006 , , 641 , 190 narayanan , a. , misawa , t. , charlton , j. c. , & kim , t. 2007 , , 660 , 1093 nestor , d. b. , turnshek , d. a. , & rao , s. m. 2005 , , 628 , 637 nielsen , n. m. , churchill , c. w. , & kacprzak , g. g. 2012 , , in preparation oesch , p. a. , bouwens , r. j. , carollo , c. m. , illingworth , g. d. , magee , d. , trenti , m. , stiavelli , m. , franx , m. , labb , i. , & van dokkum , p. g. 2010 , , 725 , 150 pontzen , a. , hewett , p. , carswell , r. , & wild , v. 2007 , , 381 , 99 porciani , c. , viel , m. , & lilly , s. j. 2007 , , 659 , 218 prochter , g. e. , prochaska , j. x. , & burles , s. m. 2006a , , 639 , 766 prochter , g. e. , prochaska , j. x. , chen , h. , bloom , j. s. , dessauges - zavadsky , m. , foley , r. j. , lopez , s. , pettini , m. , dupree , a. k. , & guhathakurta , p. 2006b
, , 648 , 93 rao , s. m. , & turnshek , d. a. 2000 , , 130 , 1 reddy , n.a .
, & steidel , c. c. 2009 , , 692 , 778 rigby , j. r. , charlton , j. c. , & churchill , c. w. 2002 , , 565 , 743 sargent , w. l. w. , steidel , c. c. , & boksenberg .
a. 1988 , , 334 , 22 schneider , d. p. , 1993 ,
apjs , 87 , 45 shakura , n. i. , & sunyaev , r. a. 1973 , , 24 , 337 steidel , c. c. , dickinson , m. , meyer , d. m. , adelberger , k. l. , & sembach , k. r. 1997 , , 480 , 586 steidel , c. c. , dickinson , m. , & persson , s. e. 1994 , , 437 , l75 steidel , c. c. , & sargent , w. l. w. 1992 , , 80 , 1 stocke , j. t. , & rector , t. a. 1997 , , 489 , 17 sudilovsky , v. , savaglio , s. , vreeswijk , p. , ledoux , c. , smette , a. , & greiner , j. 2007 , , 669 , 741 sudilovsky , v. , smith , d. , & savaglio , s. 2009 , , 699 , 56 tejos , n. , lopez , s. , prochaska , j. x. , bloom , j. s. , chen , h. , dessauges - zavadsky , m. , maureira , m. j. 2009 , , 706 , 1309 tejos , n. , lopez , s. , prochaska , j. x. , chen , h. , & dessauges - zavadsky , m. 2007 , , 671 , 622 vergani , s. d. , petitjean , p. , ledoux , c. , vreeswijk , p. , smette , a. , meurs , e. j. a. 2009 , , 503 , 771 veron - cetty , m. p. , & veron , p. 2001
, , 374 , 92 wyithe , j. s. b. , oh , s. p , & pindor , b. 2011 , , 414 , 209 | we have identified 469 doublet systems having @xmath0 in 252 keck / hires and uves / vlt quasar spectra over the redshift range @xmath1 . using the largest sample yet of 188 weak systems ( @xmath2 @xmath3 ) , we calculate their absorber redshift path density , @xmath4 .
we find clear evidence of evolution , with @xmath4 peaking at @xmath5 , and that the product of the absorber number density and cross section decreases linearly with increasing redshift ; weak absorbers seem to vanish above @xmath6 .
if the absorbers are ionized by the uv background , we estimate number densities of @xmath7 per mpc@xmath8 for spherical geometries and @xmath9 per mpc@xmath8 for more sheetlike geometries .
we also find that @xmath4 toward intrinsically faint versus bright quasars differs significantly for weak and strong ( @xmath10 ) absorbers .
for weak absorption , @xmath4 toward bright quasars is @xmath11 higher than toward faint quasars ( 10 @xmath12 at low redshift , @xmath13 , and 4 @xmath12 at high redshift , @xmath14 ) . for strong absorption
the trend reverses , with @xmath4 toward faint quasars being @xmath15 higher than toward bright quasars ( also 10 @xmath12 at low redshift and 4 @xmath12 at high redshift ) .
we explore scenarios in which beam size is proportional to quasar luminosity and varies with absorber and quasar redshifts .
these do not explain @xmath4 s dependence on quasar luminosity . |
the effective field theory ( eft ) approach provides a powerful framework that exploits the separation of scales in physical systems .
only low - energy ( or long - range ) degrees of freedom are included explicitly , with the rest parametrized in terms of the most general local ( contact ) interactions .
this procedure exploits the fact that a low - energy probe of momentum @xmath0 can not resolve structures on scales smaller than @xmath1 . in this talk . ] using renormalization , the influence of short - distance physics on low - energy observables is captured in a small number of low - energy constants .
thus , the eft describes universal low - energy physics independent of detailed assumptions about the short - distance dynamics .
all physical observables can be described in a controlled expansion in powers of @xmath2 , where @xmath3 is the characteristic low - energy length scale of the system .
the size of @xmath3 depends on the system under consideration : for a finite - range potential , e.g. , it is given by the range of the potential .
for the systems discussed here , @xmath3 is of the order of the effective range @xmath4 or the van der waals scale @xmath5 .
effective field theories can be obtained by applying a renormalization group ( rg ) transformation to a ( more ) fundamental theory .
the rg transformation integrates out high momentum modes from the fundamental theory and leads to a description of low - energy physics in terms of low - energy degrees of freedom only .
it can be understood as a change of the resolution scale of the eft .
the transformation generates a flow in the space of coupling constants that determine the effective lagrangian @xmath6 .
if the rg transformation is continuous , this flow can be expressed by a differential equation for the coupling constants @xmath7 : @xmath8 where @xmath9 is the ultraviolet cutoff which corresponds to the inverse resolution scale .
the function @xmath10 determines the behavior of @xmath7 under the rg transformation .
the rg equation ( [ eq : rgeq ] ) can have many types of solutions .
the two simplest ones are ( a ) renormalization group fixed points and ( b ) renormalization group limit cycles .
fixed points are ubiquitous in condensed matter and particle physics .
they play an important role , e.g. , in critical phenomena and the scale dependence of coupling constants in high energy physics .
renormalization group limit cycles have been suggested by ken wilson already in 1971 @xcite , but their phenomenological importance has only been realized very recently . an important signature of a limit cycle is the appearance of a discrete scaling symmetry , the invariance of physical observables under changes of the resolution scale by a discrete scale transformation with a preferred scaling factor @xmath11 .
the prime example of a rg limit cycle is the efimov effect @xcite for 3-body systems with large s - wave scattering length @xmath12 .
vitaly efimov discovered that in the limit @xmath13 there can be infinitely many 3-body bound states ( trimers ) with an accumulation point at the scattering threshold @xcite .
these trimers are called efimov states .
one remarkable feature of efimov states is that they have a geometric spectrum with preferred scaling factor @xmath14 : @xmath15 where @xmath16 is the binding wavenumber of the branch of efimov states labeled by @xmath17 .
the efimov effect is just one example of the universal phenomena characterized by the discrete scaling symmetry in the 3-body system @xcite .
these universal properties persist also for finite values of the scattering length as long as @xmath18 . for a review of these phenomena , which we refer to as efimov physics ,
see ref .
@xcite . for a generic system ,
the scattering length is of the same order of magnitude as the low - energy length scale @xmath3 .
only a very specific choice of the parameters in the underlying theory ( a so - called fine tuning ) will generate a large scattering length .
nevertheless , systems with large scattering length can be found in many areas of physics .
examples are the s - wave scattering of nucleons and of @xmath19he atoms . for alkali atoms close to a feshbach resonance , @xmath12 can be tuned experimentally by adjusting an external magnetic field .
this is particularly interesting , since it allows to experimentally test the dependence of physical observables on the scattering length . in this talk , we discuss some applications of an eft for few - body systems with large scattering length @xmath18 @xcite . in this theory , the limit cycle is manifest in the rg behavior of the 3-body interaction which makes it ideally suited to study the efimov effect and its consequences for other 3-body observables .
we first give a very brief review of the eft for few - body systems with large scattering length @xmath12 , focusing on s - waves .
a more detailed treatment is given in ref .
@xcite . for typical momenta @xmath20 ,
the eft expansion is in powers of @xmath21 so that higher order corrections are suppressed by powers of @xmath21 .
we consider a 2-body system of nonrelativistic bosons ( referred to as atoms ) with large scattering length @xmath12 and mass @xmath22 . at sufficiently low energies
, the most general lagrangian may be written as : @xmath23 where the @xmath24 and @xmath25 are nonderivative 2- and 3-body interaction terms , respectively .
the strength of the @xmath24 term is determined by the scattering length @xmath12 , while @xmath25 depends on a 3-body parameter to be introduced below .
the dots represent higher - order terms which are suppressed at low - energies . for momenta @xmath0 of the order of the inverse scattering length @xmath26 , the problem is nonperturbative in @xmath27 .
the exact 2-body scattering amplitude can be obtained analytically by summing the so - called bubble diagrams with the @xmath24 interaction term .
the @xmath25 term does not contribute to 2-body observables .
after renormalization , the resulting amplitude reproduces the leading order of the well - known effective range expansion for the atom - atom scattering amplitude : @xmath28 where the total energy is @xmath29 . if @xmath30 , @xmath31 has a pole at @xmath32 corresponding to a shallow dimer with the universal binding energy @xmath33 .
higher - order interactions are perturbative and give the momentum - dependent terms in the effective range expansion .
we now turn to the 3-body system . at leading order
, the atom - dimer scattering amplitude is given by the integral equation shown in fig .
[ fig : ineq ] .
a solid line indicates a single atom and a double line indicates an interacting two - atom state ( including rescattering corrections ) .
the integral equation contains contributions from both the 2-body and the 3-body interaction terms .
the inhomogeneous term is given by the first two diagrams on the right - hand side : the one - atom exchange diagram and the 3-body term .
the integral equation simply sums these diagrams to all orders . after projecting onto s - waves , we obtain the equation @xmath34 for the off - shell atom - dimer scattering amplitude with the inhomogeneous term @xmath35 the first term is the s - wave projected one - atom exchange , while the second term comes from the 3-body interaction .
the physical atom - dimer scattering amplitude @xmath36 is given by the solution @xmath37 evaluated at the on - shell point : @xmath38 where @xmath39 .
the 3-body binding energies @xmath40 are given by those values of @xmath41 for which the homogeneous version of eq .
( [ eq : bhvk ] ) has a nontrivial solution .
note that an ultraviolet cutoff @xmath9 has been introduced in ( [ eq : bhvk ] ) .
this cutoff is required to insure that eq .
( [ eq : bhvk ] ) has a unique solution .
all physical observables , however , must be invariant under changes of the cutoff , which determines the behavior of @xmath42 as a function of @xmath9 @xcite : @xmath43 \over \cos [ s_0 \ln ( \lambda/ \lambda _ * ) - \arctan s_0]}\ , , \label{h - lambda}\end{aligned}\ ] ] where @xmath44 is a transcendental number and @xmath45 is a 3-body parameter introduced by dimensional transmutation .
this parameter can not be predicted by the eft and must be taken from experiment .
it is evident that @xmath46 is periodic and runs on a limit cycle .
when @xmath9 is changed by the preferred scaling factor @xmath47 , @xmath42 returns to its original value .
note that the definition of the 3-body parameter is not unique .
the parameter @xmath45 from eq .
( [ h - lambda ] ) arises naturally in the eft description , while the parameter @xmath16 from eq .
( [ kappa - star ] ) is defined via the 3-body efimov spectrum in the limit @xmath48 .
both definitions are related by a constant factor : @xmath49 @xcite . in summary ,
two parameters are required in the 3-body system at leading order in @xmath21 : the scattering length @xmath12 ( or the dimer binding energy @xmath50 ) and the 3-body parameter @xmath45 or @xmath16 @xcite .
the eft reproduces the universal aspects of the 3-body system first derived by efimov @xcite and is a very efficient calculational tool to calculate those properties .
we now turn to some applications of this eft to systems of cold atoms .
first we discuss universal scaling functions .
since only two parameters enter at leading oder , different 3-body observables show correlations .
these correlations must appear in all 3-body systems with short - range interactions and large scattering length . in the left panel of fig .
[ fig : scaling ] , we display the scaling function relating the trimer excited and ground state energies @xmath51 and @xmath52 , respectively @xcite .
the data points give various calculations using realistic @xmath19he potentials while the solid line gives the universal prediction from eft .
different points on this line correspond to different values of @xmath16 .
the small deviations of the potential calculations from the universal curve are mainly due to effective range corrections and can be calculated at next - to - leading order in eft @xcite .
the calculation corresponding to the data point far off the universal curve can easily be identified as problematic . in the right panel of fig .
[ fig : scaling ] , we display a similar correlation between the triton binding @xmath53 and the spin - doublet @xmath54-wave neutron - deuteron scattering length @xmath55 from nuclear physics taken from @xcite , which is known as the phillips line . versus @xmath56 for @xmath19he atoms @xcite .
right panel : the scaling function @xmath55 versus @xmath53 from nuclear physics @xcite .
the dash - dotted and dashed lines exclude and include the leading correction in @xmath21 , respectively .
the data points are calculations using realistic potentials.,title="fig : " ] versus @xmath56 for @xmath19he atoms @xcite .
right panel : the scaling function @xmath55 versus @xmath53 from nuclear physics @xcite .
the dash - dotted and dashed lines exclude and include the leading correction in @xmath21 , respectively .
the data points are calculations using realistic potentials.,title="fig:",width=283 ] the dash - dotted and dashed lines give the results excluding and including the leading corrections in @xmath21 , respectively .
it is evident that nuclear and atomic systems show the same universal scaling behavior .
while these systems have very different structure at short distances , all that matters here is that @xmath57 but not the details of the mechanism leading to the large scattering length .
there are also observable features directly related to the occurence of a limit cycle in the 3-body system .
the discrete scale invariance is manifest in the log - periodic dependence of 3-body observables on the scattering length @xmath12 .
this dependence can be tested in experiments with cold atoms close to a feshbach resonance . for this purpose
we consider 3-body recombination , which is the process when three atoms scatter to form a dimer and the third atom balances energy and momentum .
this is one of the main loss processes for trapped atoms and condensates of atoms near a feshbach resonance .
the event rate @xmath58 per unit time and unit volume can be parametrized as @xmath59 , where @xmath60 is the density of the atoms and @xmath61 is the recombination constant .
unfortunately , heavy alkali atoms , such as rb and cs , form many deeply - bound diatomic molecules .
therefore , efimov states are resonances rather than sharp states , because they can decay into a deep molecule and an energetic atom .
the presence of deep molecules also affects other 3-body observables , but using unitarity their influence can be accounted for by one real parameter @xmath62 @xcite .
this parameter describes how much probability is lost by scattering into the deep bound states . for @xmath63 the observables are not modified , while for @xmath64 all probability is lost into the deep states .
the consequences of the limit cycle become totally washed out for @xmath65 .
the 3-body recombination coefficient @xmath61 was first calculated in refs .
analytical expressions for @xmath61 as a function of @xmath12 , @xmath16 , and @xmath62 have been obtained in refs .
@xcite . in a recent experiment with cold @xmath66cs atoms ,
the innsbruck group has presented the first experimental evidence for efimov physics @xcite .
they used a feshbach resonance to control the scattering length of @xmath66cs atoms .
since inelastic 2-body losses were energetically forbidden , the dominant loss mechanisms were inelastic 3-body losses . by varying the external magnetic field , they were able to change the scattering length from @xmath67 through 0 to @xmath68 , where @xmath69 is the bohr radius .
they observed a giant loss feature at a magnetic field that corresponds to the scattering length @xmath70 .
they also measured the 3-body recombination rate for positive values of @xmath12 reached by increasing the magnetic field through a zero of the scattering length .
they observed a local minimum in the inelastic loss rate at a magnetic field that corresponds to a scattering length of 210(10 ) @xmath69 .
their data could well described by the universal expressions from refs .
@xcite , therefore giving the first indirect evidence for efimov states in cold atoms @xcite . in the left panel of fig .
[ fig:133cs ] , we compare the universal prediction for negative scattering length at @xmath71 with the innsbruck data for the loss coefficient @xmath72 at @xmath73 nk @xcite . for @xmath74 and @xmath75 ,
one obtains a good fit .
both parameters are well determined by the resonance : @xmath16 by the position of the efimov resonance and @xmath62 by the height of the peak . in @xmath66cs for @xmath76 .
the curves are for @xmath77 and three different values of @xmath62 .
the data points are for @xmath73 nk @xcite .
right panel : @xmath78 in @xmath66cs for @xmath30 .
the curves are for @xmath79 and three different values of @xmath62 .
the data points are for @xmath80 nk @xcite.,title="fig:",width=264 ] in @xmath66cs for @xmath76 .
the curves are for @xmath77 and three different values of @xmath62 .
the data points are for @xmath73 nk @xcite .
right panel : @xmath78 in @xmath66cs for @xmath30 .
the curves are for @xmath79 and three different values of @xmath62 .
the data points are for @xmath80 nk @xcite.,title="fig:",width=264 ] the positive scattering length data from ref .
@xcite are shown in the right panel of fig .
[ fig:133cs ] .
one complication here is that the lowest temperature that was reached for positive @xmath12 was 250 nk . to take account
the thermal effects properly would require knowing the 3-body recombination rate as a function of the collision energy .
we compare the universal expression for @xmath71 with the data for @xmath80 nk @xcite . for @xmath81 and @xmath75 ,
one obtains a good fit to the data above @xmath82 .
however , the value of @xmath62 is not well determined by the data and only the upper bound @xmath83 can be given @xcite .
we note that the values of @xmath16 and @xmath62 for positive and negative scattering length need not be the same in the present case since both regions are divided by a nonuniversal region with small scattering length
. moreover the universal expressions are rigorously only valid for @xmath84 with @xmath85 for cs atoms .
therefore the minimum around @xmath86 is outside the region of validity of the universal theory .
we have discussed the universal properties of few - body systems with large scattering length @xmath18 . in the 3-body system
these properties include the efimov effect and a discrete scaling symmetry leading to log - periodic dependence of 3-body observables on the scattering length .
these features can be understood as manifestations of a rg limit cycle in the 3-body system .
moreover , we have presented an eft that is designed to exploit the separation of scales in few - body systems with large scattering length . in this eft ,
the limit cycle is manifest through the rg behavior of the 3-body interaction required for proper renormalization .
the eft is very general and has applications in few - body systems from atomic to nuclear and particle physics . as an example , we have shown universal scaling functions for @xmath19he atoms and the phillips line . furthermore , we have discussed the recent 3-body recombination data for cold @xmath66cs atoms by the innsbruck group that provided the first evidence of efimov states in cold atoms @xcite .
future challenges include the extension of the eft to the four - body system @xcite and the proper calculation of finite temperature effects @xcite .
e. braaten and h .- w .
hammer , phys .
428 ( 2006 ) 259 .
f. bedaque , h .- w . hammer , and u. van kolck , phys .
82 ( 1999 ) 463 ; nucl .
phys . a 646 ( 1999 ) 444 .
e. braaten and h .- w .
hammer , phys .
rev . a 67 ( 2003 ) 042706 .
l. platter and d.r .
phillips , arxiv : cond - mat/0604255 .
f. bedaque , g. rupak , h.w .
griehammer and h .- w .
hammer , nucl .
phys . a 714 ( 2003 ) 589 .
e. braaten and h .- w .
hammer , phys .
lett . 91 ( 2003 ) 102002 .
e. braaten and h .- w .
hammer , phys .
rev . a 70 ( 2004 ) 042706 .
e. nielsen and j.h .
macek , phys .
lett . 83 ( 1999 ) 1566 .
p. f. bedaque , e. braaten , and h .- w .
hammer , phys .
85 ( 2000 ) 908 ; e. braaten and h .- w .
hammer , phys .
lett . 87 ( 2001 ) 160407 .
t. kraemer , m. mark , p. waldburger , j.g .
danzl , c. chin , b. engeser , a.d .
lange , k. pilch , a. jaakkola , h .- c .
ngerl , and r. grimm , nature 440 ( 2006 ) 315 . | physical systems with a large scattering length have universal properties independent of the details of the interaction at short distances . such systems can be realized in experiments with cold atoms close to a feshbach resonance .
they also occur in many other areas of physics such as nuclear and particle physics .
the universal properties include a geometric spectrum of three - body bound states ( so - called efimov states ) and log - periodic dependence of low - energy observables on the physical parameters of the system .
this behavior is characteristic of a renormalization group limit cycle .
we discuss universality in the three- and four - body sectors and give an overview of applications in cold atoms . |
traditionally the ( semi ) phenomenological high precision two - nucleon ( nn)potentials av18 @xcite , cd bonn @xcite and nijm i , ii @xcite go together with the nonrelativistic operator for the kinetic energy @xmath2 in the nn c.m.system .
nevertheless , as is well known , this nonrelativistic schrdinger equation @xmath3 can be related to an underlying relativistic nn schrdinger equation @xmath4 by a simple algebraic step@xcite,@xcite .
applying @xmath5 to ( [ 2 ] ) from the left one obtains @xmath6 which can be identically rewritten into ( [ 1 ] ) if one defines @xmath7 with @xmath8 .
( we use @xmath9 in order to distinguish the momentum operator from the number @xmath10 ) . therefore adjusting @xmath11 in ( [ 1 ] ) to the nn phase shift and mixing parameters from a phase shift analysis and relating the c.m .
momentum @xmath12 to the lorentz invariant lab energy @xmath13 via @xmath14 ( a relation identically valid for relativistic and nonrelativistic kinematics ) one has in fact solved a relativistic equation .
we also see that @xmath15 equals @xmath16 .the question remains , what is @xmath17 given @xmath18 ?
the formal solution of that quadratic equation ( [ 4 ] ) is @xmath19 why is @xmath17 of interest ?
if one turns to the 3n system and would like to investigate relativistic effects @xcite,@xcite the knowledge of @xmath17 is very useful .
it defines together with the relativistic kinetic energy the interacting nn mass operator , which is a key ingredient for building the interactive 3n mass operator @xcite .
therefore we focus in this paper on the determination of @xmath20 related to the high precision nn potentials via ( [ 4 ] ) or ( [ 6 ] ) . in @xcite,@xcite
a potential @xmath20 has been determined directly fitting ( [ 2 ] ) to nn phase shifts .
thereby the urbana @xmath21 potential has been readjusted achieving a fair fit ( though not of the quality of the high precision potentials ) . in @xcite a momentum scale transformation
@xmath22 has been introduced which provides an analytical relation between @xmath11 and @xmath20 and guarantees that the s - matrix related to ( [ 1 ] ) at c.m.momentum k equals the s - matrix related to ( [ 2 ] ) at c.m .
momentum q. in other words the relativistic and nonrelativistic s - matrices agree at the same energy .
this , however , is misleading since the equality of the two s - matrices should hold at the same c.m.momenta @xcite . a better , though still approximate approach to relate @xmath11 and @xmath20 has been given in @xcite .
on the other hand there is the possibility to add an interaction to the square of the free nn mass operator @xmath23 .
then @xmath24 has exactly the same form as ( [ 1 ] ) with @xmath25 provided we identify @xmath26 with the mass operator of the interacting nn system @xcite , @xcite .
this is of course also obvious from the relations ( [ 1 ] ) - ( [ 4 ] ) . the construction of the relativistic 3n hamiltonian requires , however , the 3n mass operator @xmath27 rather than its square .
therefore our aim here is to solve ( [ 4 ] ) and ( [ 6 ] ) exactly for @xmath20 .
this is outlaid in section ii .
the validity of the resulting @xmath20 is verified by demonstrating that it provides exactly the same phase shift parameters using ( [ 2 ] ) as the underlying @xmath11 using ( [ 1 ] ) .
next we regard the 3n mass operator where nn c.m .
forces enter in the form @xcite @xmath28 with @xmath29 the total nn momentum .
the @xmath29-dependence arises since in a 3-body system the nn subsystems are not at rest . in section iii
we propose a simple manner to determine @xmath30 in a numerical precise way .
this opens now the door to use @xmath31 which are equivalent to the underlying high precision potentials in a relativistic context in 3n bound and scattering problems .
we summarize in section iv .
a technical derivation is given in the appendix .
the determination of @xmath20 using eq .
( [ 6 ] ) can be achieved by a spectral decomposition .
one can proceed in close analogy to the representation derived in @xcite for @xmath30 given in eq .
( [ 8 ] ) .
we regard a specific partial wave state ( or coupled ones ) with given orbital angular momentum(a ) , total spin and total angular momentum . for the sake of a simpler notation
we will not show these quantum numbers explicitely . using the completeness relation of bound and scattering states for the potential @xmath18 one obtains @xmath33 where @xmath34 is the nonrelativistic deuteron wave function of ( [ 1 ] ) and @xmath35 the mass of the deuteron .
here we introduced a different definition of the binding energy , namely an implicit one : @xmath36 in lowest order it agrees with the usual one @xmath37 .
this new definition of the binding energy has in addition the feature that it can naturally be written as @xmath38 in agreement with the form of the energy eigenvalue of ( [ 1 ] ) at the bound state pole @xmath39 of the s - matrix .
the expression ( [ 9 ] ) can be identically rewritten into the form @xmath40\cr & - & 2 \omega(k~ ' ) ' \re [ t ( k , k ' ; { { k'}^2 \over m } ) ] \ } \cr & + & { m^2 \over { k ^2 - { k ' } ^2 } } \times \cr & \ { & { \cal p } \int_0^\infty d k '' ( k~'')^2 { 2 \omega(k~ '' ) \over { { { k '' } ^2 - { k } ^2 } } } \cr & & t ( k , k '' ; { { k''}^2 \over m } ) t^ { * } ( k ' , k '' ; { { k''}^2 \over m } ) \cr & - & { \cal p } \int_0^\infty d k '' ( k~'')^2 { 2 \omega(k~ '' ) \over { { k '' } ^2 - { k ' } ^2 } } \cr & & t ( k , k '' ; { { k''}^2 \over m } ) t^ { * } ( k ' , k '' ; { { k''}^2 \over m } ) \}.
\label{11 } \end{aligned}\ ] ] where @xmath41 is the standard nn t - matrix related to @xmath42 via the nonrelativistic lippmann schwinger equation .
the derivation of that form is defered to the appendix
. a numerical implementation has not yet been performed , but we expect no problem . a second more simple way is to directly solve the quadratic operator equation ( [ 4 ] ) . in momentum space
it reads @xmath43 or @xmath44 we verified numerically that for all the realistic high precision potentials av18 , cd bonn , and nijm i , ii the following very simple iterative scheme works @xmath45 @xmath46 with @xmath47 . for certain partial waves the iteration to converge requires an additional step , namely @xmath48 where the constants @xmath49 and @xmath50 are typically 1 .
we display in table [ tab:1 ] an example documenting the convergence . in figs
[ fig.1]-[fig.3 ] we show for an example the original nonrelativistic potential @xmath51 , the resulting relativistic potential @xmath52 and the difference @xmath53 .
we see in that example that @xmath54 .
av18 in the state @xmath55 .
[ fig.1 ] ] related to av18 in the state @xmath55.[fig.2 ] ] in the state @xmath55 .
[ fig.3 ] ] to further characterize the difference between @xmath56 and @xmath57 one can regard the asymptotic behavior @xmath58 against @xmath59 as examples , for av18 we find @xmath60 , @xmath61 , whereas cd bonn delivers @xmath62 , @xmath63 .
the value of @xmath64 for cd bonn can easily be understood @xcite : @xmath65 with the two @xmath66 s resulting from the meson propagator and the choice of the strong form factor , the @xmath67 arising from transforming the blankenbeclar - sugar equation to the nonrelativistic lippmann schwinger equation and the @xmath68 from the two dirac spinors . in both cases , av18 and cd bonn
, @xmath69 is larger than @xmath64 by one unit , which is suggested by eq.([12 ] ) .
.convergence of @xmath70 to the iteration in eq .
( [ 14 ] ) .
+ we choose the coupled partial waves ( @xmath71-@xmath72 ) of the argonne v18 potential@xcite .
the momenta @xmath10 and @xmath73 are 1.0 @xmath74 and the potential unit is [ @xmath75 .
[ tab:1 ] [ cols="^,^,^,^,^",options="header " , ]
if one defines @xmath76 then using ( [ 4 ] ) eq .
( [ 8 ] ) can be written @xmath77 or @xmath78 between free states it yields @xmath79 this has the same structure as ( [ 13 ] ) replacing @xmath80 by @xmath81 . the iteration procedure described in section ii works equally well .
we compare in fig [ fig.4 ] @xmath82 and @xmath83 .
we see a weakening of @xmath84 against @xmath20 .
this is a fact known from previous calculations @xcite and from the approximate ( but very useful ) expression @xcite @xmath85 this is demonstrated in fig [ fig.4 ] , where also an even simpler approximation @xmath86 is shown .
that latter approximate version , however , is somewhat worse and is not recommended . against @xmath87 for av18 in the state @xmath55 for fixed @xmath88 @xmath74 .
the solid curve shows the relativistic potential @xmath89 ( @xmath90 ) .
the other curves show @xmath91 for @xmath92@xmath74 .
the long dashed curve is the exact one , the short dashed and dotted curves show the approximations for @xmath30 given in eqs.([approx1 ] ) and ( [ approx2 ] ) , respectively . [ fig.4 ] ]
relativistic calculations in the instant form of dynamics proposed in @xcite requires an interacting 3n mass operator .
as has been shown in @xcite and used in @xcite,@xcite,@xcite , @xcite , @xcite , and @xcite the nn potential in a moving frame @xmath30 enters in the form given in eq.([8 ] ) , where the nn force in the nn c.m.system @xmath20 enters into the relativistic nn schrdinger equation ( [ 2 ] ) .
we showed that the quadratic operator relation ( [ 4 ] ) for @xmath89 can be solved directly in an iterative manner and this very precisely .
this has been documented by evaluating nn phase shift and mixing parameters using the standard nonrelatvistic schrdinger equation ( [ 1 ] ) and the relativistic nn schrdinger equation ( [ 2 ] ) .
this opens the way to get @xmath20 s related to any nn potential @xmath42 adjusted in a nonrelativistic frame work like the high precision nn potentials . by the same iterative procedure also
@xmath30 can be gained .
it turned out that @xmath17 is smaller in magnitude than @xmath18 and @xmath30 is smaller than @xmath89 .
applications to the 3n bound and scattering states are planned . the numerical calculations has been performed on the ibm regatta p690 + of the nic in jlich , germany .
derivation of eq.([11 ] ) we start from ( [ 9 ] ) and use the well known decomposition @xmath93 to arrive at @xmath94 the integral requires some care and we keep the limiting processes for the two scattering states separately by putting @xmath95 this allows us to perform one limit firstly with the result @xmath96 thus we get for some well behaved function @xmath97 @xmath98 the principal value prescription is denoted as `` @xmath99 '' . in our case @xmath100 and
therefore @xmath101 this is part of the unitary relation @xmath102 ( note we used the symmetry of the off- the- energy- shell t - matrix ) .
consequently @xmath103 \label{a10}\end{aligned}\ ] ] and @xmath104 + \omega ( k ' ) \im [ t ( k~ ' , k ; { { k'}^2 \over m } ) ] \right ) \label{a11}\end{aligned}\ ] ] combined with eq .
( [ a2 ] ) certain terms cancel and one arrives at eq.([11 ] ) . | the potentials @xmath0 in the nonrelativistic ( relativistic ) nucleon - nucleon ( nn ) schrdinger equation are related by a quadratic equation .
that equation is numerically solved , thus providing phase equivalent @xmath1 - potentials related for instance to the high precision nn potentials , which are adjusted to nn phase shift and mixing parameters in a nonrelativistic schrdinger equation .
the relativistic nn potentials embedded in a three - nucleon ( 3n ) system for total nn momenta different from zero are also constructed in a numerically precise manner .
they enter into the relativistic interacting 3n mass operator , which is needed for relativistic 3n calculations for bound and scattering states . |
the local cluster substructure survey ( locuss see smith et al . 2003 , 2005 ) sample of clusters contains 164 clusters with redshifts between 0.142 and 0.295 .
the locuss aims to measure the relationship between the structure of galaxy clusters and the evolution of the hot gas and galaxies that inhabit them using gravitational lensing data and other observations spanning the electromagnetic spectrum from the radio to x - ray .
example locuss analysis papers relevant to this work are @xcite and @xcite .
we have imaged a subset of locuss in the sunyaev - zeldovich ( sz ; @xcite ) effect with the arcminute microkelvin imager ( ami ; see e.g. @xcite ) centred at 16 ghz .
the sz signal arises from the inverse compton scattering of cosmic microwave background ( cmb ) photons by the hot cluster plasma ( see e.g. @xcite and @xcite ) , with a surface brightness that is independent of redshift and dependencies on plasma density and temperature that are different to those for e.g. x - ray bremsstrahlung emission .
hereafter , we assume a concordance @xmath5cdm cosmology with @xmath6 = 0.3 , @xmath7 = 0.7 and h@xmath8 = 100kms@xmath9mpc@xmath9 .
the dimensionless hubble parameters are defined as @xmath10 = h@xmath8/(x kms@xmath9mpc@xmath9 ) .
all coordinates are given at equinox j2000 .
in this paper we focus on locuss clusters with a declination greater than 20@xmath11 and an x - ray luminosity ( @xmath12 ) greater than 11 @xmath13w over the 0.1 - 2.4 kev band in the cluster rest frame ( according to ebeling et al .
1998 , 2000 , using @xmath14 ) .
radio source contamination can make it difficult to observe the sz effect at 16 ghz and we have not studied the clusters with sources brighter than 10 mjy / beam within 10@xmath15 of the cluster x - ray centre . note that our redshifts correspond to those cited in ebeling et al .
( 1998 , 2000 ) .
we present results from the analysis of eight galaxy cluster systems ; table [ tab : cluster_coords ] shows the coordinates , redshifts and x - ray luminosities of our selected locuss clusters .
a companion paper ( @xcite ) presents results from 11 locuss clusters with an x - ray luminosity in the range 7 - 11 @xmath13w ( @xmath14 ) . [ cols="<,^,^,^,^,^ " , ] our ami sa maps and the parameters we measure are presented in figure [ fig : a586 ] .
we have overlayed our map on an x - ray _ chandra _ image ; the cluster centroids match and we observe an extension of the cluster towards the south .
the sz effect from abell 586 has previously been observed with ovra / bima by @xcite and @xcite .
laroque et al .
apply an isothermal @xmath0-model to sz and _ chandra _ x - ray observations and find @xmath16 and @xmath17 respectively ( using @xmath18 ) .
in addition , they determine an x - ray spectroscopic temperature of the cluster gas of @xmath19kev between a radius of 100kpc and @xmath20 ; @xmath20 is the radius at which the average cluster density falls to 2500 times the critical density at that redshift and is determined from _
observations by @xcite . in comparison , @xcite use _
subaru _ to calculate the cluster mass from weak lensing by applying a navarro , frenk & white ( nfw ; @xcite ) profile .
they find @xmath21 , whilst at large radii they find @xmath22 ( using @xmath18 ) .
abell 586 has been studied extensively in the x - ray e.g. @xcite and @xcite .
a recent analysis of the temperature profile @xcite shows how the temperature falls from @xmath23 kev at the cluster centre to @xmath24kev at a radius @xmath25 .
cypriano et al . have used the gemini multi - object spectrograph together with x - ray data taken from the _ chandra _ archive to measure the properties of abell 586 .
they compare mass estimates derived from the velocity distribution and from the x - ray temperature profile and find that both give very similar results , @xmath26 within 1.3@xmath27mpc .
they suggest that the cluster is spherical and relaxed with no recent mergers .
the elongation of the sz signal on our map ( figure [ fig : a586 ] ) suggests non - sphericity .
abell 586 we present ami sa maps , large - scale cluster parameters and a _ chandra _ image of abell 611 in figure [ fig : a611 ] .
the @xcite analysis of the _ chandra _ achive data shows that the x - ray isophotes are quite circular , the surface brightness profile is smooth and the brightest cluster galaxy lies at the centre of the x - ray emission .
these results indicate that the cluster is relaxed . from the x - ray data ,
the cluster mass was estimated using an nfw profile , spherical symmetry and hydrostatic equilibrium to be [email protected]@xmath29 ( within a radius of [email protected] mpc ) .
however , the donnarumma et al .
analysis of strong lensing data indicates that the cluster mass could be closer to [email protected]@xmath29 ( within a radius of [email protected] mpc .
note that the values quoted from donnarumma et al .
are an example of their mass estimates ; from fitting different models they find estimated mass varies significantly ( between 9.3211.11@xmath29 for the x - ray mass and between 4.016.32@xmath29 for the lensing mass ) .
their mass estimates use @xmath18 .
several other analyses of _ chandra _ data produce comparable mass estimates ( e.g. @xcite , @xcite , @xcite and @xcite ) .
@xcite perform a weak lensing analysis of abell 611 using data from the large binocular telescope . with an nfw profile they estimate @xmath30 = 47@xmath29 and @xmath31kpc , for @xmath18
these are in agreement with the values obtained from _ subaru
_ weak lensing observations by okabe et al .
using gmrt observations @xcite concluded that abell 611 has no radio halo at 610mhz .
abell 611 has also previously been observed in the sz at 15 ghz by @xcite and zwart et al .
( 2010 ) , and at 30 ghz by @xcite , @xcite and laroque et al . from our analysis of the ami sa observations of abell 611 presented in this paper we find that @xmath30 = 4.0@xmath32@xmath33 .
we note that the mass obtained is significantly smaller than the result given in zwart et al .
( 2010 ) ; however , their @xmath34 estimates are biased high .
the bias occurs because they used a low - radius x - ray temperature as a constant temperature throughout the cluster , as is explained by them and in olamaie et al .
the sz maps presented in this paper are similar to those in zwart et al .
( 2010 ) ; both sets of observations indicate that the cluster is extended in the nw direction .
however , the analysis presented in this paper differs from that by zwart et al .
( 2010 ) who sample from temperature and @xmath35 and derive @xmath36 under the additional assumption of hydrostatic equilbrium .
instead we sample from @xmath30 and @xmath37 and calculate @xmath38 using the m - t scaling relation given in rodrguez - gonzlvez et al .
the differences between these two models are described in detail by olamaie et al . who demonstrate that the mass estimated using the technique in this paper produces a more reliable value and that the zwart et al .
( 2010 ) analysis underestimates the values for @xmath35 and @xmath37 .
the values of @xmath0 and @xmath39kpc presented here agree with those in the zwart et al .
( 2010 ) analysis .
we find no significant contamination from radio sources and detect the cluster with a high signal - to - noise ratio .
a comparison of the sz - effect image and the _ chandra _ map shows that the centres of the sz and x - ray emission are coincident .
a611 in figure [ fig : a773 ] we show the ami sa maps of abell 773 , a _
chandra _ x - ray map and the cluster parameters derived from our analysis .
the sz effect associated with abell 773 has been observed several times ( @xcite , @xcite , @xcite , laroque et al . and @xcite ) .
most recently , zwart et al .
( 2010 ) observed the cluster and found a cluster mass of @xmath40 = 1.9@xmath32@xmath41 using @xmath18 ; however , their @xmath34 estimates are biased high see section [ sec : a611 ] .
inspection of a 10@xmath42 region of the sloan digital sky survey ( sdss ) centred on abell 773 reveals a complex galaxy distribution with some ew extension .
our observations support this extension , but there is no detailed correspondence between the galaxy and gas distributions .
the _ chandra _
observations appear to show little if any such extension .
we find no significant contamination from radio sources and detect the cluster with a high signal to noise ratio .
for this cluster , present an intensive study of the optical data from the telescopio nazionale galileo ( tng ) telescope and x - ray data from the _ chandra _ data archive .
they find two peaks in the velocity distribution of the cluster members which are separated by 2@xmath15 along the e - w direction .
two peaks can also be seen in the x - ray , although these are along the ne - sw direction .
barrena et al .
estimate the virial mass of the entire system to be @xmath43 = 1.2 - 2.7@xmath44 .
@xcite reported the existence of a radio halo in abell 773 .
this feature , typical of cluster mergers , was confirmed with 1.4 ghz vla observations by @xcite .
@xcite used _ xmm - newton _ to study abell 773 and found @xmath45 = 8.3 @xmath282.5 @xmath33 , where @xmath2 = 1.33mpc ; they assumed isothermality , spherical symmetry and @xmath18 . @xcite
present a _
chandra _ temperature map and an x - ray image of abell 773 ; they estimate a mean temperature of [email protected] within a radius of 800kpc ( @xmath18 ) .
abell 773 ami sa maps , derived parameters and a _ chandra _ observation of the abell 781 cluster are presented in figure [ fig : a781 ] . from x - ray observations with _
chandra _ and _ xmm - newton _ ( @xcite ) it is apparent that abell 781 is a complex cluster system .
the main cluster is surrounded by three smaller clusters , two to the east of the main cluster and one to the west .
they estimate the mass of the clusters assuming a nfw matter density profile ; the results indicate that the cluster mass of abell 781 within @xmath2 is @xmath46@xmath47@xmath48 from _ xmm - newton _ and__chandra _ _ x - ray observations or @xmath49@xmath47@xmath48 from the kitt peak mayall 4-m telescope lensing observations ( where @xmath2 is @xmath50 and @xmath51 respectively ) .
alternatively , zhang et al .
use xmm newton observations to estimate @xmath45 = 4.5 @xmath281.3 @xmath33 , where @xmath2 = 1.05mpc , assuming isothermality and spherical symmetry .
both zhang et al . and sehgal et al . use @xmath18 .
the main cluster of abell 781 is also known to contain a diffuse peripheral source at 610mhz ; this was observed with the gmrt by @xcite .
a781 we present the sa maps before and after source subtraction in figure [ fig : a1413 ] .
we also show the derived cluster parameters and we overlay our sz image on a _ chandra _ x - ray map .
abell 1413 has been observed in the x - ray by _ xmm - newton _ ( e.g. @xcite ) , _ chandra _ ( e.g. @xcite and @xcite ) and most recently by the low background _
satellite ( @xcite ) ; sz images have been made of abell 1413 with the ryle telescope at 15 ghz ( @xcite ) and with ovro / bima at 30 ghz ( laroque et al . and @xcite ) .
these analyses indicate that abell 1413 is a relaxed cluster with no evidence of recent merging despite its elliptical morphology . between the x - ray observations there is good agreement in the temperature and density profiles of the cluster out to half the virial radius .
hoshino et al .
measure the variation of the electron temperature with radius , finding a temperature of 7.5kev at the centre and 3.5kev at @xmath1 .
they assume spherical symmetry , an nfw density profile and hydrostatic equilibrium to calculate @xmath30 = [email protected]@xmath52 ; where @xmath1 = @xmath53mpc .
zhang et al .
xmm - newton _ to study abell 1413 and find @xmath45 = 5.4 @xmath281.6 @xmath33 , where @xmath2 = 1.18mpc ; they assume isothermality , spherical symmetry and @xmath18 . recent vla observations ( @xcite ) indicate that there is diffuse 1.4-ghz emission associated with the cluster this may be due to a mini - halo around the cluster . from the sa observations we detect the sz decrement at high significance .
we determine the cluster mass to be @xmath30 = 4.0@xmath54@xmath55 and @xmath1 = 1.14@xmath56@xmath57kpc ; both these values are comparable with the hoshino et al . results .
the source environment around the cluster at 16ghz is reasonable : the brightest source is 14mjy ( 11:55:36.63 + 23:13:50.1 ) , but this is 700@xmath58 from the cluster x - ray centre .
after this bright source is subtracted from our data we are left with a low level of residual flux density on the map ; it is unlikely that this residual flux significantly contaminates our cluster detection or parameters .
both the x - ray map and the sz image indicate that the cluster is elliptical and extended in the n - s direction .
abell 1413 an analysis of _ rosat _ images clearly shows that this system consists of two interacting clusters , abell 1758a and abell 1758b , separated by 8@xmath59 ( @xcite ) .
in figure [ fig : a1758a ] we present a single map that contains combined data from observations towards both clusters and the derived parameters for cluster abell 1758a .
we present the derived parameters for cluster abell 1758a and x - ray maps from both the _ chandra _ data archive and _ _ rosat__. in figure [ fig : a1758b ] we show the derived parameters for cluster abell 1758b .
a detailed analysis of _ xmm - newton _ and _ chandra _ by @xcite indicates that the clusters abell 1758a and abell 1758b are likely to be in an early stage of merging and that both of these clusters are also undergoing major mergers with other smaller systems .
a recent analysis of spitzer / mips 24@xmath60 m data by @xcite classifies abell 1758 as the most active system they have observed at that wavelength . they also identify numerous smaller mass peaks and filamentary structures , which are likely to indicate the presence of infalling galaxy groups , in support of the david & kempner observations .
zhang et al .
xmm - newton _ to study abell 1758a and found @xmath45 = 1.1 @xmath280.3 @xmath41 , where @xmath2 = 1.43mpc .
they assume isothermality , spherical symmetry and @xmath18 .
abell 1758a abell 1758b we detect several sources close to the cluster centre a point source with a flux density of 7.97 mjy at 14:56:59.11 + 22:18:55.97 and a 4.67 mjy source at 14:57:25.38 + 22:37:33.03 .
we do not detect an sz effect from this cluster even though we would expect to , given the low noise levels of our sa maps .
the sa maps and derived parameters are shown in figure [ fig : zw7160 ] .
the derived parameters for this non - detection are as expected : we find that @xmath30 approaches our lower prior limit ( @xmath61 @xmath47 @xmath62 ) and that @xmath63 shows simliar behaviour ; both @xmath1 and @xmath3 are well constrained because both these parameters are derived from @xmath30 which itself is well constrained at the value of its lower prior limit .
zhang et al . used xmm newton to study zw1454.8 + 2233 and found @xmath45 = 2.4 @xmath280.7 @xmath33 , where @xmath2 = 0.87mpc .
they assumed isothermality , spherical symmetry and @xmath18 .
@xcite observed the cluster with the gmrt at 610mhz and found that the cluster contains a core - halo source .
this is in agreement with the value obtained from _
weak lensing observations by okabe et al .
chandra _ x - ray observations ( @xcite ) also reveal that zw1454.8 + 2233 is a cooling core cluster ; these are often associated with core - halos .
zw1454.8 + 2233 at 16 ghz the source environment around the cluster is challenging : in our la data we detect a 3.9 mjy source at the same position as the cluster , and several other sources with comparable flux densitites @xmath64 from the cluster centre .
however , using our bayesian analysis we are able to accurately model the positions , flux densities and spectral indices of these sources such that , after they are subtracted from our sa maps , we see a significant decrement .
the ami sa maps before and after source subtraction are shown in figure [ fig : rxj1720.1 + 2638 ] , as are the derived cluster parameters and a _ chandra _ image of the cluster .
our sz - effect map shows that the cluster may have an irregular shape ; we see low signal - to - noise emission to the se and nw of the cluster x - ray centroid ; however , the centre of the sz emission is coincident with the x - ray centroid .
_ chandra _ observations ( @xcite ) indicate that , although the cluster does not have an irregular shape or elongation , it has discontinuities in its density profile ; this may indicate it is in the latter stages of merging .
the largest discontinuity is observed in the se sector of the cluster and is noted to have a structure similar to a cold front observed in other merging systems such as abell 2142 and abell 3667 .
mazzotta et al .
determined the mass profile for the cluster assuming hydrostatic equilibrium using two distinct regions ( se and nw ) to model the cluster density profile : each region was separately analysed and used to calculate @xmath65 = 4 @xmath28 10 @xmath66 .
rxj1720.1 + 2638 following on from rodrguez - gonzlvez et al .
( 2011 ) , in figure [ temp_comp1 ] we compare the ami sa observed cluster temperatures within @xmath1 ( @xmath3 ) with large - radius x - ray values ( @xmath4 ) from _ chandra _ or _ suzaku _ that we have been able to find in the literature .
we use large radius x - ray temperature values as these ignore the complexities of the cluster core and are representative of the average cluster temperature within @xmath671mpc which is measured by ami . before comment on these
we deal with two technical points .
first , for abell 611 , we have plotted two x - ray values ( from _ chandra _ data ) ; one from the accept archive ( @xcite ) which is higher than our ami sa measurement , while the second x - ray measurement from _ chandra _ ( donnarumma et al . )
is consistent with our measurement .
secondly , the accept archive r= 475 - 550kpc temperature for abell 1758a is 16@xmath287kev and for clarity is not included on the plot . evidently abell 586 , abell 611 ( with the donnarumma et al .
x - ray temperature ) and abell 1413 have corresponding sz and x - ray temperatures while abell 773 , abell 1758a and rxj1720.1 + 2638 have x - ray temperatures significantly higher than their sz temperature .
the position is made clearer by combining the values with those in rodrguez - gonzlvez et al .
the combined data are shown in figure [ temp_comp2 ] , in which there is reasonable correspondence between sz and x - ray temperatures at lower x - ray luminosity , with excess ( over sz ) x - ray temperatures at higher x - ray luminosity .
an exception to this is abell 1413 which despite its high x - ray luminosity is in good agreement with our sz value , but for this cluster we have been able to use _ suzuka _ measurements over r= 700 - 1200kpc .
it is noteworthy that abell 773 , abell 1758a and rxj1720.1 + 2638 are major mergers , and we emphasize the @xmath68 weighting of x - ray temperature measurements .
\i ) we have good sz detections for eight clusters and a non - detection for zw1454.8 + 2233 .
+ ii ) for the seven detected clusters with @xmath69@xmath13w ( @xmath14 ) , we fit @xmath0 profiles to the cluster signals and find @xmath35 values of 1.7 - 4.3@xmath70 and values @xmath36 of 2.0 - 5.1@xmath71 .
+ iii ) for abell 611 and abell 773 our values of @xmath35 and @xmath36 are lower than those in zwart et al . ( 2010 ) which they thought to be biased high , because they use a high value for @xmath3 ( estimated from a low - radius x - ray measurement ) and assume this value to be constant throughout the cluster .
+ iv ) for the six clusters in the work of this paper for which we have found large - radius @xmath72kpc x - ray spectroscopic temperatures in the literature , we find that @xmath4 and @xmath3 values correspond reasonably well for abell 586 , abell 611 ( with the donnarumma et al .
x - ray temperature rather than the accept archive value ) and abell 1413 , but that correspondence falls away for abell 773 , abell 1758a and rxj1720.1 + 2638 which have a high @xmath4 , for these , @xmath3 is less than @xmath4 .
+ v ) the picture seems to become clearer although all of this work involves only very small numbers when we add in the data of rodrguez - gonzlvez et al .
we find that there is reasonable @xmath4:@xmath3 correspondence for the six clusters at lower @xmath4 but that the correspondence breaks down at high @xmath4 .
however , two points are evident .
the more general one is that the breakdown of the @xmath4:@xmath3 correspondence tends to be associated with high @xmath12 and with major mergers .
the more specific one is that abell 1413 has values of @xmath4 and @xmath3 that correspond yet has high @xmath12 : but we have used @xmath4 measured by the _
suzaku _ at very high radius .
+ vi ) we suspect this points to agreement between large - radius sz estimates and larger - radius spectroscopic temperature measurements , but that substantial mergers bias @xmath4 measurements more than @xmath3 ; however we stress again that our sample from that and our companion paper is very small .
we thank pparc / stfc for support of ami and its operations .
we are grateful to the staff of the cavendish laboratory and the mullard radio astronomy observatory for the maintenance and operation of ami .
crg , mo , mps and tws acknowledge pparc / stfc studentships .
this work was performed using the darwin supercomputer of the university of cambridge high performance computing service ( http://www.hpc.cam.ac.uk/ ) , provided by dell inc . using strategic research infrastructure funding from the higher education funding council for england , and the altix 3700 supercomputer at damtp , university of cambridge supported by hefce and stfc .
we are grateful to stuart rankin and andrey kaliazin for their computing assistance . | we present observations from the small array of the arcminute microkelvin imager ( ami ) of eight high x - ray luminosity galaxy cluster systems selected from the local cluster substructure survey ( locuss ) sample .
we detect the sunyaev - zeldovich ( sz ) effect in seven of these clusters . with the assumptions that galaxy clusters are isothermal , have a density profile described by a spherical @xmath0-model and obey the theoretical m - t relation , we are able to derive cluster parameters at @xmath1 from our sz data . with the additional assumption of hydrostatic equilibrium we are able to derive parameters at @xmath2 .
we present posterior probability distributions for cluster parameters such as mass , radius and temperature ( @xmath3 ) .
combining our sample with that of ami consortium : rodrguez - gonzlvez et al .
( 2011 ) and using large - radius x - ray temperature estimates ( @xmath4 ) from _ chandra _ and _ suzaku _ observations , we find that there is reasonable correspondence between @xmath4 and @xmath3 values at low @xmath4 , but that for clusters with @xmath4 above around 6kev the correspondence breaks down with @xmath4 exceeding @xmath3 ; we stress that this finding is based on just ten clusters .
cosmology : observations - sunyaev - zeldovich galaxies : clusters x - ray cosmic microwave background galaxies : clusters : individual ( abell 586 , abell 611 , abell 773 , abell 781 , abell 1413 , abell 1758 , zw1454.8 + 2233 and rxj1720 + 2638 ) |
the belle experiment @xcite , located on the kekb @xcite asymmetric - energy @xmath1 collider , is primarily devoted to the study of cp violation in the @xmath2 meson system .
kekb has shown a very stable operation with increasing luminosity over the years .
it has turned to so - called `` continuous injection mode '' last january , thus allowing a gain in the integrated luminosity of about 30% . in this mode , the beam particle losses are compensated by continuously injecting beam from the linear accelerator , without interruption of data taking . kekb has reached a world record peak luminosity of @xmath3 @xmath4s@xmath0 and an integrated luminosity of about 1 fb@xmath0 per day . in the meanwhile ,
belle has accumulated a total integrated luminosity of 290 fb@xmath0 ( about 275 million @xmath2 meson pairs ) , among which 127 fb@xmath0 was collected between october 2003 and july 2004 ( svd2 run ) .
.belle data acquisition and processing figures .
numbers in parentheses correspond to data accumulated during svd2 run . [
cols="<,<,<",options="header " , ] [ failure - tab ] in addition , a number of limitations to the processing speed have been observed : database access is one of the main issues , since processing makes heavy use of the database , in particular at the start of a job .
using more dedicated servers will solve this issue .
the limited cpu power of tape servers which distribute the data the dbasf clusters was identified as the present bottleneck .
faster tape servers will be used in the future .
the network bandwidth between the input server and basf nodes may eventually limit the processing power .
this does not seem to be a major issue in the near future .
none of these limitations , however , seriously hampered the processing speed .
the new offline processing scheme started running on may 18 , 2004 . in 39 days of stable
running , it successfully processed the 3 billion events collected by kekb with a maximum rate of 5/fb / day , 5 times faster than the data acquisition and the previous scheme . with the expected increase of the kekb luminosity , however , the dst processing will face further challenges room for improvement of this scheme still remains . | the offline processing of the data collected by the belle detector has been recently upgraded to cope with the excellent performance of the kekb accelerator .
the 127 fb@xmath0 of data ( 120 tb on tape ) collected between autumn 2003 and summer 2004 has been processed in 2 months , thanks to the high speed and stability of the new , distributed processing scheme .
we present here this new processing scheme and its performance . |
full counting statistics@xcite ( fcs ) of electron transport through mesoscopic system has attracted considerable attention both experimentally and theoretically because it can provide a deeper insight into the nature of electron transport mechanisms , which can not be obtained from the average current blanter , nazarov , gustavsson , fujisawa , flindt , fricke01,ubbelohde , fricke02,maisi .
for instance , the shot noise measurements can be used to probe the dynamical in an open double quantum dots ( qds ) @xcite , the coherent coupling between serially coupled qds @xcite , the evolution of the kondo effect in a qd @xcite , and the conduction channels of quantum conductors @xcite . in particular , shot noise characteristics can provide information about the feature of the pseudospin kondo effect in a laterally coupled double qds @xcite , the spin accumulations in a electron reservoir @xcite , and the charge fractionalization in the @xmath0 quantum hall edge milletari .
in addition , the degree of entanglement of two electrons in the double qds @xcite , the dephasing rate in a closed qd @xcite , the internal level structure of single molecule magnet xuejap10,xuepla11 can be characterized by the super - poissonian shot noise . on the other hand ,
the quantum coherence in coupled qd system , which is characterized by the off - diagonal elements of the reduced density matrix of the qd system within the framework of the density matrix theory@xcite , plays an important role in the electron tunneling processes and has a significant influence on electron transport gurvitz , braun , wunsch , djuric01,djuric02,harbola , pedersen , begemann , darau , schultz , schaller .
in particular , theoretical studies have demonstrated that the high - order cumulants , e.g. , the shot noise , the skewness , are more sensitive to the quantum coherence than the average current in the different types of qd systems @xcite and the quantum coherence information in a side - coupled double qd system can be extracted from the high - order current cumulants @xcite .
in fact , the non - markovian dynamics of the qd system also plays an important role in the non - equilibrium electron tunneling processes .
however , the above studies on current noise or fcs were mainly based on the different types of markovian master equations .
although the influence of non - markovian effect on the long - time limit of the fcs in the qd systems has received some attention schaller , braggio01,flindt01,zedler01,emary01,flindt02,marcos , emary02,zedler02 , how the non - markovian effect affects the fcs is still an open issue , especially the influence of the interplay between the quantum coherence and non - markovian effect on the long - time limit of the fcs has not yet been revealed .
the aim of this report is thus to derive a non - markovian fcs formalism based on the exact time - convolutionless ( tcl ) master equation and study the influences of the quantum coherence and non - markovian effect on the fcs in qd molecule systems .
it is demonstrated that the non - markovian effect manifests itself through the quantum coherence of the considered qd molecule system , and has a significant impact on the fcs in the high quantum - coherent qd molecule system , which depends on the coupling of the considered qd molecule system with the incident and outgoing electrodes .
consequently , it is necessary to consider the influence of the non - markovian effect on the full counting statistics of electron transport in a high quantum - coherent single - molecule system .
we now study the influences of the quantum coherence and non - markovian effect on the fcs of electronic transport through the qd molecule system . in order to facilitate discussions effectively
, we consider three typical qd systems , namely , single qd without quantum coherence , serially coupled double qds and side - coupled double qds with high quantum coherence in a certain parameter regime ( see fig .
1 ) . in addition
, we assume the bias voltage ( @xmath1 ) is symmetrically entirely dropped at the qd - electrode tunnel junctions , which implies that the levels of the qds are independent of the applied bias voltage even if the couplings are not symmetric , and choose mev as the unit of energy which corresponds to a typical experimental situation@xcite . in this subsection , we consider a single qd weakly coupled to two ferromagnetic electrodes .
the hamiltonian of the considered system is described by the @xmath2 .
the qd hamiltonian @xmath3 is given by @xmath4where @xmath5 ( @xmath6 ) creates ( annihilates ) an electron with spin @xmath7 and on - site energy @xmath8 ( which can be tuned by a gate voltage @xmath9 ) in this qd system .
@xmath10 is the intradot coulomb interaction between two electrons in the qd system . the relaxation in the two ferromagnetic electrodes is assumed to be sufficiently fast , so that their electron distributions can be described by equilibrium fermi functions .
the two electrodes are thus modeled as non - interacting fermi gases and the corresponding hamiltonians can be expressed as @xmath11where @xmath12 ( @xmath13 ) creates ( annihilates ) an electron with energy @xmath14 , spin @xmath15 and momentum @xmath16 in @xmath17 ( @xmath18 ) electrode , and @xmath19 denotes the majority ( minority ) spin states with the density of states @xmath20 .
the polarization vectors @xmath21 ( left lead ) and @xmath22 ( right lead ) are parallel to each other , and their magnitudes are characterized by @xmath23 .
the tunneling between the qd and the electrodes is described by @xmath24where spin - up @xmath25 and spin - down @xmath26 are defined to be the majority spin and minority spin of the ferromagnet , respectively .
the qd - electrode coupling is assumed to be sufficiently weak , thus , the sequential tunneling is dominant and can be well described by the quantum master equation of reduced density matrix spanned by the eigenstates of the qd .
the particle - number - resolved tcl quantum master equation for the reduced density matrix of the considered single qd is given by @xmath27 , \label{master1}\end{aligned}\]]for more details , see methods section . here
, the complete basis @xmath28 , @xmath29 , @xmath30 , @xmath31 is chosen to describe the electronic states of this single qd system , and the single qd system parameters are chosen as @xmath32 , @xmath33 , @xmath34 and @xmath35 .
figure 2 shows the first four current cumulants as a function of the bias voltage for different ratios @xmath36 describing the left - right asymmetry of the qd - electrode coupling .
we found that the non - markovian effect has no influence on the current noise behaviors of the single qd considered here , see fig .
2 . scrutinizing eq .
( [ master1 ] ) , it is found that for the non - markovian case the elements of the reduced density matrix are equivalent to that for the markovian case because there are not the off - diagonal elements of the reduced density matrix .
thus , the equations of motion of the four elements of the reduced density matrix can be expressed as @xmath37 \left\langle 0,0\right\vert \rho_{s}\left ( t\right ) \left\vert 0,0\right\rangle \nonumber\\ & -\left [ \gamma_{l\downarrow}f_{l,+}\left ( \epsilon_{\downarrow}\right ) + \gamma_{r\downarrow}f_{r,+}\left ( \epsilon_{\downarrow}\right ) \right ] \left\langle 0,0\right\vert \rho_{s}\left ( t\right ) \left\vert 0,0\right\rangle \nonumber\\ & + \left [ \gamma_{l\uparrow}f_{l,-}\left ( \epsilon_{\uparrow}\right ) + \gamma_{r\uparrow}f_{r,-}\left ( \epsilon_{\uparrow}\right ) e^{i\chi } \right ] \left\langle 0,\uparrow\right\vert \rho_{s}\left ( t\right ) \left\vert \uparrow,0\right\rangle \nonumber\\ & + \left [ \gamma_{l\downarrow}f_{l,-}\left ( \epsilon_{\downarrow}\right ) + \gamma_{r\downarrow}f_{r,-}\left ( \epsilon_{\downarrow}\right ) e^{i\chi } \right ] \left\langle \downarrow,0\right\vert \rho_{s}\left ( t\right ) \left\vert 0,\downarrow\right\rangle , \label{zero1}%\end{aligned}\ ] ] @xmath38 \left\langle 0,0\right\vert \rho_{s}\left ( t\right ) \left\vert 0,0\right\rangle \nonumber\\ & -\left [ \gamma_{l\uparrow}f_{l,-}\left ( \epsilon_{\uparrow}\right ) + \gamma_{r\uparrow}f_{r,-}\left ( \epsilon_{\uparrow}\right ) \right ] \left\langle 0,\uparrow\right\vert \rho_{s}\left ( t\right ) \left\vert \uparrow,0\right\rangle \nonumber\\ & -\left [ \gamma_{l\downarrow}f_{l,+}\left ( \epsilon_{\uparrow\downarrow } -\epsilon_{\uparrow}\right ) + \gamma_{r\downarrow}f_{r,+}\left ( \epsilon_{\uparrow\downarrow}-\epsilon_{\uparrow}\right ) \right ] \left\langle 0,\uparrow\right\vert \rho_{s}\left ( t\right ) \left\vert \uparrow,0\right\rangle \nonumber\\ & + \left [ \gamma_{l\downarrow}f_{l,-}\left ( \epsilon_{\uparrow\downarrow } -\epsilon_{\uparrow}\right ) + \gamma_{r\downarrow}f_{r,-}\left ( \epsilon_{\uparrow\downarrow}-\epsilon_{\uparrow}\right ) e^{i\chi}\right ] \left\langle \downarrow,\uparrow\right\vert \rho_{s}\left ( t\right ) \left\vert \uparrow,\downarrow\right\rangle , \label{oneup1}%\end{aligned}\ ] ] @xmath39 \left\langle 0,0\right\vert \rho_{s}\left ( t\right ) \left\vert 0,0\right\rangle \nonumber\\ & -\left [ \gamma_{l\downarrow}f_{l,-}\left ( \epsilon_{\downarrow}\right ) + \gamma_{r\downarrow}f_{r,-}\left ( \epsilon_{\downarrow}\right ) \right ] \left\langle \downarrow,0\right\vert \rho_{s}\left ( t\right ) \left\vert 0,\downarrow\right\rangle \nonumber\\ & -\left [ \gamma_{l\uparrow}f_{l,+}\left ( \epsilon_{\uparrow\downarrow } -\epsilon_{\downarrow}\right ) + \gamma_{r\uparrow}f_{r,+}\left ( \epsilon_{\uparrow\downarrow}-\epsilon_{\downarrow}\right ) \right ] \left\langle \downarrow,0\right\vert \rho_{s}\left ( t\right ) \left\vert 0,\downarrow\right\rangle \nonumber\\ & + \left [ \gamma_{l\uparrow}f_{l,-}\left ( \epsilon_{\uparrow\downarrow } -\epsilon_{\downarrow}\right ) + \gamma_{r\uparrow}f_{r,-}\left ( \epsilon_{\uparrow\downarrow}-\epsilon_{\downarrow}\right ) e^{i\chi}\right ] \left\langle \downarrow,\uparrow\right\vert \rho_{s}\left ( t\right ) \left\vert \uparrow,\downarrow\right\rangle , \label{onedown1}%\end{aligned}\ ] ] @xmath40 \left\langle 0,\uparrow\right\vert \rho_{s}\left ( t\right ) \left\vert \uparrow,0\right\rangle \nonumber\\ & + \left [ \gamma_{l\uparrow}f_{l,+}\left ( \epsilon_{\uparrow\downarrow } -\epsilon_{\downarrow}\right ) + \gamma_{r\uparrow}f_{r,+}\left ( \epsilon_{\uparrow\downarrow}-\epsilon_{\downarrow}\right ) e^{-i\chi}\right ] \left\langle \downarrow,0\right\vert \rho_{s}\left ( t\right ) \left\vert 0,\downarrow\right\rangle \nonumber\\ & -\left [ \gamma_{l\uparrow}f_{l,-}\left ( \epsilon_{\uparrow\downarrow } -\epsilon_{\downarrow}\right ) + \gamma_{r\uparrow}f_{r,-}\left ( \epsilon_{\uparrow\downarrow}-\epsilon_{\downarrow}\right ) \right ] \left\langle \downarrow,\uparrow\right\vert \rho_{s}\left ( t\right ) \left\vert \uparrow,\downarrow\right\rangle \nonumber\\ & -\left [ \gamma_{l\downarrow}f_{l,-}\left ( \epsilon_{\uparrow\downarrow } -\epsilon_{\uparrow}\right ) + \gamma_{r\downarrow}f_{r,-}\left ( \epsilon_{\uparrow\downarrow}-\epsilon_{\uparrow}\right ) \right ] \left\langle \downarrow,\uparrow\right\vert \rho_{s}\left ( t\right ) \left\vert \uparrow,\downarrow\right\rangle . \label{two1}%\end{aligned}\ ] ] here , @xmath41 is the fermi function of the electrode @xmath42 , and @xmath43 .
the detailed procedure for calculation of the equation of motion of a reduced density matrix , see methods section . within the framework of the density matrix theory ,
the off - diagonal elements of the reduced density matrix characterize the quantum coherence of the considered qd system .
thus , the influence of the non - markovian effect on the fcs may be associated with the quantum coherence of the considered qd system . in order to confirm this conclusion
, we take serially coupled and side - coupled double qds for illustration in the following two subsection .
we now consider two serially coupled double qds weakly connected to two metallic electrodes , see fig .
for the sake of simplicity , the spin degree of freedom has not been considered .
the double - qd is described by a spinless hamiltonian @xmath44where @xmath45 ( @xmath46 ) creates ( annihilates ) an electron with energy @xmath47 ( which can be tuned by a gate voltage @xmath9 ) in @xmath48th qd .
@xmath10 is the interdot coulomb repulsion between two electrons in the double qd system , where we consider the intradot coulomb interaction @xmath49 so that the double - electron occupation in the same qd is prohibited .
the last term of @xmath3 describes the hopping coupling between the two dots with @xmath50 being the hopping parameter .
the two metallic electrodes are modeled as non - interacting fermi gases and the corresponding hamiltonians are given by @xmath51where @xmath52 ( @xmath53 ) creates ( annihilates ) an electron with energy @xmath14 and momentum @xmath16 in @xmath17 ( @xmath18 ) electrode .
the tunneling between the double qds and the two electrodes is described by @xmath54for the case of the weak qd - electrode coupling , the particle - number - resolved tcl quantum master equation for the reduced density matrix of the considered serially double - qd system reads @xmath55 , \label{master2}\end{aligned}\]]here , we can diagonalize the serially coupled double qds hamiltonian @xmath56 in the basis represented by the electron occupation numbers in the qd-1 and qd-2 denoted respectively by @xmath57 and @xmath58 , namely , @xmath59 , and obtain the corresponding four eigenstates of the considered serially coupled double qds system@xcite @xmath60with @xmath61and@xmath62here , we focus on the regime @xmath63 , where the hopping coupling between the two qds strongly modifies the internal dynamics , and the off - diagonal elements of the reduced density matrix play an essential role in the electron tunneling processesgurvitz , stoof96,aguado04,luojy11 . in the following numerical calculations , thus , the parameters of the serially coupled double qds system are chosen as @xmath64 , @xmath65 , @xmath66 and @xmath67 . when the coupling of the qd-2 with the right ( drain ) electrode is stronger than that of the qd-1 with the left ( source ) electrode , namely , @xmath68 , we plot the first four current cumulants as a function of the bias voltage for different values of the qd-2-electrode coupling @xmath69 at @xmath70 in figs .
3(a)-3(d ) .
we found that the non - markovian effect has a very weak influence on the fcs .
interestingly , the high - order current cumulants the skewness and the kurtosis can still show the tiny differences , see figs . 3(c ) and 3(d ) . whereas for the @xmath71 case , the non - markovian effect has a significant impact on the fcs , see fig .
4 . especially , for a relatively large value of the ratio @xmath72 and the coupling of the qd-1 with the left electrode being stronger than the hoping coupling , namely , @xmath73 , the non - markovian effect can induce a strong negative differential conductance ( ndc ) and super - poissonian noise , see figs .
4(e ) and 4(f ) .
in addition , in the case of @xmath71 and @xmath73 , the transitions of the skewness and the kurtosis from positive ( negative ) to negative ( positive ) values are observed , see the dotted line in fig .
4(c ) , the dotted and dash - dot - dotted lines in fig .
4(d ) , and the dash - dot - dotted line in fig .
it is well known that the skewness and the kurtosis ( both its magnitude and sign ) characterize , respectively , the asymmetry of and the peakedness of the distribution around the average transferred - electron number @xmath74 during a time interval @xmath75 , thus that provides further information for the counting statistics beyond the shot noise . to discuss the underlying mechanisms of the current noise clearly , for the system parameters considered here ,
the two singly - occupied eigenstates and eigenvalues can be expressed as @xmath76here we have utilized the equations @xmath77 and @xmath78 . in this situation ,
the equations of motion of the six elements of the reduced density matrix are given by @xmath79 \left\langle 0,0\right\vert \rho _ { s}\left ( t\right ) \left\vert 0,0\right\rangle \nonumber\\ & + \frac{1}{2}\left [ \gamma_{l}f_{l,-}\left ( \epsilon\right ) + \gamma _ { r}f_{r,-}\left ( \epsilon\right ) e^{i\chi}\right ] \left\langle 1\right\vert ^{+}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{+}\nonumber\\ & -\frac{1}{2}\left [ \gamma_{l}f_{l,-}\left ( \epsilon\right ) -\gamma _ { r}f_{r,-}\left ( \epsilon\right ) e^{i\chi}\right ] \left\langle 1\right\vert ^{+}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{-}\nonumber\\ & -\frac{1}{2}\left [ \gamma_{l}f_{l,-}\left ( \epsilon\right ) -\gamma _ { r}f_{r,-}\left ( \epsilon\right ) e^{i\chi}\right ] \left\langle 1\right\vert ^{-}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{+}\nonumber\\ & + \frac{1}{2}\left [ \gamma_{l}f_{l,-}\left ( \epsilon\right ) + \gamma _ { r}f_{r,-}\left ( \epsilon\right ) e^{i\chi}\right ] \left\langle 1\right\vert ^{-}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{- } , \label{zero2}%\end{aligned}\]]@xmath80 \left\langle 0,0\right\vert \rho_{s}\left ( t\right ) \left\vert 0,0\right\rangle \nonumber\\ & -\frac{1}{2}% % tcimacro{\dsum \limits_{\alpha = l , r}}% % beginexpansion { \displaystyle\sum\limits_{\alpha = l , r } } % endexpansion \gamma_{\alpha}\left [ f_{\alpha,+}\left ( \epsilon+u\right ) + f_{\alpha , -}\left ( \epsilon\right ) \right ] \left\langle 1\right\vert ^{\pm}\rho _ { s}\left ( t\right ) \left\vert 1\right\rangle ^{\pm}\nonumber\\ & \pm\frac{1}{2}\frac{\gamma_{l}}{2\pi}\left ( i\phi_{l}\mp\pi f_{l}\right ) \left\langle
1\right\vert ^{+}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{-}\mp\frac{1}{2}\frac{\gamma_{r}}{2\pi}\left ( i\phi_{r}% \mp\pi f_{r}\right ) \left\langle 1\right\vert ^{+}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{-}\nonumber\\ & \mp\frac{1}{2}\frac{\gamma_{\alpha}}{2\pi}\left ( i\phi_{l}\pm\pi f_{l}\right ) \left\langle 1\right\vert ^{-}\rho_{s}\left ( t\right )
\left\vert 1\right\rangle ^{+}\pm\frac{1}{2}\frac{\gamma_{r}}{2\pi}\left ( i\phi_{r}\pm\pi f_{r}\right ) \left\langle 1\right\vert ^{-}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{+}\nonumber\\ & + \frac{1}{2}\left [ \gamma_{l}f_{l,-}\left ( \epsilon+u\right ) + \gamma _ { r}f_{r,-}\left ( \epsilon+u\right ) e^{i\chi}\right ] \left\langle 1,1\right\vert \rho_{s}\left ( t\right ) \left\vert 1,1\right\rangle , \label{add2}%\end{aligned}\]]@xmath81
\left\langle 0,0\right\vert \rho_{s}\left ( t\right ) \left\vert 0,0\right\rangle \nonumber\\ & \pm\frac{1}{2}\frac{\gamma_{l}}{2\pi}\left ( i\phi_{l}\mp\pi f_{l}\right ) \left\langle
1\right\vert ^{+}\rho_{s}^{\left ( n\right ) } \left ( t\right ) \left\vert 1\right\rangle ^{+}\mp\frac{1}{2}\frac{\gamma_{r}}{2\pi}\left ( i\phi_{r}\mp\pi f_{r}\right ) \left\langle 1\right\vert ^{+}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{+}\nonumber\\ & -\frac{1}{2}% % tcimacro{\dsum \limits_{\alpha = l , r}}% % beginexpansion { \displaystyle\sum\limits_{\alpha = l , r } } % endexpansion \gamma_{\alpha}\left [ f_{\alpha,+}\left ( \epsilon+u\right ) + f_{\alpha , -}\left ( \epsilon\right ) \right ] \left\langle 1\right\vert ^{\pm}\rho _ { s}\left ( t\right ) \left\vert 1\right\rangle ^{\mp}\mp2ij\left\langle 1\right\vert ^{\pm}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{\mp } \nonumber\\ & \mp\frac{1}{2}\frac{\gamma_{l}}{2\pi}\left ( i\phi_{l}\pm\pi f_{l}\right ) \left\langle 1\right\vert ^{-}\rho_{s}^{\left ( n\right ) } \left ( t\right ) \left\vert 1\right\rangle ^{-}\pm\frac{1}{2}\frac{\gamma_{r}}{2\pi}\left ( i\phi_{r}\pm\pi f_{r}\right ) \left\langle 1\right\vert ^{-}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{-}\nonumber\\ & + \frac{1}{2}\left [ \gamma_{l}f_{l,-}\left ( \epsilon+u\right ) -\gamma _ { r}f_{r,-}\left ( \epsilon+u\right ) e^{i\chi}\right ] \left\langle 1,1\right\vert \rho_{s}\left ( t\right ) \left\vert 1,1\right\rangle , \label{addsub2}%\end{aligned}\]]@xmath82 \left\langle 1\right\vert ^{+}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{+}\nonumber\\ & + \frac{1}{2}\left [ \gamma_{l}f_{l,+}\left ( \epsilon+u\right ) -\gamma _ { r}f_{r,+}\left ( \epsilon+u\right ) e^{-i\chi}\right ] \left\langle 1\right\vert ^{+}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{-}\nonumber\\ & + \frac{1}{2}\left [ \gamma_{l}f_{l,+}\left ( \epsilon+u\right ) -\gamma _ { r}f_{r,+}\left ( \epsilon+u\right ) e^{-i\chi}\right ] \left\langle 1\right\vert ^{-}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{+}\nonumber\\ & + \frac{1}{2}\left [ \gamma_{l}f_{l,+}\left ( \epsilon+u\right ) + \gamma _ { r}f_{r,+}\left ( \epsilon+u\right ) e^{-i\chi}\right ] \left\langle 1\right\vert ^{-}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{-}\nonumber\\ & -\left [ \gamma_{l}f_{l,-}\left ( \epsilon+u\right ) + \gamma_{r}% f_{r,-}\left ( \epsilon+u\right ) \right ] \left\langle 1,1\right\vert \rho_{s}\left ( t\right ) \left\vert 1,1\right\rangle , \label{two2}%\end{aligned}\ ] ] where @xmath83 , @xmath84 $ ] ( @xmath85 is the digamma function ) and @xmath86 .
compared with the markovian case , it is obvious that the non - markovian effect manifests itself through the off - diagonal elements of the reduced density matrix , namely , the quantum coherence of the considered qds system . in fig .
5(a ) , we plot the functions @xmath87 @xmath88 , @xmath89 @xmath90 and @xmath91 @xmath92 as a function of bias voltage .
it is clearly evident that the values of the functions @xmath87 and @xmath89 show significant variations with increasing bias voltage , especially in the vicinity of the bias voltages @xmath93 and @xmath94 because the new transport channels begin to participate in quantum transport ; while @xmath95 has a gentle variation .
consequently , the non - markovian effects in the @xmath71 case have a remarkable impact on the fcs , see fig .
4 . moreover , for @xmath96 case , the non - markovian effect has a more significant on the fcs than the @xmath97 case , which originates from the qd-2-electrode coupling @xmath69 is weaker than the hoping coupling @xmath50 , where the electron tunneling from qd-1 can not tunnel out qd-2 very quickly and still influence the internal dynamics . in order to illustrate whether the non - markovian effect has a weak influence on the fcs in a relatively small quantum - coherent qd system , we consider the regime @xmath98 @xmath99 , where the off - diagonal elements of the reduced density matrix have little influence on the electron tunneling processes .
we find that for the @xmath100 case the diagonal elements of the reduced density matrix play a major role in the electron tunneling processes , and the non - markovian effect in this case indeed has little impact on the fcs , see figs .
3(e)-3(h ) .
consequently , the influence of the non - markovian effect on the fcs depends on the quantum coherence of the considered qd system . to prove whether this conclusion is universal or not , we take side - coupled double qds for further illustration in the following subsection .
we consider here a side - coupled double qds system . in this case , the qd-1 is only weakly coupled to the two electrodes , see fig .
the qd - electrode tunneling is thus described by @xmath101 in the case of the qd - electrode weak coupling , the particle - number - resolved tcl quantum master equation for the side - coupled double qds can be expressed as @xmath102 .
\label{master3}\end{aligned}\]]here , the eigenstates and eigenvalues of the side - coupled double qds system are the same as the serially coupled double qds system . in the following numerical calculations , the parameters of the side - coupled qds system are chosen as @xmath64 , @xmath65 , @xmath33 and @xmath103 .
for the present side - coupled qds system with high quantum coherence , we find that for @xmath71 case the non - markovian effect has a more remarkable impact on the fcs than that in the serially coupled double qds system , but the ndc does not appear , see figs . 4 and 6 .
for instance , in the case of @xmath73 and @xmath97 , the non - markovian effect can further enhance the super - poissonian shot noise , see the dotted and dash - dot - dotted lines in fig .
6(b ) ; and the transitions of the skewness and the kurtosis from a relatively small positive to a large negative values take place , especially for a relatively large value @xmath104 the kurtosis can be further decreased to a very large negative value , see the dotted and dash - dot - dotted lines in figs .
6(c ) and 6(d ) .
while for the @xmath73 and @xmath96 case the non - markovian effect can enhance the shot noise to a super - poissonian value , see the dotted and dash - dot - dotted lines in fig .
6(f ) , and the transition of the kurtosis from small positive to large negative values only takes place , see the dotted and dash - dot - dotted lines in fig .
for the system parameters considered here , namely , in the limit of @xmath78 , the equations of motion of the six elements of the reduced density matrix read @xmath79 \left\langle 0,0\right\vert \rho _ { s}\left ( t\right ) \left\vert 0,0\right\rangle \nonumber\\ & + \frac{1}{2}\left [ \gamma_{l}f_{l,-}\left ( \epsilon\right ) + \gamma _ { r}f_{r,-}\left ( \epsilon\right ) e^{i\chi}\right ] \left\langle 1\right\vert ^{+}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{+}\nonumber\\ & -\frac{1}{2}\left [ \gamma_{l}f_{l,-}\left ( \epsilon\right ) + \gamma _ { r}f_{r,-}\left ( \epsilon\right ) e^{i\chi}\right ] \left\langle 1\right\vert ^{+}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{-}\nonumber\\ & -\frac{1}{2}\left [ \gamma_{l}f_{l,-}\left ( \epsilon\right ) + \gamma _ { r}f_{r,-}\left ( \epsilon\right ) e^{i\chi}\right ] \left\langle 1\right\vert ^{-}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{+}\nonumber\\ & + \frac{1}{2}\left [ \gamma_{l}f_{l,-}\left ( \epsilon\right ) + \gamma _ { r}f_{r,-}\left ( \epsilon\right ) e^{i\chi}\right ] \left\langle 1\right\vert ^{-}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{- } , \label{zero3}%\end{aligned}\]]@xmath105 \left\langle 0,0\right\vert \rho_{s}\left ( t\right ) \left\vert 0,0\right\rangle \nonumber\\ & -\frac{1}{2}% % tcimacro{\dsum \limits_{\alpha
= l , r}}% % beginexpansion { \displaystyle\sum\limits_{\alpha = l , r } } % endexpansion \gamma_{\alpha}\left [ f_{\alpha,+}\left ( \epsilon+u\right ) + f_{\alpha , -}\left ( \epsilon\right ) \right ] \left\langle 1\right\vert ^{\pm}\rho _ { s}\left ( t\right ) \left\vert 1\right\rangle ^{\pm}\nonumber\\ & \pm\frac{1}{2}% % tcimacro{\dsum \limits_{\alpha = l , r}}% % beginexpansion { \displaystyle\sum\limits_{\alpha = l , r } } % endexpansion \frac{\gamma_{\alpha}}{2\pi}\left ( i\phi_{\alpha}\mp\pi f_{\alpha}\right ) \left\langle 1\right\vert ^{+}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{-}\nonumber\\ & \mp\frac{1}{2}% % tcimacro{\dsum \limits_{\alpha = l , r}}% % beginexpansion { \displaystyle\sum\limits_{\alpha = l , r } } % endexpansion \frac{\gamma_{\alpha}}{2\pi}\left [ i\phi_{\alpha}\pm\pi f_{\alpha}\right ] \left\langle 1\right\vert ^{-}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{+}\nonumber\\ & + \frac{1}{2}\left [ \gamma_{l}f_{l,-}\left ( \epsilon+u\right ) + \frac { 1}{2}\gamma_{r}f_{r,-}\left ( \epsilon+u\right ) e^{i\chi}\right ] \left\langle 1,1\right\vert \rho_{s}\left ( t\right ) \left\vert 1,1\right\rangle , \label{add3}%\end{aligned}\]]@xmath106 \left\langle 0,0\right\vert \rho_{s}\left ( t\right ) \left\vert 0,0\right\rangle \nonumber\\ & \pm\frac{1}{2}% % tcimacro{\dsum \limits_{\alpha = l , r}}% % beginexpansion { \displaystyle\sum\limits_{\alpha = l , r } } % endexpansion \frac{\gamma_{\alpha}}{2\pi}\left [ i\phi_{\alpha}\mp\pi f_{\alpha}\right ] \left\langle 1\right\vert ^{+}\rho_{s}^{\left ( n\right ) } \left ( t\right ) \left\vert
1\right\rangle ^{+}\nonumber\\ & -\frac{1}{2}% % tcimacro{\dsum \limits_{\alpha = l , r}}% % beginexpansion { \displaystyle\sum\limits_{\alpha = l , r } } % endexpansion \gamma_{\alpha}\left [ f_{\alpha,+}\left ( \epsilon+u\right ) + f_{\alpha , -}\left ( \epsilon\right ) \right ] \left\langle 1\right\vert ^{\pm}\rho _ { s}\left ( t\right ) \left\vert 1\right\rangle ^{\mp}\mp2ij\left\langle 1\right\vert ^{\pm}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{\mp } \nonumber\\ & \mp\frac{1}{2}% % tcimacro{\dsum \limits_{\alpha = l , r}}% % beginexpansion { \displaystyle\sum\limits_{\alpha = l , r } } % endexpansion \frac{\gamma_{\alpha}}{2\pi}\left [ i\phi_{\alpha}\pm\pi f_{\alpha}\right ] \left\langle 1\right\vert ^{-}\rho_{s}^{\left ( n\right ) } \left ( t\right ) \left\vert
1\right\rangle ^{-}\nonumber\\ & + \frac{1}{2}\left [ \gamma_{l}f_{l,-}\left ( \epsilon+u\right ) + \gamma _ { r}f_{r,-}\left ( \epsilon+u\right ) e^{i\chi}\right ] \left\langle 1,1\right\vert \rho_{s}\left ( t\right ) \left\vert 1,1\right\rangle , \label{addsub3}%\end{aligned}\]]@xmath82 \left\langle 1\right\vert ^{+}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{+}\nonumber\\ & + \frac{1}{2}\left [ \gamma_{l}f_{l,+}\left ( \epsilon+u\right ) + \gamma _ { r}f_{r,+}\left ( \epsilon+u\right ) e^{-i\chi}\right ] \left\langle 1\right\vert ^{+}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{-}\nonumber\\ & + \frac{1}{2}\left [ \gamma_{l}f_{l,+}\left ( \epsilon+u\right ) + \gamma _ { r}f_{r,+}\left ( \epsilon+u\right ) e^{-i\chi}\right ] \left\langle 1\right\vert ^{-}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{+}\nonumber\\ & + \frac{1}{2}\left [ \gamma_{l}f_{l,+}\left ( \epsilon+u\right ) + \gamma _ { r}f_{r,+}\left ( \epsilon+u\right ) e^{-i\chi}\right ] \left\langle 1\right\vert ^{-}\rho_{s}\left ( t\right ) \left\vert 1\right\rangle ^{-}\nonumber\\ & -\left [ \gamma_{l}f_{l,-}\left ( \epsilon+u\right ) + \gamma_{r}% f_{r,-}\left ( \epsilon+u\right ) \right ] \left\langle 1,1\right\vert \rho_{s}\left ( t\right ) \left\vert 1,1\right\rangle .
\label{two3}%\end{aligned}\ ] ] from the above four equations , we find that these characteristics also originate from the quantum coherence of the side - coupled double qds , and can also be understood in terms of the functions @xmath107 and @xmath108 , which have considerable variations in the vicinity of the bias voltages @xmath93 and @xmath109 because the new transport channels begin to enter the bias voltage window , see the solid and dashed lines in fig .
5(b ) . as for the @xmath110 case the non - markovian effect
has a slightly influence on the fcs because the function @xmath111 has a gentle variation with increasing the bias voltage , see the dotted line in fig .
5(b ) , which is the same as the serially coupled double qds system , see figs .
3(a)-3(d ) and 7 .
in addition , it should be pointed out that for @xmath97 the non - markovian effect has a stronger impact on the fcs than that for @xmath112 case , which is contrary to the case of the serially coupled double qds system . for the the side - coupled double qds system
, the quantum coherence originates from the quantum interference between the direct electron tunneling process , namely , the conduction - electron tunneling into the qd-1 and then directly tunneling out of the qd-1 onto the drain electrode , and the indirect tunneling process , namely , the conduction - electron from the source electrode first tunneling from the qd-1 to the qd-2 , then tunneling back into the qd-1 , and at last tunneling out of the qd-1 .
thus , the fast direct tunneling process in the @xmath113 case can be suppressed compared with the @xmath114 case , which leads to the non - markovian effect has a relatively strong impact on the fcs in the @xmath97 case .
we have developed a non - markovian fcs formalism based on the exact tcl master equation , and studied the influence of the interplay between the quantum coherence and non - markovian effect on the long - time limit of the fcs in three qd systems , namely , single qd , serially coupled double qds and side - coupled double qds .
it is demonstrated that the non - markovian effect manifests itself through the quantum coherence of the considered qd molecule system , and especially has a significant impact on the fcs in the high quantum - coherent qd molecule system , which depends on the coupling of the considered qd molecule system with the source and drain electrodes .
for the single qd system without quantum coherence , the non - markovian effect has no influence on the current noise properties ; whereas for the serially coupled and side - coupled double qds systems with high quantum coherence , that has a remarkable impact on the fcs when the coupling of the considered qd molecule with the incident electrode is equal to or stronger than that with the outgoing electrode .
for instance , for the high quantum - coherent serially coupled double qds system , the non - markovian effect can induce a strong ndc and change the shot noise from the sub - poissonian to super - poissonian distribution in the case of @xmath115 and @xmath116 ; while for the high quantum - coherent side - coupled double qds system , that can remarkably enhance the super - poissonian noise or the sub - poissonian noise for the @xmath71 case . moreover , the non - markovian effect can also lead to the occurrences of the skewness and kurtosis from small positive to large negative values .
these results indicated that the influence of the non - markovian effect on the long - time limit of the fcs should be considered in a highly quantum - coherent single - molecule system .
we consider a general transport setup consisting of a single - level qd molecule weakly coupled to the two electrodes , see fig .
1 , which is described by the following hamiltonian @xmath117here , the first term @xmath118 stands for the hamiltonians of the two electrodes , with @xmath119 being the energy dispersion , and @xmath120 @xmath121 the annihilation ( creation ) operators in the @xmath122 electrode . the second term @xmath123 , which may contain vibrational or spin degrees of freedom and different types of many - body interaction , represents the qd molecule hamiltonian , where @xmath124 @xmath125 is the creation ( annihilation ) operator of electrons in a quantum state denoted by @xmath126 . the third term @xmath127 describes the tunneling coupling between the qd molecule and the two electrodes , which is assumed to be a sum of bilinear terms that each create an electron in the qd molecule and annihilate one in the electrodes or vice versa .
the qd - electrode coupling is assumed to be sufficiently weak , so that @xmath128 can be treated perturbatively . in the interaction representation ,
the equation of motion for the total density matrix reads @xmath129
\equiv \mathcal{l}% \left ( t\right ) \rho ^{i}\left ( t\right ) , \label{interaction}\]]with @xmath130\]]where @xmath131 and @xmath132 . in order to derive an exact equation of motion for the reduced density matrix @xmath133 of the qd molecule system , it is convenient to define a super - operator @xmath134 according to @xmath135 \otimes \rho _ { b}=\rho _ { s}\otimes \rho _ { b } , \label{pdefinition}\]]with @xmath136 being some fixed state of the electron electrode . accordingly
, a complementary super - operator @xmath137 reads @xmath138for a factorizing initial condition @xmath139 , @xmath140 , and @xmath141 . using the tcl projection operator method @xcite
, one can obtain the second - order tcl master equation @xmath142the eq .
( [ second - order ] ) is the starting point of deriving the particle - number - resolved quantum master equation .
using eqs .
( interaction ) and ( [ pdefinition ] ) , after some algebraic calculations we can rewrite eq .
( [ second - order ] ) as @xmath143 \notag \\ & & -\sum_{\alpha \mu \nu } \int_{-\infty } ^{t}dt_{1}\text{tr}_{b}\left [ d_{\mu } ^{\dag } \left ( t\right ) f_{\alpha \mu } \left ( t\right ) f_{\alpha \nu } ^{\dag } \left ( t_{1}\right ) d_{\nu } \left ( t_{1}\right ) \rho _ { i , s}\left ( t\right ) \otimes \rho _ { b}\right ] \notag \\ & & + \sum_{\alpha \mu \nu } \int_{-\infty } ^{t}dt_{1}\text{tr}_{b}\left [ f_{\alpha \mu } ^{\dag } \left ( t\right ) d_{\mu } \left ( t\right ) \rho _ { i , s}\left ( t\right ) \otimes \rho _ { b}d_{\nu } ^{\dag } \left ( t_{1}\right ) f_{\alpha \nu } \left ( t_{1}\right ) \right ] \notag \\ & & + \sum_{\alpha \mu \nu } \int_{-\infty } ^{t}dt_{1}\text{tr}_{b}\left [ d_{\mu } ^{\dag } \left ( t\right ) f_{\alpha \mu } \left ( t\right ) \rho _ { i , s}\left ( t\right ) \otimes \rho _ { b}f_{\alpha \nu } ^{\dag } \left ( t_{1}\right ) d_{\nu } \left ( t_{1}\right ) \right ] + h.c .. \label{intereqm}\end{aligned}\ ] ] in order to fully describe the electron transport problem , we should record the number of electrons arriving at the drain electrode , which emitted from the source electrode and passing through the qd molecule .
we follow li and co - authors @xcite and introduce the hilbert subspace @xmath144 @xmath145 corresponding to @xmath146 electrons arriving at the drain electrode , which is spanned by the product of all many - particle states of the two isolated electrodes , and formally denoted as @xmath147 span@xmath148 .
then , the entire hilbert space of the two electrodes can be expressed as @xmath149 . with this classification of the electrode states , the average over states in the entire hilbert space @xmath150 in eq .
( [ intereqm ] ) should be replaced with the states in the subspace @xmath151 , and leading to a conditional tcl master equation @xmath152 \notag \\ & & -\sum_{\alpha \mu \nu } \int_{-\infty } ^{t}dt_{1}\text{tr}_{b^{\left ( n\right ) } } \left [ d_{\mu } ^{\dag } \left ( t\right ) f_{\alpha \mu } \left ( t\right ) f_{\alpha \nu } ^{\dag } \left ( t_{1}\right ) d_{\nu } \left ( t_{1}\right ) \rho _ { i , s}\left ( t\right ) \otimes \rho _ { b}\right ] \notag \\ & & + \sum_{\alpha \mu \nu } \int_{-\infty } ^{t}dt_{1}\text{tr}_{b^{\left ( n\right ) } } \left [ f_{\alpha \nu } ^{\dag } \left ( t_{1}\right ) d_{\nu } \left ( t_{1}\right ) \rho _ { i , s}\left ( t\right ) \otimes \rho _ { b}d_{\mu } ^{\dag } \left ( t\right ) f_{\alpha \mu } \left ( t\right ) \right ] \notag \\ & & + \sum_{\alpha \mu \nu } \int_{-\infty } ^{t}dt_{1}\text{tr}_{b^{\left ( n\right ) } } \left [ d_{\mu } ^{\dag } \left ( t\right ) f_{\alpha \mu } \left ( t\right ) \rho _ { i , s}\left ( t\right ) \otimes \rho _ { b}f_{\alpha \nu } ^{\dag } \left ( t_{1}\right ) d_{\nu } \left ( t_{1}\right ) \right ] + h.c .. \label{ndensity}\end{aligned}\ ] ] to proceed , two physical considerations are further implemented .
( i ) instead of the conventional born approximation for the entire density matrix @xmath153 , the ansatz @xmath154 is proposed , where @xmath155 being the electrode density operator associated with @xmath146 electrons arriving at the drain electrode . with this ansatz for the entire density operator , tracing over the subspace @xmath151 , the eq .
( [ ndensity ] ) can be rewritten as @xmath156 \rho _ { i , s}^{\left ( n\right ) } \left ( t\right ) d_{\nu } \left ( t_{1}\right ) d_{\mu } ^{\dag }
\left ( t\right ) \notag \\ & & -\sum_{\alpha \mu \nu } \int_{-\infty } ^{t}dt_{1}\text{tr}_{b^{\left ( n\right ) } } \left [ f_{\alpha \mu } \left ( t\right ) f_{\alpha \nu } ^{\dag } \left ( t_{1}\right ) \rho _ { b}\right ] d_{\mu } ^{\dag } \left ( t\right ) d_{\nu } \left ( t_{1}\right ) \rho _ { i , s}^{\left ( n\right ) } \left ( t\right ) \notag \\ & & + \sum_{\mu \nu } \int_{-\infty } ^{t}dt_{1}\text{tr}_{b^{\left ( n\right ) } } % \left [ f_{l\mu } \left ( t\right ) f_{l\nu } ^{\dag } \left ( t_{1}\right ) \rho _ { b}\right ] d_{\nu } \left ( t_{1}\right ) \rho _ { i , s}^{\left ( n\right ) } \left ( t\right ) d_{\mu } ^{\dag } \left ( t\right ) \notag \\ & & + \sum_{\mu \nu } \int_{-\infty } ^{t}dt_{1}\text{tr}_{b^{\left ( n\right ) } } % \left [ f_{r\mu } \left ( t\right ) f_{r\nu } ^{\dag } \left ( t_{1}\right ) \rho _ { b}\right ] d_{\nu } \left ( t_{1}\right ) \rho _ { i , s}^{\left ( n-1\right ) } \left ( t\right ) d_{\mu } ^{\dag } \left ( t\right ) \notag \\ & & + \sum_{\mu \nu } \int_{-\infty } ^{t}dt_{1}\text{tr}_{b^{\left ( n\right ) } } % \left [ f_{l\nu } ^{\dag } \left ( t_{1}\right ) f_{l\mu } \left ( t\right ) \rho _ { b}\right ] d_{\mu } ^{\dag } \left ( t\right ) \rho _ { i , s}^{\left ( n\right ) } \left ( t\right ) d_{\nu } \left ( t_{1}\right ) \notag \\ & & + \sum_{\mu \nu } \int_{-\infty } ^{t}dt_{1}\text{tr}_{b^{\left ( n\right ) } } % \left [ f_{r\nu } ^{\dag } \left ( t_{1}\right ) f_{r\mu } \left ( t\right ) \rho _ { b}\right ] d_{\mu } ^{\dag } \left ( t\right ) \rho _ { i , s}^{\left ( n+1\right ) } \left ( t\right ) d_{\nu } \left ( t_{1}\right ) + h.c .. \label{ndensitymodified}\end{aligned}\]]here we have used the orthogonality between the states in different subspaces .
( ii ) the extra electrons arriving at the drain electrode will flow back into the source electrode via the external closed transport circuit .
moreover , the rapid relaxation processes in the electrodes will bring the electrodes to the local thermal equilibrium states quickly , which are determined by the chemical potentials .
consequently , after the procedure done in eq .
( [ ndensitymodified ] ) , the electrode density matrices @xmath157 and @xmath158 should be replaced by @xmath159 . in the schrdinger representation , the eq .
( [ ndensitymodified ] ) can be expressed as @xmath160 \notag \\ & & -\sum_{\alpha \mu \nu } \int_{-\infty } ^{t}dt_{1}c_{\alpha \nu \mu } ^{\left ( + \right ) } \left ( t_{1}-t\right ) \rho _ { s}^{\left ( n\right ) } \left ( t\right ) e^{-ih_{s}\left ( t - t_{1}\right ) } d_{\nu } e^{ih_{s}\left ( t - t_{1}\right ) } d_{\mu } ^{\dag } \notag \\ & & -\sum_{\alpha \mu \nu } \int_{-\infty } ^{t}dt_{1}c_{\alpha \mu \nu } ^{\left ( -\right ) } \left ( t - t_{1}\right ) d_{\mu } ^{\dag } e^{-ih_{s}\left ( t - t_{1}\right ) } d_{\nu } e^{ih_{s}\left ( t - t_{1}\right ) } \rho _ { s}^{\left ( n\right ) } \left ( t\right ) \notag \\ & & + \sum_{\mu \nu } \int_{-\infty } ^{t}dt_{1}c_{l\mu \nu } ^{\left ( -\right ) } \left ( t - t_{1}\right ) e^{-ih_{s}\left ( t - t_{1}\right ) } d_{\nu } e^{ih_{s}\left ( t - t_{1}\right ) } \rho _ { s}^{\left ( n\right ) } \left ( t\right ) d_{\mu } ^{\dag } \notag \\ & & + \sum_{\mu \nu } \int_{-\infty } ^{t}dt_{1}c_{r\mu \nu } ^{\left ( -\right ) } \left ( t - t_{1}\right ) e^{-ih_{s}\left ( t - t_{1}\right ) } d_{\nu } e^{ih_{s}\left ( t - t_{1}\right ) } \rho _ { s}^{\left ( n-1\right ) } \left ( t\right ) d_{\mu } ^{\dag } \notag \\ & & + \sum_{\mu \nu } \int_{-\infty } ^{t}dt_{1}c_{l\nu \mu } ^{\left ( + \right ) } \left ( t_{1}-t\right ) d_{\mu } ^{\dag } \rho _ { s}^{\left ( n\right ) } \left ( t\right ) e^{-ih_{s}\left ( t - t_{1}\right ) } d_{\nu } e^{ih_{s}\left ( t - t_{1}\right ) } \notag \\ & & + \sum_{\mu \nu } \int_{-\infty } ^{t}dt_{1}c_{r\nu \mu } ^{\left ( + \right ) } \left ( t_{1}-t\right ) d_{\mu } ^{\dag } \rho _ { s}^{\left ( n+1\right ) } \left ( t\right ) e^{-ih_{s}\left ( t - t_{1}\right ) } d_{\nu } e^{ih_{s}\left ( t - t_{1}\right ) } + h.c .. \label{nqme}\end{aligned}\]]where the correlation function are defined as @xmath161 = \left\langle f_{\alpha \mu } ^{\dagger } \left ( t\right ) f_{\alpha \nu } \left ( t_{1}\right ) \right\rangle , \notag \\
c_{\alpha \mu \nu } ^{\left ( -\right ) } \left ( t - t_{1}\right ) & = \text{tr}_{r}% \left [ f_{\alpha \mu } \left ( t\right ) f_{\alpha \nu } ^{\dag } \left ( t_{1}\right ) \rho _ { b}\right ] = \left\langle f_{\alpha \mu } \left ( t\right ) f_{\alpha \nu } ^{\dagger } \left ( t_{1}\right ) \right\rangle .
\label{correfun}\end{aligned}\]]introducing the following super - operators @xmath162then , the eq .
( [ nqme ] ) can be rewritten as a compact form @xmath160 \notag \\ & & -\sum_{\mu } \left\ { \rho _ { s}^{\left ( n\right ) } \left ( t\right ) a_{\mu } ^{\left ( + \right ) } \left ( t\right ) d_{\mu } ^{\dag } + d_{\mu } ^{\dag } a_{\mu } ^{\left ( -\right ) } \left ( t\right ) \rho _ { s}^{\left ( n\right ) } \left ( t\right ) \right .
\notag \\ & & -a_{l\mu } ^{\left ( -\right ) } \left ( t\right ) \rho _ { s}^{\left ( n\right ) } \left ( t\right ) d_{\mu } ^{\dag } -a_{r\mu } ^{\left ( -\right ) } \left ( t\right ) \rho _ { s}^{\left ( n-1\right ) } \left ( t\right ) d_{\mu } ^{\dag } \notag \\ & & \left . -d_{\mu } ^{\dag } \rho _ { s}^{\left ( n\right ) } \left ( t\right ) a_{l\mu } ^{\left ( + \right ) } \left ( t\right ) -d_{\mu } ^{\dag } \rho _ { s}^{\left ( n+1\right ) } \left ( t\right ) a_{r\mu } ^{\left ( + \right ) } \left ( t\right ) + h.c.\right\ } .
\label{nqmefinal}\end{aligned}\]]where @xmath163 .
the above equation is the starting point of the non - markovian fcs calculation . in this subsection
, we outline the procedure to calculate the non - markovian fcs based on eq .
( [ nqmefinal ] ) .
the fcs can be obtained from the cumulant generating function ( cgf ) @xmath164 which related to the probability distribution @xmath165 by@xcite @xmath166 , where @xmath167 is the counting field .
the cgf @xmath164 connects with the particle - number - resolved density matrix @xmath168 by defining @xmath169 .
evidently , we have @xmath170tr@xmath171 $ ] , where the trace is over the eigenstates of the qd molecule system . since eq .
( [ nqmefinal ] ) has the following form @xmath172 , then , @xmath173 satisfies @xmath174 , where @xmath175 is a column matrix , and @xmath176 , @xmath177 and @xmath178 are three square matrices .
the specific form of @xmath179 can be obtained by performing a discrete fourier transformation to the matrix element of eq .
( [ nqmefinal ] ) . in the low frequency limit , the counting time , namely , the time of measurement is much longer than the time of tunneling through the qd molecule system . in this case
, @xmath164 is given byflindt01,bagrets , flindt02,flindt03,kieblich01,groth @xmath180 , where @xmath181 is the eigenvalue of @xmath179 which goes to zero for @xmath182 . according to the definition of the cumulants
one can express @xmath183 as @xmath184 .
the low order cumulants can be calculated by the rayleigh
schrdinger perturbation theory in the counting parameter @xmath185 . in order to calculate the first four current cumulants we expand @xmath179 to four order in @xmath167
@xmath186and define the two projectors @xcite @xmath187 and @xmath188 , obeying the relations @xmath189 and @xmath190 . here
, @xmath191 is the right eigenvector of @xmath192 , i.e. , @xmath193 , and @xmath194 is the corresponding left eigenvector . in view of @xmath192 being regular , we also introduce the pseudoinverse according to @xmath195 , which is well - defined due to the inversion being performed only in the subspace spanned by @xmath196 . after a careful calculation ,
@xmath197 is given by @xmath198 \chi ^{2 } \notag \\ & + \frac{1}{3!}\left [ \left\langle \left\langle \tilde{0}\right .
\right\vert l_{3}\left\vert \left .
0\right\rangle \right\rangle -3\left\langle \left\langle \tilde{0}\right .
\right\vert \left ( l_{2}rl_{1}+l_{1}rl_{2}\right ) \left\vert \left .
0\right\rangle \right\rangle \right .
\notag \\ & \left .
-6\left\langle \left\langle \tilde{0}\right .
\right\vert l_{1}r\left ( rl_{1}p - l_{1}r\right ) l_{1}\left\vert \left .
0\right\rangle \right\rangle \right ] \chi ^{3}+ \notag \\ & + \frac{1}{4!}\left [ \left\langle \left\langle \tilde{0}\right .
\right\vert l_{4}\left\vert \left .
0\right\rangle \right\rangle -6\left\langle \left\langle \tilde{0}\right .
\right\vert l_{2}rl_{2}\left\vert \left .
0\right\rangle \right\rangle \right .
\notag \\ & -4\left\langle \left\langle \tilde{0}\right .
\right\vert \left ( l_{3}rl_{1}+l_{1}rl_{3}\right ) \left\vert \left .
0\right\rangle \right\rangle \notag \\ & -12\left\langle \left\langle \tilde{0}\right .
\right\vert l_{2}r\left ( rl_{1}p - l_{1}r\right ) l_{1}\left\vert \left .
0\right\rangle \right\rangle \notag \\ & -12\left\langle \left\langle \tilde{0}\right .
\right\vert l_{1}r\left ( rl_{2}p - l_{2}r\right ) l_{1}\left\vert \left .
0\right\rangle \right\rangle \notag \\ & -12\left\langle \left\langle \tilde{0}\right .
\right\vert l_{1}r\left ( rl_{1}p - l_{1}r\right ) l_{2}\left\vert \left .
0\right\rangle \right\rangle \notag \\ & -24\left\langle \left\langle \tilde{0}\right .
\right\vert l_{1}r\left ( r^{2}l_{1}pl_{1}p - rl_{1}pl_{1}r - l_{1}r^{2}l_{1}p\right .
\notag \\ & \left .
-rl_{1}rl_{1}p+l_{1}rl_{1}r\right ) l_{1}\left\vert \left .
0\right\rangle \right\rangle \right ] \chi ^{4}+\cdots .
\label{matrixlambda}\end{aligned}\]]from eq .
( [ matrixlambda ] ) we can identify the first four current cumulants : @xmath199@xmath200 /i^{2 } , \label{shot noise}\]]@xmath201 /i^{3}. \label{skewness}\end{aligned}\]]@xmath202 /i^{4}.
\label{kurtosis}\end{aligned}\]]here , it is important to emphasize that the first four cumulants @xmath203 are directly related to the transport characteristics . for example , the first - order cumulant ( the peak position of the distribution of transferred - electron number ) @xmath204 gives the average current @xmath205 .
the zero - frequency shot noise is related to the second - order cumulant ( the peak - width of the distribution ) @xmath206 .
the third - order cumulant @xmath207 and four - order cumulant @xmath208 characterize , respectively , the skewness and kurtosis of the distribution . here , @xmath209 . in general , the shot noise , skewness and kurtosis
are represented by the fano factor @xmath210 , @xmath211 and @xmath212 , respectively .
this work was supported by the nkbrsfc under grants nos .
2011cb921502 , 2012cb821305 , nsfc under grants nos . 11204203 , 61405138 , 11004124 , 11275118 , 61227902 , 61378017 , 11434015 , sklqoqod under grants no .
kf201403 , sprpcas under grants no . xdb01020300 .
h. b. x. conceived the idea and designed the research and performed calculations .
h. j. j , j. q. l. and w. m. l. contributed to the analysis and interpretation of the results and prepared the manuscript .
the authors declare no competing financial interests .
correspondence and requests for materials should be addressed to hai - bin xue or wu - ming liu .
xue , h. b. , nie , y. h. , li , z. j. & liang , j. q. effect of finite coulomb interaction on full counting statistics of electronic transport through single - molecule magnet .
a _ 375 , 716 ( 2011 ) .
begemann , g. , darau , d. , donarini , a. & grifoni , m. symmetry fingerprints of a benzene single - electron transistor : interplay between coulomb interaction and orbital symmetry .
b _ 77 , 201406 ( r ) ( 2008 ) .
flindt , c. , novotn , t. , braggio , a. & jauho , a. p. counting statistics of transport through coulomb blockade nanostructures : high - order cumulants and non - markovian effects .
b _ 82 , 155407 ( 2010 ) . | full counting statistics of electron transport is a powerful diagnostic tool for probing the nature of quantum transport beyond what is obtainable from the average current or conductance measurement alone . in particular , the non - markovian dynamics of quantum dot molecule plays an important role in the nonequilibrium electron tunneling processes .
it is thus necessary to understand the non - markovian full counting statistics in a quantum dot molecule .
here we study the non - markovian full counting statistics in two typical quantum dot molecules , namely , serially coupled and side - coupled double quantum dots with high quantum coherence in a certain parameter regime .
we demonstrate that the non - markovian effect manifests itself through the quantum coherence of the quantum dot molecule system , and has a significant impact on the full counting statistics in the high quantum - coherent quantum dot molecule system , which depends on the coupling of the quantum dot molecule system with the source and drain electrodes .
the results indicated that the influence of the non - markovian effect on the full counting statistics of electron transport , which should be considered in a high quantum - coherent quantum dot molecule system , can provide a better understanding of electron transport through quantum dot molecules . |
one of the major challenges in nuclear many - body theory is to develop a microscopic and systematic approach that would account for the most relevant correlations in the description of nuclear properties .
most existing methods can be divided into two major classes : the self - consistent mean field ( scmf ) approximation @xcite , also known as density functional theory ( dft ) , and the configuration - interaction shell model ( cism ) approach @xcite .
scmf theories are often cast in terms of an energy density functional that is minimized to obtain the ground - state solution of the system . there
exist parameterizations of this energy density functional that are valid globally through the table of nuclei .
such parametrizations are usually based on the zero - range skyrme force @xcite or on the gogny interaction @xcite .
the cism approach requires as input a large number of interaction matrix elements in a truncated space .
while effective cism interactions can sometimes be traced back to the bare nucleon - nucleon interaction , it is often necessary to adjust them empirically . since the cism method requires a truncation to a finite number of shells , the single - particle energies and interaction matrix elements are specific to the mass region under consideration .
thus the cism approach lacks the global validity of the scmf . on the other hand ,
the cism approach has the advantage that the underlying nuclear wave functions ( both ground and excited states ) are fully correlated .
complete @xmath5 major shell calculations in @xmath6- @xcite , @xmath1- @xcite and @xmath7-shell @xcite nuclei have successfully described many of the observed properties of these nuclei . in contrast , the scmf approach does not account for all correlations present in the nuclear ground state .
partial correlations in a mean - field approach are often accounted for by the spontaneous symmetry breaking of the ground state ( e.g. , the breaking of rotational symmetry by a deformed ground state ) @xcite .
additional correlations beyond the mean - field approximation are included by restoration of this broken symmetry ( e.g. , by angular momentum projection ) @xcite , and by configuration mixing of symmetry - projected states via the generator coordinate method ( gcm ) @xcite .
the gcm is also useful in providing information on excited states of the nucleus .
angular - momentum and particle - number projected gcm has been applied successfully in global studies of binding energies @xcite and of spectroscopic properties of the first excited @xmath8 state @xcite in even - even nuclei . in a recent work
@xcite we initiated an approach that takes advantage of the global validity of the scmf method and the higher accuracy of the cism approach .
it is based on the idea of mapping an scmf theory onto an effective cism hamiltonian that is defined in a suitably chosen truncated model space .
such an approach would enable us to start from an scmf theory with a global parametrization and construct effective cism hamiltonians in different mass regions . in heavy nuclei ,
the required truncated shell model spaces are expected to be too large for conventional diagonalization of the cism hamiltonian . in such cases
, one can apply methods that have been developed to treat very large model spaces , such as the shell model monte carlo ( smmc ) method @xcite , the monte carlo shell model @xcite approach , and direct matrix methods @xcite .
alternatively , one can use tractable approximations that go beyond the mean - field approximation such as the rpa @xcite . a mean - field approximation , based on a skyrme force ,
was used in ref .
@xcite to determine the single - particle energies of a cism hamiltonian . in ref .
we carried out the first step of mapping an scmf theory onto the cism and constructed an effective hamiltonian that includes quadrupolar correlations .
the map was tested successfully in @xmath0 @xmath1-shell nuclei , using the hartree - fock ( hf ) approximation of the so - called usd hamiltonian @xcite as our starting point for an scmf theory .
the usd hamiltonian was constructed empirically and is known to give a good description of the low - energy spectroscopy of @xmath1-shell nuclei . using the mapped cism hamiltonian , we found that quadrupolar correlations alone account for more than 50% of the full correlation energies contained in the usd hamiltonian .
we then applied our mapping procedure to study quadrupolar correlations in @xmath0 @xmath1-shell nuclei starting from the sly4 parametrization of the skyrme energy functional as the scmf theory @xcite . here
we explore the scmf to cism mapping idea further by including pairing correlations in the mapped effective cism hamiltonian in addition to quadrupolar correlations .
the starting point is an scmf theory that includes pairing correlations .
such a theory can be solved in the hf plus bcs approximation [ or more generally in the hartree - fock - bogolyubov ( hfb ) approach ] to provide the ground - state single - particle hf wave functions and energies , as well as the bcs occupation amplitudes .
a similar hf+bcs approximation can be used to determine a spherical single - particle basis .
this mean - field output is used to construct an effective cism hamiltonian in a truncated spherical model space that contains both quadrupolar and monopole pairing correlations . after testing the method using the hf+bcs approximation of the usd interaction as our initial scmf theory ,
we apply it to deformed @xmath0 @xmath1-shell nuclei in the context of dft .
the latter is based on the sly4 parameterization of the skyrme force in the particle - hole channel @xcite plus a zero - range density - dependent force in the pairing channel @xcite .
it is solved in the hf+bcs approximation using the brussels - paris computer code ev8 @xcite .
the mapping constructed here could also be useful in global studies of level densities within the smmc approach @xcite .
both quadrupolar and pairing correlations play an important role in the microscopic calculations of level densities . pairing correlations are natural for spherical nuclei . here , however , we consider the addition of pairing correlations on top of quadrupolar correlations in the cism hamiltonian .
we therefore focus on nuclei whose hf+bcs ground state is deformed . in principle , the method presented here can be extended to spherical nuclei by applying an external quadrupole field , thus effectively deforming the nucleus .
the outline of this paper is as follows . in sec .
[ formalism ] we discuss our mapping procedure when pairing correlations are included in the scmf theory .
the ground state is solved in the hf+bcs approximation and its unique description requires both a density matrix and a pairing tensor ( or an anomalous density matrix ) .
the cism hamiltonian is constructed in a truncated spherical space to include both quadrupolar and pairing correlations . in sec .
[ usd ] , we test our mapping procedure using the hf+bcs approximation of the usd hamiltonian as our initial scmf theory for deformed @xmath0 @xmath1-shell nuclei ( @xmath2ne , @xmath3 mg and @xmath4ar ) . in sec .
[ map - dft ] we map a dft that is based on the sly4 skyrme interaction plus a zero - range density - dependent pairing force onto an effective cism hamiltonian for deformed @xmath0 @xmath1 shell nuclei .
the correlation energies extracted from the mapped hamiltonian compare well with the correlation energies obtained with other methods .
we conclude with a brief discussion of possible future directions in sec .
[ conclusion ] .
in this section , we discuss the theoretical framework of mapping an scmf theory into a cism theory in the case where the scmf includes pairing correlations .
we assume that the scmf theory is solved in the hf+bcs approximation @xcite .
this is for example the case of skyrme - type density functional theories in which pairing correlations are considered through the addition of a zero - range density - dependent force in the pairing channel @xcite .
the hf+bcs mean - field theory provides us with a set of single - particle hf orbitals @xmath9 and their corresponding single - particle energies @xmath10 and bcs occupation amplitudes @xmath11 , as well as the total energy @xmath12 of the hf+bcs ground state . for simplicity
we assume that this solution , obtained as the global minimum of an energy density functional , is deformed . using the same scmf theory
( i.e. , hf+bcs ) we can also find the spherical solution .
our plan is to use this output of the scmf theory to construct an effective cism hamiltonian that is defined within a truncated shell model space . in particular
, we will consider an cism hamiltonian that contains a quadrupole - quadrupole interaction and a monopole pairing interaction .
the cism is defined in a spherical basis , and we determine such a basis from the spherical solution of the scmf theory .
a spherical solution could be found by iterations when the initial guess for the density matrix is chosen to be spherically symmetric , and as long as the spherical symmetry is preserved in the iteration procedure . in ref .
we used the hf approximation , in which case it was necessary ( for an open shell nucleus ) to use the uniform filling approximation or add a pairing - like interaction with a fixed gap to preserve spherical symmetry in the next iteration . here
, the bcs approximation guarantees that the density matrix in the next iteration remains spherical .
the construction of a standard spherical basis with good angular momentum @xmath13 and projection @xmath14 was discussed in sec .
iv of ref . .
here we use @xmath15 as a generic label for the spherical basis , i.e. , @xmath15 includes the quantum numbers @xmath16 as well as other quantum numbers . to denote spherical states and lowercase roman letters @xmath17 to denote deformed states . ] the transformation matrix between the deformed single - particle basis @xmath18 and the spherical single - particle basis @xmath15 is denoted by @xmath19 .
we have @xmath20 where @xmath21 is the creation operator in a spherical state @xmath15 and @xmath22 is the creation operator in a deformed state @xmath18 .
the deformed ground state in the hf approximation is a slater determinant @xmath23 which is uniquely described by a density matrix @xmath24 .
however , in the hf+bcs approximation , the ground state @xmath25 is defined by the bcs amplitudes @xmath26 ( the product runs over half the number of single - particle states @xmath27 and @xmath28 is the state conjugate to @xmath18 ) .
the unique description of this bcs ground state requires both a density matrix @xmath24 and a pairing tensor ( or anomalous density ) @xmath29 @xcite . in the deformed basis @xmath30 and @xmath31 , where @xmath32 denotes the expectation value in the hf+bcs ground state @xmath33 . in the spherical basis
, we can similarly define @xmath34 and @xmath35 . using the unitary transformation ( [ u - trans ] ) from the deformed to the spherical basis
, we have [ trans ] @xmath36 or , equivalently , in a matrix notation [ trans - mat ] @xmath37 according to ( [ trans - mat ] ) , @xmath24 transforms as a linear operator under the unitary transformation @xmath38 while @xmath29 transforms as an _ antilinear _ operator . just as in the hf approximation
, the trace of @xmath24 must be equal to the total number of particles @xmath39 of a given type ( protons or neutrons ) @xmath40 in the framework of the hf+bcs approximation , this condition is satisfied by using a suitable chemical potential @xmath41 ( for each type of particle ) . in the hf approximation , the ground state is a slater determinant and the corresponding density matrix must be a projector @xmath42 with all eigenvalues either 1 or 0 .
however , in the hf+bcs ground state the matrix @xmath24 is no longer a projector and its eigenvalues @xmath43 are between 0 and 1 .
the matrices @xmath24 and @xmath29 , which characterize uniquely the hf+bcs ground state , satisfy now the relations [ new - conditions ] @xmath44 in the following , we assume that the scmf theory has time - reversal symmetry so that the single - particle orbitals come in degenerate time - reversed pairs @xmath45 and @xmath46 ( @xmath47 is the time - reversal operator ) . in the following we choose the @xmath48 orbitals to be the orbitals with positive @xmath49-signature @xcite .
the set @xmath50 with @xmath48 spans the complete single - particle space .
we also assume that the ground state solution is axially symmetric , in which case @xmath51 is a good quantum number ( @xmath49 is the symmetry axis ) , and the positive @xmath49-signature orbitals have @xmath52 . similarly , the spherical states appear in degenerate time - reversed pairs @xmath53 with @xmath54 denoting the positive @xmath49-signature states . the matrices @xmath24 and @xmath29 have the following simple form in the deformed hf basis ( for @xmath55 ) [ hf - basis ] @xmath56 the transformation matrix @xmath57 is real ( and thus orthogonal ) and the matrices @xmath24 and @xmath29 remain real in the spherical basis .
since the spherical states also have good @xmath49-signature , the transformation @xmath57 does not mix states with different @xmath49-signature and for @xmath58 we have @xmath59 . applying the time - reversal operator to this relation and using the fact that @xmath57 is real , we find that the time - reversed states transform in exactly the same way @xmath60 .
we therefore have @xmath61 it follows that the density and anomalous density matrices have the block form @xmath62 where @xmath63 and @xmath64 are real matrices whose dimension is half the total number of orbitals .
it follows from eqs .
( [ trans ] ) and ( [ hf - basis ] ) that ( for @xmath65 ) [ a - b ] @xmath66 the matrix @xmath24 is real symmetric while the matrix @xmath29 is real antisymmetric .
the form ( [ block - form ] ) is equivalent to [ spherical - basis ] @xmath67 it is convenient to define the density and anomalous density _ operators _ by [ operator - form ] @xmath68 such that their matrix representation in the corresponding basis gives us the matrices @xmath24 and @xmath29 , i.e. , @xmath69 and @xmath70
. equations ( [ a - b ] ) and ( [ spherical - basis ] ) follow then immediately from the operator form ( [ operator - form ] ) . the basic relations ( [ new - conditions ] ) defining an hf+bcs state can be rewritten in terms of the matrices @xmath63 and @xmath64 as [ a - b - conditions ] @xmath71 a more compact way of representing the relations ( [ new - conditions ] ) or ( [ a - b - conditions ] ) is to introduce a generalized density matrix @xmath72 whose dimension is twice the total number of orbitals @xcite @xmath73 where @xmath74 is the identity matrix in a space whose dimension is half the number of orbitals . relations ( [ new - conditions ] ) or ( [ a - b - conditions ] ) are equivalent to the condition that @xmath72 is a projector @xmath75 thus the eigenvalues of @xmath72 must be 1 or 0 . this condition is formally similar to the one satisfied by @xmath24 in the hf approximation except that there are now twice as many eigenvalues . in our case ,
the matrix @xmath72 is real symmetric and its eigenvalues are pairwise degenerate .
as long as the spherical basis @xmath15 is complete , * u * is unitary and we have just generated a different representation of the same ground - state density matrix and pairing tensor .
however , in constructing a shell model space , it is necessary to truncate to a valence subspace .
we define such a subspace in terms of a projector @xmath76 where @xmath77 denotes a sum over a subset of the single - particle orbitals @xmath15 . the density operator @xmath78 and anomalous density operator @xmath79 are then replaced by projected densities @xmath80 the corresponding matrices in the truncated spherical space are @xmath81 and @xmath82 . in choosing a suitable truncated subspace , we require the projected densities to preserve approximately their basic properties in the complete space . in the hf+bcs approximation ,
the trace of the density matrix is constraint to give the total number of particles [ see eq .
( [ trace - rho ] ) ] , and we require that a similar relation is satisfied approximately by the projected density @xmath83 with the number of particles replaced by the valence number of particles @xmath84 another requirement is that the conditions ( [ new - conditions ] ) are satisfied approximately by the projected densities @xmath85 and @xmath86 . alternatively , we require the matrices @xmath87 and @xmath88 to approximately satisfy ( [ a - b - conditions ] ) .
it is convenient to define the projected @xmath72 matrix @xmath89 and require that it satisfies approximately the condition ( [ r - proj_first ] ) , namely that its eigenvalues are close to either 1 or 0 .
formally , this condition is similar to the one we require from the projected density matrix @xmath24 at the hf level @xcite , except that the projected @xmath72 matrix is defined in twice the number of dimensions . in general , we can calculate the expectation value of a one - body observable @xmath90 in the truncated cism space using the projected density @xmath91 where the sum is restricted to the truncated spherical space .
the ratio between the expectation value of @xmath92 in the full and truncated spaces provides us with simple rescaling factors of one - body operators in the truncated cism space . in this section
we construct the cism hamiltonian given the hf+bcs ground - state solution .
in the self - consistent basis @xmath18 of the hf+bcs , the hf hamiltonian is @xmath93 , where @xmath94 are the energies of the deformed hf single - particle orbitals .
we then proceed as in ref . .
the hf hamiltonian matrix @xmath95 is transformed to the spherical space and projected on the truncated cism space .
the projected one - body hamiltonian is @xmath96 next , we expand @xmath97 in multipoles @xmath98 and construct the second quantized one - body tensor operators @xmath99 of rank @xmath100 [ see eqs . ( 10 ) and ( 8) in ref . ] .
the lowest multipole is the monople @xmath101 , which we use to define the spherical single - particle hamiltonian , while the @xmath102 multipole is a quadrupole operator @xmath103 , which we use to define a quadrupole - quadrupole interaction .. ] we now consider the following effective cism hamiltonian @xmath104 where @xmath105 denotes normal ordering and @xmath106 is the monopole pair creation operator @xmath107 defined separately for protons and neutrons . here
@xmath108 and @xmath109 are , respectively , quadrupole and pairing coupling constants . in comparison with the cism effective hamiltonian considered in ref . , eq . ( [ effective - h ] ) contains an additional monopole pairing interaction .
we note that one might consider an isovector monopole pairing ( which contains also a proton - neutron component ) .
however , in the dft applications we discuss in sec .
[ map - dft ] , the resulting hf hamiltonian does not conserve isospin symmetry ( because of the coulomb interaction ) and we only consider here an hamiltonian of the form ( [ effective - h ] ) .
we found that an isospin - invariant interaction constructed from an isovector monopole pair operator gives only a small correction to the correlation energy when compared with a hamiltonian of the form ( [ effective - h ] ) . ]
both quadrupole - quadrupole and pairing interactions in ( [ effective - h ] ) are attractive ( i.e. , @xmath110 ) , leading to an cism hamiltonian that has a good monte carlo sign in the smmc approach @xcite .
the method outlined above determines the effective hamiltonian ( [ effective - h ] ) up to the two coupling constants @xmath111 and @xmath112 . in ref .
we determine the coupling constant of the quadrupolar interaction by matching the deformation energy of the scmf theory with the deformation energy of the cism effective hamiltonian when the latter is solved in the hf approximation .
when the effective hamiltonian has the form ( [ effective - h ] ) , it is necessary to match two energy scales in order to determine the two coupling constants @xmath111 and @xmath112 .
one energy scale is the deformation energy , here redefined in the hf+bcs approximation @xmath113 where @xmath114 is the energy of the spherical hf+bcs solution and @xmath12 is the hf+bcs energy of the deformed ground - state solution .
the deformation energy in ( [ deformation - energy ] ) is usually smaller than the deformation energy calculated in the hf approximation . to introduce a second energy scale ,
we define @xmath115 to be the average interaction in the pairing channel @xmath116 where @xmath117 is the pairing gap matrix . , where @xmath118 is the anti - symmetrized two - body interaction . ]
the energy @xmath115 can be calculated in both the spherical and deformed solutions and we define the pairing energy excess to be @xmath119 since @xmath115 is lower ( i.e. , more negative ) in the spherical solution , the energy difference in ( [ delta - e - pair ] ) is usually positive .
the two coupling parameters @xmath108 and @xmath109 are determined by matching simultaneously the two energy scales ( [ deformation - energy ] ) and ( [ delta - e - pair ] ) between the scmf theory and the cism hamiltonian when the cism is treated in the hf+bcs approximation .
in general , the coupling parameters should be renormalized depending on the truncation used for the cism space .
these renormalization effects are included implicitly in our procedure by matching the energy scales calculated in the scmf approach in the complete space with similar energy scales calculated in the truncated cism space within the hf+bcs approximation .
an important application of our mapping is in the calculation of correlation energies .
the correlation energy @xmath120 is defined as the difference between the ground - state energy in the hf+bcs approximation to the cism hamiltonian and the fully correlated ground - state energy @xmath121 of the cism @xmath122 quadrupolar and pairing correlations compete with each other ; the quadrupole - quadrupole interaction favors deformation in which pairing is suppressed , while pairing favors the spherical configuration .
if we consider the spherical configuration alone , then the bcs approximation already takes into account some of the gain in pairing correlation energy .
however , in the deformed nuclei considered here we effectively gain more in mean - field energy by deformation . in the deformed ground state ,
bcs correlations are often significantly smaller than in the spherical solution ( or even vanish ) .
nevertheless , the correlation energy described by ( [ e - corr ] ) is expected to increase when a pairing term is added to the quadrupolar cism hamiltonian .
this increase arises from fluctuations of the pairing fields beyond the bcs approximation .
in particular , these fluctuations sample the spherical configuration in which pairing correlations are significant .
even in spherical nuclei ( which we do not treat here ) we expect to gain a finite correlation energy in ( [ e - corr ] ) when adding a pairing term to the cism hamiltonian , i.e. , there are pairing correlations beyond those described by the spherical bcs solution .
the bcs approximation is valid in the limit when the pairing gap is much larger than the mean single - particle level spacing , and studies of pairing effects in ultra - small metallic grains revealed deviations from bcs theory in the crossover to the fluctuation - dominated regime @xcite .
nuclei belong to this crossover regime , in which the pairing gap is comparable to the mean - level spacing @xcite . in the remaining part of this section
we further motivate our choice of the second energy scale ( [ delta - e - pair ] ) .
as already noted , pairing effects are weaker in a deformed configuration and could vanish altogether ( i.e. , @xmath123 ) in the deformed ground state .
for example , in our hf+bcs studies of the usd interaction in sec .
[ map - usd ] , we observe such a collapse of the bcs solution in all three deformed @xmath0 @xmath1-shell nuclei .
we then have @xmath124 , and only the spherical solution is relevant for determining the second energy scale .
an alternative second energy scale that could be used in such cases is @xmath125 , the decrease in spherical energy in the hf+bcs approximation as compared with the hf approximation . in general @xmath126 where @xmath127 is the spherical one - body density and
@xmath118 is the anti - symmetrized two - body interaction .
the gain in pairing energy from @xmath128 is thus canceled in part by the increase @xmath129 of the one - body energy and the increase @xmath130 of the average interaction in the particle - hole channel . despite this partial cancelation , @xmath131 remains well correlated with @xmath125 and , as long as the deformed solution is unpaired , we find that both energy scales ( [ delta - e - pair ] ) and ( [ spherical - pairing ] ) lead to cism hamiltonians with similar coupling constants
. however , the deformed ground state can generally have a non - vanishing ( though weaker ) pairing solution , as we observe in our dft studies of sec .
[ scmf - results ] . in such cases ,
we find that matching the energy scale ( [ spherical - pairing ] ) results in a correlation energy ( [ e - corr ] ) of the cism ground state that is unphysically large .
we therefore adopt ( [ delta - e - pair ] ) as the second energy scale to be matched when the deformed solution is paired .
the scale ( [ delta - e - pair ] ) also has the obvious advantage that it is defined within a single scmf theory ( i.e. , hf+bcs ) and does not require the use of two different mean - field theories ( i.e. , hf and hf+bcs ) .
in this section we test the mapping procedure discussed in sec .
[ formalism ] by considering the hf+bcs solution of the usd hamiltonian @xcite as our starting point of an scmf theory .
the usd interaction provides a rather good description of binding energies , spectra and electromagnetic transition intensities of @xmath1-shell nuclei @xcite . in these studies
we compare the correlation energies found from the mapped cism hamiltonian with the full correlation energies of the usd hamiltonian . in sec .
[ map - dft ] , we apply the mapping formalism to a dft that is defined by an sly4 skyrme force @xcite in the particle - hole channel plus a zero - range density - dependent force in the pairing channel @xcite . here and in sec . [ map - dft ]
we focus on @xmath0 nuclei , in which isovector quadruple - quadrupole correlations could be ignored .
we also restrict our studies to nuclei whose self - consistent hf+bcs ground state is deformed . in the @xmath1 shell , this excludes @xmath132s whose self - consistent hf+bcs ground state is spherical .
spherical nuclei can be studied by a similar method once a deformation is induced by a quadrupole constraining field .
we also restrict our studies to nuclei in which the spherical configuration supports a bcs solution ( so that @xmath133 ) .
otherwise , it is necessary to add a constraining pairing field . in the @xmath1 shell
, this excludes @xmath134si for which both spherical and deformed solutions are found to have no pairing solution .
a similar situation occurs in the framework of the dft application discussed in sec .
[ map - dft ] .
therefore , the @xmath0 @xmath1-shell nuclei we consider here and in sec .
[ map - dft ] are @xmath2ne , @xmath3 mg and @xmath4ar . we have carried out hf+bcs calculations using the usd interaction .
the resulting hf+bcs energies are listed in table [ table1 ] for both spherical and deformed ground - state configurations .
we find the ground - state solutions to be prolate for @xmath2ne and @xmath3 mg , and oblate for @xmath4ar , with average mass quadrupole moments @xmath135 given in table [ table1 ] . the hf+bcs deformation energies ( [ deformation - energy ] ) are calculated to be @xmath136 and @xmath137 mev for @xmath2ne , @xmath3 mg and @xmath4ar , respectively .
it is interesting to compare these energies with the corresponding hf deformation energies @xcite of @xmath138 and @xmath139 mev for the same nuclei .
the reduction in deformation energies originates in a negative @xmath140 that lowers the spherical hf energy .
.hf+bcs results for the deformed @xmath0 @xmath1-shell nuclei @xmath2ne , @xmath3 mg and @xmath4ar , using the usd interaction and the effective cism hamiltonian ( [ effective - h ] ) .
energies are in mev , the average quadrupole moment @xmath141 is in @xmath142 ( with @xmath143 being the oscillator radius ) , @xmath111 is in mev/@xmath144 and @xmath109 is in mev . [
cols="^,^,^,^,^,^,^,^,^,^,^,^",options="header " , ] other relevant one - body quantities are the occupations @xmath145 of the spherical orbitals . in table
[ table4 ] we compare their dft values @xmath146 with the respective spherical occupations of the cism when solved in the hf+bcs approximation .
we also list in table [ table4 ] the cism occupations as calculated by the shell model code oxbash .
results are shown for @xmath2ne and @xmath3 mg at @xmath147 mev-@xmath148 , and @xmath4ar at @xmath149 mev-@xmath148 . even in cases
where the deformed ground state is unpaired , the cism occupations are expected to differ from their values in ref . .
the reason is that the cism hamiltonian studied here contains an additional pairing term .
the occupations computed in all three methods are in rather close agreement .
these various spherical occupations are also shown in fig . [ occupations ] versus the corresponding spherical single - particle energy @xmath150 .
= 0.9 = 0.6 finally , we compare the bcs occupations @xmath43 of the deformed solution of the mapped cism hamiltonian ( defined in the truncated space ) with the @xmath43 of the original scmf ground state ( defined in the complete space ) .
figure [ v2-bcs ] makes this comparison for @xmath2ne at @xmath151 mev-@xmath148 .
the @xmath43 of the mapped theory are slightly enhanced ( suppressed ) below ( above ) the fermi energy when compared with their dft values . in the dft ,
core orbitals below the @xmath1 shell have @xmath152 , while orbitals above the valence @xmath1 shell have @xmath153 .
an important application of the dft to cism map is the calculation of correlation energies .
we calculate the ground - state energy @xmath121 of the mapped hamiltonian ( [ effective - h ] ) in the shell model approach and determine the correlation energy using eq .
( [ e - corr ] ) .
the correlation energies for the three nuclei @xmath2ne , @xmath3 mg and @xmath4ar are tabulated in table [ table2 ] . in general , the correlation energy of the mapped hamiltonian is found to increase with @xmath154 . in fig .
[ corr - dft ] we show the correlation energy versus mass number @xmath155 for @xmath147 mev @xmath156 .
we compare our results ( solid squares ) with the correlation energies obtained in ref . where the effective interaction includes only a quadrupolar component ( open squares ) .
the inclusion of pairing correlations provides an additional gain in correlation energy and the observed trend versus mass number @xmath155 is similar to the trend observed in the usd results shown in fig .
[ corr - usd ] .
= 0.57 our correlation energies of @xmath157 mev and @xmath158 mev for the nuclei @xmath3 mg and @xmath4ar , calculated for @xmath149 mev @xmath156 , are in very close agreement with the values of @xmath157 mev and @xmath159 mev obtained in ref .
starting from the same dft but using an angular - momentum projected gcm .
the correlation energy of @xmath160 mev we find for @xmath2ne ( using a somewhat reduced value of @xmath154 ) is also consistent with the value of 3.1 mev reported in ref . .
in this work we have mapped an scmf theory onto an effective cism hamiltonian in the presence of both quadrupolar and pairing correlations . when compared with ref .
, this is an important step forward in constructing a general map that would take advantage of the global validity of the energy density functional and the higher accuracy of the cism approach .
the main differences compared with the study initiated in ref .
@xcite are : ( i ) the scmf is now based on an hf+bcs approximation ( rather than hf ) , and the ground - state consistency conditions are recast in terms of both the density matrix and the pairing tensor .
( ii ) the effective hamiltonian depends now on two coupling strengths , and consequently two energy scales must be matched to determine these coupling parameters .
one energy scale is the deformation energy , similar to the one used previously but now redefined in terms of the hf+bcs energies .
the second scale is determined by the difference of average pairing energy in the deformed ground state and spherical configurations .
the new map constructed here provides improved correlation energies that are consistent with other methods that include quadrupolar and pairing correlations beyond the mean field .
the dft application discussed here is based on a skyrme force in the particle - hole channel plus a zero - range density - dependent interaction in the pairing channel .
one could alternatively start from a mean - field calculation with a finite - range gogny force .
such an interaction includes pairing correlations consistently and is usually treated in the hartree - fock - bogolyubov ( hfb ) method .
the map constructed here can be easily generalized to the gogny dft by replacing the hf+bcs approximation by the hfb approximation , both when solving the dft and in the mean - field solution of the cism hamiltonian . as already discussed in ref .
@xcite much remains to be done .
nuclei that have no spherical pairing solution can be treated by introducing a constraining pairing field .
the addition of an interaction term built from an isovector quadrupole operator would enable us to generalize our studies to @xmath161 nuclei and carry out systematic studies .
a major challenge is the application to the heavy deformed nuclei .
although the method presented here can in principle be adapted to such nuclei , outstanding issues are the ability of a spherical shell model to describe nuclei with large deformation in a truncated space and the practical issue of solving the cism hamiltonian in large model spaces .
99 for a review see m. bender , and p .- h .
heenen and p.g .
reinhard , rev .
phys . * 75 * , 121 ( 2003 ) . for
a review see b.a .
brown and b.h .
wildenthal , annu .
nucl . part .
sci . * 38 * , 29 ( 1988 ) .
skyrme , phil mag . * 1 * , 1043 ( 1956 ) ; nucl . phys . * 9 * , 615 ( 1959 ) .
d. gogny , proc .
int . conf . on _ nuclear self - consistent fields _ ,
g. ripka and m.porneuf , eds . , north holland , amsterdam , 1975 .
s. cohen and d. kurath , nucl . phys . *
73 * , 1 ( 1965 ) .
wildenthal , prog .
. phys . * 11 * , 5 ( 1984 ) . w. a. richter , m.g .
vandermerwe , r.e .
julies , and b.a .
brown , nucl . phys . *
a523 * , 325 ( 1991 ) ; g. martinez - pinedo , a. p. zuker , a. poves and e. caurier , phys . rev . *
c 55 * , 187 ( 1997 ) ; a. poves , j. sanchez - solano , e. caurier , and f. nowacki , nucl .
a * 694 * , 157 ( 2001 ) .
p. ring and p. schuck , _ the nuclear many body problem _ , springer , berlin ( 1980 ) .
rodriguez - guzmn , j.l .
egido and l.m .
robledo , phys .
b * 474 * , 15 ( 2000 ) ; phys . rev .
c*62 * , 054308 ( 2000 ) .
rodriguez - guzmn , j.l .
egido and l.m .
robledo , phys .
c * 65 * , 024304 ( 2002 ) ; nucl . phys .
a * 709 * , 201 ( 2003 ) .
m. bender and p .- h .
heenen , nucl .
a * 713 * , 390 ( 2003 ) .
m. bender , p. bonche , t. duguet and p .- h .
heenen , phys .
c * 69 * , 064303 ( 2004 ) .
m. bender , g.f . bertsch and p .- h .
heenen , phys .
lett . * 94 * , 102503 ( 2005 ) .
bertsch , b. sabbey and m. uusnkki , phys .
c * 71 * , 054311 ( 2005 ) . m. bender , g. f. bertsch and p .- h .
heenen , phys .
c * 73 * , 034322 ( 2006 ) .
b. sabbey , m. bender , g.f . bertsch and p .- h .
heenen , phys .
c * 75 * , 044305 ( 2007 ) .
y. alhassid , g.f .
bertsch , l. fang and b. sabbey , phys .
rev c * 74 * , 034301 ( 2006 ) .
g. h. lang , c. w. johnson , s. e. koonin , and w. e. ormand , phys .
c * 48 * , 1518 ( 1993 ) .
y. alhassid , d.j .
dean , s.e .
koonin , g. lang and w.e .
ormand , phys .
* 72 * , 613 ( 1994 ) .
t. otsuka , t. mizusaki , and m. honma , j. phys .
g * 25 * 699 ( 1999 ) .
e. caurier , m. rejmund , and h. grawe , phys .
c * 67 * 054310 ( 2003 ) . i. stetcu and c. w. johnson , phys .
c * 66 * 034301 ( 2002 ) .
brown and w.a .
richter , phys .
c * 58 * 2099 ( 1998 ) .
e. chabanat , p. bonche , p. haensel , j. meyer and r. schaeffer , nucl .
a * 635 * , 231 ( 1998 ) ; nucl .
a * 643 * , 441(e ) ( 1998 ) .
j. terasaki , p .- h .
heenen , h. flocard and p. bonche , nucl .
a * 600 * , 371 ( 1996 ) .
p. bonche , h. flocard and p .- h .
heenen , comp . phys
. comm . * 171 * , 49 ( 2005 ) .
p. bonche , h. flocard , p.h .
heenen , s.j . krieger and m.s .
weiss , nucl .
a * 443 * , 39 ( 1985 ) .
h. nakada and y. alhassid , phys .
79 * , 2939 ( 1997 ) ; y. alhassid , s. liu and h. nakada , phys . rev
. lett . * 83 * , 4265 ( 1999 ) .
y. alhassid , g.f . bertsch and l. fang ,
c * 68 * , 044322 ( 2003 ) .
valatin , phys . rev . * 122 * , 1012 ( 1961 ) . j. von delft and d.c .
ralph , phys .
345 * , 61 ( 2001 ) , and references therein .
y. alhassid , l. fang and s. schmidt , cond - mat/0702304 .
y. alhassid , nucl .
a * 788 * , 357c ( 2007 ) , and references therein .
brown , a. etchegoyen , and w.d.m .
rae , msu - nscl report 524 ( 1988 ) . | we describe a procedure for mapping a self - consistent mean - field theory ( also known as density functional theory ) into a shell model hamiltonian that includes quadrupole - quadrupole and monopole pairing interactions in a truncated space .
we test our method in the deformed @xmath0 @xmath1-shell nuclei @xmath2ne , @xmath3 mg and @xmath4ar , starting from the hartree - fock plus bcs approximation of the usd shell model interaction . a similar procedure
is then followed using the sly4 skyrme energy density functional in the particle - hole channel plus a zero - range density - dependent force in the pairing channel . using the ground - state solution of this density functional theory at the hartree - fock plus bcs level
, an effective shell model hamiltonian is constructed .
we use this mapped hamiltonian to extract quadrupolar and pairing correlation energies beyond the mean field approximation .
the rescaling of the mass quadrupole operator in the truncated shell model space is found to be almost independent of the coupling strength used in the pairing channel of the underlying mean - field theory . |
operational methods , developed within the context of the fractional derivative formalism @xcite , have opened new possibilities in the application of calculus .
even classical problems , with well - known solutions , may acquire a different flavor if viewed from such a perspective .
this , if properly pursued , may allow for further progress , disclosing new avenues of study and generalizations . as it is well known , the operation of integration is the inverse of that of derivation .
such a statement by itself does not enable any means to establish `` practical '' rules to handle integrals and derivatives on the same footing .
an almost natural environment for this specific assertion are the technicalities associated with the formalism of _ real order _ derivatives ( i.e. not necessarily positive ) , in which the distinction between integrals and derivatives becomes superfluous .
the use of the real order formalism offers new computational and conceptual tools allowing e.g. the extension of the concept of integration by parts .
within such an approach the integral of a function can be written in terms of the infinite sum @xmath0 where @xmath1 denotes the @xmath2th derivative of the integrand . an elementary way to verify eq .
is to integrate by parts its left hand side , written down as @xmath3 , with @xmath4 .
first two integrations furnish @xmath5 from which , after infinite number of steps , the general form of eq .
follows . the equivalent proof is based on rewriting eq .
( [ eq : eq.1 ] ) as @xmath6 , \\ { _ { a}\hat{d}_{x}^{-1}}[h(x ) ] & = \int_a^x h(\xi ) d\xi , \quad g(x ) = 1 , \end{split}\end{aligned}\ ] ] where @xmath7 denotes the negative derivative operator @xcite .
the natural extension of eq . defines the operator @xmath8 _ via _
@xmath9 = { _ { a}d_{x}^{-1}}\left[{_{a}d_{x}^{-1 } } [ h(x)]\right].$ ] the previous definition of the integral of the product of two functions suggests the use of the following , appropriately generalized , leibniz formula : @xmath10= \sum_{s=0}^{\infty } \binom{-1}{s } { _ { 0}\hat{d}_{x}^{-s-1}}[g(x ) ] f^{(s)}(x).\ ] ] the superscript denotes the order of the derivative , be it negative or positive . the identity in eq .
( [ eq : eq.1 ] ) follows from eq .
( [ eq : eq.3 ] ) for @xmath11 , for which @xmath12 = \frac{x^{s+1}}{(s+1)!},\ ] ] and @xmath13 .
the interesting property of aforementioned relations is that they allow the evaluation of the primitive of a function in terms of an automatic procedure , analogous to that used in the calculation of the derivatives of a function . at the same time it marks the conceptual , even though not formal , difference between the two operations .
the evaluation of the primitive of a function , using the generalized leibniz rule , gives rise to a computational procedure involving , most of the times , an infinite number of steps .
( [ eq : eq.3 ] ) becomes a truly practical tool if e.g. the function @xmath14 has special properties under the operation of derivation , like being cyclic or vanishing after a number of steps .
let us combine the above formalism with the properties of the two - variable hermite - kamp de frit polynomials @xcite : @xmath15 where @xmath16 are the conventional hermite polynomials .
the @xmath17 of eq .
satisfy the following relations : @xmath18 using eqs .
( [ eq : eq.1 ] ) , ( [ eq : eq . 6 ] ) and , we get @xmath19 and @xmath20 observe that eqs . and imply that eqs .
, , , and contain finite sums .
furthermore , by taking into account the identity @xcite @xmath21 we find @xmath22 or @xmath23 eq . is a representation of known integrals ( compare eq .
( 2.33.3 ) for @xmath24 in @xcite ) in terms of series involving the hermite polynomials .
these are just a few examples that underline the versatility of this method which merges new and old concepts . in the forthcoming section
we will perform a step forward in this direction by combining this formalism with others approaches of the umbral and operational nature . in this way we will reveal some potentialities of this noticeable computational tool , and demonstrate that they are amenable for further implementations .
in a number of previous papers @xcite it has been established that the umbral image of the bessel function is the gaussian .
indeed , if we define the shift operator @xmath25 satisfing @xmath26 and @xmath27 then the relevant series expansion yields @xmath28 it is important to emphasize that the core of the previous formalism can be traced back to the work of mikusiski @xcite .
the @xmath29th order bessel functions are , in the same spirit , defined as @xmath30 such a restyling allows noticeable simplifications in the theory of bessel functions themselves @xcite .
other practical outcome concerns the handling of the associated integrals and of many other technicalities related to the ramanujan master theorem ( rmt ) @xcite .
furthermore , the same formalism provides the possibility of recovering a large body of the properties of bessel functions and of other special functions as well , using genuine algebraic tools @xcite . without entering into details of the applications of the rmt
we note that , for the purposes of this paper , it can be linked to a kind of general rule , which we call _ criterion of permanence _
, stated as it follows : _ if an umbral correspondence is established between two different functions , such a correspondence can be extended to other operations , including derivatives and integrals .
_ accordingly , since the umbral correspondence between gaussian and bessel functions holds , we can use the criterion of permanence for an appropriate `` translation '' of the well - known identities of the gaussian integrals into an umbral language .
in fact , using @xmath31 , @xmath32 , we get @xmath33 the extension of the criterion of permanence , under umbral correspondence , to the properties of the gamma function yields the identities @xcite @xmath34 see eq .
( 2.12.2.2 ) of @xcite , and @xmath35 other definite integrals can be obtained by a judicious application of the same principle .
for example , @xmath36 furthermore , the use of eq . for @xmath37 enables us to obtain the @xmath29th derivative of the bessel function @xcite @xmath38 consequently , substituting eq .
( [ eq : eq . 20 ] ) into eq .
( [ eq : eq.1 ] ) , we get @xmath39.\ ] ] eq .
is the third alternative to the formulas ( 1.10.1.1 ) and ( 1.10.1.2 ) in @xcite . likewise
is the second alternative to the formula ( 1.8.1.14 ) in @xcite .
the extension of such an approach to the theory of hankel transform @xcite is particularly illuminating . limiting ourselves to the @xmath40th order hankel transform ( denoted by @xmath41 ) we set @xmath42 \equiv \int_{0}^{\infty } x f(x )
j_{0}(x y ) dx = \int_{0}^{\infty } x f(x ) e^{-c_{z}(xy/2)^{2 } } dx\ , \frac{1}{\gamma(1+z)}\big\vert_{z=0},\ ] ] and see that the hankel transform of the function @xmath43 , reduces to the integral evaluated in eq .
: @xmath44 & = \int_{0}^{\infty } e^{-(1+y^{2}c_{z}/4)x^{2 } } dx\ , \frac{1}{\gamma(1+z)}\big\vert_{z=0 } \nonumber \\ & = \frac{\sqrt{\pi}/2}{\sqrt{1 + y^{2}c_{z}/4}}\ , \frac{1}{\gamma(1+z)}\big\vert_{z=0}.\end{aligned}\ ] ] other integral transforms can be framed within the same context .
the well - known identities ( see formula ( 2.12.18.7 ) of @xcite ) @xmath45 can be easily restated as it follows : @xmath46 by noting that @xmath47 with @xmath48 being the @xmath40-th order tricomi function , satisfying the identity @xcite @xmath49 where @xmath50 we find @xmath51 compare eq .
( 2.12.18.13 ) of @xcite . by a straightforward application of this method
we also obtain @xmath52 which is eq .
( 2.12.34.1 ) of @xcite .
we hope to have demonstrated by now the far - reaching potentialities of the method .
further examples showing its reliability will be discussed in the forthcoming sections .
in the previous sections we have introduced a technique of the umbral type , which , roughly speaking , consists in the formal `` downgrading '' of bessel functions , namely higher order transcendental functions , to the elementary gaussian . a quite natural next step forward , in this process of reduction , is downgrading of the gaussian to a rational function , according to the formal prescription : @xmath53 that gives @xmath54 which represents the umbral image of the error function @xmath55 . the use of the rational image of the gaussian provided by eq .
( [ eq : eq . 30 ] ) , allows us to introduce the new functions : @xmath56 \quad \text{and } \quad sn_{\frac{1}{2}}(x ) = -im\left[\varepsilon_{\frac{1}{2}}\left(ix\right)\right],\ ] ] where @xmath57 it is evident that @xmath58 where @xmath59 is the imaginary error function . in this manner
we have introduced the gaussian @xmath60 and its complementary function @xmath61 , whose behavior as functions of @xmath62 is shown in fig .
[ fig1 ] .
gaussian function @xmath60 ( red continuous line ) and its complement @xmath61 ( blue dashed line ) . ] @xmath60 and @xmath61 can be viewed as trigonometriclike functions and this can be strengthened by the formal identities , which follow from eqs . and : @xmath63 whose successive derivatives define new two - parameter functions : @xmath64 in general we can write @xmath65 where @xmath66 the second equality in eq .
can be obtained using the properties of the gamma function and its left hand side is equal to @xmath67 , where we have used the kummer relation eq .
( 7.11.1.2 ) of @xcite for the confluent hypergeometric function @xmath68 .
then , the use of ( 7.11.1.19 ) of @xcite permits one to rewrite it as @xmath69 which is precisely the right hand side of eq . .
examples of the behavior of the previous functions for different values of the indices are reported in figs .
2 . ( blue dashed ) and
@xmath70 ( red continuous ) for @xmath71 in fig .
[ fig2]a and @xmath72 in fig .
[ fig2]b.,title="fig : " ] ( blue dashed ) and @xmath70 ( red continuous ) for @xmath71 in fig .
[ fig2]a and @xmath72 in fig .
[ fig2]b.,title="fig : " ]
the theory of integral transforms is one of the pillars of operational calculus @xcite . the heaviside operational calculus @xcite received its rigorous ( at least for mathematicians ) foundation within the context of laplace transform theory and the already quoted mikusinski operational theory relied on the formalism of convolution quotients @xcite .
in addition , many integral transforms have been shown to be expressible in terms of exponential operators , as it is the case of the fractional fourier and airy transforms @xcite . in this section
we make further steps in both directions . when dealing with the borel transform ( bt ) @xcite we will develop , using the formalism of exponential operators , new analytical tools to reformulate the relevant findings .
we remind that the bt of a function @xmath73 is defined by the integral @xcite @xmath74 \equiv f_{b}(x ) = \int_{0}^{\infty } e^{-t}f(t x ) dt,\ ] ] already exploited tacitly in eqs . and .
such a transform plays a significant role in the treating of the series resummation in the quantum field theory @xcite . in this paper
we will consider its properties through the use of a different point of view , which will allow a fairly natural link to the previously discussed umbral methods .
the use of the identity @xcite @xmath75 allows one to write @xmath76 , \quad \text{where}\quad \hat{b}^{(1 ) } = \int_{0}^{\infty}e^{-t}t^{x\partial_{x}}dt = \gamma(1 + x\partial_{x}),\ ] ] which is the operator form of the borel transform .
the above formula also appears in connection with the inversion problem of the laplace transform , see @xcite , theorem 2.14 , eq .
( 4.74 ) .
according to our procedure it follows that the borel transform of the @xmath40th order tricomi function of eq .
is provided by @xmath77 \nonumber \\ & = \sum_{r=0}^{\infty } \gamma(1 + r ) \frac{(-x)^{r}}{(r!)^{2 } } \nonumber \\ & = e^{-x}.\end{aligned}\ ] ] the successive application of the borel operator to the tricomi function yields the identity @xmath78 \nonumber \\ & = \hat{b}^{(1 ) } e^{-x } \nonumber \\ & = \sum_{r=0}^{\infty } \gamma(1 + r ) \frac{(-x)^{r}}{r ! } \nonumber \\ & = \frac{1}{1+x } , \quad \text{for}\quad |x|<1,\end{aligned}\ ] ] while the further application of @xmath79 yields a divergent series , namely @xmath80 the pattern behind the successive applications of the borel operator to the tricomi function @xmath48 is the already mentioned `` downgrading '' procedure , which `` reduces '' a higher transcendental function to an elementary function . up to now
we have interchanged the borel operators and series summation without taking too much caution . in the case of eq .
( [ eq : eq . 41 ] ) such a procedure is fully justified , whereas in applying to eq .
( [ eq : eq . 42 ] ) the method is limited to the convergence region . in the case of eq .
( [ eq : eq . 43 ] ) it is not justified since it gives rise to a divergent series . in the following we will adopt some flexibility in handling these problems and include in our treatment also the case of divergent series . in the previous exposition
the repeated application of bt has been associated with the borel operator raised to some integer power .
now , we explore the possibility of defining the _ fractional _ bt .
for that purpose we introduce the operator @xmath81 which will be referred to as the borel operator of index @xmath82 .
for example , @xmath83 applied to @xmath84 gives @xmath85 & = \gamma\left(1 + \frac{1}{2}x\partial_{x}\right)\sum_{r=0}^{\infty}\frac{(-1)^{r}}{(r!)^{2}}\left(\frac{x}{2}\right)^{2r } \nonumber \\ & = \sum_{r=0}^{\infty}\frac{(-1)^{r}}{r!}\left(\frac{x}{2}\right)^{2r } \nonumber \\ & = e^{-(x/2)^{2}}.\end{aligned}\ ] ] assuming that the operator @xmath86 exists and satisfies @xmath87 , and using eq .
, we get @xmath88 a more rigorous definition of the inverse operator @xmath86 may be achieved through the use of the hankel contour integral @xcite , namely @xmath89 which can be extended to write @xmath90 * proposition 1 . *
_ given the function @xmath14 having the integral @xmath91 , then @xmath92 dx = k \gamma(1 - \alpha ) , \quad |\alpha| < 1,\ ] ] and @xmath93 dx = \frac{k}{\gamma(1 - \alpha ) } , \quad |\alpha| < 1.\ ] ] _ the proof of eq .
is fairly straightforward ; applying the previous definitions eq
. we find @xmath94 dx & = \int_{-\infty}^{\infty}\left(\int_{0}^{\infty } e^{-t } f(t^{\alpha}x ) dt \right)dx \nonumber \\ & = \int_{0}^{\infty } e^{-t } \left(\int_{-\infty}^{\infty } f(t^{\alpha}x ) dx\right ) dt \nonumber \\ & = \int_{0}^{\infty } e^{-t } t^{-\alpha } dt \int_{-\infty}^{\infty } f(\sigma ) d\sigma \nonumber\\ & = k\gamma(1 - \alpha).\end{aligned}\ ] ] the same procedure can be carried out for the inverse borel transform .
we shall present now how the above formalism can be employed for explicit derivation of generating functions .
we use the iconic example of @xmath29th order tricomi function as a benchmark . first note that from eq .
it follows for all @xmath29 ( compare eq .
( 2.12.9.3 ) of @xcite ) that @xmath95 and consequently @xmath96 if we assume that the summation and integration may be interchanged and if we set @xmath97 with the unknown function @xmath98 , then eq .
can be rewritten in the form @xmath99 according to eq .
( [ eq : eq . 41 ] ) and eq .
( 5.7.6.1 ) of @xcite , eq . allows the conclusion @xmath100 a slight variation of the previous arguments applies to the standard bessel functions as well . the use of eq .
( 2.12.9.3 ) of @xcite for @xmath101 and @xmath102 , i.e. @xmath103 yields @xmath104 regarding again @xmath105 as an unknown function , we obtain from eqs .
( [ eq : eq . 60 ] ) and ( [ eq : eq . 59 ] ) @xmath106 which can easily be translated into the well - known generating function of @xmath107 , see eq .
( 5.7.6.1 ) of @xcite : @xmath108 further comments on these last points will be provided in the forthcoming sections .
the application of borel transform techniques is a frequently used tool in analysis and in applied science @xcite .
its use in the treatment of perturbative series in quantum field theory is tricky albeit effective and proceeds as follows : the inverse bt is used to accelerate the convergence of the perturbative expansion ; the function obtained is then bt transformed to recover the sum of the series . to better clarify the relevance of borel resummation methods to the topics of the present article we critically review what we did in the previous section .
the successive application of the bt to a function with non - zero radius of convergence has led to a function whose series coefficients grow factorially with the order of expansion . its range of convergence is therefore vanishing . in quantum field theory one
is faced with the opposite problem , namely that of recovering a function with non - zero radius of convergence starting from a divergent series .
the problem is `` cured '' by dividing each term in the expansion by a factor @xmath109 .
this is called a borel sum ; if it can be summed and analytically continued over the whole real axis , then the initial expansion is called borel summable @xcite . for @xmath110 ,
let us calculate @xmath111 : @xmath112 using the umbral method this can be rewritten as @xmath113 as an example we calculate the action of @xmath114 on the gaussian .
that gives @xmath115 which is the inversion of eq . .
the laguerre polynomials @xcite can be framed within the same context .
indeed , it is easy to obtain eq .
( 7.414.6 ) of @xcite : @xmath116 where @xmath117 are two - variable laguerre polynomials , related to the standard laguerre polynomials by @xmath118 @xcite . a consequence of eqs .
and is @xmath119 accordingly , the ordinary generating function of the laguerre polynomials is the inverse bt of the geometric series .
namely , from eq .
we get @xmath120 \nonumber \\ & = \frac{1}{(1 - y \xi ) + \xi c_{z } x}\ , \frac{1}{\gamma(1+z)}\big\vert_{z=0 } \nonumber \\ & = \frac{1}{1 - y \xi } \exp\left(- \frac{x\xi}{1-y\xi}\right).\end{aligned}\ ] ] the last equality in eq
. follows from eq . .
if the borel operator acts on the @xmath121 variable we obtain @xmath122 the function @xmath123 denotes the bessel truncated polynomials @xcite . the use of eq .
( 5.11.1.5 ) of @xcite ( see also @xcite ) @xmath124 yields the generating function for the bessel truncated polynomials @xmath125 = \frac{c_{0}(x \xi)}{1 - y\xi}.\ ] ] we employ previous results to obtain the ordinary generating functions of _ lacunary _ laguerre polynomials .
( let us stress that the following example is included solely for illustrative purposes , as its result is immediately obtained from eq . . ) . according to our formalism we have @xmath126 \nonumber \\ & = \frac{1}{2 } \left[(1 - \sqrt{t } y)^{-1 } \left(1 + \frac{c_{z}\sqrt{\xi } x}{1 - \sqrt{t}y}\right)^{-1 } + ( 1 + \sqrt{t } y)^{-1 } \left(1 - \frac{c_{z}\sqrt{t } x}{1 + \sqrt{t}y}\right)^{-1}\right ] \frac{1}{\gamma(1+z)}\big\vert_{z=0 } \nonumber \\ & = \frac{1}{2}\left[\frac{\exp\left(-\frac{\sqrt{t } x}{1 - \sqrt{t}y}\right)}{1 - \sqrt{t}y } + \frac{\exp\left(\frac{\sqrt{t } x}{1 + \sqrt{t}y}\right)}{1 + \sqrt{t}y}\right].\end{aligned}\ ] ] the exponential generating function for lacunary laguerre polynomials @xmath127 through the umbral representation gives @xmath128 which reproduces eq .
( 40 ) in @xcite .
we shall present now the derivation of eq . using the bt approach .
the use of eq . gives @xmath129 inverting eq . and using eq .
for @xmath130 we find @xmath131 which reproduces eq . . in the concluding remarks we will further comment on the lacunary laguerre generating functions and on the relevant link with previous research @xcite
. to make further progress in our exposition we introduce the integral transform , known in the literature as the borel - leroy ( b - l ) transform @xcite , @xmath132
thus , the associated two - parameter differential operator can be written as @xmath133 the relevant action on the @xmath134th order tricomi function yields @xmath135 where @xmath136 which converges for @xmath137 .
the inverse b - l transform of the exponential yields the bessel - wright function @xmath138 @xcite , namely @xmath139 going back to eq .
( [ eq : eq . 59 ] ) we also find that @xmath140 whose the right hand side can serve for the direct computation of the generating function in eq . ( [ eq : eq . 60 ] ) . as a further generalization we discuss the @xmath141-borel - leroy transform defined by @xmath142 which , upon the use of the euler @xmath143 function @xmath144 @xcite , can be transformed into the differential form @xmath145 it is interesting to note that the previous operator when acting on the exponential transforms it into a mittag - leffler function @xcite , namely @xmath146 ( remark that eq .
is of operational nature and any modification of the argument of the exponential should be preceded by a detailed evaluation of eq . ) .
generalizing * proposition 1 * with @xmath14 such that @xmath91 to the @xmath141-borel - leroy transform , we get @xmath147 where we used eq . and eq . ( 8.380.1 ) of @xcite .
. for @xmath148 gives @xmath149 , \left[\frac{\alpha + \beta}{2 } , \frac{\alpha+\beta+1}{2}\right ] ; -x^2\right ) dx\\[0.3\baselineskip ] & = \sqrt{\pi } \frac{\gamma(\alpha-1)\gamma(\beta)}{\gamma(\alpha-1+\beta)},\end{aligned}\ ] ] see eq .
( 2.22.1.1 ) in @xcite , where @xmath150 is the generalized hypergeometric function
. these are just few examples of the various possibilities offered by the present formalism further applications will be discussed in the forthcoming section .
in the previous sections we have provided a comprehensive analysis of different formulations of the operational techniques usually adopted to treat problems associated with the handling of the properties of special functions and polynomials .
we have seen that the use of the bt is a powerful unifying tool yielding the appropriate environment for a more rigorous treatment of the umbral techniques . in this paragraph
we collect examples connected with orthogonal polynomials .
the derivation of the ordinary generating function of @xmath151-_lacunary _ laguerre polynomials ( @xmath152 ) in the language of bt goes as follows @xmath153^{-1}\nonumber \\ & = \frac{1}{p } \sum_{k=0}^{p-1 } ( 1 - t^{1/p } e^{2\pi i k / p } y)^{-1 } \left(1 + \frac{t^{1/p } e^{2\pi i k / p } c_{z } x}{1 - t^{1/p } e^{2\pi i k / p } y}\right)^{-1 } \frac{1}{\gamma(1+z)}\big\vert_{z=0 } \nonumber \\ & = \frac{1}{p } \sum_{k=0}^{p-1 } ( 1 - t^{1/p } e^{2\pi i k / p } y)^{-1 } \sum_{n=0}^{\infty } \left(-\frac{t^{1/p } e^{2\pi i k / p } x}{1 - t^{1/p } e^{2\pi i k / p } y}\right)^{\ !
n } c_{z}^n \frac{1}{\gamma(1+z)}\big\vert_{z=0 } \nonumber \\ & = \frac{1}{p } \sum_{k=0}^{p-1 } \frac{\exp\left(-\frac{t^{1/p } e^{2\pi i k / p } x}{1 - t^{1/p } e^{2\pi i k / p } y}\right)}{1 - t^{1/p } e^{2\pi i k / p } y}.\end{aligned}\ ] ] the generalization to the case @xmath154 , @xmath155 is straightforward . for completeness
it is worth to touch on the possibility of treating the hermite polynomials in a fashion borrowed from the treatment of laguerre s . for this purpose
we need another choice of the function @xmath156 in eq . , namely @xmath157 .
then the two - variable hermite polynomials @xmath158 can be defined as @xmath159 the right hand side of eq .
gives @xmath160 \
, \ulamek{(2\sqrt{y})^z}{\sqrt{\pi } } \gamma(\ulamek{1+z}{2 } ) |\cos(z\ulamek{\pi}{2})|\big\vert_{z=0 } \nonumber \\ & = \sum_{k=0}^n \binom{n}{k } x^{n - k } \frac{(2\sqrt{y})^k}{\sqrt{\pi } } \gamma\left(\frac{1+k}{2}\right ) \left|\cos\left(k\frac{\pi}{2}\right)\right| \nonumber \\ &
= \sum_{r=0}^{\lfloor n/2\rfloor } \binom{n}{2r } x^{n-2r } \frac{4^r y^r}{\sqrt{\pi } } \gamma\left(\frac{1}{2 } + r\right ) \nonumber \\ & = n ! \sum_{r=0}^{\lfloor n/2\rfloor } \frac{x^{n-2r } y^r}{(n-2r ) ! } \frac{4^r \gamma(\frac{1}{2 } + r)}{\sqrt{\pi } ( 2r ) ! } \nonumber \\ & = n ! \sum_{r=0}^{\lfloor n/2\rfloor } \frac{x^{n-2r } y^r}{r ! ( n-2r)!},\end{aligned}\ ] ] compare eq . .
the operator formula eq .
is an effective tool to quickly derive a number of known and less known summation formulas involving the hermite polynomials .
we start with @xmath161 \ , \ulamek{(2\sqrt{y})^z}{\sqrt{\pi } } \gamma(\ulamek{1+z}{2 } ) |\cos(z\ulamek{\pi}{2})|\big\vert_{z=0 } \nonumber \\ & = \sum_{r=0}^{\infty } \frac{(4yx^{2})^r}{\sqrt{\pi } } \frac{\gamma(\ulamek{1}{2 } + r)}{(2r ) ! } \nonumber \\ & = \sum_{r=0}^{\infty } \frac{(y x^2)^r}{r ! }
= e^{y x^2 } , \end{aligned}\ ] ] and @xmath162 \ , \ulamek{(2\sqrt{y})^z}{\sqrt{\pi } } \gamma(\ulamek{1+z}{2 } ) |\cos(z\ulamek{\pi}{2})|\big\vert_{z=0 } \nonumber \\ & = \frac{1}{\sqrt{\pi } } \int_{-\infty}^{\infty } e^{-\xi^2}e^{y ( 2\xi\sqrt{x})^2 } d\xi \nonumber\\ & = \frac{1}{\sqrt{1 - 4yx } } , \quad \text{for } \quad |x| < \frac{1}{4 |y|}.\end{aligned}\ ] ] combining the previous results one can show @xmath163 \ , \ulamek{(2\sqrt{y})^z}{\sqrt{\pi } } \gamma(\ulamek{1+z}{2 } ) |\cos(z\ulamek{\pi}{2})|\big\vert_{z=0 } \nonumber \\ & = e^{t(x + c_z)^2 } \ , \ulamek{(2\sqrt{y})^z}{\sqrt{\pi } } \gamma(\ulamek{1+z}{2 } ) |\cos(z\ulamek{\pi}{2})|\big\vert_{z=0 } \nonumber \\ & = \frac{1}{\sqrt{1 - 4yt}}e^{\frac{x^{2}t}{1 - 4yt } } , \quad \text{for } \quad |t| < ( 4|y|)^{-1},\end{aligned}\ ] ] see eq .
( 5.12.1.4 ) of @xcite .
an alternative procedure has been put forward by gessel and jayawant @xcite , who discussed a _
triple lacunary _ generating function for hermite polynomials using umbral and combinatorial techniques . moreover , it is interesting to note that from eqs .
( [ eq : eq . 83 ] ) and we can derive another operational definition of the hermite polynomials : @xmath164 which , on account of @xmath165 yields the well - known operational identity @xmath166 also used as a primary definition of @xmath17 @xcite . according to the previous identities the derivation of the classical mehler formula can be given rapidly .
using eqs .
and we get @xmath167^n}{n ! } h_{n}(u , v)\ , \ulamek{(2\sqrt{y})^z}{\sqrt{\pi } } \gamma(\ulamek{1+z}{2 } ) |\cos(z\ulamek{\pi}{2})|\big\vert_{z=0 } \nonumber \\ & = e^{t x u + v ( t x)^2 } e^{vt^2 c_z^2 + ( 2v x t^2 + tu ) c_z}\ , \ulamek{(2\sqrt{y})^z}{\sqrt{\pi } } \gamma(\ulamek{1+z}{2 } ) |\cos(z\ulamek{\pi}{2})|\big\vert_{z=0 } \nonumber \\ & = e^{-u^2/(4 v)}e^{v t^2 [ x + u/(2 v t ) + c_z]^2 } \ , \ulamek{(2\sqrt{y})^z}{\sqrt{\pi } } \gamma(\ulamek{1+z}{2 } ) |\cos(z\ulamek{\pi}{2})|\big\vert_{z=0 } \nonumber \\ & = \frac{e^{y(ut)^2}}{\sqrt{1 - 4yvt^2 } } \exp\left(\frac{u t x + v t^2 x^2}{1 - 4yvt^2}\right ) , \quad 1 - 4yvt^2>0.\end{aligned}\ ] ] we conclude this paper with a further remark , concerning the possibility of extending the method to hybrid bilateral generating functions , involving products of laguerre and hermite polynomials .
this problem has already been discussed in @xcite using the procedure exploiting the gauss transform representation of hermite polynomials .
the derivation we propose here is significantly simpler : @xmath168^{n}}{n ! } h_{n}(u , v)\ , \frac{1}{\gamma(1+z)}\big\vert_{z=0 } \nonumber \\ & = e^{t(y - c_{z}x)u + [ t(y - c_{z}x)]^{2 } v}\ , \frac{1}{\gamma(1+z)}\big\vert_{z=0 } \nonumber \\ & = e^{(ty)u + ( ty)^{2}v } e^{- t(xu + 2tyxv)c_{z } + ( tx)^{2 } v c_{z}^{2}}\ , \frac{1}{\gamma(1+z)}\big\vert_{z=0 } \nonumber \\ & = e^{(ty)u + ( ty)^{2}v}\,\sum_{r=0}^{\infty } \frac{x^r}{(r!)^2 } h_r(-tu-2yvt^2 , vt^2).\end{aligned}\ ] ] most of the topics treated in this paper may deserve a deeper and more thorough treatment .
however , we believe that in spite of the heterogeneity of the argument we touched upon many implications offered by the formalism . we hope that they may provide the hints for further speculations . very recently v. strehl @xcite commented on the identities of @xcite and stressed their relevance with the remark : `` all identities are very combinatorial , and combinatorics can help to systematize and extend them considerably '' . furthermore , the following question has been addressed and answered in @xcite : `` what the lacunary laguerre series really count ? '' .
evidently it is very tempting and challenging to reinterpret ensemble of the above results from the combinatorial perspective .
the authors express their sincere appreciation to dr .
d. babusci for interesting and enlightening discussions on the topics treated in this paper .
it is also a pleasure to recognize the interest and the encouragement of prof .
v. strehl .
k. g. , a. h. and k. a. p. were supported by the pan - cnrs program for the french - polish collaboration .
moreover , k. g. thanks for the support from mnisw , warsaw ( poland ) , under `` iuventus plus 2015 - 2016 '' , program no ip2014 013073 .
g. dattoli , p. l. ottaviani , a. torre , and l. vzquez , _ evolution operator equations : integration with algebraic and finite - difference methods . applications
to physical problems in classical and quantum mechanics and quantum field theory _ , riv .
20 * , 1 ( 1997 ) .
j. mikusiski , _ sur les fondements du calcul operatoire _ , studia math . * 11 * , 41 ( 1949 ) ; `` operational calculus '' , ( pergamon press , london , 1959 ) ; s. okamoto , _ a simplified derivation of mikusiski s operational calculus _ , proc .
japan acad .
a * 55*(1 ) , 1 ( 1979 ) .
t. amdeberhan and v. h. moll , _ a formula for a quartic integral : a survey of old proofs and some new ones _ , ramanujan j. * 18 * , 91 ( 2009 ) ; t. amdeberhan , o. espinosa , i. gonzalez , m. harrison , v. h. moll , and a. straub , _
ramanujan s master theorem _ , ramanujan j. * 29 * , 103 ( 2012 ) .
a. a. zakharov , _ borel summation method _
, in `` encyclopedia of mathematics '' , m. hazewinkel ( ed . ) , ( springer , berlin , 2001 ) ; s. weinberg , `` the quantum theory of fields .
ii '' , ( cambridge university press , new york , 2005 ) . | integro - differential methods , currently exploited in calculus , provide an inexhaustible source of tools to be applied to a wide class of problems , involving the theory of special functions and other subjects .
the use of integral transforms of the borel type and the associated formalism is shown to be a very effective mean , constituting a solid bridge between umbral and operational methods .
we merge these different points of view to obtain new and efficient analytical techniques for the derivation of integrals of special functions and the summation of associated generating functions as well . |
an introduction to the problem of characterizing the psf of a natural guide star ( ngs ) adaptive optics ( ao ) system is given in @xcite ( hereafter referred to as paper i ) .
in an ngs observation a bright star is used to guide the ao system while the telescope is pointed at a scientific target which is typically 10 - 30 away , and perhaps as distant as 1 . because the performance of the ao system degrades for fainter guide stars - correction is poorer as signal - to - noise ( s / n ) on the wavefront sensor ( wfs ) decreases - any subsequent calibration observation must use a guide star of similar brightness .
even so , the delivered correction depends directly on quickly changing seeing conditions ( measured by the fried parameter @xmath1 ) and is therefore not likely to be the same during the two observations . further aggravating this
, the spatial dependence of the phase error is dependent on the distribution of turbulence strength along the line of sight ( given by the @xmath2 profile , where @xmath3 is height ) , and thus psf anisoplanatism can be strong and change with time . therefore a calibration measurement done with the same offset but at a later time may provide little information about the psf at the scientific target . in paper
i we discussed a method which may improve the determination of the ao off - axis psf .
it uses a mosaicked observation of a dense star field to characterize the field dependence of the psf .
the entire mosaic is deconvolved with the image of the guide star in order to generate a kernel map " .
each deconvolved star is a function which can convert the guide star image into a psf at that off - axis position .
the kernel map is combined with on - axis psf information during the target observation in order to produce a psf estimate that is potentially more accurate for both the target position and time of the scientific observation .
the technique we outlined in paper i provided modest but useful improvement in the accuracy of ngs ao off - axis psf reconstruction - reducing a 60% error in the prediction of fwhm at 25 offset ( if one used just the guide star image alone ) down to about 19% by our method . for many scientific programs
this level of accuracy would be helpful .
for example , it would assist in subtracting the pointlike core from the image of a quasar host or radio galaxy .
the method would be particularily helpful in this case because one is very unlikely to find a star sufficiently bright to guide ao within a few arcseconds of any given galaxy .
worse , even if a usable star is within 30 , it is likely that there is still no true
psf star " , that is , a star within a few arcseconds of the target sufficiently bright to provide a simultaneous measurement of the target psf .
the advent of laser beacons for ao promises to overcome the problem of guide star availability by providing a bright artificial star " which can be positioned near any target . having a pointable guide star does not , however , solve the problem of psf characterization .
in those cases where no psf star is available - just as in the case of ngs ao - some other form of psf characterization will be needed . in laser guide star ( lgs ) ao , the artificial beacon is produced either by rayleigh scattering or the excitation of atmospheric sodium at high altitude .
the outgoing laser beam is usually sighted along the optical axis of the telescope , which means that in a typical observation both are pointed at the scientific target .
the laser spot " which it produces is entirely optical emission , and thus after reflecting off the deformable mirror ( dm ) a dichroic is used to send this light to a wfs and allow the nir light to pass through to the scientific camera .
these optics are essentially unchanged from the ngs ao case .
however , the laser beam wanders by an unknown amount on the way up and thus the absolute position of the laser spot on the sky is unknown , and it can only be used to sense the high - order aberrations of the incoming wavefront .
an ngs is still needed to probe global tilts across the pupil .
this can be done by means of a separate quad - cell sensor for the ngs , for example .
the radial dependence of anisoplanatism for low zernike orders is weak , and thus the ngs and lgs need not be at the same position on the sky .
in fact , it is highly desirable to guide with the laser pointed at the scientific target - because it takes advantage of the best correction - and choose a convenient nearby ( less than @xmath4 away ) star for the quad - cell . in this paper
we only consider data where the lgs is coincident with the telescope optical axis , and so we will consistently refer to the location of laser spot as on - axis .
one could still obtain a direct on - axis psf measurement by interleaving scientific observations with short exposures of the tip - tilt star with the laser ( and telescope ) pointed at it .
but if the offset is larger than a few arcseconds this method will provide a poor estimate of the off - axis psf , especially if anisoplanatism is strong , as is the case for @xmath5 profiles skewed to high altitude .
a better method would incorporate some means of modeling this anisoplanatism into a corrective off - axis psf kernel .
the results could be limited by how quickly and often the tip - tilt guide star calibration exposures were taken , however , because the seeing could change significantly over the course of a scientific observation .
the differences between the optical configurations of lgs and ngs ao suggest that the semiempirical method of paper i for characterizing the psf is not applicable for lgs observations , for two reasons .
first , the field dependence of the psf variation will be different .
second , a typical lgs ao scientific observation will provide no simultaneous on - axis measurement with which to calibrate the seeing variations of the psf .
the difference in field dependence for lgs ao is easy to understand .
it is partly due to the finite altitude of the artificial beacon , which is perhaps only 15 km for one produced by rayleigh backscatter , or 90 km for emission from excited atmospheric sodium .
that is , the volume of turbulent air traversed by the return beam does not include the entire atmosphere nor is it a cylinder - as it would be if the lgs were effectively at infinity .
this cone effect " degrades the on - axis lgs strehl ratio , but it is not the most important difference in field dependence compared to ngs ao , and it is a weaker effect with smaller aperture . a more important issue is the angular separation of the lgs and ngs on the sky . even though the ngs can be much fainter than that needed for sensing high - order aberrations , in practice the nearest available star may be up to 1 away . clearly , the larger the separation between the lgs and ngs the more complicated the field dependence will be .
the separation of the lgs and ngs also causes a more serious problem with the application of our previous method . in an lgs observation
the on - axis psf is now at the position of the scientific target , which may not be a star .
the laser spot is invisible in the nir and therefore provides no measurement of the on - axis psf in the scientific image .
we describe six techniques for determining the lgs ao off - axis psf and apply these methods to observations made with the shane telescope ao system at lick observatory .
this is a continuation of our program funded by the center for adaptive optics to study ao psfs , which in paper i dealt with ngs data only .
the results of the lgs methods are of accuracy similar to the semiempirical method of paper i , @xmath020% accuracy in strehl ratio for offsets up to 28 . in sections [ strategy ] and [ map ] we propose an observing strategy for lgs - mode psf characterization and discuss the functional form of the kernel map . in section [ observations ]
we describe the observations and data reduction .
analytical calculations and computer simulations were used to further characterize the data , and the psf prediction methods were applied to them .
these analyses are found in section [ analysis ] , and our conclusions follow in section [ conclusions ] .
we have described how lgs and ngs ao observations differ , but at least their high - order corrections should be similar . in paper
i we noted that for perfect correction of @xmath6 zernike modes the strehl ratio at offset @xmath7 from the guide star should be given by @xmath8 , \eqno(1)\ ] ] with the exponent tending toward two for very partial correction .
instrumental " isoplanatic angle @xmath9 depends on the fried parameter @xmath1 and mean height of turbulence @xmath10 as @xmath11 where @xmath12 is the zenith angle . if we neglect tip - tilt anisoplanatism , equations 1 and 2 should still describe the high - order lgs ao correction ( with @xmath7 now refering to radius from the lgs ) because cone effect is folded into the value of @xmath13 .
in reality , wavefront tip and tilt are sensed differently in lgs compared to ngs mode , so the analogy is not exact .
if we assume cone effect and tip - tilt anisoplanatism are negligible , a typical lgs ao observation then differs from an ngs one only by the absence of an on - axis psf star .
we can we still determine the on - axis strehl ratio . a simple model of the lgs ao on - axis
strehl ratio might have it degraded by a wavefront fitting error , a delay error , and photon statistics .
this is @xmath14},\eqno(3)\ ] ] assuming the errors are uncorrelated , where @xmath15 is the number of actuators , @xmath16 is the delay time , @xmath17 is the guide source flux , and the following definitions are used ( see , for e.g. , roddier 1999 ) : @xmath18 @xmath19 and @xmath20 for telescope diameter @xmath21 , mean windspeed @xmath22 , guide - star magnitude @xmath23 , throughput @xmath24 , and wavelength @xmath25 . here
@xmath24 is the total throughput of the telescope plus ao system .
note that even this
simple " model requires several atmospheric measurements and ao system telemetry to estimate @xmath13 .
we can still try to use it , but it makes more sense to find @xmath13 by fitting equation 1 to the star - field data , as we did in paper i. this will account for cone effect by interpolating an estimate of @xmath13 .
fitting equation 1 to the star - field data will also provide an estimate of @xmath9 . to characterize @xmath26 we must fit equation 1 to a further set of star - field mosaics - this time with high - order correction turned off . a proposed observing strategy for
the lgs ao psf characterization is similar to that described in paper i , with one new step .
the observer obtains images of a dense star field once or at most a few times per night ; the core of a globular cluster would work best .
the ao system is guided on a suitable tip - tilt guide star .
the goal is to obtain unsaturated images of a roughly uniform distribution of stars from which to determine the isoplanatic angle . to keep the analysis straightforward
, the same offsets between the lgs and tip - tilt guide star used in the scientific observations should be repeated here .
the contribution of tip - tilt anisoplanatism to the psf is measured in a separate mosaic by repeating the mosaic observations with the high - order correction turned off .
obtaining this tip - tilt - only mosaic is an extra step compared to paper i. the tip - tilt - only and high - order mosaics are used together to determine the psf variation by fitting each independently with equation 1 .
the off - axis psf during a scientific observation is provided by applying an appropriate kernel to either the observed psf at the position of the tip - tilt guide star or one generated using wfs telemetry .
unlike in paper i , to obtain a two - dimensional model of the lgs ao psf , we must theorize the functional form of the kernel map . for simplicity ,
let us assume for now that tip - tilt anisoplanatism is negligible .
the off - axis psf then has the form @xmath27 where @xmath28 is a position in the focal plane , k is the kernel , and @xmath29 indicates a convolution .
note that since @xmath30 is the on - axis psf we need not assume that cone effect is negligible .
the optical tranfer function ( otf ) is the fourier transform of the psf and will have associated with it an
anisoplanatic transfer function " ( atf ) , the fourier transform of the kernel . equation 7 in terms of spatial frequencies @xmath31 is then @xmath32 our goal is to find the functional form of either k or atf .
the results of paper i suggest that the ngs ao kernel is approximately gaussian with a fwhm varying with @xmath7 .
let us assume that the lgs ao kernel also has this shape .
that is , @xmath33}. \eqno(9)\ ] ] the volume of the kernel must be unity everywhere , which gives @xmath34}. \eqno(10)\ ] ] thus if we assume that the kernel has a width of one pixel on - axis we have @xmath35 where fwhm is in pixels .
this gaussian kernel " is obviously a simplification of the kernel shape because a gaussian of one pixel in width is not a true delta function .
let us just assume for now that this is a reasonable approximation .
we have not assumed that cone effect is negligible .
for any non - zero offset @xmath7 , equation 1 combined with equations 9 and 11 then suggests that the kernel should be @xmath36}]}. \eqno(12)\ ] ] it is much more mathematically complex , but one can also determine the kernel from theory .
we can decompose the turbulent phase onto a zernike polynomial basis and see how this decomposition changes with increasing pupil offset . for a given
offset the phase can be written as @xmath37 where @xmath38 is the zernike coefficient at offset @xmath7 .
we will assume that the ao system perfectly corrects @xmath6 zernike modes .
the residual phase is then @xmath39 @xcite have shown in this circumstance that the atf can be written as @xmath40 } , \eqno(15)\ ] ] where sf is the residual phase structure function for wavelength @xmath25 given by @xmath41u_{i , j}(\lambda{\mathbf f } ) , \eqno(16)\ ] ] and the @xmath42 are functions defined as @xmath43[z_j({\mathbf r})-z_j({\mathbf r}+\lambda{\mathbf f})]p({\mathbf r})p({\mathbf r}+\lambda{\mathbf f})\over{\int { \rm d}{\mathbf r}p({\mathbf r})p({\mathbf r}+\lambda{\mathbf f } ) } } , \eqno(17)\ ] ] for position @xmath44 over the range of pupil function @xmath45 .
the angular correlations @xmath46 of this theoretical kernel " have been computed by @xcite and @xcite by assuming particular @xmath5 profiles , and implicitly assuming that cone effect is negligible .
these correlations are @xmath47 where @xmath48 is a height sufficient to be beyond atmospheric turbulence .
the integrals @xmath49 can be calculated either by numerical means or by the method of mellin transforms ; the latter can be found in @xcite . here
@xmath50 and @xmath51 correspond to the radial orders of zernike modes @xmath52 and @xmath53 .
although the gaussian kernel seems to be applicable to both ngs and lgs modes , some error will be committed by neglecting cone effect in the theoretical kernel .
even so , for 3-meter aperatures it is a small error . @xcite
have calculated the residual phase variance for perfect correction of all zernike modes ( @xmath54 ) over a circular aperture when the incoming wavefronts are spherical , that is , propagating from a height of 15 or 90 km . for a hufnagel profile with a mean turbulence height of @xmath55 km and an lgs at 90 km
they find a residual phase variance of @xmath56 if one applies the marechal approximation this suggests perfect lgs ao correction on a 3 meter could at best achieve an on - axis strehl ratio of @xmath57 at @xmath58 , which is very close to unity .
put another way , everything else being equal , this system should achieve about 4% less strehl ratio than an ngs system .
a loss of strehl ratio of 4% due to the cone effect is not a serious concern for our data , where strehl ratios are more typically 30% .
this would not be true for ao on a 10 meter , however , because a similar calculation suggests that the cone effect results in a degradation of about 24% in strehl ratio .
thus the theoretical kernel may not be directly applicable to keck ao without first accounting for the cone effect .
this is important because the integrals in equation 18 require significant computational effort .
we will show that for our data this complex analytical kernel is not needed , and the simpler gaussian kernel - which accounts for cone effect - gives comparable results .
so far we have modeled our kernel maps by assuming that tip - tilt anisoplanatism is negligible .
a less severe and more plausible assumption is that the tip - tilt and high - order loops are independent .
this will allow us to develop two auxiliary kernels with which we may conceivably account for tip - tilt anisoplanatism .
equation 12 should apply equally well to ngs mode , so the gaussian tip - tilt gaussian kernel " has an exponent of two , according to equation 1 .
also , equation 18 is actually a calculation for ngs mode .
so it follows that the tip - tilt theoretical kernel " would be given by equation 16 with @xmath59 .
note that @xmath7 now refers to the offset from the ngs . to account for the tip - tilt component of the field variation
we should be able to degrade the tip - tilt guide - star psf with either of these modified kernels .
we could next apply the high - order lgs gaussian or theoretical kernels to this map as a second step , and see it this reproduces the observed lgs field variation .
observations were made with the 3.0 m shane telescope of the lick observatory using the lawrence livermore national laboratory ( llnl ) ao system in lgs mode .
we discussed the operation of the ao system in ngs mode in paper i. see @xcite and @xcite for further discussion of the ngs system ; @xcite and @xcite for a description of the lgs configuration .
the artificial beacon is produced by a pulsed solid - state pumped - dye laser tuned to excite the 589 nm resonance line of atomic sodium , which is naturally present in an atmospheric layer at about 90 km in altitude .
the laser typically operates at about 10 watts output , although up to 15 watts has been achieved .
the return power depends on seasonal variation of density and thickness of the sodium layer , nightly variation of intervening thin cloud , and target airmass .
the laser launch telescope is attached to the side of the shane and produces a spot along the optical axis , aligned with the same ccd - based shack hartmann sensor used in the ngs system . during our observations
the lgs was sufficiently bright to permit ccd framerates of 100 hz , about twice as fast as during our ngs run .
we always used a relatively bright ( @xmath60 ) star for global tip and tilt measurement ( @xmath61 is possible ) .
the tip - tilt quad cell can patrol within a region roughly 2-square , which permits telescope offsets of up to 55 in any direction .
the quad cell drives a tip - tilt mirror ; higher order aberrations are corrected with the same 61 active element piezostack mirror used in ngs mode .
our lgs observations of m15 were obtained through a @xmath62 filter with the infrared camera for adaptive optics at lick ( ircal ) @xcite in august 2001 .
the same tip - tilt guide star was used for all observations - the same source used for the ngs observations of this field in september 2000 - which had the advantage of being in a region of sky already mapped by us .
archival @xmath23 , @xmath63 , and @xmath64-band hubble space telescope ( hst ) wide - field planetary camera 2 ( wfpc2 ) images were also available .
the large 2.6 @xmath65 2.6 fov of wfpc2 helped to provide astrometry for registration of the mosaicked ao images .
our previous ngs observations had shown that the core of the cluster , 45 to the southeast , was too crowded to obtain reliable strehl ratio measurements for ao .
this region was therefore avoided .
three individual pointings of the telescope produced a chevron shaped mosaic roughly 35 across along each arm .
figure [ figure_m15 ] is an image of the resulting field .
the crosses indicate the position of the laser spot in each pointing , and the tip - tilt guide star is the bright star at upper right .
notice the triangular arrangement of bright stars in this field .
one can see that we could have obtained a direct measurement of the on - axis psf in every frame by centering each observation on a vertex of this constellation . to allow a test that will be explained in section [ generating_artificial_images ] ,
we instead opted to offset the laser either 8 or 20 from the tip - tilt star so as to maintain the same 8 separation between the laser pointing and each psf star .
reducing the largest offset from 25 to 20 also made our observing sequence faster .
the exposure time at each pointing was 100 seconds , and each mosaic took only approximately 10 minutes to complete .
every third or fourth pass was observed with the high - order correction turned off , that is , it was obtained with tip - tilt only correction .
a total of 25 mosaics , including the tip - tilt only fields , was obtained over two nights .
a journal of the observations is given in table [ table_journal_of_observations ] .
no independent measurements of seeing or the turbulence profile were available at lick .
we were still able , however , to determine @xmath1 during the run .
wavefront - sensor telemetry was recorded at intervals by the ao system and used to determine @xmath66 in a manner similar to @xcite .
each mosaic observation was bracketed by @xmath1 measurements , and the averages of those are given in table [ table_journal_of_observations ] .
note that the seeing during the first night ( @xmath67 cm ) was almost a factor of two better than the second night ( @xmath68 cm ) .
the output laser power at the beginning of the first night was 13.4 watts .
no output power was recorded for the second night , but it was probably similar .
variation in the returned signal over the course of each night can be inferred from the mean wfs illumination .
the standard deviation of the mean return power was 7% for the first night and 35% for the second .
some unknown part of this variation could have come from changes in output power .
variable thin cloud could also have played a role , but apparently a small one .
the flux from the tip - tilt guide star was measured in each of the mosaics and was stable over both nights to within 4% ( std .
all observations were at an airmass less than about 1.3 , which leaves only fluctuations in the concentration of atmospheric sodium as a remaining possible source of lgs variability .
perhaps this accounts for the larger changes in return power on the second night .
the windspeed was recorded by an anemometer on the roof of a nearby telescope building , situated about 100 m from the shane dome .
the anemometer measurements may be affected by the sheltering of buildings and terrain , and thus the resulting average @xmath69 almost certainly underestimates the true windspeed of the bulk of turbulence .
an atmospheric profiling instrument such as a scintillation detection and ranging ( scidar ) device , if it had been available , could have provided an independent measurement of @xmath10 and estimates of the windspeed at higher altitudes .
all the globular - cluster data were compared to hst wfpc2 observations of the same fields in order to determine the plate scale and orientation of the cameras .
we obtained good relative astrometry of stars , with sub - pixel accuracy .
the data were flat - fielded , and aperture photometry was carried out on all the sky - subtracted images .
strehl ratios were calculated in the same manner as described in paper i. the accuracy of these values is similar to the photometric accuracy , which was better than 4% .
we obtained an estimate of the on - axis strehl ratio @xmath13 and the instrumental isoplanatic angle @xmath9 by a least - squares fit of equation 1 to the measurements of strehl ratios for stars in each frame of the mosaics . for the tip - tilt
only correction frames , the fitted values of @xmath13 agreed well with the direct measurement of the guide star .
this direct measurement was not available in the the lgs observations .
the values of both @xmath13 and @xmath9 corrected for zenith angle are given in table [ table_journal_of_observations ] .
it is somewhat surprising that @xmath13 in most of the 20 offset fields are higher than in 8 offset fields taken at almost the same time , although this is not always the case .
one would expect exactly the opposite due to tip - tilt anisoplanatism .
there are fewer stars in the 8 pointing , which does contribute to a noisier fit with equation 1 .
there are also correspondingly fewer stars very close to on - axis , which might lead to a small systematic underestimate of @xmath13 , because the fit will be poorest at the most sharply falling part of the curve . the instrumental isoplanatic angles ( at 2.2 @xmath70 ) were approximately 13 for lgs and 32 for tip - tilt on average .
we alternated between tip - tilt only and full correction mosaics during the observations and thus both should sample roughly the same @xmath10 .
this suggests that @xmath71 . combining these values of @xmath9 with the mean value of @xmath1 yields an estimate of @xmath72 km via equation 2 .
the shape of the mosaics should allow us to detect azimuthal variation in anisoplanicity , if it were present , by comparing @xmath13 in the eastern and southern arms .
this second - order anisoplanicity might be expected if the windspeed @xmath22 were sufficiently fast to outpace the bandwidth of the ao system more in one direction than another .
we do not detect a strong azimuthal anisoplanatism here .
the average relative difference in @xmath13 between eastern and southern arms is only about 7% ( an absolute difference in strehl ratio of @xmath73% ) for the first night , which is well within the observational uncertainties .
this difference increased to about 19% ( @xmath74% ) for the second night , again not significantly larger than the observational uncertainty .
let us first compare the data to the expected performance given by equations 1 and 3 .
we will assume @xmath75 , @xmath76 , @xmath77 , @xmath78 s , and that @xmath22 was either the mean value that we recorded on the ground or a factor of 2 faster . to simplify the calculations we approximated the laser beacon by a star of magnitude @xmath79 .
an a - type star of this brightness provides roughly the same number of photons per subaperture as the laser at its typical output of 10 watts .
note that @xmath24 is the total throughput of the entire ao system , and thus here it folds in the efficiency of the lgs return .
a range of @xmath24 should therefore give an optimistic and pessimistic estimate of @xmath13 because we did not record the true brightness of the the laser spot for every frame .
the lgs ao on - axis strehl ratio data as a function of @xmath1 are shown in the upper panel of figure [ figure_theory ] as filled circles . for comparison , the tip - tilt only data are indicated by open circles .
the model prediction assuming @xmath80 is overplotted as a dotted line . to achieve a strehl ratio at or above
this line would require either perfect throughput from the ao system or that the lgs was brighter than 10 watts .
therefore the one noisy data point at @xmath81 m may be an outlier .
all of the remaining data are reasonably well bounded by the optimistic , @xmath82 estimate ( upper solid line ) and pessimistic , @xmath83 estimate ( lower solid line ) of @xmath13 which the model provides .
we can not distinguish between a model with @xmath84 ( solid lines ) and one with @xmath85 ( dashed lines ) .
the lower panel of figure [ figure_theory ] shows the measurements of instrumental isoplanatic angle .
the symbols are the same as the upper panel .
the lgs data appear to be well bounded by models with @xmath86 km line ( lower solid line ) and @xmath87 km line ( upper solid line ) .
the flatness of the @xmath9 distribution may be caused by a difference in @xmath10 between the nights .
this seems plausible because there was a sharp division at @xmath88 m separating the first and second nights .
the dashed line indicates a model with @xmath89 km , which is within the observational uncertainties of most of the tip - tilt data .
we confirm that the instrumental isoplanatic angle for tip - tilt correction was large , about a factor of 2 or 3 larger than for full lgs operation .
it is unclear , however , how important the effect of partial tip - tilt correction is for the lgs results because our simple model implicitly assumes perfect tip - tilt correction
. it may be that we also can not ignore the effect that the finite angular separation of lgs and tip - tilt guide star has on the instrumental isoplanatic angle . only a more sophisticated model - one which includes imperfect
tip - tilt correction - can determine the importance of this effect .
a description of one such model follows .
a computer simulation was developed to characterize the data .
it is similar in principle to the simulation described in paper i , but necessarily somewhat more complex .
here we require the simulation of two ao systems : the high - order ao loop , and the tip - tilt loop .
we used the code for adaptive optics systems ( caos ) software package @xcite , which is available from the observatorio astrofisico di arcetri website . like our previous simulation
, it is written in idl .
we chose caos because it provides an intuitive graphical interface with which a realistic virtual ao system can be assembled quickly .
the user constructs the desired optical configuration - including the appropriate offset between the ngs and lgs , for example - by arranging prebuilt modules in the gui .
system parameters such as detector pixel - scale and efficiency , framerate , and dwell time can be set .
an artificial atmosphere is generated in a manner similar to that described in paper i. the user supplies the desired number of turbulent layers , their heights , windspeeds and fractional contribution to @xmath1 .
the software then automatically determines the appropriate system interaction matrices and constructs the simulation code .
we assembled the simplest possible virtual ao system which still incorporated all of the essential components of the real one . that is , we simulated a system with a 10 watt sodium laser beacon , a square shack - hartmann wfs , a 61 active - element piezostack deformable mirror , and a quad cell driving a tip - tilt mirror .
we assumed that the @xmath90 of the high - order wfs was degraded only by poisson noise and a readnoise of 10 @xmath91 rms .
the total system throughput ( telescope @xmath92 ao optics @xmath92 wfs detector ) could be adjusted between 0.33 and 0.67 .
we assumed that the quad - cell was noiseless .
the simulated dm and tip - tilt mirror were also very simple .
we specified a dm that could perfectly reproduce the shapes of up to the first 14 zernike polynomials .
we operated the dm without time delay , which means that all of the high - order wavefront error was caused purely by wfs noise and fitting error .
the tip - tilt mirror had a framerate limited to 100 hz , which means that all of the low - order wavefront error was due exclusively to finite bandwidth .
the resulting ao system is almost identical to the one which was successful in describing our ngs - mode data in paper i , with the exception that tip - tilt and higher - order correction have been separated .
this separation allows us to adjust the total throughput of the wfs and independently control the bandwidth of the tip - tilt loop .
the simplest possible atmospheric model was used .
we allowed a single turbulence layer at @xmath93 , 5 , or 7 km with @xmath94 or 20 cm moving with a speed of 1.82 m @xmath95 .
the values of @xmath1 chosen were the observed mean values for each night . the goal is to see if a single - layer atmospheric model is a reasonable match to the lgs data , despite the addition of a tip - tilt offset .
here we have treated servo lag in the dm like a fitting error , which is admittedly a very simple treatment of the high - order ao loop .
the reason for this is that the theoretical kernel treats anisoplanatism in the same way - as perfect correction of @xmath6 zernike modes .
if we can achieve a good fit with the simulation this should provide guidance for a value of @xmath6 in equation 16 . because it was not included in our version of caos
, we developed a new module which provides the integrated output psf at any point in the field .
this is essentially the output of the scientific camera .
however , we did not take into account any additional aberrations due to this camera .
therefore , the outputs from our code should be the average values of the psf strehl ratio for the ao system alone , assuming perfect camera optics .
we ran simulations with @xmath96 , 9 , and 14 .
for fixed @xmath6 , raising or lowering @xmath10 has the expected result of decreasing or increasing @xmath9 .
the results for @xmath75 and @xmath72 km are presented in table [ table_computer_modeling ] .
the averages of the 8 and 20 arcsec offset models are in good agreement with the data , especially for the @xmath97 cm model and the second night .
models with @xmath98 are not favoured because to obtain @xmath99 with @xmath96 would require @xmath3 greater than 7 km .
similarily , models with @xmath100 produce isoplanatic angles that are too small , even for @xmath93 .
we would expect to obtain the best correction when the tip - tilt star and the lgs are coincident .
the models confirm this , and reproduce values similar to those predicted with equation 3 .
recall that equation 3 only applies for observations without offset , but we did not obtain any data for this configuration .
even so , we can see that models without offset also plausibly fit the data of the first night , although adding an offset brings @xmath13 into better agreement .
this is also true of the second night , although there the no - offset models predict isoplanatic angles which are too small .
this suggests that although tip - tilt anisoplanatism is a weak effect in our data it is not negligible . in summary ,
the anisoplanatism predicted with the computer model is a reasonably good match to the data .
the best fit models have @xmath101 and @xmath72 km .
this does not preclude variation in @xmath3 from night to night , or even within a night , but does suggest that a single layer of turbulence is a plausible model for both nights .
this model predicts a mean no - offset on - axis strehl ratio of about 0.3 ( @xmath102 ) over both of our nights .
note that if we had observed in this configuration , our on - axis strehl ratio at a tip - tilt offset of between 8 and 20 was just 0.18 .
this illustrates the necessity for good psf characterization for lgs ao observations .
one can not rely on even a fairly nearby star to determine the local psf . according to our model
if one were to use an observation of the tip - tilt star ( by aligning the lgs with it ) to predict the on - axis psf at at tip - tilt to lgs offset of 20 , one would commit an error of at least 67% in strehl ratio .
using it to predict the psf within a radius of 8 from this position would increase the error beyond a factor of two .
we now discuss three basic types of off - axis psf predictions : empirical , semiempirical , and semianalytical .
variations within these catagories provide eight distinct methods .
we have just discussed the simplest empirical method , using the image of the tip - tilt guide star , which gives poor results .
one semiempirical method was dealt with in paper i and applies only to ngs data .
this leaves six lgs methods to be tested .
each carries with it assumptions about the importance of cone effect and tip - tilt anisoplanatism .
table [ table_model_summary ] gives a summary . for each of the methods we generated artificial images for every lgs mosaic pointing .
we did the same to the lick ngs data discussed in paper i , for those methods which are applicable . to provide a closer match to the lgs data we limited ourselves to those ngs data with on - axis strehl ratio greater than 0.05 . for comparison
, we repeated the technique of paper i on these better ngs data .
images were generated as follows . for each pointing a blank 2-dimensional array with the same dimensions as the original field
was created . for each star in the field , an artificial psf with the same flux was generated .
these were then inserted into the blank array at the same position as the original star . the propitious configuration of our lgs mosaics - each laser pointing 8 from the vertex of a trianglular constellation of bright stars , 20 long on each side - provided a means of predicting the psf at two locations in the field .
we simply used the image of the tip - tilt guide star in the first pointing to predict the psf at the location of the bright star in the other two pointings .
this is plausible because all three of the bright stars are the same radius from their respective laser pointings .
thus one can think of this first - order approach as having applied a corrective offset " to account for the high - order anisoplanatism .
it assumes , however , that tip - tilt anisoplanatism is negligible . to improve on the corrective offset method we added the second order effect of tip - tilt anisoplanatism by applying a kernel ( either the gaussian or theoretical one ) to the tip - tilt guide - star image .
these are the tip - tilt kernel " methods .
figure [ figure_kernel_tiptilt_20 ] shows the gaussian and theoretical tip - tilt kernels for offsets of 20 .
the two kernels represent their respective averages of all the tip - tilt mosaic pointings .
notice how similar they are , especially in the core .
the observed instrumental isoplanatic angle for each field was used to determine the appropriate tip - tilt isoplanatic angle , by applying the @xmath103 correction mentioned in section [ observations ] . from this
we can infer @xmath10 from equation 2 , which provides a single - layer model of @xmath5 for equation 18 .
next , we tested the effect of just the high - order aberrations .
this is called the high - order gaussian kernel " method . here
we have attempted to account for tip - tilt anisoplanatism by predicting the psf based on the interpolated strehl ratio at the center of the frame , that is , tip - tilt anisoplanatism should already be included in the prediction of on - axis strehl ratio , and we need only correct for higher - order anisoplanatism . to do this , we first took the interpolated values from equation 1 for each of the off - axis pointings to find the peak strehl ratio at the center of the frame .
we then constructed a psf appropriate for the location of the laser spot , because no star was available at this location .
this psf was estimated by convolving the telescope diffraction pattern with a gaussian kernel assuming @xmath104 ( the ao - corrected core ) , and then adding the remaining flux ( @xmath105 ) in a gaussian of width @xmath106 ( the seeing - limited halo ) .
we then degraded this on - axis psf for off - axis positions by further convolving it with the gaussian kernel , using the observed instrumental isoplanatic angle .
note that the on - axis psf should already incorporate the degradation due to tip - tilt anisoplanatism at the location of the laser spot .
even though the high - order gaussian kernel method requires a model psf , one might still think of it as a semiempirical case because it is based on a fit to strehl ratios in the image .
this would not be the case with the theoretical kernel , for example , which would require a choice for the value @xmath6 in order to predict the on - axis psf .
figure [ figure_kernels_laser_8 ] shows the average gaussian high - order kernel at a radius of 8 from the lgs . for comparison , the theoretical kernel with @xmath75 is also plotted .
the gaussian and theoretical kernels appear to be very similar .
this suggests that a logical next step might be to predict both the on - axis psf and the psf spatial variation analytically . finally , we employed two semianalytical methods .
these approaches are similar to the high - order gaussian kernel case , but here we have modeled the on - axis psf as if there were no tip - tilt guide - star offset . in the case of the gaussian kernel , for each frame we estimated the on - axis strehl ratio from equation 3 assuming an overall throughput ( @xmath102 ) and an estimated lgs brightness ( @xmath79 ) , and then applying the observed fried parameter and windspeed .
the theoretical kernel used the measurement of the fried parameter and @xmath9 , and assumed a value of @xmath6 .
we were guided by the computer simulations to use @xmath75 .
we then degraded this psf due to both tip - tilt and higher - order anisoplanatism with either the gaussian or theoretical kernels .
we label these semianalytical methods , as opposed to semiempirical ones , because here the on - axis strehl ratio is predicted with a theoretical model and wfs measurements , not just from information in the image itself .
in fact , if an independent measurement of @xmath5 were available no information at all would have been required from the image , and we might then call the method a purely analytical one .
no current ao telescope routinely provides an independent @xmath5 profile , but our lick data should at least show if relying on wfs telemetry and a single - layer @xmath5 model can in principle provide useful results .
the fake images were measured in exactly the same manner described in section [ observations ] and compared with the observations .
table [ table_psf_predictions ] lists the mean percent error in the predicted average strehl ratio @xmath107 , fwhm , and ellipticity @xmath108 for the various models at the positions of the three bright stars . in some cases
these predictions are only relevant at the locations of two stars , one each in the two arms of the mosaic . in those cases , to simplify the analysis we present the average of the two results . in all cases
we give the absolute percent error , that is , we assume either an overprediction or underprediction of @xmath107 , fwhm , or @xmath108 is equally undesirable and have thus dropped the sign .
we will deal with the lgs - mode results first , and then compare them with the ngs results at the end .
the corrective offset is an improvement in observational technique , albeit a modest one .
the error in strehl ratio committed by this method was 27% ( i.e. @xmath109 ) .
the error in fwhm was 26% and the error in ellipticity 2% .
the semiempirical approaches improve slightly on the empirical result . applying a tip - tilt gaussian kernel
reduces the errors in strehl ratio and fwhm by a small amount : to 20% and 12% respectively , with no change in ellipticity .
similar results were achieved using the theoretical kernel .
the results of the high - order gaussian method are similar .
the mean error in the predicted strehl ratio is 20% and fwhm 16% .
note that we did not incorporate anisotropy into the gaussian kernel model which is why it has done a poor job of predicting ellipticity .
somewhat better results were achieved , at least for predictions of strehl ratio , by a semianalytical approach .
the gaussian kernels improved the predicted off - axis strehl ratio to an accuracy of 14% , and the theoretical kernels further reduced this to 13% . for a tip - tilt guide star offset as small as 8 , the results are even better ; just 2% .
we also applied the same methods while assuming that the height of turbulence for every observation was the same over both nights ( @xmath72 km ) .
this did not significantly affect the results , only increasing the error in off - axis strehl ratio at large offsets to 16% ( gaussian ) and 15% ( theoretical ) .
this suggests that obtaining a measurement of @xmath10 for every observation , although helpful , is not necessary .
useful improvement in psf characterization may be achieved by obtaining perhaps just one star field mosaic during a night .
it should be noted that neither of the semianalytical models benefit from an empirical image of the psf .
this empirical information would be the only way to incorporate non - common path errors or camera aberrations into the psf . without
it the halo is likely to be underpredicted , which is probably why these methods do a better job than the empirical ones at predicting strehl ratio , but do no better in predicting fwhm .
this effect can be seen in figure [ figure_results ] . here
we show the data , an empirical model which uses an image of the tip - tilt guide star without any corrective kernel , the semianalytical gaussian model and semianalytical theoretical model , and their respective residuals .
note the diffuse haloes remaining in the residuals of the semianalytical models .
these are less apparent in the purely empirical treatment , although in that case variation in the core has not been accounted for , which is why the semianalytical results are better overall . where they were applicable we tried all of the new techniques on our lick ngs data , and redid the analysis of paper i. note that the percent error in fwhm which we quote in table [ table_psf_predictions ] ( 14% ) for the original semiempirical technique is somewhat better than that which we gave in paper i ( 19% ) .
this is because in our reanalysis we used only the 5 ngs mosaics with the highest on - axis strehl ratios to better compare to the higher strehl ratio lgs data . for large offsets ( @xmath110 )
we obtained similar , if somewhat better results with all the new methods .
note that all of our data include a corrective offset , which made the analysis more straightforward .
an ao observer need not apply this offset .
conceivably , the only impact would be the addition of an extra step to the first two semiempirical approaches . after applying the tip - tilt kernel to the ngs psf
, a high - order kernel would also be applied to account for the lgs ao field variation .
the other methods would be unchanged .
we have discussed observations of a dense star field for the purpose of characterizing the variation in the lgs ao psf at lick observatory .
the results of our simulations confirm the expectation that the field dependence of the psf is complicated , and that the effect of tip - tilt anisoplanatism can not be ignored .
however , a simple atmospheric model with a thin layer of turbulence is still a useful model for our data .
a dominant layer at @xmath05 km above the telescope reproduces the anisoplanicity seen in the data .
one simple empirical way to characterize lgs data in the absence of a convenient psf star would be to use the psf of the ngs obtained with the laser pointed directly at the tip - tilt star .
this might match the wings well but would underestimate the width of the core badly for offsets of 8 or more . according to our simulations , this would have introduced an error greater than 67% in strehl ratio ( i.e. @xmath109 ) for our data .
a potential improvement in observational technique would be to offset the tip - tilt guide star by the same amount ( and in the opposite direction ) as the scientific observations .
this somewhat improves the accuracy in the psf prediction , reducing the error in strehl ratio in our data 27% .
we advocate that lgs ao observers obtain at least one star - field mosaic each night using the same offsets as their target observations , and repeat this with high - order correction turned off . just as in ngs - mode observations ,
the mosaicked images must be completed quickly before variations in delivered strehl ratio ruin the measurements . by using just the information in the tip - tilt - only mosaic and applying a corrective kernel to the empirical psfs at the tip - tilt guide star position , strehl ratio uncertainty
was reduced to 20% ( tip - tilt gaussian kernel ) and 18% ( tip - tilt theoretical kernel ) for our data .
another semiempirical method of comparable accuracy ( strehl ratio to within 20% ) is possible by instead using information from the high - order lgs ao mosaic .
all of these semiempirical techniques inherently account for cone effect and should therefore be applicable for ao on 10-meter apertures .
results with the semianalytical methods were more comparable to ngs ao .
errors in prediction of either strehl ratio were better than 14% , and even as good as a few percent if the offset is small . however , their application is complicated by the intertwined effects of @xmath1 , @xmath10 , @xmath22 , @xmath24 , and @xmath6 .
better results could be realized by separating these effects through independent measurement of @xmath5 .
we do not find that the significant computational effort in applying the theoretical kernels was worthwhile for our data because the prediction of fwhm with the gaussian kernels is equally good .
we would like to thank the staff of lick observatory for making these observations possible .
we thank summer student q. konopacky for her assistance with the observing log .
we also thank t. fusco and g. molodij for their help and original notes concerning the zernike polynomial autocorrelations .
this work was supported by the national science foundation science and technology center for adaptive optics , managed by the university of california at santa cruz under cooperative agreement no .
ast-9876783 .
carbillet , m. , fini , l. , femenia , b. , riccardi , a. , esposito , s. , viard , e. , delplancke , f. , hubin , n. 2001 , asp conf .
series 238 , 249 bauman , b.j . ,
gavel , d.t . , waltjen , k.e . , freeze , g.j .
, keahi , k.a . ,
kuklo , t.c . ,
lopes , s.k . , newman , m.j . , & olivier , s.s .
1999 , , 3762 , 194 bauman , b.j . , gavel , d.t .
, waltjen , k.e . , freeze , g.j . ,
hurd , r.l . , gates , e.l . ,
max , c.e . , olivier , s.s . , & pennington , d.m .
2002 , , 4494 , 19 chassat f. , rousset , g. & primot j. 1989 , , 1114 , 14 fini , l. , carbillet , m. , & riccardi , a. 2001 , asp conf .
series 238 , 253 fusco , t. , conan , j .-
m . , mugnier , l.m . ,
michau , v. , & rousset , g. 2000 , , 142 , 149 gavel , d.t . , olivier , s.s . ,
bauman , b.j . ,
max , c.e .
, & macintosh , b.a 2000 , , 4007 , 63 lloyd , j.p . ,
liu , m.c . ,
macintosh , b.a . ,
severson , s.a .
, deich , w.t . , & graham , j.r .
2000 , , 4008 , 814 molodij , g. & rousset , g. 1997 , josa a , 14(8 ) , 1949 molodij , g. & rayrole , j. 1998 , , 128 , 229 noll , r.j .
1976 , josa , 66 , 3 , 207 olivier , s.s . , gavel , d.t . , friedman , h.w.,max , c.e . , an , j.r . ,
avicola , k. , bauman , b.j . ,
brase , j.m . ,
campbell , e.w . ,
carrano , c.j . ,
cooke , j.b . ,
freeze , g.j . , gates , e.l . ,
kanz , v.k . ,
kuklo , t.c . ,
macintosh , b.a . ,
newman , m.j . ,
pierce , e.l . ,
waltjen , k.e . & watson , j.a . 1999 , , 3762 , 2 roddier , f. 1999 , in adaptive optics in astronomy , cambridge university press , new york , ed
. f. roddier steinbring , e. , faber , s.m . ,
hinkley , s. , macintosh , b.a . , gavel , d. , gates , e.l . ,
christou , j.c .
, le louarn , m. , raschke , l.m . , severson , s.a .
, rigaut , f. , crampton , d. , lloyd , j.p . , graham , j.r .
2002 , , 114 , 1267 ( paper i ) valley , g.c . &
wandzura , s.m .
1979 , josa , 69 , 5 vran , j .-
, rigaut , f. , matre , h. , & rouan , d. 1997 , josa a , 14(11 ) , 3057 lcccccccccccc 1 & aug 16 & 08:19 & 1.11 & 18.6 & @xmath111 & @xmath112 & & & & & & + 2 & & 08:27 & 1.11 & 19.3 & @xmath113 & @xmath114 & & @xmath115 & @xmath116 & & & + 3 & & 08:42 & 1.12 & 19.5 & @xmath117 & @xmath118 & & @xmath119 & @xmath120 & & & + 4 & & 08:51 & 1.13 & 18.6 & & & & & & & @xmath121 & @xmath122 + 5 & & 09:04 & 1.14 & 17.2 & @xmath123 & @xmath124 & & @xmath125 & @xmath120 & & & + 6 & & 09:21 & 1.16 & 18.0 & @xmath126 & @xmath127 & & @xmath128 & @xmath127 & & & + 7 & & 09:30 & 1.19 & 19.0 & & & & & & & @xmath129 & @xmath130 + 8 & & 09:45 & 1.22 & 21.2 & @xmath131 & @xmath132 & & @xmath133 & @xmath134 & & & + 9 & & 09:57 & 1.25 & 22.2 & @xmath135 & @xmath136 & & @xmath137 & @xmath138 & & & + 10 & & 10:09 & 1.27 & 21.8 & @xmath139 & @xmath118 & & @xmath140 & @xmath141 & & & + 11 & & 10:24 & 1.33 & 21.1 & & & & & & & @xmath142 & @xmath143 + 12 & aug 17 & 07:50 & 1.10 & 13.3 & @xmath144 & @xmath145 & & @xmath146 & @xmath147 & & & + 13 & & 08:10 & 1.11 & 12.0 & @xmath129 & @xmath148 & & @xmath149 & @xmath150 & & & + 14 & & 08:20 & 1.11 & 13.1 & @xmath144 & @xmath134 & & @xmath151 & @xmath112 & & & + 15 & & 08:27 & 1.11 & 14.3 & & & & & & & @xmath152 & @xmath153 + 16 & & 08:34 & 1.12 & 14.3 & @xmath154 & @xmath155 & & @xmath156 & @xmath127 & & & + 17 & & 08:50 & 1.13 & 11.4 & @xmath154 & @xmath157 & & @xmath129 & @xmath158 & & & + 18 & & 09:02 & 1.15 & 9.8 & @xmath152 & @xmath159 & & @xmath160 & @xmath150 & & & + 19 & & 09:09 & 1.16 & 9.8 & & & & & & & @xmath161 & @xmath162 + 20 & & 09:25 & 1.18 & 10.1 & & & & @xmath163 & @xmath164 & & & + 21 & & 09:33 & 1.20 & 10.1 & & & & @xmath165 & @xmath166 & & & + 22 & & 09:42 & 1.22 & 10.4 & @xmath167 & @xmath168 & & @xmath169 & @xmath134 & & & + 23 & & 09:49 & 1.24 & 10.9 & & & & & & & @xmath170 & @xmath130 + 24 & & 10:01 & 1.27 & 11.9 & @xmath171 & @xmath172 & & @xmath173 & @xmath158 & & & + 25 & & 10:11 & 1.30 & 14.7 & @xmath174 & @xmath175 & & @xmath115 & @xmath141 & & & + lcccccccc @xmath13 & @xmath176 & @xmath113 & & 0.28 & 0.40 & & 0.12 & 0.18 + @xmath9 ( arcsec ) & @xmath177 & @xmath172 & & 16 & 14 & & 12 & 9 + @xmath13 & & & & 0.32 & 0.41 & & 0.13 & 0.22 + @xmath9 ( arcsec ) & & & & 13 & 12 & & 9 & 7 + llccc empirical & & & & + no correction & uses image of the ngs to predict off - axis psf ; not tested here & yes & no &
no + corrective offset & same as above , but with ngs shifted to correct for target offset & yes & no & yes + semiempirical & & & & + deconvolved kernel & kernel deconvolved from mosaic ; applies only to ngs mode & & & + tip - tilt gaussian kernel & improves on corrective offset by applying a tip - tilt kernel & yes & yes & yes + tip - tilt theoretical kernel & same as above , but using a theoretical kernel & yes & yes & yes + high - order gaussian kernel & on - axis psf given by fit to mosaic ; off - axis kernel is gaussian & yes & yes & yes + semianalytical & & & & + both gaussian kernels & predicts ngs psf from telemetry ; kernels are gaussian & no & yes & yes + both theoretical kernels & same as above , but using theoretical kernels & no & yes & yes + lccccccccccc empirical & & & & & & & & & & & + no correction & 42 & 29 & 2 & & & & & & & & + corrective offset & & & & & 27 & 26 & 2 & & & & + semiempirical & & & & & & & & & & & + deconvolved kernel & 14 & 14 & 2 & & & & & & & & + tip - tilt gaussian kernel & 13 & 5 & 1 & & 20 & 12 & 2 & & & & + tip - tilt theoretical kernel & 10 & 4 & 2 & & 18 & 16 & 2 & & & & + high - order gaussian kernel & & & & & 20 & 16 & 13 & & 4 & 3 & 16 + semianalytical & & & & & & & & & & & + both gaussian kernels & 9 & 14 & 10 & & 14 & 16 & 13 & & 5 & 1 & 16 + both theoretical kernels & 8 & 4 & 7 & & 13 & 18 & 6 & & 2 & 2 & 1 + | most current astronomical adaptive optics ( ao ) systems rely on the availability of a bright star to measure the distortion of the incoming wavefront .
replacing the guide star with an artificial laser beacon alleviates this dependency on bright stars and therefore increases sky coverage , but it does not eliminate another serious problem for ao observations . this is the issue of psf variation with time and field position near the guide star .
in fact , because a natural guide star is still necessary for correction of the low - order phase error , characterization of laser guide star ( lgs ) ao psf spatial variation is more complicated than for a natural guide star alone .
we discuss six methods for characterizing lgs ao psf variation that can potentially improve the determination of the psf away from the laser spot , that is , off - axis .
calibration images of dense star fields are used to determine the change in psf variation with field position .
this is augmented by ao system telemetry and simple computer simulations to determine a more accurate off - axis psf .
we report on tests of the methods using the laser ao system on the lick observatory shane telescope .
we observed with offsets typical of separations between dim science targets and the nearest suitably bright tip - tilt guide star , up to 20 .
if the tip - tilt guide star is used as the psf reference , the predicted strehl ratio within an 8 radius of the lgs can be more than a factor of two too high , due to not taking into account the field dependent anisoplanatism .
a better result may be obtained by an improved empirical approach : repositioning the tip - tilt guide star to compensate for the additional 8 offset .
this would result in a 27% relative error in strehl ratio .
a simple gaussian model of psf variation can reduce the error in the prediction to @xmath020% .
some further improvement may by obtained semianalytically , by using telemetry from the ao system ( 14% error ) plus a more sophisticated theoretical model of psf variation ( 13% error ) .
it should be noted , however , that all but the first method achieve comparable accuracy in predicting psf full - width at half maximum .
this is important because the last method , unlike the others , is not directly applicable to 10-meter apertures without further computational complexity . |
the standard model of particle physics is believed to be an effective theory valid up to energies close to 1 tev .
however , no new physics beyond the standard model ( sm ) has been observed yet , and it is critical to explore a wide range of possible signatures .
a promising avenue lies in final states that involve the heaviest of the particles presumed to be elementary : the top quark .
this document describes two searches for new phenomena involving top quarks using the atlas detector @xcite at the cern large hadron collider ( lhc ) .
the first search is carried out for a pair - produced exotic top partner ( @xmath1 ) , decaying to a top - antitop pair and two stable , neutral weakly - interacting particles ( @xmath2 , which in some models may be its own anti - particle ) @xcite . in most models , the @xmath1 has typical quark - like quantum numbers , and is produced through @xmath3 annihilation and gluon fusion .
the final state for such a process ( @xmath4 ) is identical to @xmath5 , though with a larger amount of missing transverse momentum ( ) from the undetected @xmath2 s . in supersymmetry models with @xmath6-parity conservation @xmath1 refers to the stop squark and @xmath2 refers to the lightest supersymmetric particle .
the @xmath7 @xcite signature appears in a general set of dark matter - motivated models , as well as in other sm extensions , such as the mentioned supersymmetry models , little higgs models with @xmath1-parity conservation @xcite , models of universal extra dimensions ( ued ) with kaluza - klein - parity @xcite , models in which baryon and lepton number conservation arises from gauge symmetries @xcite or models with third generation scalar leptoquarks @xcite .
many of these models provide a mechanism for electroweak symmetry breaking and predict dark matter candidates , which can be identified indirectly through their large signature . the search is carried out in the @xmath8 lepton+jets channel where one @xmath9 boson from the top quark decay decays leptonically ( including @xmath10 decays to @xmath11 or @xmath12 ) and the other @xmath9 boson decays hadronically , resulting in a final state with an isolated lepton of high transverse momentum , four or more jets and large .
a second search is carried out for new heavy particles decaying to @xmath8 pairs @xcite . using the lepton+jets
channel the reconstructed @xmath8 mass spectrum , based on three or four jets , an electron or muon , and a neutrino , is used to search for a signal .
previous searches for @xmath8 were most recently carried out by the cdf @xcite and d0 @xcite collaborations at run ii of the fermilab tevatron collider , and the atlas @xcite and cms @xcite collaborations at the cern lhc .
no evidence for new particles was uncovered and 95% confidence level ( c.l . )
limits were set on the mass of a leptophobic topcolour @xmath13 boson @xcite at @xmath14 @xcite gev as well as on the coupling strength of a heavy colour - octet vector particle .
this analysis is very similar to the previous atlas analysis @xcite but uses data collected in 2011 .
the benchmark model used to quantify the experimental sensitivity to narrow resonances is a topcolour boson @xcite arising in models of strong electroweak symmetry breaking through top quark condensation @xcite .
the specific model used is the leptophobic scenario , model * iv * in ref .
@xcite with @xmath15 and @xmath16 and a width of 1.2% of the @xmath17 boson mass .
this is the same set of parameters that was used by the d0 collaboration @xcite .
the model used for wide resonances is a kaluza - klein gluon @xmath18 , which appears in randall - sundrum ( rs ) models with a warped extra dimension in which particles are located in the extra dimension @xcite .
the concrete model used is described in detail in ref .
@xcite and implemented in the madgraph @xcite event generator .
the couplings to quarks take the `` standard '' rs values @xcite : @xmath19 for light quarks including charm , @xmath20 for bottom quarks and @xmath21 for the top quark . in this case ,
the resonance is predicted to be significantly wider than the detector and reconstruction algorithm s resolution .
this model is taken as a proxy for coloured resonances .
the atlas detector @xcite consists of an inner detector tracking system ( i d ) surrounded by a superconducting solenoid providing a @xmath22 t magnetic field , electromagnetic and hadronic calorimeters , and a muon spectrometer ( ms ) .
the i d consists of pixel and silicon microstrip detectors inside a transition radiation tracker ( trt ) which provide tracking in the region @xmath23 .
the electromagnetic calorimeter is a lead / liquid - argon ( lar ) detector in the barrel ( @xmath24 ) is measured around the beam axis and the polar angle @xmath25 is the angle from the beam axis .
the pseudorapidity is defined as @xmath26 . ] and the endcap ( @xmath27 ) regions .
hadron calorimetry is based on two different detector technologies .
the barrel ( @xmath28 ) and extended barrel ( @xmath29 ) are lar / copper .
the forward calorimeters ( @xmath30 ) are instrumented with lar / copper and lar / tungsten , providing electromagnetic and hadronic energy measurements , respectively .
the ms consists of three large superconducting toroids with 24 coils , a system of trigger chambers , and precision tracking chambers which provide muon momentum measurements out to @xmath31 of 2.7 .
the search for a pair - produced exotic top partner is based on data recorded by the atlas detector in 2010 using 35 pb@xmath32 of integrated luminosity .
the search for @xmath8 resonances is based on data recorded in 2011 using 200 pb@xmath32 .
the data were collected using electron and muon triggers
. requirements of good beam conditions , detector performance and data quality are imposed .
monte carlo ( mc ) event samples with full atlas detector simulation @xcite based on the geant4 program @xcite and corrected for all known detector effects are used to model the signal process and most of the backgrounds .
the qcd multi - jet background is modeled using data control samples rather than the simulation .
the @xmath5 and single top samples are produced with mc@nlo @xcite , while the @xmath9+jets and @xmath33+jets samples are generated with alpgen @xcite .
herwig @xcite is used to simulate the parton shower and fragmentation , and jimmy @xcite is used for the underlying event simulation .
the diboson background is simulated using herwig .
the inclusive @xmath9+jets and @xmath33+jets cross sections are normalized to nnlo predictions @xcite , and the cross sections of the other backgrounds are normalized to nlo predictions @xcite . madgraph @xcite is used to simulate the signal process , and pythia @xcite is used to simulate the parton shower and fragmentation .
a grid of @xmath1 and @xmath2 masses is generated with 250 gev @xmath34 350 gev and 10 gev @xmath35 100 gev . in this search , only on - shell tops are considered , and therefore grid points where the mass difference between the @xmath1 and @xmath2 approaches the top quark mass are excluded .
each sample is normalized to the cross section calculated at approximate nnlo in qcd using hathor @xcite , ranging from 20.5 pb for a @xmath1 mass of 250 gev to 3.0 pb for a @xmath1 mass of 350 gev .
signal samples for bosons decaying to @xmath36 are generated using pythia allowing all top quark decay modes .
samples of rs gluons are generated with madgraph , and showered with pythia .
electron and muon candidates are selected as for other recent top quark studies using the lepton+jets signature @xcite .
jets are reconstructed using the anti-@xmath37 @xcite algorithm with a distance parameter @xmath6 , defined as @xmath38 , of 0.4 . to take into account the differences in calorimeter response to electrons and hadrons , a @xmath39- and @xmath40-dependent factor , derived from simulated events and validated with data ,
is applied to each jet to provide an average energy scale correction @xcite back to particle level . in the calorimeter ,
the energy deposited by particles is reconstructed in three - dimensional clusters .
the energy of these clusters is summed vectorially , and the projection of this sum in the transverse plane corresponds to the negative of the @xcite .
corrections to the hadronic and electromagnetic energy scales , dead material and out - of - cluster energy are applied , and in the case of reconstructed muons , an additional correction is included .
events are selected with exactly one electron or muon , where each lepton is required to pass the following selection criteria .
electrons are required to satisfy @xmath41 gev ( @xmath42 gev for the resonance search due to tighter trigger requirements in 2011 ) and @xmath43 .
electrons in the transition region between the barrel and the endcap electromagnetic calorimeters ( @xmath44 ) are removed .
muon candidates are required to satisfy @xmath45 gev and @xmath23 .
events with four or more reconstructed jets with @xmath45 gev ( @xmath46 gev for the resonance search ) and @xmath47 are selected . to reduce the @xmath9+jets background , events are required to have @xmath48 gev and @xmath49 @xmath50 gev , where @xmath49 is the transverse mass of the lepton and missing energy , where @xmath51 is the ( ) of the muon ( electron ) and @xmath52 ( @xmath53 ) is the azimuthal angle of the lepton ( ) . ] .
events with either a second lepton candidate satisfying looser selection criteria or an isolated track with @xmath54 gev are rejected in order to reduce the contribution from @xmath5 dilepton events .
in particular the isolated track veto is useful for eliminating single - prong hadronic @xmath10 decays in @xmath5 dilepton events . in the electron channel , must be larger than 35 gev and @xmath55 gev . in the muon channel , @xmath56 gev and + @xmath57 gev is required .
at least one of the selected jets must be tagged as a @xmath58-jet .
jets originating from @xmath58-quarks are selected by exploiting the long lifetimes of @xmath59-hadrons ( about 1.5 ps ) leading to typical flight paths of a few millimeters , which are observable in the detector .
the sv0 @xmath58-tagging algorithm @xcite used in this analysis explicitly reconstructs a displaced vertex from the decay products of the long - lived @xmath59-hadron .
two - track vertices at a radius consistent with the radius of one of the three pixel detector layers are removed , as these vertices likely originate from material interactions .
a jet is considered @xmath58-tagged if it contains a secondary vertex , reconstructed with the sv0 tagging algorithm , with @xmath60 , where @xmath61 is the decay length and @xmath62 its uncertainty .
this operating point yields a 50% @xmath58-tagging efficiency in simulated events .
the sign of @xmath63 is given by the sign of the projection of the decay length vector on the jet axis .
a summary of the background estimates and a comparison with the observed number of selected events passing all selection criteria is shown in table [ t : bg ] . a total yield of 17.2 @xmath64 2.6 events
is expected from sm sources , and 17 events are observed in data .
the background composition is similar in the electron and muon channels .
.summary of expected sm yields including statistical and systematic uncertainties compared with the observed number of events in the signal region .
[ cols=">,^,^,^,^,^,^,^,^",options="header " , ] [ t : bg ] the dominant background arises from @xmath5 dilepton final states , in which one of the leptons is not reconstructed , is outside the detector acceptance , or is a @xmath10 lepton . in all such cases ,
the @xmath5 decay products include two high- neutrinos , resulting in large and @xmath49 tails . in mc ,
the second lepton veto removes 45% of the dilepton @xmath5 and 10% of the single - lepton @xmath5 in the signal region .
the veto performance is validated in the data in several control regions both enhanced and depleted in dilepton @xmath5 , and in all cases the veto efficiencies in mc and data agree within 10% .
the next largest backgrounds come from single - lepton sources , including both @xmath9+jets and @xmath5 with one leptonic @xmath9 decay .
both the normalization and the shape of the @xmath49 distribution for this combined background are extracted from the data .
first , the yield of the single lepton background estimated from simulation is normalized in the control region 60 gev @xmath65 @xmath49 @xmath65 90 gev to the data .
next , the shape of the @xmath49 distribution in mc is compared with data in various control regions , where events satisfy the signal event selection but have fewer than four jets .
further , events with identified @xmath58-jets , based on lifetime @xmath58-tagging @xcite , are rejected in order to reduce the dilepton background , such that the samples are dominated by @xmath9+jets events ; the corresponding loss of single - lepton from this @xmath58-jet veto is accounted for in the systematic uncertainties .
good agreement is observed between data and mc and based on this agreement an uncertainty of 15% is assigned on this background . in general
, qcd multi - jet events do not produce large and therefore fail the kinematic requirements for the signal region .
although the qcd multi - jet background is expected to be small , it is difficult to model with simulation .
therefore , data - driven techniques similar to those described in @xcite are used to estimate this background . in both lepton channels the contribution to the signal region is consistent with zero .
other electroweak processes , single top , diboson production ( @xmath66 , @xmath67 , and @xmath68 ) , and @xmath33+jets are estimated using mc simulation , normalized to the theoretical cross section and total integrated luminosity .
they have small production cross sections compared to @xmath5 and @xmath9+jets , and are further suppressed by the multiple - jet , and @xmath49 selection criteria .
the following data - driven relative scale factors are applied to the simulated @xmath9+light parton multiplicity samples : 0-parton exclusive sample : 0.978 @xmath64 0.004 ; 1-parton exclusive sample : 1.107 @xmath64 0.015 ; 2-parton exclusive sample : 1.147 @xmath64 0.047 ; 3-parton exclusive sample : 0.86 @xmath64 0.16 ; 4-parton exclusive sample : 1.63 @xmath64 0.44 ; 5-parton inclusive sample : 0.95 @xmath64 0.58 .
these are determined by fitting the observed jet multiplicity distribution in a @xmath9+jets - dominated data sample selected by : exactly one lepton with @xmath69 20 gev , veto on the presence of any other lepton with @xmath69 10 gev , 30 @xmath65 @xmath65 80 gev , 40 @xmath70 80 gev , and rejection of events in which at least one of the four hardest jets is @xmath58-tagged .
the latter cut ensures that the sample is orthogonal to the signal selection .
the sample is expected to contain over 95% @xmath9+jets events , and remaining contributions from other standard model processes are subtracted using mc predictions prior to the fit .
note that the resulting scale factors are heavily anticorrelated , which is taken into account in the evaluation of systematic uncertainties .
the uncertainties in the scale factors arise from the limited size of the data sample .
this background s normalization uncertainty in the signal sample is 35% , driven by the uncertainty in the relative event tagging probabilities between events with two and four or more jets .
the qcd multi - jet background is estimated in the same way as described in section [ bg1 ] .
the and the other electroweak backgrounds are estimated using mc simulation , normalized to the theoretical cross section and total integrated luminosity .
due to the small size of the data sample , and in order to preserve as much model - independence as possible , a simple cut - and - count analysis is carried out in the high tails of the @xmath9 transverse mass and distributions .
figure [ f : basickin ] shows that the @xmath9 transverse mass and tails are well modeled by sm contributions in two different control samples .
the dilepton veto carries an uncertainty of 15% , determined in control regions in data described in section [ bg1 ] .
the single lepton shape correction , @xmath71 , was determined to be consistent with unity , but carries an uncertainty of 15% from the spread in control regions .
the single lepton normalization is taken from the @xmath9 mass range ( 60 - 90 gev ) just outside the signal region .
this normalization has an uncertainty of 10% , which is mostly the statistical uncertainty , but also includes the effect of varying the mass range used .
the jet energy scale and resolution have some uncertainty @xcite , which is propagated through the analysis by correcting the jet multiplicity , @xmath72 , and .
the per - jet uncertainty ranges from more than 10% at low @xmath73 to roughly 5% at high @xmath73 .
the monte carlo is adjusted for measured differences in lepton i d and trigger efficiencies , and these corrections each contribute some uncertainty , typically 4 - 5% per lepton .
the integrated luminosity has a relative uncertainty of 3.4% . comparing with the data , good agreement with the combined background prediction
is observed : 17.2 background events are expected , and 17 events are observed .
figure [ f : dataplots ] shows some distributions in the signal region .
no evidence of an excess or of a mis - modeled background is observed .
given a theoretical cross section of 12 pb , an additional 12.4 @xmath64 3.1 signal events are expected from the 275 gev @xmath1 ( 50 gev @xmath2 ) mass point .
similarly , the 300 gev @xmath1 ( 10 gev @xmath2 ) model would predict 11.7 @xmath64 3.0 extra events from a cross section of 7.3 pb @xcite .
the event yield distribution is studied from pseudo - experiments , assuming gaussian systematics and including correlations in signal and background systematics ( _ e.g. _ , luminosity , theoretical cross sections ) , for both the signal and background - only hypotheses , and it is determined that both of these models can be excluded with confidence greater than 95% .
the samples with @xmath74 ( @xmath75 ) of 250 ( 10 ) gev and 275 ( 10 ) gev are excluded as well .
implicitly , all other mass points with a lighter @xmath2 are also excluded .
to reconstruct the @xmath36 mass , the neutrino s longitudinal momentum ( @xmath76 ) is determined by imposing the @xmath9-boson mass constraint .
if the discriminant of the quadratic equation is negative , the missing transverse energy is adjusted to get a null discriminant @xcite .
if there are two solutions , the smallest @xmath76 solution is chosen .
the dominant source of long , non - gaussian tails in the mass resolution is the use of a jet from initial- or final - state radiation in the place of one of the jets directly related to a top quark decay product . to reduce this contribution
, the @xmath77 algorithm @xcite considers the four leading jets with @xmath69 20 gev and @xmath78 2.5 , and excludes a jet if its angular distance to the lepton or closest jet satisfies @xmath79 , where @xmath80 is the jet s mass .
( if more than one jet satisfies this condition , the jet with the largest @xmath77 is excluded . ) if a jet was discarded and more than three jets remain , the procedure is iterated . then @xmath81 is reconstructed from the lepton , and the leading four jets , or three jets if only three remain .
the @xmath82 cut removes jets that are `` far '' from the rest of the activity in the event .
furthermore , by only requiring three jets in the mass reconstruction , the method allows one of the jets from top quark decay to be outside the detector acceptance , or merged with another jet .
the reconstructed invariant masses and corresponding resolutions obtained with the @xmath77 algorithm are shown for three different simulated @xmath13 boson masses in fig .
[ fig : drres ] .
systematic uncertainties that only affect the normalization of the different backgrounds come from the uncertainty on the integrated luminosity ( 4.5% ) , background normalizations ( @xmath36 : @xmath83% @xcite , single top : 10% , @xmath9+jets : 35% , diboson : 5% , qcd multi - jet(e ) : 30% , qcd multi - jet(@xmath12 ) : 50% ) , and lepton trigger and reconstruction efficiencies ( @xmath84 1.5% ) .
the dominant shape uncertainties arise from the @xmath58-tagging efficiency ( 11% variation in the event yields ) , jet energy scale including pileup effects ( 9% ) @xcite , and modelling of initial and final state radiation ( 7% ) .
the first two have been determined from data by comparing results from different methods and/or data samples , while the latter has been estimated from mc simulations in which the relevant parameters were varied .
other uncertainties arising from mc modelling as well as lepton identification and momentum measurements have a substantially smaller impact .
the results of this search are obtained by comparing the top quark pair invariant mass ( @xmath81 ) distribution with background - only and signal - plus - background hypotheses . in practice ,
the search is done in two steps : in a first step the data is compared to the standard model prediction , i.e. the null hypothesis , using the bumphunter @xcite algorithm . since no excess is found , in a second step a limit is set using a bayesian approach @xcite on the maximum allowed cross - section times branching ratio for new physics as a function of @xmath81 . in this limit - setting step ,
40 gev - wide bins are used , a value close to the mass resolution and limiting bin - by - bin statistical fluctuations .
a single bin contains all events with @xmath85 2.96 tev .
the observed limits on narrow and wide resonances using the @xmath77 mass reconstruction method are shown in fig .
[ fig : dr - sensitivity ] together with the predicted cross - section times branching ratio for the models considered and the expected sensitivity of the analysis .
the observed ( expected ) limit on @xmath86 br(@xmath87 ) ranges from 38 ( 20 ) pb at @xmath88 gev to 3.2 ( 2.2 ) pb at @xmath89 gev .
while narrow resonances with production cross - sections predicted by the leptophobic topcolour model can not be excluded , the analysis is already able to probe the few picobarn range for masses close to 1 tev .
the observed ( expected ) limit on @xmath90 br(@xmath91 ) ranges from 32 ( 24 ) pb at @xmath92 gev to 6.6 ( 2.9 ) pb at @xmath93 gev , which excludes @xmath18 resonances with mass below 650 gev at 95% c.l .
two searches for new phenomena involving top quarks are presented : a search for a top partner in events with large missing transverse momentum , and a search for resonances in 35 and 200 of data , respectively .
no evidence for a signal is observed and 95% c.l .
limits are set on benchmark models .
both analyses are currently being updated with more luminosity @xcite .
n. arkani - hamed , a. g. cohen and h. georgi , phys .
b * 513 * , 232 ( 2001 ) ; h. c. cheng and i. low , j. high energy phys . * 0309 * , 051 ( 2003 ) ; h. c. cheng and i. low , j. high energy phys . *
0408 * , 061 ( 2004 ) ; h. c. cheng , i. low and l. t. wang , phys .
d * 74 * , 055001 ( 2006 ) . | two searches for new phenomena involving top quarks are presented : a search for a top partner in events with large missing transverse momentum , and a search for resonances in proton - proton collisions at a center - of - mass energy of 7 tev .
the measurements are based on 35 and 200 of data collected with the atlas detector at the lhc in 2010 and 2011 , respectively .
no evidence for a signal is observed .
the first limits from the lhc are established on the mass of a top partner , excluding a mass of 275 gev for a neutral particle mass less than 50 gev and a mass of 300 gev for a neutral particle mass less than 10 gev . using the reconstructed mass spectrum ,
limits are set on the production cross - section times branching ratio to for narrow and wide resonances . for narrow models , the observed 95% c.l .
limits range from approximately 38 pb to 3.2 pb for masses going from @xmath0 500 gev to @xmath0 1300 gev . in randall - sundrum models , kaluza - klein gluons with masses below 650 gev
are excluded at 95% c.l . |
quasars located at high redshifts provide excellent probes of the intervening universe .
as their light travels through the universe , it is partially absorbed by neutral hydrogen or metals such as magnesium an effect detectable in the quasars optical and uv spectra .
these absorption lines provide an effective way of probing the evolution of metal - bearing gas over a large range of redshifts , as the sensitivity to absorption is largely independent of redshift , depending only on the background source s surface brightness .
the absorption systems in particular are most likely associated with foreground galaxies @xcite , based on the velocity structure within the systems , the clustering properties of the absorbers , and the presence of metals that have most likely been produced locally .
when identified , most of these galaxies appear to be spirals ( e.g. * ? ? ?
* ) , although absorbers have been found that span the range from late - type spirals to galaxies resembling present - day ellipticals @xcite .
the observational challenge then is to correctly identify the absorbing galaxies and build up a picture of the absorbing system .
such identifications allow one to track the evolution of metal - bearing gas and its relationship to galactic environment .
the identification of the object responsible for an absorption system involves finding a galaxy close to the line - of - sight to the quasar that has a redshift matching that of the absorption system .
as such galaxies are often faint , to examine many in the vicinity of a given quasar requires relatively large amounts of telescope time , or a large degree of multiplexing .
the advent of multi - object spectrographs on large telescopes has enabled more systems and galaxies to be examined , greatly assisting the identification of absorbing systems .
the identification of nearby galaxies close on the sky to a high - redshift quasar also allows us to address the question of gravitational lensing .
the high redshifts of these quasars on face value imply very large luminosities , placing them at the high end of the luminosity function , where the slope of the function is steepest .
if a small amount of magnification due to gravitational lensing is taking place , the intrinsic luminosity of the quasar will be reduced , changing the shape of the luminosity function .
this is particularly important for the highest redshift quasars ( such as those at @xmath2 , e.g. @xcite ) , but also for those around @xmath3 .
in this paper we address these issues by targeting pks 2126@xmath0158 , a luminous high - redshift quasar that exhibits a number of metal - line absorption systems and has a large number of galaxies close to its line of sight , using the multiplexing capabilities of gmos on the gemini south telescope .
we use imaging and low - resolution spectroscopy to obtain redshifts for many of the galaxies in the field , focussing on the @xmath4 environment .
we describe this quasar and our observations of it and the surrounding field in section [ sec - obs ] , while the results of the gemini / gmos observations are presented in section [ sec - results ] .
the implications of these results for the absorption line systems seen in the quasar s spectrum are discussed in section [ sec - absn ] , and the possibility of magnification of this quasar due to gravitational lensing is discussed in section [ sec - lens ] . a summary of the results is found in section [ sec - summary ] . note that throughout this paper we assume a standard @xmath5cdm cosmology , with @xmath6 , @xmath7 and @xmath8 @xcite .
the quasar pks 2126@xmath0158 is a radio - loud , flat - spectrum quasar at a redshift of @xmath9 .
it was first identified by @xcite in optical follow - up of flat - spectrum sources from the parkes 2700 mhz surveys , and at the time its redshift was measured it was only the fifth radio quasar with @xmath10 @xcite .
pks 2126@xmath0158 is known to be a very bright object at radio ( @xmath11 , @xcite ) , optical / near - infrared ( @xmath12 , @xcite ) and x - ray ( @xmath13 , @xcite ) frequencies .
these fluxes , combined with its relatively high redshift , mean it is among the most luminous quasars known .
the optical / near - infrared spectrum of pks 2126@xmath0158 has a blue power - law shape for wavelengths longer than @xmath14 band @xcite , where the ly@xmath15 line falls .
the flux drops off at shorter wavelengths due to absorption by the intervening gas of the lyman alpha forest .
absorption systems with strong metal lines are observed at the redshifts indicated in table [ tab - lines ] .
the relatively strong system at @xmath16 is particularly relevant for this paper , and is discussed in more detail in section [ sec - abssystem ] .
the peculiar concentration of objects in the vicinity of pks 2126@xmath0158 was first noticed from @xmath17-band images taken on 6 june 1993 by @xcite with iris @xcite on the anglo - australian telescope in the course of identification of sources in the parkes half - jansky flat - spectrum sample ( phfs ) .
there are @xmath18 sources visible within 30 arcsec of the quasar , principally toward the east and south .
@xcite obtained spectra of the two bright sources to the west ( designated therein as c1 and c2 , at distances of @xmath19 and @xmath20 arcsec west of the quasar respectively ) .
they were able to measure the redshift of c2 ( an emission line galaxy ) as @xmath21 , while the spectrum of c1 was inconclusive .
this is the only redshift information available from the literature for objects in the vicinity of pks 2126@xmath0158 .
.metal - line systems observed in the spectrum of pks 2126@xmath0158 @xcite . [ cols="<,<,<",options="header " , ] the measured redshifts for galaxies in the field of pks 2126@xmath0158 are shown in table [ tab - results ] .
the galaxies are numbered according to their position on the image , and sorted in the table in order of decreasing redshift .
the @xmath22 magnitudes shown are measured from the pre - imaging data ( i.e. the image shown in fig [ fig - gmos ] ) , calibrated on the landolt standard fields sa95 and tphe , using the transformations from @xcite .
we have converted these magnitudes into absolute magnitudes , @xmath23 .
the k - corrections were calculated using the template seds from @xcite and the filter transmission function for filter i_g0302 provided on the gemini web site .
filter at gemini north , believed to be similar to the one used at gemini south .
the transmission function for the gemini south filter was not available at time of writing . ]
we find 8 galaxies with redshifts in the range @xmath24 , straddling the low redshift metal - line absorption systems seen by @xcite .
the spectra of these galaxies are shown in fig .
[ fig - spectra ] . 158 . ]
the velocity dispersion of these eight galaxies is @xmath25 , centred on a redshift of @xmath26 .
three of these galaxies , however , are at large spatial spearations from those close to the quasar . if we include just the four closest to the quasar
# 17 , # 18 , # 19 and # 20 as well as # 14 , which is just close enough to be considered associated , we get @xmath27 and @xmath28 .
this velocity dispersion is characteristic of a group environment , rather than a more dense cluster environment which would show @xmath29 .
it is , in fact , a relatively poor group , with only five members brighter than @xmath30 .
there are at least four fainter galaxies for which we have not obtained spectra that are also candidates for inclusion in the group ( see sec .
[ sec - absn ] for further discussion ) .
the remaining galaxies at @xmath1 ( # 24 , # 03 and # 01 ) are part of the same large - scale structure that the group is embedded in , but not necessarily directly associated with the group itself .
in addition to this structure , we find several other features in redshift - space .
there are two other main structures seen in the field , at @xmath31 and @xmath32 ( and possibly @xmath33 ) , although these are not as tightly grouped on the sky as the main group at @xmath1 .
there are a number of field galaxies at other redshifts as well .
we are thus seeing several features in the large - scale structure along the line - of - sight to pks 2126@xmath0158 .
the spectra of the remaining galaxies with measured redshifts are shown in figs .
[ fig - spectra - hi ] & [ fig - spectra - lo ] , and the distribution of galaxy redshifts can be seen in fig [ fig - zdist ] . of the remainder , six objects were identified as stars , and a further five had spectra of too low quality for a redshift measurement . of these latter objects ,
two ( # 11 and # 15 ) have extended morphologies in the @xmath22 gmos image , and @xmath34 colours redder than all the confirmed group members ( @xmath35 and @xmath36 respectively , cf.@xmath37 for the group members ) .
they are thus either red galaxies in the @xmath1 group ( more likely in the case of # 15 ) , or background galaxies ( and redder due to their greater redshift ) .
as seen in section [ sec - results ] , we find a concentration of galaxies in the redshift range @xmath24 .
this coincides in redshift - space with a metal - line absorption system detected by @xcite and @xcite .
the system consists of four components in velocity , with three species detected and , plus in the two highest column density components .
the redshifts of the components are @xmath38 , 0.6628 , 0.6634 and 0.6637 @xcite .
this system is quite strong : the two strongest components ( at @xmath39 and @xmath38 ) have column densities of @xmath40 @xmath41 and @xmath42 @xmath41 respectively @xcite . using the doppler parameters quoted by @xcite , we can convert these to equivalent widths of 0.44 and 0.41 respectively ( for the @xmath43 line in the rest frame ) . a key question is whether this system can be associated with the group of galaxies detected in our gmos observations .
the nearest of the galaxies to pks 2126@xmath0158 is galaxy # 17 , which lies at a distance of 51.2kpc to the quasar s line - of - sight ( hereinafter qlos ) .
this places the qlos in the outer parts of the galaxy s halo .
this separation is comparable to those seen in other absorption systems ( see , for example , @xcite ) .
this galaxy , however , does show a large velocity offset from the absorbing system , with velocity differences @xmath44 with respect to the four components ranging from @xmath45 .
galaxy # 18 provides a much closer match to the velocities ( @xmath46 ) , and is only 61.2kpc from the qlos .
it is also the brightest galaxy in the group , with @xmath47 .
a relationship has been noted by @xcite between the equivalent width @xmath48 of the absorption system and the impact parameter @xmath49 of the associated galaxy , where @xmath50 . by converting the column density and doppler parameters of the absorption lines ,
we can obtain the equivalent widths for the system : the four @xmath51 lines from @xcite yield rest - frame values of @xmath52 and 0.13 ( in order of increasing redshift ) .
comparing with the @xcite relationship , we find that the two stronger systems have @xmath48 values corresponding to @xmath53 kpc , while the two weaker systems correspond to @xmath54 kpc .
the galaxies we have observed to be in the group all lie at impact parameters between these two values . there is , however , an object closer than the closest observed galaxy to the qlos ( # 17 ) , which , should it prove to be at the same redshift , may be a counterpart to one of the stronger systems ( see section [ sec - altgal ] for more discussion on this source ) .
if we focus just on the confirmed group members , the @xcite relationship taken at face value indicates that none of these galaxies are suitable counterparts for the absorption system(s ) .
recent studies , however , such as that presented in @xcite , indicate that there is much more scatter in the @xmath55 relationship than the @xcite relationship would suggest .
this calls into doubt its utility in making
_ predictions _ about the likelihood of potential absorbers .
the @xcite results suggest that the observed galaxies can indeed be considered potential hosts of the absorber , even with their relatively large impact parameters .
the identity of the absorber then is still an open question .
there are three objects closer to pks 2126@xmath0158 than galaxy # 17 .
these can be seen in fig .
[ fig - closeup ] , along with some of the group members .
a spectrum of the object west of pks 2126@xmath0158 was observed by @xcite ( their c1 ) .
they were unable to measure its redshift , but its optical and near - infrared colours are similar to their c2 ( the next galaxy to the west ) , which they found to be at @xmath56 .
it is therefore likely that this galaxy is foreground to the group at @xmath1 .
subset of the gmos @xmath22 image surrounding pks 2126@xmath0158 , marked by q. objects mentioned in the text are labeled .
north and east are indicated by the arrows .
the 5 arcsec scale bar at bottom is equivalent to 34.9 kpc at the redshift of the galaxy group . ]
the remaining two nearby objects were not able to be observed in our gmos observations ( due to the limitations of slit placement ) .
object x1 appears to be stellar in morphology , and has colours @xmath57 not typical of a galaxy at the same redshift as the group .
similarly , the more distant ( from the qlos ) object x3 has a stellar morphology , and even bluer colours than x1 ( @xmath58 ) , and thus is likely to be a star .
the closest object to the quasar , x2 , has an extended , non - stellar morphology , and redder colours than x1 or x3 : @xmath59 , similar to the colours of the galaxies in the group at @xmath1 .
it may thus be a galaxy in the same system .
this putative galaxy would then lie at a projected distance of @xmath60 from the qlos , making it a possible candidate as a host of the absorbing system .
its impact parameter provides a good match to the stronger absorption systems , if one uses the @xmath55 relationship of @xcite .
a further point in its favour is its extended , somewhat elongated morphology suggestive of a disc galaxy , matching the morphology of many -absorbing galaxies .
is the absorbing gas necessarily associated with a galaxy ?
the fact that we have a group environment means that the velocity dispersion is comparable to the rotational velocities of the individual galaxies .
galaxy interactions are then more likely to strip gas from galaxies and leave it in the intra - group medium
. such interactions could date from the formation epoch of the group , or from more contemporary encounters .
an potential example of such an interaction , based on imaging data , is the pair # 17 and # 18 .
the velocity difference between the two ( @xmath61 ) , is comparable to the velocity dispersion of the group , and so is feasible for an interaction . a close - up of this pair of galaxies is shown in fig .
[ fig - gal - pair ] .
there is evidence for a connection in flux between the two , with the flux in the `` bridge '' region reaching @xmath62 above the background .
there is certainly no spectroscopic sign of any strong interaction taking place , with both galaxies exhibiting simple passive elliptical spectra .
however , sensitive narrow - band imaging may be able to detect h@xmath15 emission from ejected gas in the vicinity of the galaxies , and would prove a good test of this interaction hypothesis . .
contour levels are at @xmath63 above the mean background .
field of view is 6.13arcsec , and the direction to pks 2126@xmath0158 is indicated . ]
pks 2126@xmath0158 is observed to have a strong absorption system at the same redshift as the group of galaxies detected through our gmos spectroscopy .
all the nearby galaxies that have spectroscopy are observed to be ellipticals : each shows a passive elliptical galaxy spectrum , and none shows evidence for on - going star - formation ( for instance , through the presence of an emission line ) , which would indicate the presence of large amounts of gas .
many absorbers are identified with spiral galaxies ( e.g. * ? ? ?
* ) , although this is by no means an exclusive statement .
@xcite found that absorption could be found in galaxies resembling present - day ellipticals as well as late - type spirals , @xcite found absorption associated with two s0 galaxies , and recent studies @xcite have found a small fraction of elliptical absorbers .
the latter studies also indicate that the absorption is also relatively patchy , and so one ( or more ) of the galaxies could have isolated areas of absorbing gas that happen to fall across the qlos , but are not themselves an important overall part of the galaxy .
this may be similar to the situation observed by @xcite , who failed to find absorption associated with two elliptical galaxies despite their small impact parameters .
it is possible that there is intra - group gas responsible for the absorption , although there is little direct evidence for this at this stage .
the best candidate , however , for the absorber is probably galaxy x2 .
it is closer to the qlos , and shows a much less uniform morphology than the identified galaxies .
ultimately , a spectroscopic confirmation of its redshift is the only way to be sure future observations will hopefully resolve this question .
pks 2126@xmath0158 is rather luminous at all frequencies from radio to x - ray .
we would like to know whether it is intrinsically luminous , or has been magnified through gravitational lensing by some intermediate mass distribution .
strong lensing can be ruled out , as pks 2126@xmath0158 is not multiply - imaged even at vlbi resolution ( @xmath64 mas ) , it shows a distinct core - jet structure @xcite .
magnification due to gravitational lensing is still possible without multiple images being detected .
there are two contexts in which this can take place : either a single image is produced , or there are indeed multiple images but at either faint levels or unresolved separations .
since vlbi observations place quite strong limits on the existence of optically - unresolved images , for lensing to be taking place we would require the production of a single image only .
however , @xcite find that this is possible only when the lensing is due to massive cluster - sized haloes ( @xmath65 ) .
our observations show that the largest mass concentration along the line - of - sight is merely a galaxy group .
the total luminosity of all galaxies in the group is @xmath66 ( assuming an absolute magnitude of the sun of @xmath67 , and taking into account all possible members , including x2 and the fainter objects in fig . [ fig - closeup ] ) . for reasonable mass - to - light ratios
, the group mass will be much less than the threshold given by @xcite .
this is supported by the non - detection by @xcite of any extended x - ray emission in _ chandra _ images of pks 2126@xmath0158 , indicating the lack of hot cluster gas around or in the foreground of the quasar .
we can thus rule out the possibility that pks 2126@xmath0158 is being magnified by any significant amount .
its large apparent luminosity is then indeed indicative of its intrinsic power .
we have observed the field surrounding pks 2126@xmath0158 with gemini south + gmos in multi - object spectroscopy mode , measuring the redshifts of most of the nearby galaxies .
we find a group of galaxies at @xmath1 , at a similar velocity to a metal - line absorption system seen in the quasar s spectrum .
the group has five confirmed members , and a further four fainter galaxies are seen nearby and are possibly associated .
there are a further three galaxies close in redshift but at large separations on the sky while not part of the group , they are certainly part of the same large - scale structure . we have also made use of the multiplexing capabilities of gmos to measure the redshifts of many other galaxies in the field , and we see several distinct features in redshift - space foreground to the group .
these observations , in showing the lack of a large cluster in front of pks 2126@xmath0158 , indicate that the likelihood the quasar is lensed is small .
while the group as a whole appears to be associated with the absorption system , the fact that the redshifts of a number of nearby candidates have not been measured means some doubt remains as to the identity of the absorber .
if the absorber is associated with the identified galaxies however , this would be a rare example of an absorption system associated with an early - type galaxy .
we wish to thank the staff at gemini south for their help in obtaining the spectroscopic data , particularly acknowledging the help provided by michael ledlow , our contact scientist until his tragic passing .
mtw acknowledges the assistance provided by a newsouth global fellowship from the university of new south wales .
this work is based on observations obtained at the gemini observatory under gemini program gs-2003b - q-15 .
the gemini observatory 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 ) . | the high - redshift radio - loud quasar pks 2126@xmath0158 is found to have a large number of red galaxies in close apparent proximity .
we use the gemini multi - object spectrograph ( gmos ) on gemini south to obtain optical spectra for a large fraction of these sources .
we show that there is a group of galaxies at @xmath1 , coincident with a metal - line absorption system seen in the quasar s optical spectrum .
the multiplexing capabilities of gmos also allow us to measure redshifts of many foreground galaxies in the field surrounding the quasar .
the galaxy group has five confirmed members , and a further four fainter galaxies are possibly associated .
all confirmed members exhibit early - type galaxy spectra , a rare situation for a absorbing system .
we discuss the relationship of this group to the absorbing gas , and the possibility of gravitational lensing of the quasar due to the intervening galaxies .
[ firstpage ] quasars : individual : pks 2126@xmath0158 quasars : absorption lines galaxies : general gravitational lensing |
in this section we derive the expression for the matter hamiltonian ( 6 ) used in the article . with the assumption of a frequency - independent final state coulomb interaction , from the ladder diagrams in fig .
2 , a bethe - salpeter equation for the electron - hole bubble @xmath40 can be found along the lines of ref .
this integral equation for the vertex factor @xmath40 is given by @xmath41 the solution of this equation yields the photon polarization by the following relation @xmath42 where for simplicity we have set the photon momentum equal to zero .
the function @xmath43 is given by the matsubara summation over the internal fermionic frequency @xmath44 of the electron and hole propagators , defined in fig .
2 . because we consider a zero photon momentum , only electrons and holes with opposite momentum contribute .
hence we have @xmath45 for the electron we use the free green s function @xmath46 with @xmath47 . here
, @xmath48 is the single particle dispersion in the parabolic mass approximation and @xmath49 stands for the screened exchange interaction of the optically excited electron with the fermi sea .
also the electron chemical potential @xmath50 is renormalized due to exchange interaction . for the hole green s function
, we include the interactions with the electron fermi sea perturbatively as follows : @xmath51 here , for the dynamical self energy @xmath52 , describing hole relaxation that is responsible for a finite polariton lifetime , we use the 2d rpa result .
the free hole green s function is given by @xmath53 the hole dispersion is also assumed to be quadratic , so @xmath54 .
no exchange contribution shows up because we consider a single hole ( @xmath55 ) . in order to be able to perform the matsubara summation in eq .
exactly , we further approximate the rpa result to lowest order in the lindhard polarization ( ` 1 particle - hole bubble ' ) , so only emission of a single particle - hole pair is taken into account : @xmath56 using the 2d expression of the lindhard polarization function , the dynamical self energy is given by @xmath57 \frac{1-n_{f}\left [ \xi _ { \mathbf{k}-% \mathbf{q}}^{h}\right ] + n_{b}(\varepsilon^e_{{\bf k+q}}-\varepsilon^e_{{\bf k}})}{i\nu _
{ n}-\xi _ { \mathbf{k}-\mathbf{q}}^{h}-\varepsilon^e_{{\bf k+q}}+\varepsilon^e_{{\bf k}}}. \label{eq : lindhardpol}\end{aligned}\ ] ] here , the exchange correction to the electron energy was neglected to simplify the subsequent calculations .
furthermore , the double summation runs over the center - off - mass momentum and relative wavevector of the particle - hole pair , resp .
@xmath58.combining the expressions - we can do the matsubara sum over @xmath44 .
the result is given by @xmath59 where we have @xmath60 and @xmath61 \crcr & & \times \left\ { \begin{array}{c } \frac{1}{1-n_{f}\left ( \xi^e _ { \mathbf{k}}% \right ) } \frac{1}{-\xi _ { \mathbf{-k}}^{h}-\left ( -i\omega + \xi^e _ { % \mathbf{k } } \right ) } \frac{1}{-\xi _ { % \mathbf{-k}}^{h}-\left ( -\xi _ { \mathbf{-k+q}}^{h}+\varepsilon^e _ { \mathbf{k+q}}-\varepsilon^e _ { % \mathbf{k}}\right ) } \\ + \frac{1}{1-n_{f}\left ( \xi^e _ { \mathbf{k}}\right ) } \frac{1}{\left [ -\xi _ { \mathbf{-k}}^{h}-\left ( -\xi _ { % \mathbf{-k+q}}^{h}+\varepsilon^e _ { \mathbf{k+q}}-\varepsilon^e _ { \mathbf{k}}\right ) \right ] ^{2 } } \\ -\frac{i\omega -\xi
_ { \mathbf{-k}}^{h}-\xi^e _ { \mathbf{k } } } { 1-n_{f}\left ( \xi^e _ { \mathbf{k } } \right ) } \frac{n_{f}\left ( -\xi _ { \mathbf{-k+q}}^{h}+\varepsilon^e _ { % \mathbf{k+q}}-\varepsilon^e _ { \mathbf{k}}\right ) } { -\xi _ { \mathbf{-k+q}}^{h}+\varepsilon^e _ { % \mathbf{k+q}}-\varepsilon^e _ { \mathbf{k}}+i\omega -\xi^e _ { \mathbf{k } } } \frac{1}{\left ( -\xi _ { \mathbf{-k+q}}^{h}+\varepsilon^e _ { \mathbf{% k+q}}-\varepsilon^e _ { \mathbf{k}}+\xi _ { \mathbf{-k}}^{h}\right ) ^{2}}% \end{array}% \right\ } \end{aligned}\ ] ] upon analytical continuation @xmath62 only the last line yields an imaginary part and thus the lifetime of the hole .
the other terms correspond to the energy shift , but these are beyond the scope of this work .
one has @xmath63 \frac{% n_{f}\left ( -\varepsilon _ { \mathbf{-k+q}}^{h}+\varepsilon^e _ { \mathbf{k+q}% } -\varepsilon^e _ { \mathbf{k}}\right ) } { \left ( \varepsilon _ { \mathbf{-k+q}% } ^{h}-\varepsilon^e _ { \mathbf{k+q}}+\varepsilon^e _ { \mathbf{k}}-\varepsilon _ { % \mathbf{-k}}^{h}\right ) ^{2}}\delta \left [ \omega -\varepsilon _ { \mathbf{-k+q% } } ^{h}+\varepsilon^e _ { \mathbf{k+q}}-\varepsilon^e _ { \mathbf{k}}-\varepsilon^e _ { % \mathbf{k}}-\sigma _ { x}\left ( \mathbf{k}\right ) \right ] , \end{aligned}\ ] ] where the delta - function expresses conservation of energy and momentum .
this is in fact expression ( 7 ) used in the article .
the 4-fold summation can be worked out further to a 2-dimensional integral which we than compute numerically .
this leads to the following result ( all energies expressed in exciton rydbergs @xmath37 ) : @xmath64 where the pauli blocking factor @xmath65 is @xmath66 here the hole self energy is @xmath67 ^ 2 } \theta\left(a<0\right)\theta\left(k_f^2-a >
k^2_{\text{min}})\right ) \label{eq : sigmahole}\ ] ] whith @xmath68 the inverse screening length as defined in @xcite .
the functions @xmath69 are given by @xmath70 for the bse the inverse of the function @xmath71 was needed , see eq . .
combining eqs . , and we find @xmath72 for a small hole self energy , we can use the following taylor expansion : @xmath73 for @xmath74 . identifying @xmath75
we obtain fermi s golden rule : @xmath76 plugging this expression into .
eq . the identification with eq .
( 6 ) from the article is straightforward .
it is worth noting that the taylor series is consistent with the expansion of the rpa self energy to lowest order in the lindhard polarization .
moreover , the polariton spectrum in figure 4 from the article shows that the linewidths are much smaller than the exciton binding energy .
hence , the current results and approximations are selfconsistent .
both final state interactions and the presence of the fermi edge are necessary for the formation of the upper fermi edge polariton .
this is illustrated quantitatively in fig .
[ fig : suppl ] ( a - b ) and ( c - d ) for two electron densities and cavity detunings .
the blue lines indicate the single - particle results while the red lines contain the final state coulomb interactions .
figures a , c show the photon spectral function on resonance with the fermi edge @xmath32 ( black dashed line ) for two different electron gas densities .
a clear suppression of the spectral function linewidth when taking into account final state interactions ( fsi ) is observed as compared to the single - particle result . to estimate the importance of the fermi edge ,
the cavity mode is largely blue detuned from the absorption threshold energy to @xmath77 ( black dashed lines ) , see figures b , d. it is clear that both the single particle theory and final state interaction results now coincide ( same scales are used to facilitate comparison ) .
the fermi edge is thus responsible for the ` rabi splitting ' , meaning that it really pushes the upper polariton into the electron - hole continuum . as a last remark , for very large blue detuning of the cavity mode from the fermi edge ,
a red shift from the cavity mode is seen in figures b , d. this can be attributed to a different index of refraction in the electron - hole continuum in comparison with the excitonic region ( see also ref . | we present a theoretical study on polaritons in highly doped semiconductor microcavities . in particular , we focus on a cavity mode that is resonant with the absorption threshold ( ` fermi edge ' ) . in agreement with experimental results ,
the strong light - matter coupling is maintained under very high doping within our ladder diagram approximation .
while the lower polariton is qualitatively unaltered , it acquires a finite lifetime due to relaxation of the valence band hole if the electron density exceeds a certain critical value . on the other hand
the upper polariton has a finite lifetime for all densities , because it lies in the electron - hole continuum where no bound state exists .
our calculations show that a narrow upper polariton quasiparticle still exists as a result from the interplay between light - matter coupling and final state coulomb interaction .
microcavity polaritons are bosonic quasiparticles that appear in a semiconductor microcavity with an embedded quantum well ( qw ) when the cavity is tuned to an excitonic transition in the qw . in the last decade they have emerged as quasiparticles with many favorable properties for studying quantum optics and many - body physics in integrated photonic structures .
bose - einstein condensation of polaritons and their superfluid properties have been investigated in detail both experimentally and theoretically @xcite . in these studies ,
the tunability of the polariton properties is often exploited .
for example , the number of quantum wells is varied to change the rabi frequency ; etching , strain and surface acoustic waves have been used to create potentials for the polaritons .
a tuning parameter that has received relatively little attention is the introduction of charges in the qw that interact with the polaritons . in the case of low charge concentration ,
trionic bound states exist and a trionic polariton is observed @xcite .
an experiment with high density modulation doping was performed by gabbay _
et al .
_ @xcite .
they have shown that strong light - matter coupling is possible , even though neither an excitonic nor a trionic state could be resolved . from the theoretical side ,
the mixed electronic - polaritonic system was proposed to reach superconductivity at higher temperatures , possibly even room temperature , thanks to a strongly attractive electron - electron interaction mediated by the polaritons @xcite . for what concerns the effect of the electron gas on the polaritons ,
it has been suggested to facilitate polariton lasing thanks to the polariton - electron scattering @xcite . in this work ,
the polariton quasiparticle itself was assumed not to be affected by the electron gas . a first theoretical study on the modification of the single particle polariton properties due to the electron gas started from an estimate of the oscillator strength computed in the context of the mahan singularity @xcite .
indeed , the optical excitation of an electron gas ( metal or doped semiconductor ) is a central problem in many - body physics that was introduced by mahan @xcite and to which nozires , de dominicis @xcite and anderson @xcite made seminal contributions .
it is still an active topic of research that has recently attracted the attention of the community working on ultracold atomic gases @xcite .
the purpose of this letter is to study the different regimes of light - matter coupling in highly doped semiconductor qws , see fig .
[ fig : mc ] .
the regime of low doping density , where trionic correlations are dominant , is not in the scope of this work .
a first effect of the presence of the high density electron gas is that the low - lying momentum states can not be used for the formation of the exciton , because of the pauli principle .
more importantly , the electron gas has , in contrast to the empty qw , low lying particle - hole excitations .
they are responsible for the screening of the electron - hole coulomb interaction .
both the pauli blocking and the screening reduce the exciton binding energy @xcite .
as long as the particle - hole excitations are virtual , they do not qualitatively alter the light - matter coupling , but merely renormalize the rabi frequency .
qualitative changes occur when the polariton energy becomes sufficiently high to excite free electron - hole pairs .
a systematic approach to study the different regimes is diagrammatic perturbation theory @xcite . in this formalism
, excitonic effects are taken into account by including the electron - hole coulomb interaction through ladder diagrams ( see fig . [ fdiag ] ) .
the screening of the coulomb interaction can be taken into account by a dielectric function , that we approximate by the rpa result for the jellium model .
furthermore , we assume that the quantum well electrons and holes are purely two - dimensional . to be specific ,
we will consider n - doping , so that only a free electron gas is present and a single photo - created hole .
the primary physical quantities of interest , the absorption and photoluminescence spectrum , can be characterized through the photon spectral function , defined as @xmath0 with @xmath1 the retarded photon propagator , given by @xmath2 here @xmath3 corresponds to the cavity mode which is assumed to be lossless .
the photon momentum is taken to be zero for simplicity .
the photon self energy @xmath4 originates from the material excitations within the quantum well and thus contains all many - body physics of the electron gas .
we will approximate it by the feynman diagrams shown in fig .
[ fdiag ] . .
( b ) the photon self energy @xmath4 is obtained in the ladder diagram approximation .
( c , d ) electron @xmath5 and hole @xmath6 green s functions used in the calculations . ,
height=170 ] the coulomb interaction in these diagrams is already taken to be the rpa screened interaction , indicated by the double dashed lines . while for the electron a complete dyson series for the screened exchange interaction is taken into account , for the hole both the dyson series for the propagator and the dynamical self energy
are expanded to lowest order to keep the calculations manageable : @xmath7 @xmath8 obviously , this expansion is only valid for small self energy .
the approximation , can be shown to be equivalent to the use of fermi s golden rule to calculate the linewidth of a polariton due to the relaxation of the hole by the creation of one particle - hole excitation in the electron fermi sea ( see supplemental material ) .
the dynamical self energy of the electron is omitted , because its relaxation is suppressed by pauli blocking . under the assumption that the final state coulomb interaction ( fsi ) is frequency - independent , a bethe - salpeter equation ( bse ) for the electron - hole propagator
can be derived @xcite .
it reads @xmath9p({\bf{k}},i\omega ) = g , \label{eq : p}\ ] ] with @xmath10 and @xmath11n_{f}(-\varepsilon_{\bf{-k+q}}^{h}+\varepsilon^e_{\bf{k+q}}-\varepsilon^e_{\bf{k}})}{(\varepsilon_{\bf{-k+q}}^{h}-\varepsilon^e_{\bf{k+q}}+\varepsilon^e_{\bf{k}}-\varepsilon_{\bf{-k}}^{h})^2}\,\delta[\omega-\varepsilon_{\bf{-k+q}}^{h}+\varepsilon^e_{\bf{k+q}}-\varepsilon^e_{\bf{k}}-\varepsilon^e_{\bf{k}}-\sigma_{x}(\bf{k } ) ] .
\label{eq : sigmahole}\end{aligned}\ ] ] here , the hamiltonian @xmath12 describes the relative part of the two - particle system ; the reduced mass is @xmath13 .
to arrive at the bse , we had to make the static approximation for the fsi , for which we used the 2d rpa expression of stern @xcite .
this static approximation simplifies the subsequent calculations , but it will be interesting to explore the physical consequences of relaxing this approximation .
a well known limitation of rpa is that it is only valid for a high density electron gas @xmath14 with @xmath15 being the effective bohr radius .
specifically for this problem , it means that rpa is not able to capture trionic effects that become important at low density .
the photon self energy is finally related to the vertex factor @xmath16 as @xmath17 here @xmath18 is the light - matter coupling and @xmath19 the quantum well surface .
it can be seen from eq .
( [ eq : p ] ) that the bound excitonic levels show up as poles of @xmath16 and consequently as poles in the photon self energy @xmath4 . for an undoped qw
this can be calculated analytically @xcite ; here we implement it numerically .
the lowest lying exciton level is depicted in fig .
[ fig : evskf]a as a function of @xmath20 ( blue line ) , where @xmath21 holds .
its energy is measured in exciton rydbergs @xmath22 .
a first important energy scale in the exciton physics is the lowest energy of a free electron - hole pair , indicated as @xmath23 in fig .
[ fig : evskf]b and by a red line in panel a. to be consistent with the computation of the bound exciton , we have included the renormalization of the fermi energy due to the exchange interaction .
the zero of energy in fig .
[ fig : evskf]a was chosen as the energy of a free electron - hole pair with both particles at the fermi wave vector .
this energy corresponds to the absorption threshold @xmath24 ( ` fermi edge ' ) in fig .
[ fig : evskf]b . the exciton forms the groundstate of the optically excited quantum well .
above the critical wavevector the exciton becomes metastable ( dashed line ) and obtains a finite lifetime due to relaxation of the hole via emission of a particle - hole pair ( right figure ) .
@xmath25 is determined by @xmath26 . in the figure @xmath27
is used .
, height=188 ] below a critical density , the electron and hole form a stable bound state below @xmath23 , the well - known exciton . in this case , the hole self energy ( [ eq : sigmahole ] ) vanishes at the bound state energy , corresponding to an infinite exciton life time . beyond the critical value @xmath28 , which depends on the ratio @xmath26 ( trend indicated with a black arrow )
, the exciton crosses the free electron - hole energy and is no longer the lowest energy state , but becomes a metastable bound state ( blue dashed line ) .
it is the hole kinetic energy that is responsible for the fact that the exciton is not the ground state .
when the hole is bound to the electron in a zero momentum exciton , the momentum of the electron has to be compensated by an opposite momentum of the hole . because the low momentum electron states are pauli blocked , there is a minimal kinetic energy cost @xmath29 for the hole to form a bound state with the electron
obviously , this cost vanishes in the limit where the hole becomes infinitely heavy @xmath30 .
we then find that the exciton is stable for arbitrary density and corresponds to the bound state that always exists for an attractive potential in 2d . when the exciton is metastable , the hole self energy ( [ eq : sigmahole ] ) is nonzero at the exciton energy .
it is seen to correspond to a relaxation process through the emission of a low energetic particle - hole pair , see fig .
[ fig : evskf]b .
note that only the lifetime contribution of this process is taken into account and the associated energy shift has not been considered in this work .
the delta function in expresses energy and momentum conservation for this process , while the occupation factors describe the amount of phase space that is available for the decay process . below the critical fermi wave vector @xmath31
, this phase space factor vanishes when energy and momentum are conserved .
the relaxation of the hole by plasmon emission was considered in the plasmon - pole approximation @xcite . however
, this relaxation channel does not contribute because it does not conserve energy and momentum .
unfortunately , the hole self energy ( [ eq : sigmahole ] ) diverges for all energies higher than @xmath32 ( ` the continuum ' ) so that we can not use our previous approach to include the hole self energy .
a more sophisticated approach is necessary . as a first approximation
, we follow schmitt - rink @xcite and obtain a first meaningful result by omitting the hole self energy altogether . actually above @xmath33
, the photon self energy acquires an imaginary part even when the hamiltonian ( [ eq : hx ] ) is real .
physically , this finite lifetime comes from the conversion of photons into free electrons and holes .
therefore , as a lowest order approximation to the spectral function in this regime we calculate the spectral function exactly in the same way as before , summing up all ladder diagrams to infinite order , but the self energy for the hole is now discarded .
the solution of the bse equation can now be calculated analytically to some extent @xcite but we implemented it straightforwardly numerically . for the exciton , we recover the ` excitonic enhancement ' @xcite , an increase in the absorption close to @xmath33 .
this is a many - body modification , due to final state interactions , to the constant absorption obtained in a single particle picture .
a summary of the exciton physics is presented in fig .
[ fig : mainfigure]a .
it shows the photon spectral function as a function of energy and the electron fermi wave vector , in the limit of very large positive cavity detuning . in this limit
, we recover the exciton physics .
the exciton is blue shifted when the electron density increases .
it is also clear that the excitonic linewidth then increases .
the excitonic enhancement for @xmath34 is not very clear on this figure and is shown in more detail in the inset on fig .
[ fig : mainfigure]c - e .
it is an intriguing question whether this excitonic effect may lead to the formation of polaritons .
experimentally , this question was answered positively in ref .
@xcite where the observation of ` fermi edge polaritons ' was claimed . to answer this question theoretically , we plot in fig .
[ fig : mainfigure]b the photon spectral function when the cavity photon is in resonance with the fermi edge @xmath24 .
different detunings are shown in the supplemental material .
several cuts at fixed electron densities , indicated with white dashed lines , are shown in the panels ( c - e ) . as expected , the lower polariton is pushed below the excitonic resonance .
the shift of the polariton with respect to the exciton is however reduced due to the presence of the electron gas .
this is expected since the light - matter coupling is proportional to the electron - hole overlap , which is reduced due to pauli blocking and screening .
importantly , the linewidth of the polariton is strongly reduced with respect to the excitonic linewidth .
this narrowing can be attributed to the reduced final state phase space that slows the hole relaxation . for low electron gas densities
it is even possible for the polaritons to decrease their energy below the free electron - hole energy @xmath35 making their linewidth vanish ( on the figure a very small non - zero value is added to make the resonance visible ) .
the linewidth reduction can be seen more clearly from the various cuts taken for different values of @xmath20 ( fig .
[ fig : mainfigure]c - e ) .
qualitatively however the lower polariton is entirely similar to the one in the empty quantum well .
this is not the case for the upper polariton that can be clearly identified in fig .
[ fig : mainfigure]b , above the absorption threshold .
above @xmath33 bound states do no longer exist and one should think of the excitonic enhancement as the valence band hole interacting with all electrons in the fermi sea _
simultaneously_. nevertheless , this polaritonic resonance has a linewidth that is much smaller than the excitonic enhancement feature ( see insets of fig .
[ fig : mainfigure]c - e ) .
moreover , it is shifted from the cavity resonance by @xmath36 @xmath37 .
despite the two different approximations that were used to compute the spectral function below and above the absorption continuum ( red vertical line ) respectively , the matching of the curves is satisfactory .
it is conceptually interesting to elucidate the role of the final state interactions on this upper polariton . to this purpose
, we have turned off the final state coulomb interaction [ @xmath38 in eq .
( [ eq : hx ] ) ] in our calculations . without fsi
a much smaller energy shift from the cavity resonance is observed and a larger linewidth is obtained .
it is really the final state coulomb interactions between the photo - excited electron - hole pair that pushes the quasi - particle above the cavity frequency .
we have also checked that the ` rabi splitting ' vanishes when the cavity energy is blue detuned with respect to @xmath24 .
it is indeed the fermi edge @xmath33 that is responsible for the formation of an upper polariton in the continuum , see supplemental material .
finally , it is worth mentioning that when the light - matter coupling is increased , the upper polariton is further blue shifted and its linewidth increases . for a good visibility of the upper ` fermi edge ' polariton , it is actually advantageous if the rabi splitting is not too large ( not too many quantum wells ) .
this trend is opposite to the one of the lower polariton , where a larger rabi splitting leads to a lower energy and therefore a longer lifetime .
note that the upper polariton energy decreases in fig .
[ fig : mainfigure]b for increasing @xmath39 , even though the photon energy is constant and the exciton energy increases . in an undoped quantum well
, the opposite dependence is observed for increasing exciton energy .
we attribute the decrease of the upper polariton energy to the reduction of the electron - hole overlap when the density is increased .
our theoretical results are in qualitative agreement with the experiments in ref .
@xcite , where at high doping density both an upper and lower polariton were observed . in this experimental work , a phenomenological explanation based on the fermi edge enhancement of the optical response was suggested .
we have shown here that this enhancement is due to the usual excitonic bound state for the lower polariton , where it is of a different , collective nature for the upper polariton .
the discontinuity in the spectral function at @xmath33 is of course an artifact of our different approximations below and above this continuum energy .
it should be cured by a consistent inclusion of the self energy below and above the absorption threshold energy .
this self energy should actually contain the diagrams that are relevant for the physics of the mahan singularity so to capture the effect of this physics on the nature of the polaritons in doped quantum wells .
the present work is only a first step to such a complete description .
it will also be of great interest to investigate the interactions between polaritons in doped quantum wells .
it may be expected that interactions that are induced by excitations in the electron gas may enhance the polariton - polariton interactions with possible applications in quantum information processing .
99 i. carusotto , and c. ciuti , rev .
mod .
phys .
r. rapaport , r. harel , e. cohen , arza ron , e. linder , and l. n. pfeiffer , phys .
rev .
lett . * 84 * , 1607 ( 2000 ) .
d. bajoni , m. perrin , p. sennelart , a. lematre , b. sermage , and j. bloch , phys .
rev .
b * 73 * , 205344 ( 2006 ) .
a. gabbay , y. preezant , e. cohen , b. m. ashkinadze , and l. n. pfeiffer , phys .
rev .
lett . * 99 * , 157402 ( 2007 ) .
f. laussy , a. kavokin , and i. shelykh , phys .
rev .
lett . * 104 * , 106402 ( 2010 ) . g. malpuech , a. kavokin , a. di carlo , and j. j. baumberg , phys .
rev .
b * 65 * , 153310 ( 2002 )
. n. s. averkiev , and m. m. glazov , phys .
rev .
b * 76 * , 045320 ( 2007 ) .
n. s. averkiev , m. m. glazov , and m. m. voronov , solid .
state .
comm . * 152 * , 395 ( 2012 ) . g. d. mahan , phys . rev . *
153 * , 882 ( 1967 ) .
p. nozires , and c.t
. de dominicis , phys . rev . * 178 * , 1097 ( 1969 ) . p. anderson , phys . rev . lett . * 18 * , 1049 ( 1967 ) .
m. knap , a. shashi , y. nishida , a. imambekov , d. a. abanin , and e. demler , phys .
rev .
x * 2 * , 041020 ( 2012 ) .
g. refael , and e. demler , phys .
rev .
b * 77 * , 144511 ( 2008 ) .
a. schirotzek , c. wu , a. sommer , and m. zwierlein , phys .
rev .
lett . * 102 * , 230402 ( 2009 ) .
d. a. kleinman , phys .
rev .
b * 32 * , 3766 ( 1985 ) .
d. kulakovskii , s. gubarev and yu .
lozovik , jetp lett . * 74 * , 118 ( 2001 ) .
g. d. mahan , many - particle physics , plenum press - new york .
f. stern , phys .
rev .
lett .
* 18 * , 546 ( 1967 ) .
d. s. citrin , and j. b. khurgin , phys .
rev .
b * 68 * , 205325 ( 2003 ) .
a. w. overhauser , phys .
rev .
b * 3 * , 1888 ( 1971 ) .
s. schmitt - rink , c. ell , and h. haug , phys .
rev .
b * 33 * , 1183 ( 1986 ) .
p. hawrylak , phys .
rev .
b * 44 * , 3821 ( 1991 ) .
d. bajoni , p. senellart , a. lematre , and j. bloch , phys .
rev .
b * 76 * , 201305(r ) ( 2007 ) . * supplemental material * |
the successes of the nonrelativistic constituent quark models stimulate the study of the microscopic structure of the constituent quark .
such a structure can not only influence predictions for many observables of the nucleon but also offer a possibility to derive an important part of the effective interaction between constituent quarks .
the idea that the constituent quark is represented by a current quark surrounded by a chiral field , stems from georgi and manohar @xcite and has been further elaborated by cheng and li @xcite , and by baumgartner , pirner , knigsmann and povh @xcite in order to account for the parton counting in deep inelastic scattering ( related to the flavour asymmetry of sea quarks as given by the gottfried sum ; and to strangeness and spin content of the nucleon ) .
we offer an explicit candidate for such a constituent quark : a soliton - like solution of a coherent state of pions and @xmath0-mesons around a massless quark in the linear @xmath0-model .
the choice of a mesonic rather than gluonic description of the structure of the constituent quark is based on the experience that the scale for chiral symmetry breaking and emergence of chiral goldstone bosons appears at lower energies than the confinement scale where gluons dominate .
the proposed picture may be a good representation of constituent quarks in the way they contribute to static properties of hadrons and to the low lying energy spectrum .
it has been recently substantiated by several groups @xcite which demonstrated that the meson exchange effective potential between constituent quarks yields much more reasonable predictions for the baryon spectroscopy and the baryon - baryon interaction than the gluon exchange potential .
one of the simplest models describing the spontaneous breaking of the chiral symmetry is the linear @xmath0-model . in the non - strange sector
it involves u and d quarks , a triplet of pions and the @xmath0-meson .
the lagrangian density of the model @xcite takes the form : @xmath1 where different contributions in order represent the free quark part , the linear quark - meson interaction , the free meson part involving the @xmath0-mesons and the pions , and the meson self - interaction : @xmath2 the final term in ( [ potential ] ) is introduced to explicitly break the symmetry giving a finite mass to the pion .
the model contains only one adjustable parameter , @xmath3 .
the other three parameters , the chiral meson masses and the pion decay constant , are fixed to the corresponding experimental values ; we use @xmath4 mev and @xmath5 mev , for the mass of the @xmath0-meson we have taken a typical value , the recently suggested @xmath6 mev @xcite .
the results , however , only weakly depend on the @xmath0-meson mass .
we use coherent states to describe the pion and the @xmath0-meson clouds , and , furthermore , the `` hedgehog '' ansatz for the pion field and the current quark state in complete analogy to the approaches in which the nucleon is described as a meson cloud surrounding the three valence quarks @xcite .
the intrinsic state takes the form @xmath7 here @xmath8 and @xmath9 are the creation operators for the @xmath0-meson and the pion , respectively , in the spherical basis and @xmath10 is the third component of isospin ; @xmath11 , @xmath12 , and @xmath13 and @xmath14 represent the u ( d ) quark with spin down ( up ) , respectively .
only @xmath15-wave @xmath0-mesons and @xmath16-wave pions are coupled to the quark .
the functions @xmath17 and @xmath18 are related to the expectation values of the field operators as @xmath19 and @xmath20 , where @xmath21 and @xmath22 are the fourier transforms of @xmath17 and @xmath18 , respectively .
the physical states are obtained by performing the peierls - yoccoz projection @xmath23 using the grand spin symmetry of the hedgehog , the evaluation of physical quantities between the projected states is considerably simplified ; details of the calculation technique as well as a discussion of the validity of the hedgehog approximation can be found in @xcite .
the _ ansatz _ ( [ hedgehog],[projection ] ) leads to a system of coupled nonlinear differential equations which are solved using the package _ colsys _ @xcite .
we also impose a constraint that ensures the correct asymptotic behaviour of the pion field which is , otherwise , not automatically satisfied in this type of variational approach @xcite . we have found a selfconsistent solution with @xmath24 for the range @xmath25 ( fig .
1 ) ; we shall call this solution the _ quark soliton _ ( qs)-model with @xmath26 was first discussed by v. soni in the context of the maximum fermion mass @xcite . ] . for lower values of @xmath3 , the trivial solution @xmath27 , @xmath28 has a lower energy while for larger @xmath3 , the valence orbital already drops into the dirac sea . in the constituent quark models the quark mass is usually taken between 310 mev and 390 mev ; from fig .
1 one might naively expect to reproduce these values for @xmath29 .
however , the model possesses another nontrivial approximate solution representing the nucleon the _ nucleon soliton _ ( ns ) in which the three current quarks are surrounded by a common pion and @xmath0-meson cloud @xcite and here the experimental nucleon mass can be reproduced for @xmath3 below @xmath30 .
since the soliton energy decreases with @xmath3 higher values are not physically admissible for the qs either . for @xmath31
the energy of the soliton is rather independent of @xmath3 and is around 550 mev which is considerably higher than a typical mass of the constituent quark .
this may not be a too serious shortcoming of the model for the following two reasons : ( i ) as we shall show in the following , there exist a strong attractive interaction between the qss due to meson exchange which considerably lowers the energy of the three - qs system , ( ii ) the present solution is strongly localized and the energy of the soliton contains a large contribution of the kinetic energy of the spurious center - of - mass motion .
both effects allow the existence of a solitonic solution also for @xmath3 below 5.7 ; we may even expect that the physically sensible solution should be sought below 5 . in order to investigate point ( ii ) and to be able to obtain a stable one - quark soliton solution in the physically relevant range of @xmath3 it would be important to perform , in addition to the angular momentum , also the linear momentum projection .
techniques for linear momentum projection were developed in soliton models @xcite and for combined linear and angular momentum projection @xcite .
the resulting lowering of the soliton energy is rather strong , in the case of the qs even exaggerated .
we have found out that the strong effect is primarily due to the assumption that the negative energy dirac sea orbitals are plane waves , while in fact they are distorted by the soliton .
this distortion diminishes the effect of projection , especially since some negative energy states become localized . in order to get reliable center - of - mass corrections ,
dirac sea polarization should be implemented .
this work is in progress .
the main challenge of the model is to build hadrons from the quark solitons . in the following sections we shall present a simple model of the nucleon assuming that the qss preserve their identity in the baryon
this is a very strong assumption since it implies that the system of three interacting qss is energetically more favourable than the ns .
it would have to be investigated using more sophisticated approaches ( e.g. removal of center - of - mass motion of a single qs , solving the faddeev equations for the three quark system , @xmath32 ) which are far beyond the scope of this letter .
though we are not able in the present approach to generate stable solitons in the physically interesting range , it is still interesting to study their properties at somehow larger @xmath3 .
as we shall show in the following , several properties depend only weakly on the parameter @xmath3 and we may expect that they will , at least qualitatively , persist when more sophisticated approaches are used .
in table 1 we summarize the static properties of the qs .
the average number of pions ( @xmath33 ) is not a physical observable , nonetheless , it is an important tool in analyzing the flavour and spin content of the nucleon .
the violation of the gottfried sum rule , the drell - yan process and the deep inelastic lepton - nucleon scattering measuring various quark contributions to the nucleon spin can well be understood assuming the quarks are surrounded by a pion cloud @xcite .
the parameter of this analysis , @xmath34 , representing the probability for the @xmath35 pion in the cloud , would lead to @xmath36 if two- and more pion configurations are neglected .
since the determination of this parameter is model dependent and still very inconclusive , our values in table 1 are consistent with @xmath37 from ref .
@xcite and @xmath38 from ref .
@xcite .
.static properties of the constituent quark . for comparison ,
the nucleon properties deduced from the three - qs system ( @xmath39 ) , those calculated for the nucleon soliton ( ns ) , and the experimental values are given .
the notation @xmath40 ( @xmath41 ) refer to u ( d ) constituent quark or to proton ( neutron ) , respectively .
[ cols="^ , > , > , > , > , > , > " , ] the value for the constituent quark - pion coupling constant @xmath42 leads to a good agreement of @xmath43 with the experimental value .
the contribution of the meson self - interaction is significant ( @xmath44 % ) .
the axial current acquires an important contribution from the @xmath0 and the pion field , and using a simple relation @xmath45 leads to too large a value of @xmath46 for the nucleon .
the meson fields of constituent quarks may considerably alter in the nucleon giving rise to effective exchange ( axial ) currents which may eventually reduce this value .
the magnetic moments are too small , yet not too far from the corresponding dirac values ( e.g. @xmath47 ) .
we can therefore expect that reducing the mass of the qs in an improved computational scheme would increase the proton and the neutron magnetic moments , @xmath48 , to more acceptable values .
similarly as for @xmath46 , the effective exchange currents may yield an important contribution also in this case .
the calculated charge radii pertain to the size of the constituent quark . to obtain the nucleon charge radius we have taken into account the orbital motion of the constituent
quarks in the nucleon as discussed in the next section .
the nucleon charge radius would be considerably increased if the vector mesons were introduced ; our values are therefore not inconsistent with the experiment .
we have calculated the chiral meson exchange effective potentials between constituent quarks using the born oppenheimer approximation . while the assumption that the qs behaves as a static source of mesons can be justified in the case of the pion because of its small mass compared to that of the qs , it is certainly not appropriate for the @xmath0-meson . lacking a more consistent approach in which recoil effects are taken into account , the resulting @xmath0-exchange potential should be regarded only as a crude approximation .
we give here explicit expressions for the contributions arising from the quark - meson interaction in ( [ lagrangian ] ) which dominate . the meson self - interaction ( [ potential ] )
yields the pion and @xmath0-exchange potentials of similar shapes as the corresponding ones arising from the quark - meson interaction ; they merely increase the strength of the total potentials by some 20 % .
the quark contribution to the static @xmath0-exchange potential is simply : @xmath49 where the index @xmath34 refers to the qs positioned at the origin and @xmath50 to the qs positioned at @xmath51 , @xmath52 ( @xmath53 ) is the quark bispinor upper ( lower ) component , and @xmath54 is the angle between @xmath51 and @xmath55 . the potential , including also the contribution from ( [ potential ] ) , is attractive for all @xmath56 .
its value at the origin is @xmath57 mev and has a harmonic shape with @xmath58 gev/@xmath59 at small @xmath56 .
asymptotically it has a typical yukawa behaviour with the mass @xmath60 ( rather than @xmath61 ) as a consequence of remaining close to the chiral circle . in order to write down the pion exchange potential we first notice that the pion field of the qs can be written in the form : @xmath62 where @xmath63 and @xmath64 are , respectively , the spin and isospin operators acting on dressed ( constituent ) quarks .
similarly , the quark - pion interaction term in ( [ lagrangian ] ) can be manipulated : @xmath65 yielding the quark contribution to the pion exchange potential : @xmath66 where @xmath67 , and @xmath68 and @xmath69 now act on current quarks .
the potential ( [ pionex ] ) contains the scalar as well the tensor part . the scalar part of the pion exchange potential including also the contribution from the meson self - interaction ( [ potential ] ) can be written in the form @xmath70 ; @xmath71 is shown in fig . 2 and compared to some commonly used potentials in the constituent quark model calculations of the baryon spectra @xcite .
we reproduce well its asymptotic behaviour which reflects the fact that the pion coupling constant is also well reproduced in the model while the internal part has a too large range .
it could be reduced by going to stronger @xmath3 but this is physically not desirable .
the large range is partially due to spurious center - of - mass motion so we may expect that its size would be reduced by removing this effect .
since our calculation gives just a crude picture of the potentials , we have made only a simple estimate of the size of the nucleon and the @xmath72(1232 ) by assuming harmonic oscillator wave functions @xmath73 to describe the orbital motion of each qs and determined the parameter @xmath50 variationally .
it is interesting to notice that both baryons are bound even without including the confining potential .
using @xmath74 the rms radius of the orbital motion is @xmath75 fm for the nucleon and @xmath76 fm for the @xmath72 ; the energy difference between the nucleon and the @xmath72 is only 43 mev . including a linear confining potential @xmath77 and using a popular value @xmath78 gev / fm , reduces the rms radii to @xmath79 fm and @xmath80 fm ; while the energy difference increases to 125 mev . using the values for
the internal sizes of the u and d quark the charge radii of the proton and the neutron are estimated and displayed in table 1 .
let us stress that the strength of the attractive part of @xmath81 is almost entirely determined by the condition @xmath82 which holds for any pseudoscalar meson exchange effective interaction , and through this condition to the value of the quark - pion coupling constant which is , in turn , constrained by the experimental value of @xmath83 .
by shrinking the width of the attractive part ( see e.g. fig .
2 ) , its strength can be increased but not to the extent to bring the @xmath72-n splitting up to the experimental value .
we have shown that the linear @xmath0-model is a suitable framework for the microscopic description of constituent quarks .
it requires no new model parameters and gives sensible predictions about the effective interaction between constituent quarks as well as about different static properties of the baryon .
while the present results are still based on a rather simplified approach designed to explore the trends , their relative success encourages the use of more sophisticated techniques .
the pion - exchange effective potential between constituent quarks has a similar shape as the phenomenological potential used in ref @xcite . since
a strong attractive part has been shown @xcite to be needed to give simultaneously a good n-@xmath72 splitting as well as the correct ordering of the roper and negative parity states we anticipate that some important mechanisms are still to be included in order to model this type of interaction .
possible candidates are : removal of center - of - mass motion of each constituent quark , nonadiabadic corrections , three - body forces between constituent quarks , inclusion of @xmath84 mesons in the model .
we get a rather strong @xmath0-exchange potential .
its role in the linear @xmath0-model may be also to mimic a part of the confining potential at low energies as well as to strongly reduce the energy of the system so that the `` effective constituent mass '' of quarks in hadrons is considerably less than the calculated mass for an isolated constituent quark .
we are pleased to acknowledge stimulating and helpful discussions with l. ya .
glozman , w. plessas , d. o. riska as well as with m. banerjee and t. cohen .
we are particularly indebted to m. fiolhais for his explanation of the method of linear momentum projection .
a. manohar and h. georgi , nucl .
b234 ( 1984 ) 189 .
t. p. cheng and ling - fong li , phys .
74 ( 1995 ) 2872 .
s. baumgartner , h. j. pirner , k. knigsmann and b. povh , z. phys .
a 353 ( 1996 ) 397 .
glozman , d. o. riska , phys .
reports 286 ( 1996 ) 263 .
a. valcarce , p. gonzlez , f. fernndez , v. vento , phys .
lett b 367 ( 1996 ) 35 ; d. r. entem , a.i .
machavarini , a. valcarce , a. j. buchmann , a. faessler , f. fernndez , nucl .
a602 ( 1996 ) 308 . z. dziembowski , m. fabre de la ripelle , and g. a. miller , preprint nucl - th/9601022 . l. ya .
glozman , z. papp , w. plessas , phys .
b 381 ( 1996 ) 311 .
y. fujiwara , c. nakamoto , and y. suzuki , phys .
76 ( 1996 ) 2242 , kuns-1395 ( 1996 ) , kyoto university
. m. gell - mann and m. lvy , nuovo cim .
* 16 * ( 1960 ) 705 .
m. c. birse and m. k. banerjee , phys .
b 136 ( 1984 ) 284 ; phys . rev .
d 31 ( 1985 ) 118 .
s. kahana , g. ripka and v. soni , nucl .
phys . a 415 ( 1984 ) 351 .
n. a. trnqvist and m. roos , phys .
76 ( 1996 ) 1575 .
b. golli and m. rosina , phys .
b 165 ( 1985 ) 347 . m. c. birse , phys .
d 33 ( 1986 ) 1934 .
m. ibej , m. fiolhais , b. golli and m. rosina , j. phys . g. 18 ( 1992 ) 49 .
u. asher et al . , a. c. m. trans .
software 7 ( 1981 ) 209 .
l. amoreira , m. fiolhais , b. golli and m. rosina , j. phys .
g. 21 ( 1995 ) 1657 . v. soni , orsay preprint ipno / th86 - 85 , unpublished .
e. g. lbeck , m. c. birse , e. m. henley , and l. wilets , phys .
d 33 ( 1986 ) 234 . t. neuber and k. goeke , phys .
b 281 ( 1992 ) 202 . | we study a stable , soliton - like solution in the linear @xmath0-model where the chiral fields are coupled to a single massless nonstrange quark .
we investigate a possibility that this solution represents the constituent quark of georgi and manohar .
we show that its properties are indeed consistent with nucleon observables .
furthermore , we derive chiral meson exchange potentials between such objects which are similar to recently used potentials in the constituent quark models . and _
pacs : _
12.39.fe , 14.65.bt |
understanding the nature of the ultimate power source in ultraluminous infrared galaxies ( ulirgs ) has been a primary driver for numerous scientific investigations ever since the discovery in the 1980 s that ulirgs represent the most luminous class of galaxies in the local universe . with total power outputs comparable to those of quasars , characterizing the source of power
is of fundamental importance in understanding the composition and evolution of normal galaxies .
ulirgs are found to share the common trait that virtually all have recently or are currently experiencing major encounters with other galaxies @xcite . encounters like these may play a significant role in the process of building normal galaxies . with the discovery of faint sources at sub - millimeter wavelengths having characteristic spectral energy distributions like that typically found in ulirgs ( e.g. , arp 220 ) and redshifts @xmath7 , the importance of understanding these systems and their role in the evolution of normal galaxies has been heightened .
the present survey was designed to look for near - infrared spectral signatures of obscured active galactic nuclei ( agn ) in ulirgs , either by the presence of velocity - broadened pa@xmath2 emission or by the appearance of the high excitation [ coronal line .
a summary of the conclusions of this survey , along with a median ulirg near - infrared spectrum , are presented in @xcite .
only two of the 33 sample galaxies show clear indications for agn activity , with the rest appearing to be dominated by star formation processes .
either the agn phenomenon is relatively rare among ulirgs , or the extinction at 2 @xmath1 m is still too high to permit a proper assessment . mid infrared studies ( e.g. , * ? ? ?
* ; * ? ? ?
* ) favor the former conclusion , though it is found in this study that the extinction at 2 @xmath1 m may often be greater than four magnitudes , resulting in over one magnitude of extinction at mid - infrared wavelengths .
this paper presents the individual nuclear spectra for the sample galaxies , for both the primary and secondary nuclei where possible .
two - dimensional spectra are also presented to give a qualitative view of the line emission distributions typically found in ulirgs .
some discussion of extinction measures , molecular hydrogen excitation mechanisms , and a summary of typical line ratios found in ulirgs is included , as is a discussion of the velocity differences found among double nucleus systems .
the sample of 33 ulirgs in this survey originates from the 2jy sample of _ iras _ ( infrared astronomical satellite ) galaxies defined by @xcite .
the galaxies in the present sample are chosen to meet the following criteria : 1 . 60 @xmath1 m flux density , @xmath8(60 @xmath1 m ) @xmath9 1.94 jy 2 .
moderate or high quality iras flux measurements at 60 and 100 @xmath1 m ( as defined in the * ? ? ?
* ) 3 . @xmath10(60 @xmath1m)@xmath11(12 @xmath1m)@xmath12(25 @xmath1 m ) , where @xmath8(@xmath13 ) is the _ iras _ flux density 4 .
far - infrared luminosity @xmath14 5 . absolute galactic latitude , @xmath15 6 .
declination , @xmath16 7 .
redshift , @xmath17 the first six conditions are those used to define the northern 2 jy ulirg sample in @xcite , such that the current sample is a subset of this work .
the seventh condition places both the pa@xmath2 and br@xmath3 lines in the @xmath18 band atmospheric window .
the far - infrared luminosity is computed from the iras flux densities by the following conversion:@xmath19 l_{\odot } , \ ] ] @xcite where @xmath20 represents the luminosity distance to the source in mpc . due to the redshift constraint
, the current sample approximately represents a complete volume - limited sample , because even the galaxies with the minimum 60 @xmath1 m flux density are luminous enough to meet the sample criteria at all allowed redshifts . at the outer redshift limit of the survey , corresponding to a luminosity distance of 455 mpc using @xmath21 km s@xmath6 mpc@xmath6 and @xmath22 ( as is assumed throughout ) , the minimum 60 @xmath1 m flux density of 1.94 jy by itself accounts for @xmath23 of the total far - infrared luminosity . to make up the remaining @xmath24 necessary for sample inclusion , the 100 @xmath1 m flux density needs only be greater than 1.25 jy , implying a maximum @xmath8(60 @xmath1m)@xmath25(100 @xmath1 m ) ratio of 1.55 .
ulirgs very rarely have 60 @xmath1 m to 100 @xmath1 m flux density ratios this high , such that the volume limit imposed by the upper redshift cutoff is more stringent than the flux density cutoff , resulting in a sample that is mostly volume - limited rather than flux - limited . taking the 64 ulirgs defined in the northern 2 jy sample of @xcite ,
one finds an average @xmath8(60 @xmath1m)@xmath25(100 @xmath1 m ) ratio of @xmath26 , with a median value of 0.86 and a maximum of 1.64 ( in iras 08572@xmath273915 ) , which is the only ratio exceeding the maximum of 1.55 compatible with the volume - limited case .
the ulirg from the northern 2 jy sample with the second highest ratio is iras 05246@xmath270103 , with @xmath8(60 @xmath1m)@xmath25(100 @xmath1m)@xmath28 . both of these galaxies
are contained in the present sample .
the redshift limit excludes many of the highest luminosity ulirgs , as these are most likely found at larger distances .
for instance , mrk 1014a quasar and a ulirg would be a member of the sample if not for the upper redshift limit . of the 35 ulirgs in the redshift survey of @xcite satisfying the above conditions , two are not included here for the following reasons .
the spectra of iras 21396@xmath273623 show this galaxy to be at a redshift of @xmath29 , rather than the previously reported @xmath30 .
iras 19297@xmath310406 was free of any discernible lines , which may be the result of improper pointing , producing the spectrum of a nearby star in the very crowded field .
therefore the entire sample contains 33 galaxies , and is very nearly complete in the volume - limited sense .
a list of the sample galaxies appears in table [ tab : sample ] .
lccc iras 00262@xmath324251 & 29205 & 12.08 & s + iras 01521@xmath325224 & 23959 & 11.95 & 5.5 + iras 04232@xmath321436 & 23640 & 11.99 & s + iras 05246@xmath320103 & 29105 & 12.05 & 6.0 + iras 08030@xmath325243 & 24946 & 11.97 & s + iras 08311@xmath332459 & 30167 & 12.40 & s + iras 08344@xmath325105 & 28999 & 11.94 & s + iras 08572@xmath323915 & 17491 & 12.08 & 5.6 + iras 09061@xmath331248 & 22014 & 11.97 & 5.6 + iras 09111@xmath331007 & 16247 & 11.98 & ( 37 ) + iras 09583@xmath324714 & 25748 & 11.98 & 25.5 + iras 10035@xmath324852 & 19449 & 11.93 & 9.8 + iras 10190@xmath321322 & 22898 & 11.98 & 4.0 + iras 10494@xmath324424 & 27671 & 12.15 & s + iras 11095@xmath330238 & 31936 & 12.20 & s + iras 12112@xmath320305 & 22009 & 12.27 & 2.9 + iras
14348@xmath331447 & 24798 & 12.28 & 3.3 + iras 14352@xmath331954 & 26938 & 11.95 & s + iras 14394@xmath325332 & 31401 & 12.04 & ( 28 ) + iras 15245@xmath321019 & 22629 & 11.96 & 2.6 + iras 15250@xmath323609 & 16515 & 11.99 & s + iras 15462@xmath330450 & 30060 & 12.16 & s + iras 16487@xmath325447 & 31110 & 12.12 & 3.1 + iras 17028@xmath325817 & 31805 & 12.11 & 13.0 + iras 18470@xmath323233 & 23520 & 12.02 & ( 7.0 ) + iras 19458@xmath320944 & 29983 & 12.36 & s + iras 20046@xmath330623 & 25286 & 12.02 & s + iras 20087@xmath330308 & 31613 & 12.39 & s + iras 20414@xmath331651 & 26061 & 12.19 & s + iras 21504@xmath330628 & 23343 & 11.94 & s + iras 22491@xmath331808 & 23264 & 12.12 & 1.6 + iras 23327@xmath322913 & 32225 & 12.03 & 12.5 + iras 23365@xmath323604 & 19305 & 12.13 & s +
observations were conducted using the palomar longslit infrared spectrograph @xcite , operating on the 200-inch hale telescope and employing a 256@xmath34256 hgcdte ( nicmos 3 ) array .
observations were made over a time period beginning in 1995 july , and ending in 1997 december .
each object was observed in three grating settings , with each spanning @xmath35 @xmath1 m at a scale of @xmath36 @xmath1 m pixel@xmath6 .
typical slit widths of @xmath373.75 pixels , or 0625 , resulted in spectra with resolutions of @xmath38 , corresponding to @xmath37300 km s@xmath6 .
the grating settings were chosen to cover the pa@xmath2 line at 1.8751 @xmath1 m , the suite of br@xmath39 , h@xmath40 10 s(3 ) , and [ lines centered at 1.954 @xmath1 m , and the h@xmath40 10 s(1 ) and br@xmath3 lines centered at 2.14 @xmath1 m .
the slit was generally rotated such that spectra of both primary and secondary nuclei were obtained simultaneously when possible , or otherwise to coincide with the major axis of the galaxy , where evident .
observational details are provided for each source in table [ tab : obs ] .
individual exposure times were 300 s , with total integration times generally around 1800 s. in all cases , the galaxy was dithered between two positions on the slit , generally @xmath41 apart , with a smaller scale dither pattern employed about these points to eliminate the effects of static bad pixels . in this way
, the sky integrations were obtained simultaneously , with the effective sky position alternating by @xmath42 to either side .
wavelength calibration was provided either by the atmospheric oh airglow spectrum , or by arc lamp spectra for the h@xmath40 10 s(1)@xmath27br@xmath3 grating settings , where the oh lines become unavailable .
wavelengths reported here are in air units , with the oh wavelength data coming from @xcite .
atmospheric transmission variations as a function of wavelength were compensated via observations of the nearly featureless spectra of g dwarf stars at similar airmass , and similar telescope pointing , when possible .
the g star observations were performed either immediately prior to or following each galaxy observation .
stellar types and airmasses of the calibrators are listed for each object in table [ tab : obs ] .
the light from the calibrator star was made to uniformly fill the slit aperture by chopping the telescope secondary mirror in a triangle - wave pattern .
as such , the atmospheric calibrators perform the dual function of operating as the flat field for the spectral observations .
details about the treatment of the spectral calibration appear below . the calibration star observation for iras 12112@xmath270305
was compromised beyond repair , so the spectra for this galaxy were divided by an atmospheric template .
each spectrum is reduced pair - wise in the following manner .
pairs containing spectra at the two opposite slit positions are subtracted , with cosmic rays and static bad pixels interpolated .
the data are divided by the g star spectrum and multiplied by a planck function , described in more detail below .
a wavelength reference image is produced using either a median combination of the data frames to produce an oh airglow spectrum , or by using a combination of noble gas arc lamp spectra .
the wavelength calibration image is spatially and spectrally rectified onto a rectilinear grid using predetermined maps of the distortion in each of these directions .
a cubic function of fixed cubic coefficient and zero quadratic coefficient ( determined empirically to fit the distortion well ) is fit to the wavelength reference , providing a mapping from pixel coordinates to linear wavelength coordinates .
this mapping is combined with the predetermined distortion maps , and then applied in a single interpolation to the data frames .
once rectified , the pairs are combined into a single spectrum , using gaussian centroids for the spatial registration , or known pixel offsets for galaxies having diffuse or faint continua .
this process removes any residual sky subtraction signal , as one effectively subtracts the simultaneous sky at the alternate slit position in the process .
the noise at each pixel in the two - dimensional spectrum is computed along with the spectral data by quadrature combination of read noise and poisson shot noise based on signal level for each pixel .
pixel replacement , atmospheric calibrator division , and interpolation in the data frames are also properly treated in the noise frames throughout the process .
the same extractions applied to the spectral data are performed on the noise image , producing a one - dimensional noise array along with the extracted spectrum .
the seeing at the time of observation was assessed by taking short guided exposures
typically 10 s of nearby field stars through the wide - open slit of the spectrograph , with the spectrograph operating in imaging mode . in this way
, the point spread function ( psf ) of the telescope - plus - atmosphere during the spectral observations could be estimated .
these psf exposures were typically taken 34 times during an observation , usually at the beginnings and ends of the 3040 minute intervals spent at each grating setting .
the full - width at half - maximum ( fwhm ) along the slit direction was measured for each psf image , and the nuclear extractions of the spectra were based on this measurement . in each case , the psf fwhm was rounded up to an integral number of pixels , and extractions matching these integer widths were performed .
the centers of these apertures were not constrained , however , such that fractional pixels were summed in the production of the nuclear spectra .
the widths of the extractions , in arcseconds , are listed in table [ tab : obs ] .
lrcccccc iras 00262@xmath324251 & pa@xmath43 : & 3 jan 1996 & 1800 & @xmath44 & g1.5v & 1.02/1.05 & 1.5 + & [ si vi ] : & & 1800 & & & 1.11/1.04 & 1.7 + & h@xmath45br@xmath46 : & & 1800 & & & 1.05/1.05 & 1.5 + iras 01521@xmath325224 & pa@xmath43 : & 2 jan 1996 & 1800 & @xmath47 & g5vb & 1.06/1.09 & 2.0 + & [ si vi ] : & & 1800 & & & 1.06/1.08 & 1.8 + & h@xmath45br@xmath46 : & & 1800 & & & 1.07/1.08 & 1.7 + iras 04232@xmath321436 & pa@xmath43 : & 26 nov 1996 & 1800 & @xmath48 & g0v & 1.07/1.04 & 0.7 + & [ si vi ] : & & 1800 & & & 1.07/1.06 & 0.8 + & h@xmath45br@xmath46 : & & 1800 & & & 1.06/1.06 & 0.7 + iras 05246@xmath320103 & pa@xmath43 : & 5 dec 1996 & 1200 & @xmath49 & g1v & 1.19/1.18 & 0.8 + & [ si vi ] : & & 1800 & & & 1.21/1.21 & 0.8 + & h@xmath45br@xmath46 : & & 1800 & & & 1.18/1.17 & 0.7 + iras 08030@xmath325243 & pa@xmath43 : & 31 dec 1995 & 1800 & @xmath50 & g5v & 1.07/1.07 & 1.5 + & [ si vi ] : & & 1800 & & & 1.07/1.06 & 1.3 + & h@xmath45br@xmath46 : & & 1800 & & & 1.06/1.07 & 1.3 + iras 08311@xmath332459 & pa@xmath43 : & 8 apr 1996 & 1800 & @xmath44 & g3v & 1.91/1.99 & 1.7 + & [ si vi ] : & & 1800 & & & 1.92/1.82 & 1.7 + & h@xmath45br@xmath46 : & & 1200 & & & 1.94/1.93 & 1.0 + iras 08344@xmath325105 & pa@xmath43 : & 3 jan 1996 & 1800 & @xmath44 & g0v & 1.09/1.10 & 1.7 + & [ si vi ] : & & 1800 & & & 1.05/1.09 & 1.5 + & h@xmath45br@xmath46 : & & 1800 & & & 1.07/1.09 & 1.5 + iras 08572@xmath323915 & pa@xmath43 : & 25 nov 1996 & 2400 & @xmath51 & g0v & 1.01/1.01 & 0.8 + & [ si vi ] : & & 1200 & & & 1.03/1.02 & 0.7 + & h@xmath45br@xmath46 : & & 1800 & & & 1.01/1.01 & 0.7 + iras 09061@xmath331248 & pa@xmath43 : & 28 nov 1996 & 1800 & @xmath52 & g0v & 1.49/1.51 & 1.5 + & [ si vi ] : & & 1800 & & & 1.47/1.43 & 2.0 + & h@xmath45br@xmath46 : & & 1800 & & & 1.45/1.46 & 1.8 + iras 09111@xmath331007 & pa@xmath43 : & 11 feb 1997 & 1800 & @xmath53 & g2v & 1.39/1.45 & 1.7 + & [ si vi ] : & & 1200 & & & 1.42/1.39 & 1.5 + & h@xmath45br@xmath46 : & & 1800 & & & 1.39/1.41 & 1.7 + iras 09583@xmath324714 & pa@xmath43 : & 3 jan 1996 & 1800 & 565 & g0v & 1.03/1.03 & 1.5 + & [ si vi ] : & & 1800 & 565 & & 1.03/1.04 & 1.2 + & h@xmath45br@xmath46 : & 31 dec 1995 & 1800 & @xmath54 & & 1.04/1.03 & 0.8 + iras 10035@xmath324852 & pa@xmath43 : & 22 feb 1997 & 1200 & @xmath55 & g0v & 1.04/1.05 & 1.0 + & [ si vi ] : & 23 may 1997 & 1800 & & g1v & 1.21/1.15 & 0.8 + & h@xmath45br@xmath46 : & & 1800 & & f9v & 1.09/1.06 & 0.7 + iras 10190@xmath321322 & pa@xmath43 : & 22 may 1997 & 1800 & @xmath56 & g0v & 1.15/1.16 & 1.0 + & [ si vi ] : & & 1800 & & g1v & 1.26/1.23 & 1.2 + & h@xmath45br@xmath46 : & & 1800 & & g0v & 1.44/1.42 & 1.0 + iras 10494@xmath324424 & pa@xmath43 : & 22 feb 1997 & 1800 & @xmath57 & g2v & 1.03/1.02 & 1.0 + & [ si vi ] : & & 1200 & & g0vs & 1.15/1.10 & 0.8 + & h@xmath45br@xmath46 : & & 1800 & & & 1.09/1.10 & 0.7 + iras 11095@xmath330238 & pa@xmath43 : & 6 may 1996 & 1800 & @xmath58 & g1v & 1.24/1.26 & 1.0 + & [ si vi ] : & & 1800 & & & 1.29/1.31 & 1.0 + & h@xmath45br@xmath46 : & & 1800 & & & 1.25/1.29 & 0.8 + iras 12112@xmath320305 & pa@xmath43 : & 5 may 1996 & 1800 & @xmath59 & & 1.09/ & 1.0 + & [ si vi ] : & & 1800 & & & 1.05/ & 0.8 + & h@xmath45br@xmath46 : & & 1800 & & & 1.07/ & 1.0 + iras 14348@xmath331447 & pa@xmath43 : & 5 may 1996 & 1800 & @xmath60 & g1v & 1.51/1.47 & 1.0 + & [ si vi ] : & & 1800 & & g4v & 1.72/1.68 & 1.0 + & h@xmath45br@xmath46 : & & 1800 & & & 1.59/1.62 & 1.0 + iras 14352@xmath331954 & pa@xmath43 : & 6 may 1996 & 1800 & @xmath44 & g3v & 1.70/1.73 & 1.2 + & [ si vi ] : & & 1800 & & g0v & 1.71/1.79 & 0.8 + & h@xmath45br@xmath46 : & & 1800 & & g3v & 1.68/1.72 & 0.8 + iras 14394@xmath325332 & pa@xmath43 : & 5 may 1996 & 1800 & @xmath61 & g1v & 1.34/1.30 & 0.8 + & [ si vi ] : & 6 may 1996 & 1800 & & & 1.51/1.50 & 0.8 + & h@xmath45br@xmath46 : & & 1800 & & & 1.36/1.32 & 0.7 + iras 15245@xmath321019 & pa@xmath43 : & 22 may 1997 & 1800 & @xmath62 & g8v & 1.11/1.12 & 1.2 + & [ si vi ] : & & 1800 & & & 1.09/1.11 & 1.0 + & h@xmath45br@xmath46 : & & 1800 & & & 1.10/1.10 & 1.0 + iras 15250@xmath323609 & pa@xmath43 : & 21 may 1997 & 1800 & @xmath63 & g2v & 1.03/1.01 & 1.0 + & [ si vi ] : & & 2400 & & & 1.01/1.02 & 1.0 + & h@xmath45br@xmath46 : & & 1800 & & & 1.00/1.01 & 0.7 + iras 15462@xmath330450 & pa@xmath43 : & 7 apr 1996 & 1800 & @xmath64 & g2.5v & 1.32/1.35 & 1.0 + & [ si vi ] : & 8 apr 1996 & 1800 & & & 1.41/1.36 & 0.8 + & h@xmath45br@xmath46 : & 7 apr 1996 & 1800 & & & 1.39/1.37 & 1.0 + iras 16487@xmath325447 & pa@xmath43 : & 21 may 1997 & 1800 & @xmath65 & g5v & 1.08/1.11 & 0.7 + & [ si vi ] : & & 1800 & & & 1.11/1.12 & 0.8 + & h@xmath45br@xmath46 : & & 1800 & & g0va & 1.17/1.20 & 0.7 + iras 17028@xmath325817 & pa@xmath43 : & 22 may 1997 & 1800 & @xmath66 & g2v & 1.14/1.13 & 0.8 + & [ si vi ] : & & 1800 & & & 1.18/1.14 & 1.0 + & h@xmath45br@xmath46 : & & 1800 & & g0va & 1.24/1.26 & 0.7 + iras 18470@xmath323233 & pa@xmath43 : & 31 jul 1996 & 1800 & @xmath67 & g8v & 1.05/1.00 & 0.8 + & [ si vi ] : & & 1800 & & & 1.00/1.00 & 1.0 + & h@xmath45br@xmath46 : & & 1800 & & & 1.01/1.00 & 1.0 + iras 19458@xmath320944 & pa@xmath43 : & 2 aug 1996 & 1800 & @xmath68 & f5v+g0v & 1.09/1.09 & 0.7 + & [ si vi ] : & & 1200 & & & 1.09/1.12 & 0.7 + & h@xmath45br@xmath46 : & & 1800 & & & 1.09/1.09 & 0.7 + iras 20046@xmath330623 & pa@xmath43 : & 12 aug 1997 & 1800 & @xmath69 & g2v & 1.37/1.40 & 1.0 + & [ si vi ] : & & 1200 & & g1v & 1.31/1.27 & 1.0 + & h@xmath45br@xmath46 : & & 1200 & & & 1.31/1.27 & 1.0 + iras 20087@xmath330308 & pa@xmath43 : & 11 aug 1997 & 1800 & @xmath70 & g1v & 1.36/1.35 & 1.0 + & [ si vi ] : & & 1800 & & g0v & 1.51/1.43 & 0.8 + & h@xmath45br@xmath46 : & 12 aug 1997 & 1800 & & g1v & 1.39/1.43 & 1.3 + iras 20414@xmath331651 & pa@xmath43 : & 1 aug 1996 & 1800 & @xmath71 & g2v & 1.56/1.56 & 1.0 + & [ si vi ] : & & 1200 & & g1v & 1.71/1.70 & 0.8 + & h@xmath45br@xmath46 : & & 1800 & & g2v & 1.62/1.56 & 0.8 + iras 21504@xmath330628 & pa@xmath43 : & 3 aug 1996 & 1800 & @xmath51 & g0v & 1.30/1.33 & 0.8 + & [ si vi ] : & & 1200 & & g8v & 1.54/1.62 & 2.0 + & h@xmath45br@xmath46 : & & 1800 & & g0v & 1.34/1.33 & 0.7 + iras 22491@xmath331808 & pa@xmath43 : & 2 aug 1996 & 1800 & @xmath72 & g8v & 1.62/1.61 & 1.0 + & [ si vi ] : & & 1800 & & & 1.63/1.62 & 1.2 + & h@xmath45br@xmath46 : & & 1800 & & & 1.60/1.61 & 1.0 + iras 23327@xmath322913 & pa@xmath43 : & 11 aug 1997 & 1800 & @xmath73 & g1.5v & 1.00/1.01 & 1.0 + & [ si vi ] : & & 1200 & & & 1.03/1.01 & 0.7 + & h@xmath45br@xmath46 : & 10 jul 1995 & 1200 & & g0v & 1.00/1.00 & 0.8 + iras 23365@xmath323604 & pa@xmath43 : & 26 nov 1996 & 1800 & @xmath74 & g1.5v & 1.01/1.01 & 0.8 + & [ si vi ] : & 28 nov 1996 & 1800 & & g2v & 1.00/1.01 & 2.0 + & h@xmath45br@xmath46 : & 26 nov 1996 & 1800 & & g1.5v & 1.03/1.02 & 0.8 + the most significant stellar absorption feature present in the 2 @xmath1 m window for the atmospheric calibrator stars is the br@xmath3 line at 2.1655 @xmath1 m . while a template stellar spectrum may be adequate for removal of this line , we instead interpolated over the line in the two - dimensional g star calibration / flat - field spectrum before dividing the galaxy data by the calibrator data .
multiplying the corrected galaxy spectrum by a planck blackbody function with a temperature matching that of the calibrator rectifies the spectrum to a proper representation in flux density . to remove the effects of the weaker absorption lines in the atmospheric calibrator ,
the galaxy spectrum was multiplied by a normalized solar spectrum , smoothed and re - binned to the same resolution . for the long wavelength grating setting covering the h@xmath40 10 s(1 ) and br@xmath3 lines ,
the co band structure becomes significant when using late - type g stars for the atmospheric calibration .
when the atmospheric calibrator was of type g4g8 , a linear combination of g3 v and k0 v template spectra from @xcite was used instead of the solar spectrum . again
, this template was smoothed and re - binned to match the individual galaxy spectra .
the chief results from this spectroscopic survey that agn appear to be rare among ulirgs has been reported previously in @xcite . in the present paper
we focus attention on the range of observed properties rather than on the collected sample as a whole .
spectral extractions for the 33 primary ulirg nuclei . ]
figure [ fig : primary ] presents spectral extractions for the primary nuclei in the 33 sample galaxies , with aperture sizes indicated in table [ tab : obs ] .
each plot is fixed in the rest frame , with the observed wavelength scale indicated at bottom .
annotations mark the locations of every common emission line , though in any individual spectrum only the lines with reasonable detections are labeled .
the location of the tick marks is based on the observed redshift of the pa@xmath2 line , as is the rest wavelength scale at top . each spectrum is composed of three independent spectral datasets . for each of the three spectral ranges , a best linear fit to the line - free continuum
is assessed for the purpose of determining an overall average slope for the spectrum .
the individual spectra are then scaled to lie on a common linear `` backbone '' defined by the average slope , and combined in a noise - weighted manner across the overlap region around 1.92 @xmath1 m .
the vertical scale is normalized such that the continuum is at one unit of flux density at 2.155 @xmath1 m , corresponding to the middle of the @xmath75 bandpass .
spectral extractions for the 14 secondary ulirg nuclei . ]
figure [ fig : secondary ] contains spectra of the secondary nuclei in the 14 sample galaxies with a nearby galactic companion , employing the same wavelength scale used for figure [ fig : primary ] .
annotations and scalings are the same as for figure [ fig : primary ] , though some of the weaker lines are not marked by default in this set .
the tick marks denoting line positions are based on the redshift of the pa@xmath2 line in the _ primary _ nucleus , so that velocity displacements between the two nuclei are more easily seen .
the numerical flux density scale for each is identical to that for the corresponding spectrum in figure [ fig : primary ] , allowing direct comparison of the relative levels for each .
the relative scaling of the three individual spectra determined for figure [ fig : primary ] are used in figure [ fig : secondary ] as well .
the choice of primary versus secondary nucleus is based solely on the continuum level , with the brighter nucleus designated as the primary .
the only exception is iras 22491@xmath311808 , for which the primary nucleus is identified with the dominant line - emitting , yet weaker continuum nucleus to the east .
the western nucleus is 30% brighter , but lacks the strong h@xmath40 emission that is associated with _
every _ other primary ulirg nucleus .
comparison of figures [ fig : primary ] and [ fig : secondary ] indicate that the primary nuclei , besides being the brightest continuum sources , also exhibit apparently stronger line emission . yet despite the weaker line strengths , the secondary nuclei often do show appreciable line emission activity , appearing relatively similar to the primary nuclei . if the observed pa@xmath2 line strength is proportional to star formation rates and therefore total luminosity then even though the dominant luminosity can be associated with one nucleus , the lesser nucleus still contributes a substantial fraction of the total luminosity .
it must be realized , however , that the high levels of extinction present in ulirgs could easily lead one to misjudge the identity of the primary source of infrared luminosity based simply on apparent line strengths .
table [ tab : mofoprim ] presents the line measurements from the primary nuclear data shown in figure [ fig : primary ] and table [ tab : mofosec ] presents the same data for the secondary nuclear spectra of figure [ fig : secondary ] . for each galaxy , the continuum slope and measured pa@xmath2 fwhm
are given , as well as equivalent width , fwhm , and velocity centroid of each significant line .
in this context , weaker hydrogen recombination and molecular hydrogen vibration - rotation lines are excluded from the table as their properties follow from those of the stronger lines .
the continuum slope , @xmath2 , is reported as the power - law index , where @xmath76 . the equivalent linear slope , with @xmath77 $ ] , where @xmath78 m , is @xmath79 .
the measured line properties reported in tables [ tab : mofoprim ] and [ tab : mofosec ] are obtained by the procedures detailed in appendix [ procedures ] .
lcclccccccc + iras 00262@xmath324251 & @xmath80 & 331 & & & & & & & & + & & & ew & @xmath810.1 & @xmath820.15 & @xmath830.14 & & @xmath840.088 & @xmath850.24 & + & & & fwhm & 420@xmath86240 & 366@xmath8627 & 480@xmath8642 & & 260@xmath86160 & 380@xmath8656 & + & & & @xmath87 & @xmath8878 & @xmath8918 & @xmath9029 & & @xmath9163 & @xmath9227 & + iras 01521@xmath325224s & @xmath93 & 394 & & & & & & & & + & & & ew & & @xmath940.44 & @xmath950.14 & & & @xmath960.36 & @xmath970.49 + & & & fwhm & & 356@xmath8623 & 170@xmath8696 & & & 290@xmath86120 & 0@xmath86110 + & & & @xmath87 & & @xmath9816 & @xmath9946 & & & @xmath10034 & @xmath10176 + iras 04232@xmath321436 & @xmath102 & 183 & & & & & & & & + & & & ew & & @xmath1030.25 & @xmath1040.15 & & & @xmath1050.16 & @xmath1060.26 + & & & fwhm & & 218@xmath8618 & 195@xmath8670 & & & 203@xmath8651 & 220@xmath86100 + & & & @xmath87 & & @xmath10715 & @xmath10833 & & & @xmath10926 & @xmath11046 + iras 05246@xmath320103e & @xmath111 & 110 & & & & & & & & + & & & ew & @xmath1120.13 & @xmath1130.19 & @xmath1140.26 & @xmath1150.2 & & @xmath1160.14 & @xmath1170.24 + & & & fwhm & 497@xmath8678 & [email protected] & 575@xmath8677 & 310@xmath86260 & & 161@xmath8640 & 210@xmath86210 + & & & @xmath87 & @xmath11851 & @xmath11922 & @xmath12047 & @xmath12155 & & @xmath12238 & @xmath123110 + iras 08030@xmath325243 & @xmath124 & 322 & & & & & & & & + & & & ew & @xmath1250.13 & @xmath1260.16 & @xmath1270.11 & & @xmath1280.1 & @xmath1290.22 & @xmath1300.28 + & & & fwhm & 340@xmath86160 & [email protected] & 331@xmath8677 & & 0@xmath8691 & 270@xmath86130 & 380@xmath86110 + & & & @xmath87 & @xmath13171 & @xmath13212 & @xmath13334 & & @xmath13487 & @xmath13531 & @xmath13644 + iras 08311@xmath332459 & @xmath137 & 314 & & & & & & & & + & & & ew & & @xmath1380.29 & @xmath1390.15 & @xmath1400.088 & @xmath1410.078 & @xmath1420.092 & @xmath1430.25 + & & & fwhm & & 477@xmath8610 & 501@xmath8641 & 708@xmath8652 & 360@xmath86130 & 351@xmath8627 & 470@xmath86170 + & & & @xmath87 & & @xmath14416 & @xmath14526 & @xmath14620 & @xmath14745 & @xmath14821 & @xmath14934 + iras 08344@xmath325105 & @xmath150 & 287 & & & & & & & & + & & & ew & @xmath1510.13 & @xmath1520.26 & @xmath1530.13 & & & @xmath1540.27 & @xmath1300.36 + & & & fwhm & 60@xmath86110 & 300@xmath8613 & 196@xmath8640 & & & 310@xmath86100 & 310@xmath86140 + & & & @xmath87 & @xmath15539 & @xmath15612 & @xmath15728 & & & @xmath15830 & @xmath15935 + iras 08572@xmath323915n & @xmath160 & 340 & & & & & & & & + & & & ew & @xmath1610.13 & @xmath1620.1 & @xmath1630.062 & & & @xmath1640.027 & @xmath1650.072 + & & & fwhm & 440@xmath86110 & 334.8@xmath868 & 139@xmath8645 & & & 135@xmath8630 & 271@xmath8662 + & & & @xmath87 & @xmath16653 & @xmath16719 & @xmath16837 & & & @xmath16931 & @xmath17037 + iras 09061@xmath331248s & @xmath171 & 371 & & & & & & & & + & & & ew & @xmath1720.5 & @xmath1730.25 & @xmath1740.092 & & @xmath1750.19 & @xmath1160.13 & @xmath1760.23 + & & & fwhm & 240@xmath86280 & 320@xmath8624 & 421@xmath8699 & & 100@xmath86240 & 266@xmath8658 & 260@xmath86110 + & & & @xmath87 & @xmath177120 & @xmath17813 & @xmath17926 & & @xmath18096 & @xmath18123 & @xmath18245 + iras 09111@xmath331007 & @xmath183 & 223 & & & & & & & & + & & & ew & @xmath810.11 & @xmath1840.1 & @xmath1850.033 & & @xmath1860.052 & @xmath1870.035 & @xmath1880.062 + & & & fwhm & 290@xmath86210 & 246@xmath8613 & 189@xmath8627 & & 310@xmath8691 & 164@xmath8625 & 168@xmath8666 + & & & @xmath87 & @xmath18999 & @xmath19012 & @xmath19117 & & @xmath19251 & @xmath14620 & @xmath19326 + iras 09583@xmath324714ne & @xmath194 & 247 & & & & & & & & + & & & ew & @xmath1950.11 & @xmath1960.2 & @xmath1970.22 & & & @xmath1980.25 & @xmath1990.25 + & & & fwhm & 0@xmath8663 & [email protected] & 411@xmath8674 & & & 384@xmath8659 & 256@xmath8698 + & & & @xmath87 & @xmath16924 & @xmath20012 & @xmath10825 & & & @xmath20128 & @xmath20238 + iras 10035@xmath324852ne & @xmath203 & 237 & & & & & & & & + & & & ew & @xmath2040.049 & @xmath2050.19 & @xmath2060.13 & & @xmath2070.11 & @xmath2080.089 & @xmath2090.15 + & & & fwhm & 0@xmath8661 & 254@xmath8612 & 315@xmath8643 & & 0@xmath86240 & 256@xmath8618 & 201@xmath8671 + & & & @xmath87 & @xmath21029 & @xmath21117 & @xmath21235 & & @xmath213170 & @xmath21428 & @xmath21544 + iras 10190@xmath321322e & @xmath216 & 445 & & & & & & & & + & & & ew & & @xmath2170.13 & @xmath2180.081 & & @xmath2190.072 & @xmath2200.096 & @xmath2210.11 + & & & fwhm & & 437@xmath8616 & 412@xmath8672 & & 0@xmath86130 & 336@xmath8665 & 320@xmath8691 + & & & @xmath87 & & @xmath22219 & @xmath22337 & & @xmath224140 & @xmath18232 & @xmath22534 + iras 10494@xmath324424 & @xmath226 & 430 & & & & & & & & + & & & ew & @xmath2270.11 & @xmath2280.12 & @xmath2290.072 & & & @xmath2300.11 & @xmath2310.16 + & & & fwhm & 510@xmath86160 & [email protected] & 377@xmath8638 & & & 427@xmath8665 & 364@xmath8675 + & & & @xmath87 & @xmath16634 & @xmath23216 & @xmath21429 & & & @xmath23329 & @xmath16631 + iras 11095@xmath330238 & @xmath234 & 269 & & & & & & & & + & & & ew & @xmath2350.3 & @xmath2360.39 & @xmath2370.25 & & & @xmath2380.61 & @xmath2391 + & & & fwhm & 360@xmath86180 & 286@xmath8612 & 350@xmath8649 & & & 241@xmath8643 & 0@xmath86100 + & & & @xmath87 & @xmath24064 & @xmath24118 & @xmath24226 & & & @xmath24335 & @xmath244130 + iras 12112@xmath320305sw & @xmath245 & 159 & & & & & & & & + & & & ew & @xmath2460.61 & @xmath2470.39 & @xmath2480.086 & & @xmath2490.14 & @xmath2500.071 & @xmath2510.15 + & & & fwhm & 380@xmath86340 & 303@xmath8621 & 295@xmath8663 & & 220@xmath86290 & 0@xmath8631 & 185@xmath8653 + & & & @xmath87 & @xmath244140 & @xmath25221 & @xmath25342 & & @xmath254190 & @xmath25530 & @xmath16633 + iras 14348@xmath331447sw & @xmath256 & 227 & & & & & & & & + & & & ew & @xmath1760.15 & @xmath2570.19 & @xmath1140.12 & & @xmath2580.083 & @xmath2590.14 & @xmath2600.2 + & & & fwhm & 550@xmath86120 & [email protected] & 450@xmath8635 & & 120@xmath86210 & 334@xmath8636 & 127@xmath8680 + & & & @xmath87 & @xmath26143 & @xmath26222 & @xmath25335 & & @xmath263110 & @xmath26430 & @xmath26540 + iras 14352@xmath331954 & @xmath266 & 192 & & & & & & & & + & & & ew & @xmath2670.089 & @xmath2680.23 & @xmath2690.11 & & & @xmath2700.15 & @xmath2710.26 + & & & fwhm & 258@xmath8655 & [email protected] & 193@xmath8658 & & & 243@xmath8648 & 166@xmath8646 + & & & @xmath87 & @xmath27233 & @xmath27317 & @xmath27437 & & & @xmath26534 & @xmath27532 + iras 14394@xmath325332 & @xmath276 & 363 & & & & & & & & + & & & ew & @xmath2770.064 & @xmath2780.13 & @xmath2790.097 & & @xmath2800.08 & @xmath2810.16 & @xmath2820.27 + & & & fwhm & 558@xmath8677 & [email protected] & 490@xmath8622 & & 160@xmath86280 & 426@xmath8633 & 370@xmath86110 + & & & @xmath87 & @xmath28333 & @xmath28421 & @xmath28531 & & @xmath286170 & @xmath28733 & @xmath28847 + iras 15245@xmath321019e & @xmath289 & 283 & & & & & & & & + & & & ew & @xmath2900.069 & @xmath2910.14 & @xmath2920.067 & & @xmath2930.078 & @xmath2940.088 & @xmath2950.13 + & & & fwhm & 0@xmath8675 & [email protected] & 307@xmath8646 & & 300@xmath86210 & 247@xmath8650 & 304@xmath8651 + & & & @xmath87 & @xmath10839 & @xmath29616 & @xmath25530 & & @xmath29767 & @xmath10827 & @xmath29825 + iras 15250@xmath323609 & @xmath299 & 57 & & & & & & & & + & & & ew & @xmath3000.23 & @xmath3010.16 & @xmath3020.14 & & @xmath3030.1 & @xmath3040.059 & @xmath3050.077 + & & & fwhm & 0@xmath86180 & [email protected] & 323@xmath8688 & & 400@xmath86180 & 223@xmath8619 & 204@xmath8634 + & & & @xmath87 & @xmath306200 & @xmath30717 & @xmath27535 & & @xmath29860 & @xmath30832 & @xmath30933 + iras 15462@xmath330450 & @xmath310 & 285 & & & & & & & & + & & & ew & & @xmath3110.085 & @xmath3120.029 & & & @xmath3130.024 & @xmath3140.06 + & & & fwhm & & 535@xmath8615 & 297@xmath8653 & & & 102@xmath8643 & 360@xmath86110 + & & & @xmath87 & & @xmath31524 & @xmath31640 & & & @xmath31732 & @xmath31844 + iras 16487@xmath325447w & @xmath319 & 347 & & & & & & & & + & & & ew & @xmath3200.18 & @xmath3210.34 & @xmath3220.29 & & & @xmath3230.39 & @xmath3240.89 + & & & fwhm & 500@xmath86140 & [email protected] & 374@xmath8636 & & & 265@xmath8654 & 230@xmath86140 + & & & @xmath87 & @xmath32540 & @xmath32623 & @xmath32733 & & & @xmath20230 & @xmath32878 + iras 17028@xmath325817w & @xmath329 & 339 & & & & & & & & + & & & ew & @xmath3300.097 & @xmath3310.13 & @xmath3320.078 & & @xmath3330.13 & @xmath1160.14 & @xmath3340.25 + & & & fwhm & 320@xmath86130 & [email protected] & 250@xmath8648 & & 140@xmath86170 & 301@xmath8674 & 336@xmath8694 + & & & @xmath87 & @xmath21546 & @xmath33524 & @xmath33633 & & @xmath337130 & @xmath27539 & @xmath22545 + iras 18470@xmath323233 & @xmath338 & 228 & & & & & & & & + & & & ew & @xmath3390.4 & @xmath3400.44 & @xmath3410.16 & & @xmath3420.16 & @xmath2060.17 & @xmath820.25 + & & & fwhm & 200@xmath86120 & [email protected] & 305@xmath8668 & & 200@xmath86300 & 319@xmath8649 & 249@xmath8650 + & & & @xmath87 & @xmath34351 & @xmath34421 & @xmath34544 & & @xmath346120 & @xmath34534 & @xmath10030 + iras 19458@xmath320944 & @xmath347 & 269 & & & & & & & & + & & & ew & @xmath3480.086 & @xmath3490.075 & @xmath3500.13 & & & @xmath3510.089 & @xmath3520.29 + & & & fwhm & 390@xmath86180 & [email protected] & 324@xmath8684 & & & 237@xmath8650 & 150@xmath86110 + & & & @xmath87 & @xmath35358 & @xmath35435 & @xmath11859 & & & @xmath33647 & @xmath35553 + iras 20046@xmath330623 & @xmath356 & 244 & & & & & & & & + & & & ew & & @xmath3570.21 & @xmath1720.2 & & @xmath3580.2 & @xmath3590.44 & @xmath3601 + & & & fwhm & & 247@xmath8627 & 190@xmath86100 & & 0@xmath86140 & 0@xmath8688 & 310@xmath86170 + & & & @xmath87 & & @xmath36114 & @xmath36243 & & @xmath363140 & @xmath27560 & @xmath30980 + iras 20087@xmath330308 & @xmath364 & 595 & & & & & & & & + & & & ew & @xmath3650.093 & @xmath3660.14 & @xmath3670.067 & & & @xmath3680.15 & @xmath3690.24 + & & & fwhm & 270@xmath86150 & 574@xmath8623 & 475@xmath8649 & & & 570@xmath86110 & 280@xmath86280 + & & & @xmath87 & @xmath37048 & @xmath37120 & @xmath37233 & & & @xmath37330 & @xmath37497 + iras 20414@xmath331651 & @xmath375 & 515 & & & & & & & & + & & & ew & @xmath3760.16 & @xmath3770.21 & @xmath1530.16 & & & @xmath3780.17 & @xmath3790.63 + & & & fwhm & 400@xmath86190 & 488@xmath8620 & 287@xmath8682 & & & 460@xmath86120 & 170@xmath86180 + & & & @xmath87 & @xmath38071 & @xmath38125 & @xmath38249 & & & @xmath26546 & @xmath28392 + iras 21504@xmath330628 & @xmath383 & 256 & & & & & & & & + & & & ew & @xmath3840.14 & @xmath3850.14 & @xmath3860.14 & & & @xmath3870.099 & @xmath3520.13 + & & & fwhm & 230@xmath86100 & [email protected] & 250@xmath86130 & & & 310@xmath8643 & 144@xmath8640 + & & & @xmath87 & @xmath38852 & @xmath38926 & @xmath21544 & & & @xmath39043 & @xmath39141 + iras 22491@xmath331808e & @xmath392 & 128 & & & & & & & & + & & & ew & & @xmath3930.38 & @xmath3940.23 & & & @xmath3950.37 & @xmath3960.83 + & & & fwhm & & 131@xmath8618 & 0@xmath8652 & & & 302@xmath8660 & 430@xmath86350 + & & & @xmath87 & & @xmath39724 & @xmath39841 & & & @xmath21433 & @xmath39977 + iras 23327@xmath322913s & @xmath400 & 151 & & & & & & & & + & & & ew & @xmath3000.11 & @xmath4010.12 & @xmath3860.26 & & & @xmath4020.36 & + & & & fwhm & 170@xmath86130 & 193@xmath869 & 110@xmath86120 & & & 210@xmath86120 & + & & & @xmath87 & @xmath40376 & @xmath40418 & @xmath40577 & & & @xmath40651 & + iras 23365@xmath323604 & @xmath407 & 186 & & & & & & & & + & & & ew & & @xmath4080.083 & @xmath4090.079 & & & @xmath4100.057 & @xmath4110.094 + & & & fwhm & & 191.1@xmath865 & 245@xmath8645 & & & 219@xmath8622 & 216@xmath8639 + & & & @xmath87 & & @xmath41217 & @xmath15724 & & & @xmath41325 & @xmath34328 + lcclcccccc + iras 01521@xmath325224n & @xmath414 & 403 & & & & & & & + & & & ew & & @xmath2210.6 & & & & + & & & fwhm & & 190@xmath86230 & & & & + & & & @xmath87 & & @xmath415170 & & & & + iras 05246@xmath320103w & @xmath416 & 54 & & & & & & & + & & & ew & @xmath4170.16 & @xmath4180.31 & & & @xmath4190.29 & + & & & fwhm & 0@xmath86140 & 72@xmath8619 & & & 60@xmath86100 & + & & & @xmath87 & @xmath420120 & @xmath42123 & & & @xmath42277 & + iras 08572@xmath323915s & @xmath423 & & & & & & & & + & & & ew & & @xmath4240.21 & & & & + & & & fwhm & & 0@xmath8619 & & & & + & & & @xmath87 & & @xmath42522 & & & & + iras 09061@xmath331248n & @xmath426 & & & & & & & & + & & & ew & & @xmath4270.59 & @xmath1610.29 & & @xmath3340.39 & + & & & fwhm & & 252@xmath8676 & 0@xmath86170 & & 253@xmath8695 & + & & & @xmath87 & & @xmath42830 & @xmath420140 & & @xmath42961 & + iras 09583@xmath324714sw & @xmath430 & 186 & & & & & & & + & & & ew & & @xmath4310.16 & @xmath4190.19 & & & + & & & fwhm & & 246@xmath8641 & 320@xmath86180 & & & + & & & @xmath87 & & @xmath43231 & @xmath43376 & & & + iras 10035@xmath324852sw & @xmath434 & 424 & & & & & & & + & & & ew & @xmath1880.23 & @xmath1730.54 & @xmath4350.13 & & @xmath2700.14 & @xmath1290.17 + & & & fwhm & 50@xmath86160 & 430@xmath8631 & 320@xmath86110 & & 292@xmath8647 & 212@xmath8667 + & & & @xmath87 & @xmath420130 & @xmath43631 & @xmath43754 & & @xmath43842 & @xmath38848 + iras 10190@xmath321322w & @xmath439 & 265 & & & & & & & + & & & ew & & @xmath4400.14 & @xmath3420.092 & @xmath4410.11 & @xmath3690.13 & @xmath4420.18 + & & & fwhm & & 271@xmath8626 & 40@xmath86110 & 0@xmath86150 & 63@xmath8689 & 150@xmath86120 + & & & @xmath87 & & @xmath44321 & @xmath30664 & @xmath444170 & @xmath44556 & @xmath44668 + iras 12112@xmath320305ne & @xmath447 & 196 & & & & & & & + & & & ew & & @xmath4480.52 & @xmath4490.13 & & @xmath4500.2 & @xmath3600.39 + & & & fwhm & & 271@xmath8619 & 401@xmath8627 & & 366@xmath8649 & 278@xmath8678 + & & & @xmath87 & & @xmath45120 & @xmath45236 & & @xmath13530 & @xmath19136 + iras 14348@xmath331447ne & @xmath453 & 283 & & & & & & & + & & & ew & @xmath4540.3 & @xmath4550.35 & @xmath4560.15 & & @xmath2390.25 & @xmath1530.5 + & & & fwhm & 450@xmath86160 & 272@xmath8614 & 218@xmath8663 & & 274@xmath8676 & 0@xmath86120 + & & & @xmath87 & @xmath45764 & @xmath45823 & @xmath45935 & & @xmath46036 & @xmath461100 + iras 15245@xmath321019w & @xmath462 & 426 & & & & & & & + & & & ew & & @xmath4630.38 & @xmath4640.19 & & @xmath4650.24 & + & & & fwhm & & 310@xmath8670 & 406@xmath8674 & & 78@xmath8654 & + & & & @xmath87 & & @xmath46631 & @xmath18042 & & @xmath25441 & + iras 17028@xmath325817e & @xmath467 & 0 & & & & & & & + & & & ew & & @xmath4680.76 & & & & + & & & fwhm & & 97@xmath8633 & & & & + & & & @xmath87 & & @xmath46932 & & & & + iras
22491@xmath331808w & @xmath470 & 86 & & & & & & & + & & & ew & @xmath4710.38 & @xmath4720.2 & @xmath4730.21 & & & + & & & fwhm & 0@xmath86140 & 0@xmath8623 & 0@xmath8699 & & & + & & & @xmath87 & @xmath244100 & @xmath47423 & @xmath121160 & & & + figure [ fig : twod ] presents the two - dimensional spectra around the pa@xmath2 line for each of the sample galaxies . in each
, the horizontal axis origin is located at the center of the primary nucleus .
the apertures corresponding to the extractions displayed in figures [ fig : primary ] and [ fig : secondary ] are indicated at the top and bottom of each panel .
annotations to the left and right of each panel denote the sense of the slit orientation , the exact position angles for which are listed in table [ tab : obs ] .
the two - dimensional spectrum for iras 09583@xmath274714 has been split into two panels , though the horizontal scale reflects the real separation between the two nuclear components .
the @xmath18 band spectra of the galaxies in this sample are dominated by the pa@xmath2 line , which is the strongest line accessible to search for signs of broad emission . only in two of the sample galaxies , iras 08311@xmath312459 and
iras 15462@xmath310450 , is any such emission apparent .
both of these galaxies are optically classified as seyfert galaxies , so it seems that obscured agn in ulirgs are either extremely rare , or suffer a few magnitudes of extinction even at near - infrared wavelengths , as was also suggested by @xcite .
a few galaxies , such as iras 14348@xmath311447 and iras 14394@xmath275332 have excess emission on the blue side of the pa@xmath2 line , but in both cases there appears to be emission from contributing significantly to this flux , and no appreciable emission on the red side of the narrow pa@xmath2 profile .
furthermore , the h@xmath40 lines in iras 14394@xmath275332 have an asymmetric blue shape more typical of wind or outflow phenomena than of velocity - broadening due to an agn .
iras 20087@xmath310308 has some hint of broad emission at the base of the pa@xmath2 line , though not significantly elevated out of the noise , and again confused by . as seen in figure [ fig : twod ] , the moderate 600 km s@xmath6 width of the narrow portion of the pa@xmath2 profile in iras 20087@xmath310308 can be attributed to a steep rotation gradient on the nucleus .
the apparent broad base on the pa@xmath2 line in the southwestern nucleus of iras 12112@xmath270305 may be spurious , as this is the one galaxy in the sample whose spectra were not calibrated by a g star observation , but rather by a template atmospheric spectrum .
the excess blue emission happens to be coincident with the significant co@xmath40 atmospheric absorption at 2.002.02 @xmath1 m , so that a slight error in estimated column depth results in a large localized continuum offset in the corrected spectrum .
two dimensional spectra around the pa@xmath475 line for each galaxy .
the horizontal scale is in arcseconds , with the primary nucleus at the origin .
the spectral direction spans about 11 resolution elements , or about 3300 km s@xmath476 .
extraction apertures are indicated on the top and bottom of each frame . ]
the two - dimensional spectra of pa@xmath2 in figure [ fig : twod ] demonstrate that the atomic recombination line emission in ulirgs is almost always dominated by a nuclear component .
several ulirgs show off - nuclear emission at lower levels , and a couple of sources appear to be primarily characterized by diffuse line emission namely iras 04232@xmath271436 and iras 20046@xmath310623 .
when two nuclei are present , the secondary nucleus generally shows emission in pa@xmath2 as well .
exceptions to this are iras 16487@xmath275447 and iras 23327@xmath272913 .
the extra continuum object seen in the panel with iras 08030@xmath275243 is a nearby star .
the fraction of the total line emission that is unresolved can be estimated for each ulirg by computing the flux contribution from a gaussian spatial profile having the width of the estimated seeing and sharing the maximum amplitude of the spatial line emission distribution .
comparing this flux with the measured flux provides an upper limit to the fraction of light contained within an unresolved core .
most of the primary ulirg nuclei are consistent with being unresolved , with a median seeing of 10 and a median physical resolution of 1.4 kpc .
the scatter in the estimated fraction of unresolved flux is at the @xmath47720% level , mostly due to uncertainty in the seeing estimate .
only 8 of the 33 ulirgs have estimated unresolved contributions lower than 80% , such that the majority of ulirgs contain most if not all of their pa@xmath2 emission within the central 1 kpc radius .
only three ulirgs in the sample ( iras 04232@xmath271436 , iras 18470@xmath273233 , and iras 20046@xmath310623 ) have less than 50% of their pa@xmath2 flux in an unresolved core .
the sample was chosen to allow coverage of both the pa@xmath2 and br@xmath3 emission lines so that a comparison between these lines could be made in order to assess extinction to the line emitting source .
table [ tab : props ] lists the inferred visual extinction values for the sample galaxies .
these estimates assume an intrinsic ratio for pa@xmath4br@xmath3 of 12.1 ( * ? ? ?
* case b with @xmath478 @xmath479 and @xmath480 k ) , and @xmath481 and @xmath482 ( interpolated from the extinction law of * ? ? ?
though there exists very little leverage in wavelength between pa@xmath2 and br@xmath3 , some ulirgs are apparently obscured enough to show definite reddening across this short range .
typical 1@xmath483 errors on the extinction estimates are around 5 magnitudes , though the actual errors could be larger if disparate aperture sizes were used in the spectral extractions .
indeed , each of the four ulirgs with extinctions more negative than 1@xmath483 have aperture differences between the pa@xmath2 and br@xmath3 spectra of at least 50% .
almost half of the ulirgs have estimated extinctions greater than the 1@xmath483 limit , and 11 have at least 2@xmath483 extinction estimates typically corresponding to @xmath484 mag , with many in the range of @xmath485 mag .
histogram of estimated visual extinctions to the pa@xmath475 line emission for the 33 primary ulirg nuclei .
different shadings represent varying levels of reliability , with most of the negative estimates owing to mismatched seeing conditions and therefore significant extraction aperture differences .
the mean and median points are marked for the reliable set of measurements .
one point , corresponding to @xmath486 , is beyond the plot boundary . ]
figure [ fig : extinct ] presents the extinction measurements in histogram form .
different shading represents varying levels of confidence in the measurement .
higher extinction estimates are allowed greater variance in classifying their reliability , provided they are significantly different from zero extinction .
counting only the 22 ulirgs with reliable extinction measures , the median visual extinction of @xmath487 mag corresponds to an extinction at pa@xmath2 of approximately one magnitude .
this is a lower limit to the nuclear extinction because we can only estimate extinction to the gas we actually detect at 2 @xmath1 m , which very likely is peripheral to the nuclear region .
nonetheless , one magnitude of extinction at pa@xmath2 would not readily hide the presence of a central agn .
the very large visual extinctions of 25 mag seen in some ulirgs translate to almost 4 mag of extinction at pa@xmath2 , which almost certainly would hide an agn from view with the currently achieved signal - to - noise ratio in the ulirg spectra .
lcccc iras 00262@xmath324251 & @xmath488 & @xmath489 & @xmath490 & mixture + iras 01521@xmath325224 & @xmath491 & @xmath492 & 5000 & thermal@xmath493 + iras 04232@xmath321436 & @xmath494 & @xmath495 & & thermal + iras 05246@xmath320103 & @xmath496 & @xmath497 & & thermal + iras 08030@xmath325243 & @xmath498 & @xmath499 & & thermal + iras 08311@xmath332459 & @xmath500 & @xmath501 & & thermal + iras 08344@xmath325105 & @xmath502 & @xmath503 & @xmath504 & mixture + iras 08572@xmath323915 & @xmath505 & @xmath506 & @xmath507 & mixture + iras 09061@xmath331248 & @xmath508 & @xmath509 & 3500 & thermal@xmath493 + iras 09111@xmath331007 & @xmath510 & @xmath511 & 2700 & thermal@xmath493 + iras 09583@xmath324714 & @xmath512 & @xmath513 & & thermal + iras 10035@xmath324852 & @xmath514 & @xmath515 & & thermal + iras 10190@xmath321322 & @xmath516 & @xmath517 & & thermal + iras 10494@xmath324424 & @xmath518 & @xmath519 & 3500 & thermal@xmath493 + iras 11095@xmath330238 & @xmath520 & @xmath521 & @xmath522 & mixture + iras 12112@xmath320305 & @xmath523 & @xmath506 & 5000 & thermal@xmath493 + iras 14348@xmath331447 & @xmath524 & @xmath525 & & thermal + iras 14352@xmath331954 & @xmath526 & @xmath527 & & thermal + iras 14394@xmath325332 & @xmath528 & @xmath529 & 2800 & thermal@xmath493 + iras 15245@xmath321019 & @xmath530 & @xmath531 & & thermal + iras 15250@xmath323609 & @xmath532 & @xmath533 & @xmath534 & mixture + iras 15462@xmath330450 & @xmath535 & @xmath536 & & thermal + iras 16487@xmath325447 & @xmath537 & @xmath521 & & thermal + iras 17028@xmath325817 & @xmath538 & @xmath539 & & thermal + iras 18470@xmath323233 & @xmath540 & @xmath541 & & thermal + iras 19458@xmath320944 & @xmath542 & @xmath543 & & thermal + iras 20046@xmath330623 & @xmath544 & @xmath545 & & thermal + iras 20087@xmath330308 & @xmath546 & @xmath547 & & thermal + iras 20414@xmath331651 & @xmath548 & @xmath549 & & thermal + iras 21504@xmath330628 & @xmath550 & @xmath551 & & thermal + iras 22491@xmath331808 & @xmath552 & @xmath553 & & thermal + iras 23327@xmath322913 & @xmath554 & @xmath555 & & thermal + iras 23365@xmath323604 & @xmath556 & @xmath557 & @xmath558 & mixture + because an active galactic nucleus is believed to indicate the presence of a qso ( quasi - stellar object ) in the galaxy nucleus , one may quantify the amount of obscuration necessary to hide a central agn by estimating the extinction required to hide the line flux expected from the high - velocity wings of pa@xmath2 in a typical qso of the appropriate bolometric luminosity .
for this purpose , we use the median qso spectrum presented in @xcite to represent the appearance of an unobscured quasar . identifying the region around 1.881 @xmath1 m ( @xmath559 km s@xmath6 redward of the pa@xmath2 line center ) as being well away from the narrow pa@xmath2 emission , we measure the median qso flux density to be 22% higher than the qso continuum level .
it is found that the broad pa@xmath2 emission from an exposed qso contributing half the bolometric luminosity of a ulirg would be comparable at 1.881 @xmath1 m to the typical ulirg continuum emission ( neglecting the contribution from the qso continuum ) .
therefore , hiding the presence of the broad emission requires an attenuation factor that is comparable to the signal - to - noise ratio of the ulirg continuum . in the case of the median ulirg spectrum presented in @xcite ,
this translates to an extinction at pa@xmath2 of 4.4 mag , and therefore a visual extinction , @xmath560 mag .
the spectra of the individual ulirgs presented here typically have continuum signal - to - noise ratios around 20 , so that @xmath561 mag , or @xmath56225 mag of extinction is required to hide a qso at half the ulirg luminosity .
details on this calculation appear in appendix [ qsoextinct ] .
the estimates of extinction made in this way are conservative in two regards .
first , the aperture photometry of ulirgs used to establish typical continuum flux densities are based on 5@xmath563 apertures , such that extra flux is included from diffuse emission away from the nucleus . in smaller apertures corresponding to the spectral extractions employed in this paper ,
the ulirg continuum is diminished , while the spatially unresolved qso component would be hardly affected , thus increasing the qso - to - ulirg contrast considerably , making the qso more difficult to conceal .
the second conservative aspect is that the criterion of forcing the broad pa@xmath2 from the qso to lie at the 1@xmath483 level on the ulirg continuum would not hide the broad emission very effectively , the reason being that the 1@xmath483 level is computed for a single pixel ( one - quarter of a resolution element ) .
when summed over a few resolution elements ( i.e. , up to the narrow shoulder of the pa@xmath2 line ) , the broad signal would stand out at more than the 3@xmath483 level .
the result that , given the current level of sensitivity , at least 25 magnitudes of visual extinction are required to obscure a qso in a typical ulirg , together with the paucity of detections of low - level broad line emission especially evident in the median spectrum of @xcite leads us to conclude that agn are either rare in ulirgs , or very deeply buried .
there are five ulirgs in the present sample with measured extinctions based on the pa@xmath2/br@xmath3 ratio that are high enough to hide agn without requiring additional attenuation .
we are unable to determine the likelihood that the rest of the ulirgs have the requisite nuclear extinctions to hide the presence of agn .
the technique of comparing ulirg spectra to a typical qso spectrum can also be used to estimate the extinction to the broad - line region seen in iras 15462@xmath310405 .
based on the flux density at 1.881 @xmath1 m , the pa@xmath2 emission contributes an additional 9% to the continuum level , some of which is likely continuum emission from the qso itself . correcting for this
, one finds an excess over the assumed ulirg continuum ( 60% of the total continuum ) measuring 15% , suggesting an extinction @xmath564 mag , or @xmath565 mag .
the most abundant species of line emission in the ulirg spectra arise from the vibration - rotation spectrum of molecular hydrogen .
the molecular hydrogen emission shows up strongly in _ all _ of the primary nuclei of this sample .
three or four lines are typically present in each spectrum , most notably the odd rotation state 10 transitions .
the h@xmath40 line strengths vary by more than one order - of - magnitude relative to pa@xmath2 .
figure [ fig : pah2 ] shows the range of observed line ratios , comparing the h@xmath40 10 s(1 ) line to pa@xmath2 .
a typical ulirg has a h@xmath40 10 s(1)@xmath566pa@xmath2 ratio around 0.15 , varying from @xmath567 to 0.40 .
one galaxy , iras 00262@xmath274251 exhibits h@xmath40 emission even stronger than the pa@xmath2 line .
the bins in figure [ fig : pah2 ] are shaded to reflect the classification of the h@xmath40 excitation mechanisms , as will be discussed below .
ratio of the h@xmath568 10 s(1 ) line strength to that of pa@xmath475 for the 33 primary ulirg nuclei .
a linear scale appears at top .
typical 1@xmath569 errors are comparable to the bin width , though perhaps a little smaller . the mean is performed on the linear scale , rather than on the logarithmic scale .
shadings indicating excitation mechanisms are included for reference . ]
the apparent weakness of the even rotation transitions is an indication that the emission is thermally excited , as the thermal condition produces three times as many odd ( ortho ) states as even ( para ) states .
this so - called ortho - to - para ratio is more typically in the range of 1.01.8 for fluorescent excitation conditions @xcite .
in addition , the lack of strong emission from the 21 vibration transition indicates a thermal origin .
molecular hydrogen population diagrams for six of the ulirgs , showing the range of properties seen in the h@xmath568 excitation .
the top two show examples of mixed thermal and fluorescent excitation , while the bottom four are consistent with a single temperature characterization .
iras 10494@xmath5704244 and iras 14394@xmath5705332 may have some slight component of fluorescent excitation , judging by the elevated 21 lines .
for even ( para ) rotation states , two points are plotted , corresponding to ortho - to - para ratios of 3:1 and 1:1 , with boxes corresponding to the 3:1 case , and asterisks for 1:1 . the population diagram for iras
18470@xmath5703233 is fairly typical of most of the ulirgs in this sample in terms of excitation information and signal - to - noise ratio .
line fits are accompanied by temperatures corresponding to the slopes . ]
population diagrams were constructed for each of the primary nucleus extractions in the sample , and each characterized in terms of rotation temperature , vibration temperature ( where appropriate ) and classified in terms of thermal or mixed thermal / fluorescent excitation .
the construction of population diagrams is discussed in appendix [ append ] .
examples of these diagrams are shown in figure [ fig : h2plots ] .
the even rotation state transitions are plotted as two points representing ortho - to - para ratios of 3:1 and 1:1 .
the 3:1 points , represented by boxes in figure [ fig : h2plots ] , are almost always preferred over the 1:1 points ( asterisks ) appearing below them .
the great majority of ulirgs in this sample have no detections in the 21 transition lines , but with error bars consistent with purely thermal excitation .
solid lines in figure [ fig : h2plots ] represent rotation temperature fits , and dashed lines , where present , indicate the vibration temperature separating 10 and 21 points .
a few ulirgs appear to be definite mixtures of thermal and fluorescent excitation , with the population diagrams showing the 21 data points well above the rotation temperature line passing through the 10 points .
a number of other ulirgs have somewhat elevated 21 measures , with both the s(2 ) and s(4 ) lines above the 10 rotation temperature line , but with error bars that are still consistent with pure thermal emission .
the excitation characterizations are summarized in table [ tab : props ] . for the mixed cases ,
a vibration temperature estimate is given , assuming that the high energy states exist with a lower ortho - to - para ratio , likely around 2:1 .
this has been observed in other mixed excitation cases ( e.g. , * ? ? ?
* ) , and is explained by the fact that the enhanced 21 transition emission arises from molecules that are primarily excited by fluorescent processes , and thus reflect an ortho - to - para ratio closer to that expected for pure fluorescence .
one point that should be mentioned in the context of the line population diagrams is the possible blend of a 2.0024 @xmath1 m [ line with the h@xmath40 21 s(4 ) line at 2.0035 @xmath1 m . the expected flux ratio between the [ line at 1.9670 @xmath1 m and that at 2.0024 @xmath1 m is 23:1 . with
the 1.9670 @xmath1 m line appearing to be present in a number of ulirgs , it is to be expected that the h@xmath40 21 s(4 ) line will suffer some contamination by the [ line .
indeed , the population diagrams often show an anomalously high 21 s(4 ) line flux , with an incompatibly low 21 s(2 ) flux .
there is a good correlation between such cases and galaxies for which the 1.9670 @xmath1 m [ line is believed to be present .
rotation temperatures determined for the 33 primary ulirg nuclei .
the majority are consistent with thermal emission , for which the rotation temperature represents the ambient equilibrium temperature .
a few are definitely mixtures of thermal and fluorescent processes , with vibration temperatures in the range of 31005000 k. ] figure [ fig : h2temp ] illustrates the distribution of h@xmath40 rotation temperatures , indicating a strong clustering around 2200 k. two of the three ulirgs with estimated rotation temperatures in excess of 4000 k are classified as having agn .
it is possible in these cases that a significant fraction of the h@xmath40 emission arises from molecular gas excited by x - ray radiation from the central source .
@xcite found this mechanism to be the most viable source of h@xmath40 excitation in ngc 6240 , with the x - rays in this instance most probably arising from frequent supernovae or from merger - induced shock activity .
table [ tab : h2 ] of appendix [ append ] provides expected h@xmath40 line ratios in the thermal regime across the 1.82.4 @xmath1 m wavelength range .
referring to figure [ fig : pah2 ] , there does not appear to be an overwhelming trend between h@xmath40 line strength
relative to pa@xmath2and excitation mechanism , except for a possible slight tendency for mixtures to be seen among the brighter lines .
this could simply be a consequence of higher signal - to - noise levels for the weak 21 lines in those galaxies with bright h@xmath40 lines .
more mixed states may be revealed with greater sensitivity , most likely in those cases identified here as possible mixtures .
the lineis detected in the majority ( 27/33 ) of the primary nuclear spectra , and in a number of the secondary spectra .
the noise - weighted mean of detected @xmath566pa@xmath2 is 0.058 , with a relatively small dispersion of @xmath571 about this value .
the median @xmath566pa@xmath2 ratio is 0.063 among detected lines , and 0.056 including non - detections as zero - valued points .
the full range of detected ratios goes from 0.02 ( iras 15245@xmath273609 ) to about 0.11 ( iras 14348@xmath311447 ; iras 14394@xmath275332 ) .
lower resolution near - infrared spectra could easily mistake this line for a blue wing on the pa@xmath2 line .
the [ line at 1.9670 @xmath1 m appears in several spectra , most notably the primary nuclei of iras 00262@xmath274251 , iras 09111@xmath311007 , iras 15245@xmath271019 , and iras 15250@xmath273609 , and the secondary nucleus of iras 10190@xmath271322 .
these stronger [ emitters have [ @xmath566pa@xmath2 ratios up to 0.05 , though iras 00262@xmath274251 , with its weak pa@xmath2 , has a ratio around 0.12 .
there are many possible line detections near the noise limit that have typical [ @xmath566pa@xmath2 ratios around 0.010.02 .
lines weaker than 1% of pa@xmath2 are generally too weak to be detected with confidence .
the [ line reported for iras 08311@xmath312459 is possibly confused with the wing of [ .
a more thorough analysis of this line complex at slightly higher resolution does not find conclusive evidence for the [ line in this galaxy @xcite .
searching for the 1.9629 @xmath1 m [ line was a primary motivation for the present spectroscopic survey . with a 167 ev excitation potential , this line indicates a high probability of the presence of an active nucleus .
while it is possible to produce si@xmath572 atoms via powerful shocks , @xcite find that photoionization is the more likely excitation mechanism among a sample of nearby seyfert galaxies .
as stars are incapable of producing significant quantities of photons at this high energy , the natural photoionization source is an active nucleus with its power - law spectrum extending into the far - ultraviolet wavelength range .
only one galaxy in the sample of 33 has prominent [ emission .
iras 08311@xmath312459 has a very strong [ line , comparable in strength to the adjacent h@xmath40 10 s(3 ) line , and appearing broader than the other lines in the spectrum .
a detailed discussion of the [ line in this galaxy can be found in @xcite and @xcite . the only other galaxy in the sample with possible [ emission is iras 05246@xmath270103 .
the spectrum of this galaxy shows emission at the expected position of [ , with a signal - to - noise ratio of @xmath375 .
the adjacent h@xmath40 line appears broad on the short - wavelength side as well , calling into question the nature of the emission on the long - wavelength side of this line .
no emission appears at this position in the secondary nucleus spectrum of figure [ fig : secondary ] , ruling out a problem with atmospheric calibration or related issues .
the h@xmath40 10 s(1 ) line appears unresolved in the primary nucleus , strongly suggesting that the [ emission is real .
while iras 05246@xmath270103 has no evidence of broad line emission , it does have a peculiar radio spectrum .
@xcite find this galaxy to be a gigahertz - peaked - spectrum ( gps ) radio source , with a 6 cm flux density much higher than that measured at 20 cm .
this condition is not found in many ulirgs , with the parks ( pks ) galaxy iras 13451@xmath271232 being the only other 2 jy ulirg identified as a gps source .
this galaxy may also have [ emission , as reported by @xcite , though the blend with the h@xmath40 line at their lower resolution makes it difficult to estimate the relative contribution of [ .
the rest of the h@xmath40 lines in iras 13451@xmath271232 , as presented in @xcite , indicate a h@xmath40 rotation temperature around 2500 k , in which case the [ line would account for @xmath3730% of the total combined line flux .
iras 13451@xmath271232 does show impressive broadening of the pa@xmath2 line , clearly standing out as an agn - dominated source . while iras 05246@xmath270103 has a similar radio spectrum to iras 13451@xmath271232 , it is almost three orders of magnitude less luminous at 20 cm , according to @xcite
perhaps there is a low - luminosity active nucleus component to iras 05246@xmath270103 , with the total infrared luminosity dominated by star formation .
the double nucleus ulirgs often exhibit strong pa@xmath2 emission from both nuclei , which can be used to study orbital dynamics of the galactic pair .
table [ tab : vels ] lists the relative radial velocities for the 12 ulirgs with measured pa@xmath2 velocities on both nuclei .
the reported errors are considerably smaller than would be obtained by simply subtracting the velocities presented in table [ tab : mofosec ] from those in table [ tab : mofoprim ] , reflecting the fact that the relative velocities are not affected by systematic errors associated with wavelength calibration or slit illumination . lcc iras 01521@xmath325224 & @xmath573 & 7.6 + iras 05246@xmath320103 & @xmath574 & 9.8 + iras 08572@xmath323915 & @xmath575 & 5.8 + iras 09061@xmath331248 & @xmath576 & 7.2 + iras 09583@xmath324714 & @xmath577 & 37.5 + iras 10035@xmath324852 & @xmath578 & 11.2 + iras 10190@xmath321322 & @xmath579 & 5.3 + iras 12112@xmath320305 & @xmath580 & 3.7 + iras 14348@xmath331447 & @xmath581 & 4.7 + iras 15245@xmath321019 & @xmath582 & 3.4 + iras 17028@xmath325817 & @xmath583 & 23.0 + iras 22491@xmath331808 & @xmath584 & 2.2 + surprisingly , the velocity differences are very small , with the largest difference at 120 km s@xmath6 .
random projection effects diminish the observed relative velocities statistically by a factor of two , as the most probable observing configuration involves pure angular motion on the plane of the sky .
if all ulirgs have the same relative nuclear velocity , random orientations would produce a distribution of observed velocities described by a cosine curve , with the peak at zero observed relative velocity . binning the data from table [ tab : vels ] into 50 km s@xmath6 bins ,
the best fit to a cosine function has a maximum velocity of about 190 km s@xmath6 , shown in figure [ fig : velhist ] . using 40 km s@xmath6 bins results in a 165 km s@xmath6 maximum relative velocity .
histogram of the absolute values of projected relative nuclear velocities .
if all ulirgs had the same relative velocity , projection effects would result in our observing a cosine distribution .
the best fit cosine to the data has a 190 km s@xmath476 velocity amplitude . ]
the lack of high relative velocities among ulirgs contrasts with the predictions obtained either from simple newtonian calculations or from more sophisticated @xmath585-body models of galaxy interactions which include gas dynamics and dynamical friction .
below we discuss these two approaches , with attention to how these models might be modified to match the observations .
first , we will treat the galaxies as point masses approaching each other from far away .
these calculations assume conservative dynamics
, that all loss of potential energy is converted into kinetic energy of the galaxies . in reality some fraction of the energy is lost to dynamical friction , which is why the galaxies do eventually merge . the foregoing discussion will not explicitly treat the frictional component , but will concentrate on the more basic elements of merger dynamics that offer much greater latitude in the ultimate relative velocities calculated . in the point mass scenario ,
the galaxies would reach a relative velocity @xmath586 in excess of 1300 km s@xmath6 at a separation distance of 10 kpc , assuming total system masses of @xmath587 per galaxy .
even ten times less mass per galaxy results in over 400 km s@xmath6 of relative motion at 10 kpc . changing the initial separation from infinity to 100 kpc or 50 kpc reduces these speeds by 5% and 10% , respectively .
the high velocities computed above are obviously not consistent with the observations , but perhaps understandably given that galaxies are not well represented by point masses at close range .
galaxy masses are dominated by very large dark halos that act to soften the potential gradients i.e . ,
accelerations experienced during a close encounter .
dark halos are thought to exist based on the anomalously flat rotation curves of galaxies to very large radii .
such phenomena suggest the enclosed spherical mass is proportional to radius , such that the circular rotation velocity , @xmath588 is constant .
rotation curves of large samples of galaxies via optical emission lines ( e.g. , * ? ? ?
* ) indicate that this scaling persists at least out to 1015 kpc .
observations in find a continuation of the flat rotation curves at radii 23 times the optical radii , often to 2030 kpc ( e.g. , * ? ? ?
* ; * ? ? ?
assuming a spherical halo distribution , the total halo mass following the @xmath589 scaling is given by@xmath590 where @xmath591 is the circular velocity , and @xmath592 is the outer dimension of the halo distribution . under this prescription , two galaxies falling from infinity are moving with a relative velocity of @xmath593 by the time the halos touch , or when the galaxies are @xmath594 apart .
the galaxies will clearly accelerate even more as the separation decreases to @xmath595 kpc , even though the acceleration is somewhat moderated by the effect of overlapping halo masses .
this already presents a problem in that the observed relative velocities are less than the typical rotation velocities observed in ulirgs .
simulations of galaxy interactions , courtesy of j. c. mihos . at
left is the orbital track , at center is nuclear separation as a function of time , and at right is the relative velocity as a function of time .
the model galaxies each have a total mass of @xmath596 , and possess milky - way - like disk rotation velocities . ] the simulations by @xcite provide a means of evaluating encounter velocities via models employing realistic halo potentials and dynamical friction .
output from these models is shown in figure [ fig : mihos ] .
the model defines a halo that yields a total galaxy mass of @xmath597 @xmath598 when scaled to the milky way s rotation velocity .
the halo has a relatively small physical extent , containing most of its mass within 25 kpc , and with an effective cutoff radius at 40 kpc .
the simulations start with identical galaxies approaching from infinity ( parabolic orbits ) , resulting in time - averaged velocities of @xmath599 km s@xmath6 when the nuclei are between 425 kpc apart .
this average sums over first , second , and third encounters .
the peak velocities are in the neighborhood of 500 km s@xmath6 , as seen in figure [ fig : mihos ] .
starting the galaxies from rest at a 100 kpc separation rather than infinity reduces the peak velocities by only @xmath600% .
many ulirgs are found at times well after the first major encounter , judging by the fact that their tidal features , formed at the epoch of the first significant encounter , are often vastly more extended than the nuclei are separated .
however , some ulirgs are believed to be experiencing their first major encounter @xcite .
looking at the first - encounter and late - encounter model predictions separately , the simulation discussed above predicts first - encounter velocities averaging 475 km s@xmath6 , after which the galaxies separate by 50 kpc , followed by a final merger sequence with an average velocity around 275 km s@xmath6 , ranging from 0500 km s@xmath6 during this time .
models with larger - scale halos maintaining the same internal rotation structure demand higher total masses , resulting in increased net velocity differences between the merging galaxies . on the other hand ,
if ulirgs lacked any halo material past 10 kpc , corresponding to total masses around @xmath601 @xmath602 , the galaxies would behave much more like point masses , and dynamical friction would be less important . in this scenario
, the 10 kpc velocities would be approximately 400 km s@xmath6 .
it is very difficult to reproduce the small observed relative velocities among ulirgs with simple physical models with or without large halos .
ulirgs are massive , gas rich galaxies with rotation velocities and near - infrared absolute magnitudes characteristic of galaxies at least as massive as the milky way .
the highest observed relative velocity in this sample is a mere 120 km s@xmath6 , though the observed distribution is consistent with a typical deprojected velocity difference of 200 km s@xmath6 .
some ulirgs have been observed with velocity differences very near this value ( e.g. , iras 20046@xmath330623 ; iras 19252@xmath337245 : * ? ? ?
* ; * ? ? ?
* ) , but none are seen near the 400 km s@xmath6 value as expected from a variety of physical predictions .
one possibility deserves mention , which is that during the late - merger state , eccentric orbits may conspire with the delay in coordinating the accumulation of fuel for a nuclear starburst in such a way as to select ulirgs at the apocenters of their orbits , when velocities are low .
referring to figure [ fig : mihos ] , during the final merger process ( well after the initial encounter ) , the nuclear orbits are very eccentric , owing to the removal of angular momentum , largely by tidal debris .
the very high velocity peaks near 500 km s@xmath6 are associated with the close passages .
if the ultraluminous activity is episodic in nature , with relatively short duration bursts occurring in reaction to transitory tidal disruptions , then the ulirgs we see may represent stages shortly after these close encounters . a finite time is required to organize gas motions in response to the close encounter , with the ultraluminous activity turning on only after a significant gas concentration has been established .
it is not difficult to imagine in this scenario that the gas moving more slowly than the galaxy , but with less distance to travel will reach the nucleus at approximately the same time that the galaxies reach apocenter .
keeping in mind that we are discussing the events occurring after the second close encounter , the orbital timescale is @xmath603 yr , such that apocenter is reached a few@xmath604 yr after close approach .
this timescale is in rough agreement with expectations of the dynamical timescale within a few kpc of the nucleus the region from which the fuel for the nuclear event is gathered . during the final merger process , it is possible that several ultraluminous bursts occur following close interactions of the nuclei .
the idea of repetitive starbursts in the final stages of merging was first explored by @xcite , though in the context of cloud - cloud collisions in this case .
the scheme proposed here of delayed and potentially episodic bursting would bias late - merger ulirgs to larger separations and therefore lower velocities , quite possibly consistent with the observed distribution .
the problem with first encounter ulirgs remains , since the average first encounter velocity is around 475 km s@xmath6 .
unfortunately the current sample is not large enough to probe possible differences in the velocity distributions of early and late ulirg encounters .
near - infrared spectra in the 2 @xmath1 m window have been obtained for a nearly complete , volume - limited sample of 33 ulirgs .
lines of atomic hydrogen and helium recombination , molecular hydrogen vibration - rotation , [ , and [ are seen in the spectra .
the pa@xmath2 is predominantly seen as a nuclear event , with the majority of the emission arising in the central kpc of most ulirgs . initiated as a search for optically hidden active galactic nuclei , this survey reveals only two such galaxies , both of which are characterized as seyfert galaxies based on visible - light spectra .
it is found via comparison of the pa@xmath2 and br@xmath3 lines that significant extinction of several magnitudes persists into the near - infrared bands , such that buried active nuclei can not be ruled out in many objects .
it is estimated that with the sensitivity of the current survey , at least 25 mag of visual extinction would be needed to hide a qso with half the bolometric luminosity of the ulirg .
strong molecular hydrogen emission is seen in all of the primary nuclear spectra .
the molecular hydrogen spectrum indicates that most h@xmath40 emission in ulirgs stems from thermal excitation rather than fluorescent processes .
some ulirgs appear to have a small fraction of their molecular hydrogen emission owing to fluorescence , though none approach pure fluorescence .
the relative radial velocities found in this sample of ulirgs does not exceed @xmath605 km s@xmath6 , with an inferred maximum deprojected velocity differential of around 200 km s@xmath6 .
this relatively low velocity differential is difficult to produce from interactions between galaxies with masses comparable to that of the milky way using simple physical assumptions .
it is suggested that ulirgs in the late stages of merging may be selectively found near the time of apocenter , leading to a lower average relative velocity between nuclei .
we thank michael strauss for his role in defining the sample of ulirgs , and for participating in the early stages of the caltech effort in studying the 2 jy ulirgs .
we also thank chris mihos , gerry neugebauer , nick scoville , and sterl phinney for helpful discussions .
james graham provided information on [ transition strengths and wavelengths .
many persons accompanied us on the observing runs to palomar , most notably rob knop and james larkin , both of whom provided expert tutelage on the use of the spectrograph , and on methods of data reduction .
we thank the night assistants at palomar , juan carasco , rick burruss , skip staples , and karl dunscombe for their assistance in the observations .
this research has made use of the nasa / ipac extragalactic database ( ned ) , which is operated by the jet propulsion laboratory , caltech under contract with nasa .
is supported by the nasa graduate student researchers program , and the lewis kingsley foundation .
this research is supported by a grant from the national science foundation .
the line properties reported in tables [ tab : mofoprim ] and [ tab : mofosec ] represent measured values from the spectra , rather than gaussian fit parameters . especially for the weaker lines ,
the quality of the continuum subtraction is vital to proper line characterization .
this is done with a simultaneous multiple gaussian line fit and quadratic continuum fit . in all but a few cases , this results in a very close fit to the continuum .
for those few cases where the continuum fit is questionable , some of the weaker lines are excluded from tables [ tab : mofoprim ] and [ tab : mofosec ] .
all of the fitting procedures make use of the noise estimate computed on a pixel - by - pixel basis , as described earlier . in the end
, the noise is scaled slightly ( typically by factors of less than 30% ) in order to make the final reduced chi - squared value equal to one .
the scaled noise is used in estimating errors on the values in tables [ tab : mofoprim ] and [ tab : mofosec ] .
each of the line parameters is computed based on the following line definition procedure .
each potential line is searched for by summing two resolution elements ( 7 pixels ) of information around the line center , and determining if there is emission above the continuum exceeding 2.5@xmath483 . for lines satisfying this condition ,
cutoffs are established on the red and blue sides of the line center by the following rules : 1 .
if the spectrum extends below the continuum level , establish a cutoff , including only positive valued points .
if there exists a local minimum , followed by at least three higher points , assume that the following points belong to another line and exclude them , keeping only the minimum point .
3 . if the next two pixels are between 01@xmath483 above the continuum , include them and quit .
these procedures obviously can not practice the same discrimination that a diligent scientist might , but we find that a reasonable job is performed especially on lines whose boundaries are obvious to the eye .
nonetheless , lines with a final computed equivalent width measuring less than 2.5@xmath483 are typically excluded , though those that appear to be real in the one - dimensional spectra are preserved .
conversely , a few lines with reported detections greater than 2.5@xmath483 are excluded on the basis of either poor local continuum fit or simply less - than - believable results . because the weaker lines do not have a well defined full - width at half - maximum , the fwhm values in tables [ tab : mofoprim ] and [ tab : mofosec ] are based on a combination of line flux and peak amplitude .
if a gaussian shape is assumed , the fwhm of the line would be equal to the line flux divided by the peak amplitude , times the proportionality constant of @xmath606 .
performed in this manner , it is relatively simple to compute the error of such an estimate .
a strict measure of the fwhm is provided in tables [ tab : mofoprim ] and [ tab : mofosec ] for the pa@xmath2 line in each galaxy for comparison to the value estimated by this technique .
all of the reported fwhm values have been reduced to intrinsic line widths by subtracting the instrumental resolution in quadrature .
each spectrum uses the measured line widths in the wavelength calibration spectrum as a basis for the decomposition .
values in tables [ tab : mofoprim ] and [ tab : mofosec ] for which the measured line width is exceeded by the instrumental width are listed as unresolved lines of zero width .
almost all of the lines in this survey are resolved , typically around 200300 km s@xmath6 in width .
the velocity measurements in tables [ tab : mofoprim ] and [ tab : mofosec ] are indicated relative to pa@xmath2 , though the pa@xmath2 velocity itself is expressed as a direct quantity in km s@xmath6 .
these values are simply computed as first - moment measures in the spectral lines , with error estimates computed from the noise estimates in a straightforward way .
added to this error estimate is a wavelength calibration error , assumed to be 1015 km s@xmath6 , and a slit placement error .
the latter uncertainty attempts to estimate the wavelength offset induced by a non - uniform slit illumination . when the line source is spatially compact enough to produce a discernible peak somewhere in the slit , the location of this peak within the slit influences the ultimate wavelength associated with this emission .
for these purposes , it is assumed that the line source has a real physical size roughly half that of the continuum , which is almost always spatially resolved .
further , it is assumed that the source is centered in the slit with a gaussian 1@xmath483 placement error of one - quarter of a slit width .
only cases of exceptional seeing result in an appreciable error from this source , with a median error estimate of 15 km s@xmath6 and a maximum of 38 km s@xmath6 . despite the accounting of uncertainties ,
the velocities indicate a dispersion outside of the error budgets that is not correlated with line species or grating setting .
it may well be that the variable spectral range over which the line is summed is largely responsible for determining the offset , as the range is biased to contain only positive points , and will extend as far as possible from the line center to include such points . with the large moments generated from distant points , the estimated velocity may wander far from its true center .
thus the values of velocity in tables [ tab : mofoprim ] and [ tab : mofosec ] , especially for weaker lines , are perhaps of little use .
gaussian fit centers may do a better job of appropriately representing the line velocities .
the median qso spectrum is constructed from nine optically identified qsos in the redshift range @xmath607 , seven of which have available @xmath18 band photometry . comparing the absolute @xmath18 band magnitudes to bolometric luminosities of the quasars ,
one finds the approximate relation @xmath608 .
the median infrared luminosity of the 31 non - agn ulirgs in the present sample is @xmath609 .
a qso within a ulirg that contributes half of the bolometric luminosity most of which is reprocessed into the infrared would then have @xmath610 , or @xmath611 mag .
this establishes the @xmath18 band flux density of the qso continuum . at a rest wavelength of 1.881 @xmath1 m , corresponding to @xmath612 km s@xmath6 in velocity
, it is found that the median qso has an excess flux density due to pa@xmath2 measuring 22% of the baseline continuum level @xcite .
thus the absolute magnitude of the line emission at this velocity is @xmath613 mag .
typical ulirg continuum levels can be approximated in a similar way as that above , using 5@xmath563 aperture photometry of the sample galaxies where available @xcite .
eight of the 31 non - agn ulirgs have @xmath75 measurements , and comparing to the infrared luminosity ( 81000 @xmath1 m ) leads to the relation @xmath614 .
therefore the typical non - agn ulirg in this sample , with a median luminosity @xmath615 , has @xmath616 mag .
the expected ratio of broad pa@xmath2 to the ulirg continuum at 1.881 @xmath1 m is then 0.92 , with the 1@xmath483 range extending from 0.44 to 1.91 .
if a given ulirg has a spectrum with a signal - to - noise ratio @xmath617 , then diminishing the qso pa@xmath2 to the 1@xmath483 level requires@xmath618 where @xmath619 represents qso contributions to continuum and line flux , @xmath620 is the ulirg continuum contribution , and @xmath621 is the attenuating factor . with the additional relations that @xmath622 and @xmath623 , we find that the required attenuation is @xmath624 , or @xmath625 magnitudes at pa@xmath2 .
the visual extinction is roughly 7 times this value .
the specific mechanism responsible for the h@xmath40 emission can be investigated via population diagrams . by accounting for various degeneracies , transition probabilities , and photon energy ,
one may compute the relative populations of the upper states of each transition based on the observed line flux .
in pure thermal equilibrium , the population levels will follow an exponential profile with respect to upper state energy .
more formally , the intensity of a given transition is described by@xmath626 where the upper and lower states are denoted by the 1 and 2 subscripts , @xmath627 represents vibration state , @xmath628 the rotation state , @xmath629 the ortho / para degeneracy , @xmath630 the transition probability , @xmath631 the emitted photon frequency , @xmath632 the upper state energy , and @xmath633 the equilibrium temperature .
therefore a logarithmic plot of the line intensities divided by the terms multiplying the exponential above produces a linear relation against the upper energy scale with a slope proportional to the inverse of the temperature .
the physical parameters for selected @xmath18 band transitions are given in table [ tab : h2phys ] .
10 s(0 ) & 2.22268 & 2 & 2.53 & 6471 + 10 s(1 ) & 2.12125 & 3 & 3.47 & 6951 + 10
s(2 ) & 2.03320 & 4 & 3.98 & 7584 + 10 s(3 ) & 1.95702 & 5 & 4.21 &
8365 + 10 s(4 ) & 1.8914 & 6 & 4.19 &
9286 + 10 s(5 ) & 1.8353 & 7 & 3.96 & 10341 + 21 s(0 ) & 2.3550 & 2 & 3.68 & 12095 + 21 s(1 ) & 2.2471 & 3 & 4.98 & 12550 + 21
s(2 ) & 2.1536 & 4 & 5.60 & 13150 + 21 s(3 ) & 2.0729 & 5 & 5.77 &
13890 + 21 s(4 ) & 2.0035 & 6 & 5.57 & 14764 + 21 s(5 ) & 1.9443 & 7 & 5.05 &
15763 + 21 s(6 ) & 1.8942 & 8 & 4.30 & 16880 + 21 s(7 ) & 1.8523 & 9 & 3.38 & 18107 + 32 s(3 ) & 2.2008 & 5 & 5.63 & 19085 + when the molecular hydrogen is not in thermal equilibrium such as may be the case in fluorescent excitation by ultraviolet photons a diagram such as that described above will not show all points lying along a common line , but rather disconnected linear segments , with common vibration states each defining a line connecting the different rotation states .
the characteristic temperature of this line is referred to as the rotation temperature , @xmath634 .
ideally , each vibration state has the same characteristic rotation temperature , and these segments can be connected with another line characterized by a vibration temperature , @xmath635 .
the thermal case simply has @xmath636 .
typically , pure fluorescent conditions are characterized by @xmath6371400 k , and @xmath6386000 k @xcite .
excitation conditions associated with thermal processes are typically found to have @xmath6392400 k , with a slight difference between rotation and vibration temperatures characterized by @xmath640 @xcite . because ulirgs tend to show relatively uniform h@xmath40 emission properties namely those corresponding to thermal emission at 15002500 k it may be useful for other researchers to have a table of expected h@xmath40 line strengths in the @xmath18 band for a range of typical temperatures .
table [ tab : h2 ] provides such a reference , with line intensities normalized to the h@xmath40 10 s(1 ) line .
these estimates assume an ortho - to - para ratio of 3:1 , and @xmath642 , as is found empirically @xcite . in order to regain the pure thermal regime characterized by a single temperature , the 21 entries in table [ tab : h2 ]
should be multiplied by @xmath643 , and the 31 transition multiplied by @xmath644 , lowering these values considerably .
10 s(0 ) & 2.22268 & 0.228 & 0.211 & 0.201 & 0.187 + 10
s(1 ) & 2.12125 & 1.000 & 1.000 & 1.000 &
1.000 + 10 s(2 ) & 2.03320 & 0.336 & 0.374 & 0.398 &
0.438 + 10 s(3 ) & 1.95702 & 0.805 & 1.019 & 1.174 & 1.450 + 10 s(4 ) & 1.8914 & 0.177 & 0.261 & 0.329 &
0.468 + 10 s(5 ) & 1.8353 & 0.295 & 0.519 & 0.728 &
1.211 + 21 s(0 ) & 2.3550 & 0.015 & 0.030 & 0.045 & 0.083 + 21
s(1 ) & 2.2471 & 0.068 & 0.143 & 0.225 &
0.441 + 21 s(2 ) & 2.1536 & 0.023 & 0.053 & 0.089 &
0.191 + 21 s(3 ) & 2.0729 & 0.055 & 0.145 & 0.259 &
0.622 + 21 s(4 ) & 2.0035 & 0.012 & 0.037 & 0.072 & 0.197 + 21
s(5 ) & 1.9443 & 0.020 & 0.072 & 0.156 &
0.496 + 21 s(6 ) & 1.8942 & 0.003 & 0.014 & 0.033 & 0.124 + 21
s(7 ) & 1.8523 & 0.004 & 0.020 & 0.054 &
0.246 + 32 s(3 ) & 2.2008 & 0.003 & 0.016 & 0.045 & 0.202
+ turner , kirby - docken , & dalgarno.(1977 ) beckwith , s. , evans , n. j. , gatley , i. , gull , g. , & russel , r. w. 1983 , , 264 , 152 black , j. h. , & van dishoeck , e. f. 1987 , , 322 , 412 carico , d. p. , graham , j. r. , matthews , k. , wilson , t. d. , soifer , b. t. , neugebauer , g. , & sanders , d. b. 1990 , , 349 , l39 casertano , s. & van albada , t. s. 1990 , in _ baryonic dark matter _ , ed .
d. lynden - bell & g. gilmore ( dordrecht : kluwer ) p. 298
clements , d. l. , sutherland , w. j. , mcmahon , r. g. , & saunders , w. 1996 , , 279 , 477 crawford , t. , marr , j. , partridge , b. , & strauss , m. a. 1996 , , 460 , 225 dabrowski , i. 1984 , can .
j. phys . , 62 , 1639 dale , d. a. , giovanelli , r. , haynes , m. p. , hardy , e. , & campusano , l. e. 1999 , , 118 , 1468 draine , b. t. , & woods , d. t. 1990 , , 363 , 464 fich , m. , & tremaine , s. 1991 , , 29 , 409 genzel , r. , et al .
1998 , , 498 , 579 goldader , j. d. , joseph , r. d. , doyon , r. , & sanders , d.b .
1995 , , 444 , 97 hasegawa , t. , gatley , i. , garden , r. p. , brand , p. w. j. l. , ohishi , m. , hayashi , m. , & kaifu , n. 1987 , , 318 , l77 _ iras point source catalog , explanatory supplement _ , eds . beichman , c. a. , neugebauer , g. , habing , h. j. , clegg , p. e. , & chester , t. j. 1988 ( u.s .
gpo , washington , d. c. ) kim , d .-
1995 , ph.d .
thesis , university of hawaii kleinmann , s. g. , & hall , d. n. b. 1986 , , 62 , 501 larkin , j. e. , knop , r. a. , lin , s. matthews , k. , & soifer , b. t. 1996 , , 108 , 211 marconi , a. , moorwood , a. f. m. , salvati , m. , & oliva , e. 1994 , , 291 , 18 mihos , j. c. , & hernquist , l. 1996 , , 464 , 641 mihos , j. c. , & bothun , g. d. 1998 , , 500 , 619 murphy , t. w. , armus , l. , matthews , k. , soifer , b. t. , mazzarella , j. m. , shupe , d. l. , strauss , m. a. , & neugebauer , g. 1996 , , 111 , 1025 murphy , t. w. , soifer , b. t. , matthews , k. , kiger , j. r. , & armus , l. 1999 , , 525 , l85 murphy , t. w. , soifer , b. t. , matthews , k. , armus , l. 2000a , , october , in press ( astro - ph/0006318 ) murphy , t. w. , soifer , b. t. , matthews , k. , & armus , l. 2000b , , submitted noguchi , m. 1991 , , 251 , 360 oliva , e. , & origlia , l. 1992 , , 254 , 466 osterbrock , d. e. 1989 , _ astrophysics of gaseous nebulae and active galactic nuclei _ ( mill valley : university science books ) rieke , g. h. , & lebofsky , m. j. 1985 , , 288 , 618 rigopoulou , d. , spoon , h. w. w. , genzel , g. , lutz , d. , moorwood , a. f. m. , & tran , q. d. 1999 , , 118 , 2625 sanders , d. b. , soifer , b. t. , elias , j. h. , madore , b. f. , matthews , k. , neugebauer , g. , & scoville , n. z. 1988 , , 325 , 74 soifer , b. t. , sanders , d. b. , madore , b. f. , neugebauer , g. , danielson , g. e. , elias , j. h. , lonsdale , c. j. , & rice , w. l. 1987 , , 320 , 238 strauss , m. a. , davis , m. , yahil , a. , & huchra , j. p. 1990
, , 361 , 49 strauss , m. a. , huchra , j. p. , davis , m. , yahil , a. , fisher , k. b. , & tonry , j. 1992 , , 83 , 29 takayanagi , k. , sakimoto , k. , & onda , k. 1987 , , 318 , l81 tanaka , m. , hasegawa , t. , hayashi , s. s. , brand , p. w. j. l. , & gatley , i. 1989 , , 336 , 207 turner , j. , kirby - docken , & dalgarno , a. 1977 , , 35 , 281 veilleux , s. , sanders , d. b. , & kim , d .- c .
1997 , , 484 , 92 | we present near - infrared spectroscopy for a complete sample of 33 ultraluminous infrared galaxies at a resolution of @xmath0 .
most of the wavelength range from 1.802.20 @xmath1 m in the rest frame is covered , including the pa@xmath2 and br@xmath3 hydrogen recombination lines , and the molecular hydrogen vibration - rotation 10 s(1 ) and s(3 ) lines .
other species , such as he i , [ fe ii ] , and [ si vi ] appear in the spectra as well , in addition to a number of weaker molecular hydrogen lines .
nuclear extractions for each of the individual galaxies are presented here , along with spectra of secondary nuclei , where available .
the pa@xmath2 emission is seen to be highly concentrated on the nuclei , typically with very little emision extending beyond a radius of 1 kpc .
this survey was carried out primarily to search for signatures of active nuclei via velocity - broadened hydrogen recombination or the presence of the [ si vi ] coronal line .
these signatures are rare in the present sample , occurring in only two of the 33 galaxies .
the extinction to the hydrogen recombination lines is investigated via the pa@xmath4br@xmath3 line ratio .
it is found that visual extinctions to the nuclei in excess of 10 magnitudes are relatively common among ulirgs , and that visual extinctions greater than 25 mag are necessary to conceal a qso emitting half the total bolometric luminosity .
the ionized hydrogen regions in many ulirgs are sufficiently obscured that dust - enshrouded active galactic nuclei would remain hidden at 2 @xmath1 m at the current level of sensitivity
. the vibration - rotation lines of molecular hydrogen appear to be predominantly thermal in origin , with effective temperatures generally around 2200 k. the relative nuclear velocities between double nucleus ulirgs are investigated , through which it is inferred that the maximum deprojected velocity difference is @xmath5 km s@xmath6 . this figure is lower than the velocities predicted by physical models of strong interactions / mergers of large , gas - rich galaxies . |
prior knowledge about the shapes to be segmented is required for segmentation of images involving limited and low quality data . in many applications ,
object shapes come from multiple classes ( i.e. , the prior shape density is multimodal " ) and the algorithm does not know the class of the object in the scene .
for example , in the problem of segmenting objects in a natural scene ( e.g. , cars , planes , trees , etc . )
, a segmentation algorithm should contain a training set of objects from different classes .
another example of a multimodal density is the shape density of multiple handwritten digits , e.g. , in an optical character segmentation and recognition problem . in this paper , we consider segmentation problems that involve limited and challenging image data together with complex and potentially multimodal shape prior densities . [ cols= " < , < , < , < , < " , ] + & & & & & & + & & & & & & + & & & & & &
we have presented a mcmc shape sampling approach for image segmentation that exploits prior information about the shape to be segmented .
unlike existing mcmc sampling methods for image segmentation , our approach can segment objects with occlusion and suffering from severe noise , using nonparametric shape priors .
we also provide an extension of our method for segmenting shapes of objects with parts that can go through independent shape variations by using local shape priors on object parts .
empirical results on various data sets demonstrate the potential of our approach in mcmc shape sampling .
the implementation of the proposed method is available at spis.sabanciuniv.edu/data_code . | segmenting images of low quality or with missing data is a challenging problem .
integrating statistical prior information about the shapes to be segmented can improve the segmentation results significantly .
most shape - based segmentation algorithms optimize an energy functional and find a point estimate for the object to be segmented .
this does not provide a measure of the degree of confidence in that result , neither does it provide a picture of other probable solutions based on the data and the priors . with a statistical view , addressing these issues would involve the problem of characterizing the posterior densities of the shapes of the objects to be segmented . for such characterization , we propose a markov chain monte carlo ( mcmc ) sampling - based image segmentation algorithm that uses statistical shape priors .
in addition to better characterization of the statistical structure of the problem , such an approach would also have the potential to address issues with getting stuck at local optima , suffered by existing shape - based segmentation methods .
our approach is able to characterize the posterior probability density in the space of shapes through its samples , and to return multiple solutions , potentially from different modes of a multimodal probability density , which would be encountered , e.g. , in segmenting objects from multiple shape classes .
we present promising results on a variety of data sets .
we also provide an extension for segmenting shapes of objects with parts that can go through independent shape variations .
this extension involves the use of local shape priors on object parts and provides robustness to limitations in shape training data size . |
we consider pure bipartite states shared by alice and bob .
let @xmath4 and @xmath5 be the labels of their respective quantum systems , each with @xmath6 dimensions .
suppose alice and bob want to transform a state @xmath1 to another state @xmath2 .
they each have complete classical descriptions of the states , but they can only perform local quantum operations and classical communication ( locc ) @xcite . the transformation can be performed under many scenarios
. the final state can be exactly or approximately the desired output , and the transformation can work deterministically or with certain probability .
it is also possible to consider converting between an asymptotically large numbers of two different states .
we consider here _ only _ the scenario of exact , deterministic , and nonasymptotic transformation using locc .
if such transformation is possible , we write @xmath7 , otherwise , we write @xmath8 . a necessary and sufficient condition for @xmath7
was found by nielsen @xcite , with beautiful connections to the theory of majorization . for a pure state @xmath9 ,
let @xmath10 denote the list of eigenvalues of the reduced density matrix @xmath11 in descending order , and let @xmath12 be the @xmath13-th element in @xmath10 .
nielsen s criterion can be stated as | | _ m d _ i=1^m _ ^(i ) _ i=1^m _
^(i ) an intriguing consequence of the criterion is that , even when @xmath14 , it may still be possible to transform the state @xmath1 to the state @xmath2 as part of a larger transformation .
for example , @xmath15 may be true for some state @xmath16 , which functions as a catalyst @xcite .
it is also possible for @xmath17 @xcite . in ref .
@xcite , interesting properties about multiple - copy transformation between @xmath18 and @xmath19 under various scenarios are investigated . in the scenario of exact , deterministic , asymptotic transformation considered in this paper ,
they have derived a necessary condition for @xmath20 for some @xmath0 , @xmath21 and @xmath22 .
they have also made a very plausible conjecture , |^k |^k _ r k |^r |^r .
the conjecture is obviously true when @xmath0 divides @xmath23 .
however , we find counterexamples for @xmath24 , @xmath25 , and @xmath26 .
one example for @xmath27 is given by _
= \{0.493 , 0.284 , 0.158 , 0.035 , 0.030 } + _ = \{0.598 , 0.145 , 0.129 , 0.125 , 0.003 } and a simple example for @xmath28 is given by _
= \{0.24 , 0.22 , 0.22 , 0.19 , 0.10 , 0.03 } + _ = \{0.27 , 0.25 , 0.16 , 0.16 , 0.15 , 0.01 } the conjecture fails because of the exact and nonasymptotic nature of the transformation .
we remark that , for fixed @xmath0 , a counterexample in @xmath29 dimensions can always be turned into one in @xmath30 dimensions .
the reason is that the criteria for a counterexample involve only continuous functions of @xmath31 and @xmath32 .
therefore , the above counterexamples suffice to disprove the conjecture for @xmath24 and @xmath33 . in the following ,
we describe numerical results that provide not only counterexamples but also statistics of violations of the conjecture .
we performed a numerical search for @xmath34 and @xmath24 .
the @xmath35 case is uninteresting , as @xmath8 implies @xmath36 . for each @xmath6 ,
a large number of random pairs @xmath37 are sampled , and without loss of generality , @xmath21 .
we search for @xmath38 corresponding to @xmath39 satisfying + @xmath8 and @xmath40 + @xmath41 + the search parameters and the results can be summarized : @xmath42 \hline \rule{0pt}{2.4ex } \begin{array}{l } \mbox{sample size } \end{array } & 4 \times 10 ^ 8 & 2 \times 10 ^ 8 & 10 ^ 7 & 10 ^ 7 & 10 ^ 7 & 10 ^ 7 \\ \hline \begin{array}{l } \mbox{fraction of sample } \\ \mbox{satisfying { \bf i } } \end{array } & 8.31 \times 10^{-3 } & 1.56 \times 10^{-2 } & 2.16 \times 10^{-2 } & 2.63 \times 10^{-2 } & 3.01 \times 10^{-2 } & 3.31 \times 10^{-2 } \\ \hline \begin{array}{l } \mbox{fraction of sample } \\ \mbox{satisfying { \bf ii } } \mbox { given { \bf i } } \end{array } & 0 & 3.21 \times 10^{-7 } & 3.94 \times 10^{-4 } & 3.12 \times 10^{-4 } & 1.60 \times 10^{-4 } & 1.63 \times 10^{-4 } \\
\end{array } $ ] the fractions of sample satisfying * i * have negligible statistical uncertainties , while the fractions satisfying * ii * conditioned on * i * have about 11 - 14% uncertainties and @xmath43 for which few counterexamples are found . ] . for each @xmath44 ,
tens of counterexamples for the conjecture are found . for @xmath27 , exactly one counterexample
is found in @xmath45 random samples .
for @xmath46 no counterexample is found in @xmath47 random samples , so that the conjecture may still be true for this value of @xmath6 .
the conjecture investigated in this paper fails because of the exact , deterministic , and nonasymptotic nature of the transformation .
the conclusion drawn from our results , along with refs .
@xcite , is that even if one knows for all @xmath48 , whether @xmath49 , one still can not infer whether @xmath50 without additional information on the states .
this is in sharp contrast with the much simpler conditions for asymptotic manipulations .
in particular , asymptotically @xmath51 and @xmath52 are interconvertible with high fidelity and probability if @xmath53 where @xmath54 is the entropy function .
nielsen s criterion on transformation of pure bipartite states has led to many important results in the various possible scenarios , some of which are counterintuitive .
the results in the exact , deterministic , nonasymptotic scenario are mathematically interesting , yet their very oddity suggests that the asymptotic or probabilistic scenarios are perhaps more physically relevant .
we thank charles h. bennett for interesting discussions and the ibm watson cafeteria for enlightening coffee . | in this brief comment , we consider the exact , deterministic , and nonasymptotic transformation of multiple copies of pure states under locc .
it was conjectured in quant - ph/0103131 that , if @xmath0 copies of @xmath1 can be transformed to @xmath0 copies of @xmath2 , the same holds for all @xmath3 .
we present counterexamples to the above conjecture . |
cygnus a is the nearest example ( @xmath2=0.056 ) of a powerful fanaroff - riley type - ii radio galaxy . as such
, it contains some of the brightest radio hotspots to study these classical ` working surfaces ' of relativistic jets ( hargrave & ryle 1974 ; blandford & rees 1974 ) . although the hotspots are well - studied at radio and x - ray wavelengths ( 2 ) , cygnus a is viewed through the galactic plane , making them difficult to study at optical wavelengths ( 1.26 mag of extinction at v - band ; meisenheimer et al .
additionally , it is well known that bright foreground stars lie in the direction of cygnus a , and this has also hindered optical studies of the hotspots . to circumvent these obstacles
, we obtained new _
spitzer space telescope _ images ( figure 1 ) resulting in successful mid - infrared detections of the hotspots .
these new data , together with a wealth of published high - quality measurements at other frequencies , give us some of the most complete hotspot seds for a powerful radio galaxy .
this contribution is a summary of our recently published study ( stawarz et al .
we obtained new @xmath3/irac @xcite observations simultaneously at 4.5 and 8.0 @xmath4 m for 3.6 hrs . galactic extinction amounts to only 0.020.03 mag at these bands .
a foreground star is known to lie within an arcsecond of the position of hotspot a extrapolating its spectrum into the irac bands , its contamination to the hotspot ir fluxes is expected to be negligible ( see fig .
2 g in meisenheimer et al .
radio data were obtained from published vla and bima measurements and maps @xcite .
we focus our discussion on the two brightest hotspots ( a and d ) , whose spectral energy distributions ( seds ) are shown in figure [ fig-2 ] .
we modeled the synchrotron ( radio - to - infrared ) portion of the hotspot seds with a double power - law that decays exponentially at the higher frequencies : @xmath5 the fitted parameters ( @xmath6 , @xmath7 , @xmath8 , @xmath9 ) are ( 0.28 , 1.2 , 2.6@xmath1010@xmath11 hz , @xmath123.3@xmath1010@xmath13 hz ) for hotspot a , and ( 0.21 , 1.1 , 3.1@xmath1010@xmath11 hz , 0.9@xmath1010@xmath13 hz ) for hotspot d. the minimum frequency , @xmath14 , can be extrapolated down to 74 mhz with the vla detections of the hotspots by lazio et al .
( 2006 ) at this frequency .
the hotspots emit strong x - ray emission by the synchrotron self - compton ( ssc ) process as discovered with @xmath15 ( harris et al .
1994 ) . assuming this for the @xmath16 detection of the hotspots ( wilson et al .
2000 ; wright & birkinshaw 2004 ) , we infer magnetic field strengths of @xmath170.20.3 mg .
the emitting plasma is close to equipartition with @xmath18 ( figure 3 ) , consistent with the previous fits of the @xmath16 data .
an @xmath19 optical measurement of hotspot d ( nilsson et al . 1997 ) lies near the low energy extrapolation of the ssc spectrum ; this makes it one of the few hotspots with detected optical ssc emission ( brunetti 2004 ) .
increasing the low energy spectral index ( @xmath6 ) of hotspot d from its observed value to the ` canonical ' value of 0.5 would overproduce the observed optical and ir emissions , so the observed flat spectra are likely intrinsic . with the magnetic field and observed spectral fits , we can reconstruct the underlying electron energy spectra , @xmath20 ( @xmath21 ; figure 3 ) .
they have the following features : _
( i ) the critical break energy corresponds almost exactly to the ratio of the proton to electron masses , @xmath22 .
( ii ) the low - energy segment of the electron distributions continues down to at least @xmath170.1 @xmath23 , with flat spectra ; the power - law slopes are @xmath241.41.6 .
( iii ) the new ir data constrain the high - energy electron spectra , with indices @xmath25@xmath123 and maximum energies @xmath2650 @xmath23 . _
the `` standard '' electron spectrum expected from the diffusive ( 1st order fermi ) shock acceleration in the non - relativistic test - particle limit , @xmath27 , is not observed .
the overall curved spectra are likely to be intrinsic rather than due to any absorption / cooling effects .
terminal shocks of powerful jets , like those in cygnus a , are probably ( mildly ) relativistic , with an oblique magnetic field configuration . as such
, it should not be surprising that the diffusive acceleration process known from the non - relativistic test - particle models are not directly applicable , i.e. , the lack of a canonical shock - type spectrum @xmath28 see begelman & kirk ( 1990 ) and niemiec et al .
( 2006 ) for relevant discussions .
the break frequency corresponding to the @xmath23 mass ratio preceded by a flat - spectrum power - law is also an interesting finding , which confirms some previous evidences for such in other powerful hotspots ( see leahy et al .
1989 ) . our results ,
although inconsistent with the simplest modeling of terminal hotspots in radio galaxies as sites of the efficient 1st order fermi acceleration , are instead in good agreement with our current understanding of relativistic shock waves . in particular ,
a flat power - law electron spectrum up to @xmath23 energies is consistent with the simple ( 1d ) resonant acceleration model discussed by amato & arons ( 2006 ) for relativistic shocks .
their simulations show that as a result of resonant absorption of cyclotron emission produced by cold protons reflected from the shock front , a flat power - law electron energy distribution emerges between energies @xmath29 and @xmath30 , the latter possibly reduced by a factor of a few due to thermal dispersion in the upstream proton momenta ( @xmath31 is the bulk lorentz factor of the upstream medium ) .
the slope of this distribution is not well constrained by the simulations , but can be relatively flat and appears to depend on the plasma content ( i.e. , on the ratio of proton to electron number density ) .
although these particular simulations involve a large @xmath31 plasma , analogous processes are expected to take place also at mildly relativistic shocks considered in hotspots like cygnus a ( @xmath32 few ) , so the model results can at least be applied qualitatively .
if such an association is correct , and if the results of amato & arons ( 2006 ) are applicable , then it would automatically imply that ( _ i _ ) the jets in powerful radio sources like cygnus a are made of an electron - proton rather than electron - positron plasma , ( _ ii _ ) a significant fraction of the jet kinetic power is carried by cold protons , and ( _ iii _ ) the number density of protons within the jets is most likely lower than that of the electron - positron pairs ( see the discussion in stawarz et al .
we note that these conclusions regarding the jet content would then be consistent with those presented for quasar jets ( e.g. , sikora & madejski 2000 ) , which are widely regarded as beamed counterparts of fr ii radio galaxies like cygnus a. this work is based on observations made with the _
spitzer space telescope _ , which is operated by the jet propulsion laboratory , california institute of technology under a contract with nasa .
the nrao is operated by associated universities , inc . under a cooperative agreement with the national science foundation .
s. and m. o. were supported by mein through the research project 1-p03d-003 - 29 in years 2005 - 2008 .
the work at sao was partially supported by _
spitzer _ grant 1279229 . | we report new detections of the hotspots in cygnus a at 4.5 and 8.0 microns with the _ spitzer space telescope_. together with detailed published radio observations and synchrotron self - compton modeling of previous x - ray detections , we reconstruct the underlying electron energy spectra of the two brightest hotspots ( a and d ) .
the low - energy portion of the electron distributions have flat power - law slopes ( @xmath0 ) up to the break energy which corresponds almost exactly to the mass ratio between protons and electrons ; we argue that these features are most likely intrinsic rather than due to absorption effects . beyond the break ,
the electron spectra continue to higher energies with very steep slopes @xmath13 .
thus , there is no evidence for the ` canonical ' s=2 slope expected in 1st order fermi - type shocks within the whole observable electron energy range .
we discuss the significance of these observations and the insight offered into high - energy particle acceleration processes in mildly relativistic shocks . |
we gratefully acknowledge project management support by p. leisching .
the work was funded by the german ministry of economics and technology ( zim projects kf2303709 and kf2806204 ) . | we investigate the tuning behavior of a novel type of single - frequency optical synthesizers by phase comparison of the output signals of two identical devices .
we achieve phase - stable and cycle slip free frequency tuning over 28.1 ghz with a maximum zero - to - peak phase deviation of 62 mrad .
in contrast to previous implementations of single - frequency optical synthesizers , no comb line order switching is needed when tuned over more than one comb line spacing range of the employed frequency comb .
a single - frequency optical synthesizer ( sos ) is the analog of an electrical synthesizer in the optical domain .
it provides a single - frequency optical field whose frequency and phase can be arbitrarily adjusted within a certain spectral range and resolution while it is related to a reference signal in a phase - coherent fashion . in an ideal case
it combines the best of two worlds in a tunable way , i.e. the spectral resolution of narrow linewidth frequency stabilized lasers with the broad spectral coverage of frequency combs .
+ such an sos has applications in basic research e.g. in precision spectroscopy @xcite , frequency metrology and quantum optics but could also enable the development or improvement of a number of useful practical devices in optical metrology like high resolution optical spectrum or vector analyzers , phase stable cw thz - synthesizers @xcite , high resolution , high dynamic range optical frequency domain reflectometers based on frequency scanning , optical coherence tomography @xcite or traceable range finders and laser trackers .
+ the invention of frequency combs @xcite , which consist of a broadband fixed frequency grid with well - defined frequency and phase relations between the individual comb lines , has enabled the most sophisticated sos approaches up to date @xcite .
these implementations are based on phase - locking of a single - frequency `` clean - up '' laser to an individual comb line of a frequency comb and subsequent tuning of the repetition rate @xcite or tuning of the offset frequency between `` clean - up '' laser and comb line @xcite .
the tuning speed and range is hampered in the first approach due to the required macroscopic resonator length variation .
+ the approaches based on offset frequency tuning suffer from an ambiguity at `` critical '' optical frequencies , where the beat frequency between a specific comb line and the `` clean - up '' laser becomes zero or coincides with the beat frequency from an adjacent comb line .
this leads to forbidden frequency gaps and limits the agility of the sos .
moreover , these ambiguities need to be resolved in a complex way which imposes additional technical overhead . if forbidden frequency gaps at the critical frequencies are to be avoided , an offset frequency tuning based sos needs to take a smart action when approaching these frequencies . in @xcite for example , a second , frequency shifted beat note is generated and used as alternative error signal when critical frequencies are approached . in that case , the sos will have problems to preserve phase stability when switching between error signals .
+ the aforementioned complications can be circumvented by combining a novel method for frequency shifting the carrier frequency of frequency combs that we have reported on in an earlier paper @xcite , with a fixed offset phase lock of a `` clean - up '' laser to a single comb line . in this case , frequency tuning of the comb spectrum avoids any problematic ambiguity since all rf - beat frequencies are constant in time and conjugate beat signals never coincide .
such a universal sos consists of three elements , i.e. a mode - locked laser acting as frequency comb generator ( ofc ) , the mentioned frequency shifter for frequency combs @xcite and a single line selector @xcite which ensures single frequency operation .
the latter is implemented with a phase lock of a `` clean - up '' laser to a single comb line .
+ in this paper we present the first characterization of the tuning properties of an sos based on such a frequency shifter . in order to characterize the carrier phase fidelity of the sos during tuning action we duplicate the system and use the second sos as reference enabling the treatment in the rotating frame .
we show that we can achieve phase - stable tuning of our sos with respect to the reference over the full tuning range of the `` clean - up '' laser . + the principle of the frequency shifter relies on serrodyne frequency shifting @xcite the carrier frequency of an ofc .
it exploits the fact that the spectrum of the ofc corresponds to a periodic pulse train in the time domain and shifts the optical carrier phase between subsequent pulses of the ofc . to explain the principle , the definition of the instantaneous frequency of the monochromatic signal of the comb line with order number @xmath0 can be used .
@xmath1 the frequency evolution @xmath2 of the signal is given by a constant frequency @xmath3 plus a change of the carrier phase @xmath4 per unit time @xmath5 . in the frequency domain the frequency comb can be treated as a multitude of monochromatic fields , each experiencing the frequency evolution according to eq.([equ1 ] ) . in the time domain
the monochromatic fields interfere destructively at the dark intervals of the pulse train .
the natural time interval for an incremental phase change is thus the temporal pulse - to - pulse spacing , i.e. @xmath6 . the pulse - to - pulse change @xmath4 of the carrier phase is implemented by adding consecutively changing phase command values by means of an electro - optic phase modulator ( eom ) .
the sequence of phase command values required for a target temporal evolution of the frequency shift is determined by temporal integration of the second term in eq.([equ1 ] ) .
a simple constant carrier frequency shift thus corresponds to a linear stepwise increase of the carrier phase between subsequent pulses .
the cubic phase evolution of fig.[setup]a accordingly leads to a parabolic frequency evolution of the comb .
since the electromagnetic field is @xmath7 periodic , the phase command values can be applied modulo @xmath7 , avoiding their divergence and thus of the voltages applied to the eom .
+ the incremental phase changes @xmath4 between pulses and the @xmath7 flyback events in fig.[setup]a occur at the dark intervals of the pulse train . due to the pulsed nature of the comb electromagnetic field , only the effective phase values which are present during the optical pulse length contribute to the frequency shifting effect .
thus , possible imperfections of the phase steps , e.g. transients or ringing , are concealed and the generation of spurious frequencies is suppressed .
consequently , the phase evolution between the pulses can be designed in an arbitrary fashion . + a detailed description of the frequency shifter for frequency combs and experimental proof that frequency tuning of a comb over multiple @xmath8 is readily achieved can be found in @xcite .
+ in our experiment we use an integrated eom ( eospace ) to set the carrier phase of subsequent pulses of a frequency comb . ,
arrows : @xmath7 flybacks , solid line : possible phase evolution at eom .
b : setup of the experiment consisting of two channels cha and chb .
fpga : field programmable gate array , eom : electro - optical modulator , bpf : band pass filter ( 50 ghz ) , dl1/2 : external cavity diode laser 1 and 2 , pd : photo diode , pll : phase locked loop , ool : out of loop measurement . ]
the setup consists of two identical , but independent channels , a and b , as shown in fig.[setup]b .
a fiber splitter divides the output power of a frequency comb generator ( toptica femtoferb sync , @xmath9=1550 nm , @xmath8= 56 mhz ) in two equal parts of 30 mw average power and sends them to the two identical eoms .
the phase command values for both channels are generated with the help of two independent numerically controlled oscillators ( nco , details in @xcite ) that are implemented in one fpga ( stratix iii ) .
the light fields in both channels are thus manipulated completely autonomously by digital - to - analog converting the phase command values and sending them to the eom by means of an rf - signal .
the fpga is clocked with a harmonic of @xmath8 , hence the changes @xmath4 and the @xmath7 flybacks of the phase command value are synchronized with the comb s pulse train .
the signal gain of the phase command values is adjusted such that the @xmath7 flybacks correspond to a carrier phase shift of @xmath7 to the eom . for the sake of simplicity
we describe only channel a in the following .
the transmitted , frequency shifted light is filtered by a 50 ghz band - pass filter to improve the signal to noise ratio of the beat between a selected comb line @xmath0 and an external cavity diode laser ( dl1 , toptica dl pro design , @xmath9=1550 nm ) .
the light fields from dl1 and the filtered comb are superimposed on a photodiode ( pd1 ) using polarization maintaining fiber splitters .
the electric beat signal of the pd is used to phase - lock dl1 to the selected comb line @xmath0 .
the feedback loop is closed using a fast analog controller ( toptica falc 110 ) which acts on the piezo and current of dl1 .
additionally , the fpga presets the desired frequency evolution of dl1 by means of a baseband feed - forward signal to the piezo of the laser .
thus , the phase locked loop ( pll ) has to compensate only for imperfections of the feed - forward control , e.g. due to piezo hysteresis .
+ the remaining light from dl1 is split in two parts by a polarization maintaining fiber splitter .
one part is sent to an out - of - loop measurement ( ool ) of the laser frequency implemented with a delayed self - heterodyne interferometer of known length .
the other part is sent to a photodiode ( pd3 ) where it is superimposed with the corresponding light field from channel b , i.e. the light field from dl2 phase - locked to the same comb line @xmath0 .
the offset frequencies of the phase locks of laser dl1 and dl2 to the selected comb line @xmath0 differ by 1 mhz .
the resulting 1 mhz beat signal is detected at pd3 , digitized with a sampling rate of 7 mhz by a 16 bit daq card and stored on a computer . in a post - processing step ,
the differential phase of the two light fields during a desired frequency scan is extracted .
this phase deviation is deduced from the raw data using a software phase - retrieval algorithm @xcite .
effectively , it compares the phase of the 1 mhz beat with the differential phase between the pll offset signals .
since both channels share the same frequency comb generator , frequency and phase fluctuations of this `` master oscillator '' are ruled out via common - mode rejection and solely the phase deviation introduced by the frequency shifter is measured . + the setup in fig.[setup]b represents a two - beam interferometer with two separate paths from the eoms to pd3 .
both signal path lengths of about 11 m have to be matched to a degree of 300 @xmath10 in order to avoid moving spectral fringes during tuning .
length matching is accomplished with the help of a free space delay line in channel a. + to characterize the tuning behavior of the sos both lasers are swept simultaneously over 500 comb lines , or 28.1 ghz following a sinusoidal frequency evolution . of the beat note between dl1 and dl2 during the same sweep ( blue ) .
sidebands at 50 and 100 hz are caused by power line noise .
fft - filter function used to calculate @xmath11 ( red ) .
bottom : result of the out - of - loop frequency measurement of dl1 during the sweep .
the coarse frequency measurement has an uncertainty of @xmath12 mhz . ]
fig.[fig3 ] shows the phase deviation between dl1 and dl2 ( top ) and the frequency of dl1 ( bottom ) for this experiment . during this 2 s long sweep
the relative phase is stable within @xmath13 rad .
high frequency technical noise contributions ( @xmath14 = 55 mrad , 4 hz @xmath15 3.5 mhz ) are found together with a slow drift on the timescale of seconds .
this slow drift is caused by fluctuations of the interferometer path length difference which is not controlled .
although the frequency shifting method is in principle capable of controlling the absolute phase value of the optical output field in a deterministic way , this has not yet been implemented here .
unwanted interferometric phase drifts could be reduced , e.g. by reducing the length of the fiber pigtails of the components , temperature stabilizing the interferometer setup or ultimately by active fiber length stabilization .
+ the inset in fig.[fig3 ] shows in blue the power spectral density @xmath11 ( rf - spectrum ) of the beat note between both dls during their common , 28.1 ghz wide frequency sweep .
@xmath11 was calculated from the raw data of fig.[fig3 ] ( top ) using fft .
its 3db - width of 0.7 hz is slightly broader than the fft - filter function ( in red , 0.5 hz ) used to calculate the spectrum and the peak is 30 db suppressed with respect to the filter function .
this reflects the onset of a carrier collapse of the filtered rf - signal and is in accordance with the interferometric phase drifts found in fig.[fig3 ] ( top ) .
the spectrum represents a beat - note in the `` rotating - frame '' and is characteristic for the two - sos - scheme employed here .
+ the tuning speed of the sos reaches 45 ghz / s at the steepest slope of the 28.1 ghz sinusoidal modulation or up to 120 ghz / s for frequency modulations on the order of 200 mhz .
tuning speed and range are solely limited by the `` clean - up '' technique , i.e. by the bandwidth of the locking electronics and the mode - hop free tuning range of the diode laser .
the frequency shifting method itself , disregarding the clean - up laser , is capable of tuning speeds of up to tens of thz per second combined with a tuning range over the whole bandwidth of the frequency comb .
+ to illustrate that the sos is running cycle slip free , we intentionally provoke a cycle slip in the pll of dl1 by a modification of the locking parameters .
the cycle slip manifests itself as a @xmath7 jump in the differential phase between dl1 and dl2 as shown in fig.[fig4]a . from the absence of this signature in measurements with optimized locking parameters
, we deduce that the sos is operating cycle slip free .
+ in the following an experiment to characterize the phase deviations at the critical frequencies mentioned in the introduction is described . for our sos
the most relevant critical frequencies correspond to the frequencies of the initially unshifted comb lines .
this can be explained by small deviations of the eom driving signal from the ideal shape , e.g. non perfect-@xmath7-flybacks , which are sampled by the light pulses and lead to spurious sidebands at the optical output spectrum of the frequency shifter .
the spurs occur at the frequencies of the initial unshifted comb lines and are typically suppressed by 30 db with respect to the shifted comb lines .
these spurs in the frequency domain induce pll error signal glitches upon tuning the `` clean - up '' laser across the spur and show up as systematic phase deviations with amplitudes comparable to the technical noise of fig .
[ fig3 ] in our measurements . to measure the systematic phase deviations of an individual sos
, any common - mode suppression of the glitch during tuning transits across the spur has to be avoided .
dl1 and dl2 are thus linearly swept over the spur with a frequency shifter induced detuning of 5 mhz .
then , phase - tracking their beat note while tuning reveals the individual phase deviations at the glitch positions since the individual transits of both sos take place at different time instants .
thus , the true phase deviation of dl1 at the critical frequency is measured , while dl2 acts as an unperturbed local oscillator .
fig.[fig4]b shows the phase deviation produced when dl1 is slowly swept over such an unshifted comb line .
the trace shows the average of 100 consecutive sweeps with interferometric phase offsets subtracted .
the averaging reduces the technical noise and reveals a systematic phase glitch with a zero - to - peak deviation of 62 mrad .
the interference between the swept signal and the spur leads to an oscillatory shape of the glitch .
the deviation from a pure oscillatory shape results from the pll response .
the observed phase glitch is consistent with the expected phase error caused by spurs which are 30 db below the carrier . to the best of our knowledge
this is the first time phase stable tuning of an sos across critical frequencies is achieved and characterized . + the frequency resolution of the frequency shifter is given by the product of @xmath8= 56 mhz and the length of the nco tuning word .
the algorithm implemented in the fpga firmware uses a phase resolution of @xmath16 bits , leading to a frequency resolution of the soss center frequency of 200 nhz .
it should be noted that the smaller amplitude resolution ( 12 bit ) of the digital to analog converter used to generate the analog phase command values does not influence the frequency resolution , but determines the so called spur - free dynamic range and thus the maximum phase error .
the frequency resolution could even be improved by using more bits in the fpga implementation .
+ the short - term linewidth of dl1 locked to the free running unshifted comb corresponds to the linewidth of dl1 in the non - rotating frame and was measured to be well below 10 khz at 1550 nm using a delayed self - heterodyne interferometer .
the minimum absolute frequency uncertainty achievable with this sos depends on the choice of frequency reference for the frequency comb and can approach the values of the best optical / microwave clocks .
+ in conclusion , we have demonstrated phase - predictable tuning of two independent , common mode suppressed soss .
in contrast to previous implementations of soss , our approach allows for arbitrary , set - point - switching free tuning over the full spectrum of the used comb oscillator , limited only by the applied `` clean - up '' technique ( 28.1 ghz ) .
the maximum zero - to - peak deviation of the phase found at critical frequencies was shown to be 62 mrad and the maximum tuning speed realized was 120 ghz / s . in combination with well studied methods for self - referencing of frequency combs and widely tunable lasers
, this technique could stimulate the development of a compact sos , capable of delivering a tunable absolute optical frequency of high precision . |
the recent development of atom trapping technologies @xcite has led to the opening of new research fields of many - body physics . in particular , the possibility of controlling to a large degree the shape of the trapping potential has permitted the experimental realization of nearly isolated low dimensional interacting boson systems @xcite , and the observation of the bosonic mott insulator @xcite .
moreover , the use of fano - feshbach resonances @xcite allows experimentalist to control interaction strength in the low dimensional atomic gases @xcite .
this has permitted the observation of the girardeau @xcite fermionization of one - dimensional bosons with strongly repulsive interactions @xcite . from the theoretical point of view , fermions as well as bosons with repulsive interactions in one dimension are in the tomonaga - luttinger liquid state @xcite . the bosonization technique @xcite allows the description of their low - energy physics in terms of collective phonon - like modes as well as the computation of the various correlation functions .
it is found that in the tomonaga - luttinger liquid state at zero temperature , only quasi - long range order ( whether superfluid or density wave ) can be obtained , with correlation functions decaying as power laws , the exponent of the power law being a rational function of the tomonaga - luttinger parameter @xmath0 .
this parameter @xmath0 depends on interaction . in the case of bosons with contact interaction
@xcite , this parameter can be calculated exactly , and it is found that @xmath1 making superfluidity the dominant correlation . when the interaction becomes extremely repulsive , or the density becomes very low ,
the girardeau model is recovered @xcite and the exponent @xmath0 goes to 1 . in the case of spinless fermions , perturbative calculations @xcite show that the exponent @xmath0 is smaller than 1 for repulsive interactions ( thus favoring density waves ) and larger than 1 for attractive interactions ( thus favoring a @xmath2-wave type superfluidity ) . for finite temperatures ,
the correlation functions decay exponentially , with a correlation length inversely proportional to temperature .
such behavior gives rise to susceptibilities that diverge as a power law with temperature , with an exponent determined by the zero temperature correlation exponent .
recent experiments @xcite with trapped atoms have shown the possibility of realizing many - body systems mixing fermions and bosons . from the theoretical point of view , mixtures of bosons and spinless fermions in one dimension were considered in refs . .
it was found that sufficiently strong repulsion or attraction between fermionic and bosonic atoms could result in respectively a phase separation or a collapse , while interactions of moderate strength would lead ( in the case of incommensuration between the atom densities ) to the formation of a two - component tomonaga - luttinger liquid . in refs .
, it was proposed to describe the two - component tomonaga - luttinger liquid in terms of polarons , and various exponents of correlation functions were obtained .
an integrable model of interacting bosons and fermions was considered in refs . , in which a two - component tomonaga - luttinger liquid was found .
the mixture has also been investigated in numerical calculations @xcite .
more recent experimental work has considered mixtures of fermionic atoms .
motivated by these experiments , there has been some numerical studies of mixtures of three fermionic species @xcite .
although the one - dimensional boson fermion mixture has been studied extensively including analytical expression of the exponents@xcite of the equal time correlation function , the previous work has emphasized the polaron correlations over the density wave correlations .
it is thus worthwhile to analyze the competition between density - wave and superfluid correlation as interactions in the system are varied . besides the question of the effect of interaction on the exponents , an important question relevant for experiment is the evolution of the exponents as a function of relative concentration of fermions and bosons in the mixture .
fermions could have two opposite effects . on the one hand
, fermionic particles can give rise to retarded attractive interactions between the bosons , which can compensate the direct boson - boson repulsion and favor superfluidity and possibly collapse of the mixture .
on the other hand , fermionic particles do not form bose condensates , and may thus disfavor the formation of a superfluid state .
for the simple bose - fermi mixture model , it is useful to examine the effect of the doping on the correlated states of respective particles , e.g. , how the superfluidity is enhanced or suppressed by the interaction between boson and fermion .
a last question , in order to compare numerical and analytical result , is to develop a phenomenological bosonization scheme allowing for the calculation of the parameters of the low energy effective hamiltonian from ground state energy calculations . in the present article
, we present a study of mixtures of fermionic or bosonic atoms using bosonization and a green s function formalism . after discussing briefly in sec .
[ sec : hamiltonian ] the phenomenological bosonized hamiltonian of the model and its relation with the macroscopic observables , we focus in sec .
[ sec : correlations ] on the calculation by green s function techniques of the exponents for superfluidity and density waves .
we find a relation between the products of density wave and superfluid exponents for the bosons and the fermions .
we also give an expression of the correlation lengths at finite temperature . in sec .
[ sec : bragg - scattering ] , we use the green s function technique to obtain the bragg scattering intensity following ref . .
then , in sec . [ sec : toy - model ] , we consider the effect of the variation of the density of one of the two atom species first in the case of the model of ref . .
we show that for this model , the variation of the bosonic superfluid exponent with fermion density is not monotonous in the case of a weak interaction .
we also consider the effect of the effective current - current interaction and show that it suppresses superfluidity .
we consider a mixture of spinless fermions and bosons .
the hamiltonian reads : @xmath3 \nonumber \\ & + & \frac 1 2 \int dx dx ' [ v_{ff}(x - x ' ) \rho_f(x ) \rho_f(x ' ) \nonumber \\ & + & v_{bb}(x - x ' ) \rho_b(x ) \rho_b(x ' ) + 2 v_{bf}(x - x ' ) \rho_b(x ) \rho_f(x ' ) ] , \label{eq : hamiltonian}\end{aligned}\ ] ] where @xmath4 are the masses of ( respectively ) the fermionic and bosonic atoms , @xmath5 ( respectively ) the fermion - fermion , boson - boson and fermion - boson interactions , and @xmath6 the ( respectively ) bosonic and fermionic annihilation operators .
we have also defined the density operators : @xmath7 .
the model ( [ eq : hamiltonian ] ) can be treated by the method of bosonization @xcite .
when the boson and fermion densities are not commensurate with each other , the bosonized hamiltonian reads @xcite : @xmath8 , \end{aligned}\ ] ] where @xmath9=i \delta_{a , b } \delta(x - x')$ ] and @xmath10 .
indeed , the hamiltonian ( [ eq : hamiltonian ] ) is invariant under a parity transformation .
under such a transformation , @xmath11 and @xmath12 , so that quadratic terms @xmath13 change sign under parity , and eq .
( [ eq : bosonized ] ) is the most general hamiltonian quadratic in @xmath14 and @xmath15 for @xmath16 .
the matrices @xmath17 and @xmath18 in eq .
( [ eq : bosonized ] ) are real symmetric and are deduced from variations of the ground state energy @xmath19 of a finite system of size @xmath20 with ( respectively ) changes of boundary conditions @xmath21 and changes of particle densities @xmath22 : @xmath23 in the case of a galilean invariant hamiltonian such as eq .
( [ eq : hamiltonian ] ) , it is possible to further constrain the elements of the matrix @xmath24 with the relations ( see app .
[ app : galilean ] ) : @xmath25 so that knowing just one of the parameters @xmath26 , @xmath27 , and @xmath28 fully determines the matrix @xmath24 .
if one starts from the limit @xmath29 and applies bosonization first , and then reintroduces the interaction @xmath30 as a perturbation , to lowest order one obtains the hamiltonian : @xmath31 \nonumber \\ & + & \frac{v_{bf}}{\pi^2 } \int dx \partial_x \phi_b \partial_x \phi_f , \end{aligned}\ ] ] in which @xmath32 as a result of galilean invariance of eq .
( [ eq : hamiltonian ] ) , @xmath33 where @xmath34 is the density of particles , and @xmath35 is the compressibility . in the case of non - interacting fermions
@xmath36 , one has @xmath37 .
for hard core bosons @xcite , @xmath38 , while for the lieb - liniger model @xcite , @xmath39 , one has @xmath40 , with @xmath41 when @xmath42 and @xmath43 for @xmath44 .
note that at this low order in perturbation theory , no term @xmath28 is present .
this term is expected to appear in second order perturbation theory , along with corrections to the bare terms @xmath26 and @xmath27 . a hamiltonian with quadratic interactions similar to eq . ( [ eq : bosonized ] ) but also comprising interband tunneling terms was considered in the context of a two band model of interacting spinless fermions @xcite . a hamiltonian equivalent to eq .
( [ eq : bosonized - perturb ] ) has been studied by path integral methods as a model of one - dimensional electrons interacting with acoustic phonons @xcite . due to the quadratic character of the hamiltonian ( [ eq : bosonized ] ) , its spectrum
is readily obtained @xcite as two branches @xmath45 with linear spectrum , showing that its ground state is a two component tomonaga - luttinger liquid provided the velocities @xmath46 of both components are real .
the vanishing of the velocity @xmath47 is an indication of instability @xcite either towards phase separation ( in the case of repulsive boson - fermion interaction ) or collapse ( in the case of attractive boson - fermion interaction ) .
from the diagonalization of the hamiltonian ( [ eq : bosonized ] ) , bosonization allows to obtain the exponents of the various correlation functions @xcite . in this paper
, we will use a different ( but equivalent ) approach to compute the correlation functions of the system . by an equation of motion method
, we first obtain the green s functions : @xmath48 where @xmath49 and @xmath50 can be @xmath51 or @xmath52 . from these green s functions ( [ eq : green1 ] ) and ( [ eq : green2 ] ) , we find the correlation functions of exponential fields as : @xmath53 where @xmath54 and @xmath55 are real numbers .
such method is of course applicable to cases with more than two components , as long as the hamiltonian remains quadratic in the fields @xmath56 and @xmath57 . to derive the green s functions , we start from the equations of motions in matsubara time of the fields @xmath56 and @xmath14 read : @xmath58= -i \pi \sum_b m_{ab } \pi_b(x,\tau ) , \\
\partial_\tau \pi_a(x,\tau)&=&[h,\pi_a]= -\frac
i \pi \sum_b n_{ab } \partial^2_x \phi_b . \end{aligned}\ ] ] the equations of motions for the green s functions @xmath59 thus read : @xmath60 going to fourier space , we obtain ( @xmath61 ) : @xmath62 where @xmath63 . from eq .
( [ eq : green - phi - fourier ] ) : @xmath64 where we have introduced the cutoff @xmath65 . by using the duality transformation @xmath66 , @xmath67
, we obtain equations of motion for @xmath68 similar to eq .
( [ eq : eom - green ] ) with the roles of @xmath24 and @xmath69 exchanged .
thus , in fourier space , we have : @xmath70 the expressions ( [ eq : green - phi - fourier ] ) and ( [ eq : green - theta - fourier ] ) show that the retarded green s functions have poles for @xmath71 equal to @xmath72 where @xmath73 are the two eigenvalues of @xmath74 .
stability requires that @xmath75 and thus @xmath76 . since we know from app .
[ app : galilean ] that @xmath77 in a galilean invariant model , instabilities occur for @xmath78 . from the definition of @xmath69 , eq .
( [ eq : definition - n ] ) , such instabilities are either collapse or phase separation .
further , for zero temperature and @xmath79 , we can obtain a general form for @xmath80 using eq .
( [ eq : fourier - def ] ) .
we find : @xmath81 then equations ( [ eq : exponentials ] ) and ( [ eq : exponentials2 ] ) lead to : @xmath82 where @xmath83 .
we can define the matrices @xmath84 and @xmath85 .
these matrices yield the exponents for the exponential fields .
one can see that the duality relations become @xmath86 .
for the two component system , one has the identity : @xmath87 which can be checked by applying the right hand side of the formula to each eigenvector of @xmath74 .
we thus have : @xmath88 in the case where @xmath89 , we see that @xmath90 will have matrix elements going to infinity as @xmath91 , whereas @xmath92
. therefore , near a collapse or a phase separation the density wave exponents are divergent , while the superfluid exponents remain finite . for nonzero temperature , the sum eq .
( [ eq : fourier - def ] ) is dominated for long distances by the term with @xmath93 .
one finds that : @xmath94 so that the correlation functions decay exponentially for long distances , @xmath95 with thermal correlation lengths given respectively by @xmath96 and @xmath97 .
we note that near the instability , the correlation length @xmath98 goes to zero , in accordance with the reduction of density wave ordering found at zero temperature , while the length @xmath99 remains finite . to characterize the atomic density - wave ( adw ) ordering of the fermions , the field operator of which is given by : @xmath100 where @xmath101 is the fermi wavevector
, @xmath65 is a short - distance cutoff , and @xmath102 label the two fermi points , we have to calculate the correlation function of the @xmath103 component of the density operator @xmath104 , namely @xmath105 : @xmath106 the density wave exponent @xmath107 is : @xmath108 .\end{aligned}\ ] ] for @xmath29 , the exponent of eq .
( [ eq : exponent - adw - f ] ) reduces to @xmath109 . near the collapse ( for @xmath110 ) or the phase separation ( for @xmath111 ) which is obtained at @xmath112
, we note that the exponent @xmath113 is diverging as @xmath114 .
the fermionic density - density correlation at small wavevectors can also be obtained from bosonization .
we have : @xmath115 where we have defined @xmath116 .
for the bosonic density wave fluctuations , using the haldane expansion of the density , @xmath117 we find that the density density correlation function of the bosons reads : @xmath118 the dominant correlations are at wavevector @xmath119 and are characterized by the exponent : @xmath120 .\end{aligned}\ ] ] this exponent is obtained from the fermionic exponent by the exchange @xmath121 .
the exponent is also divergent when @xmath122 .
it is also interesting to consider the cross correlations between bosonic and fermionic density .
one has : @xmath123 so that the non - uniform components of the densities of bosons and fermions remain uncorrelated .
the cross density correlations vanish when @xmath29 and are positive when @xmath110 as a result of the boson - fermion attraction . for finite temperature , the correlation functions of eqs .
( [ eq : fermion - density - correlations ] ) and ( [ eq : boson - density - correlations ] ) decay exponentially , with correlation lengths given by : @xmath124 fermions , due to their spinless character , can only present p - wave type superfluidity .
the order parameter is @xmath125 and can be expressed using the decomposition of eq .
( [ eq : fermion - bosonized ] ) in the form : @xmath126 the order parameter for p - wave superfluidity exhibits algebraic correlations : @xmath127 with the exponent for fermion superfluidity : @xmath128 .\end{aligned}\ ] ] we now turn to the superfluid fluctuations of the bosons . the quasi - long range superfluid order is characterized by the correlation function : @xmath129
our result for the superfluid exponent @xmath130 is : @xmath131 , \end{aligned}\ ] ] the superfluid and the density wave exponents are not independent from each other . indeed , noting that @xmath132 , we have that : @xmath133 and @xmath134 .
this implies that the exponents satisfy the relation @xmath135 .
turning to the finite temperature case , the thermal lengths are : @xmath136
according to ref . , the bragg scattering intensity is proportional to the imaginary part of the retarded density - density response functions @xmath137 .
retarded density response functions @xmath138 ( @xmath139 ) can be obtained from the fourier transform of green s functions ( [ eq : green1 ] ) as : @xmath140 using the expression of the fourier transform ( [ eq : green - phi - fourier ] ) from sec .
[ sec : correlations ] we obtain for @xmath141 : @xmath142 .
\end{aligned}\ ] ] equation ( [ eq : im - chi ] ) predicts peaks at frequencies @xmath72 . the matrices @xmath24 and @xmath69 can be deduced from the spectral weight of these peaks .
one has the following sum rules : @xmath143 the first sum rule is simply a restatement of our derivation of the equal time green s function in sec . [
sec : hamiltonian ] .
the second sum rule is a kramers - kronig relation giving the real part of the zero frequency ( matrix ) density - density response function as an integral of its imaginary part . since the real part of the density density response function is the compressibility , the second result is not surprising .
the last sum rule is a consequence of current conservation . indeed , using current conservation , one can relate the density - density response function to the current - current response function . using again a kramers - kronig , the last integral is shown to be equal to the static current - current response function .
in this section , we first consider the model of eq .
( [ eq : bosonized - perturb ] ) with @xmath144 .
for the two - component case ( @xmath145 ) , the velocities are found as @xcite : @xmath146 using the green s function methods of sec .
[ sec : hamiltonian ] , we obtain the following expressions for the exponents : @xmath147 , \\ \label{eq : adw - f - toy } \eta_{adw}^{(f)}&=&\frac{u_f k_f}{u_+ u_- } \left [ u_+ + u_- - \frac{u_f^2-u_b^2}{u_++u_-}\right ] , \\ \label{eq : sf - b - toy } \eta_{s}^{(b)}&=&\frac 1 { 4 u_b k_b } \left [ u_+ + u_- - \frac{u_f^2-u_b^2}{u_++u_-}\right ] , \\ \label{eq : sf - f - toy } \eta_{s}^{(f)}&=&\frac 1 { u_f k_f } \left [ u_+ + u_- + \frac{u_f^2-u_b^2}{u_++u_-}\right ] .
\end{aligned}\ ] ] finally , using that @xmath148 and @xmath149 where @xmath150 , we can express the exponents entirely as functions of @xmath151 and @xmath152 .
expanding to second order in @xmath152 we find : @xmath153 , \nonumber \\ \\
\eta_{adw}^{(f)}&= & 2k_f\left[1 + \frac{v_{bf}^2 k_f k_b(2u_f+u_b)}{2\pi^2(u_f+u_b)^2 u_f}+ o(v_{bf}^4 ) \right].\nonumber \\ \end{aligned}\ ] ] this shows that superfluidity is enhanced by the boson - fermion interaction .
the enhancement of fermionic superfluidity becomes weaker as @xmath154 is increased . by contrast ,
when @xmath154 increased , the enhancement of bosonic superfluidity is non monotonous , being maximum for @xmath155 . at the same time
, the exponent of density wave correlations for the bosons and fermions are increased by @xmath152 , indicating a reduction of the density wave quasi - long range order .
we note that the density - wave exponents eqs .
( [ eq : adw - b - toy])([eq : adw - f - toy ] ) can be rewritten respectively as @xmath156 and @xmath157 $ ] , which are both increasing functions of @xmath158 . since from eq .
( [ eq : velocities ] ) , @xmath159 is a decreasing function of @xmath160 , the superfluid exponents are decreasing functions of @xmath160 .
in other words , boson - fermion interactions lead to an enhancement of superfluid fluctuations . in the context of electrons coupled to acoustic phonons , an analogous enhancement of fermionic superfluidity
was already noticed @xcite .
the interpretation of the result ( [ eq : fermion - sf - exponent ] ) is that the sound modes of the bosonic fluid interact with the fermionic atoms exactly as the acoustic phonons interact with the electrons in the model of ref . .
the interaction of the acoustic modes with the fermions gives rise to an effective attractive interaction that enhances superfluidity .
the enhancement of bosonic superfluidity can also be understood in terms of effective attraction between bosons generated by the sound modes of the fermi fluid .
we note that in the context of electronic systems , the collapse or phase separation that results from excessively strong @xmath152 interaction is known as the wentzel - bardeen instability @xcite .
the enhancement of superfluid correlations in mixtures could be tested by varying the strength of the boson - fermion interaction @xmath152 in a @xmath161k@xmath162rb mixture using a feshbach resonance @xcite .
superfluid correlations will present a minimum when @xmath152 is tuned to zero by interaction . with the fixed @xmath163 on the plane of @xmath164 and @xmath165 which are the dimensionless quantities defined by @xmath166 and @xmath167 , respectively . for @xmath168 ( @xmath169 ) , the quantity @xmath170 increases ( decreases ) monotonously as @xmath171 ( @xmath172 ) increases . ,
width=302 ] next , we discuss the case in the presence of both @xmath173 and @xmath28 in eq .
( [ eq : bosonized ] ) , which give the hamiltonian , @xmath174 \\ & + & \frac{n_{bf}}{\pi } \int dx \partial_x \phi_b \partial_x \phi_f + \pi m_{bf } \int dx \pi_b \pi_f . \nonumber \end{aligned}\ ] ] as shown in appendix [ app : galilean ] , we have the constraint @xmath175 from the galilean invariance of the system ( see eq .
( [ eq : m - response - bf ] ) ) , so we restrict ourselves to the case of @xmath175 in the following .
the velocity is obtained as @xmath176 thus the excitation is stable for @xmath177 from eq .
( [ eq : exponent - super ] ) , the exponent for the superfluidity for boson is calculated as @xmath178 \nonumber \\ & = & \frac{1}{2k_b } \frac { 1 } { \sqrt{1 + \xi^2 + 2 \xi \tilde{n } \tilde{m } + 2\xi \sqrt{(1-\tilde{n}^2)(1-\tilde{m}^2 ) } } } \left ( 1 + \xi \sqrt{\frac{1 - \tilde{n}^2}{1-\tilde{m}^2 } } \right ) . \label{eq : exponent - adw - b_w}\end{aligned}\ ] ] where @xmath179 and @xmath180 .
the second line shows that @xmath170 depends on @xmath164 , @xmath181 , and @xmath182 .
the stable conditions for @xmath183 are expressed as @xmath184 . at the boundary of @xmath185 ,
@xmath186 becomes infinite while it stays constant at @xmath187 @xmath188 .
since @xmath164 ( @xmath181 ) has an effect of decreasing ( increasing ) @xmath186 , they compete with each other . in fig .
[ fig : fig3 ] , the contour plot of the quantity @xmath189 is shown on the plane of @xmath164 ( horizontal axis ) and @xmath181 ( vertical axis ) with the fixed @xmath190 , where the case of @xmath191 denotes the enhancement ( suppression ) of the superfluidity .
for @xmath168 , the current - current boson - fermion coupling @xmath181 suppresses the superfluidity .
the effect of the suppression reduces and vanishes , i.e. , @xmath192 on the solid line in fig .
[ fig : fig3 ] due to the compensation effect of these two coupling constants . for small @xmath171 and @xmath172 ,
@xmath193 is obtained at @xmath194 for @xmath195 , and at @xmath196 for @xmath197 . in the region of @xmath198
, we find a novel feature for small @xmath171 .
the superfluidity is enhanced by the coupling @xmath172 , and is further enhanced due to the coupling @xmath171 .
the enhancement becomes optimized , i.e. , the minimum value of @xmath199 is obtained at a certain value of @xmath171 .
we have analyzed the mixture of bosonic and fermionic atoms in one dimension using a green s function equation of motion method starting from a phenomenological bosonized hamiltonian .
we have derived expressions of the zero temperature exponents and of the thermal correlation thermal correlation lengths for the density wave and superfluid order parameters . for the exponents ,
we have recovered the exponents previously obtained for the mixture in ref .
and we have studied their behavior as a function of the fermion density for fixed interaction .
w e have found that for weak interaction , the behavior of the superfluid exponent of the bosons could be non - monotonous , although boson superfluid correlations were always enhanced with respect to the system without fermions .
such behavior can lead to a non - monotonous dependence of the superfluid transition temperature as a function of fermion density in an array of weakly coupled mixtures generalizing the bosonic array considered in ref . .
this behavior is also in contrast with the one observed in experiments with three dimensional interacting systems @xcite , where fermion doping was seen to reduce superfluidity .
in one dimensional systems at incommensurate filling , fermion doping is seen to always enhance superfluidity , the maximal enhancement being obtained near the collapse instability . by contrast , we have seen that density wave exponents are decreasing functions of the fermion density , which recover their value in the absence of fermion - boson interaction only in the limit of large fermion density .
we have also been able to predict how the relative weights of the peaks in bragg scattering intensity depend on the parameters of the phenomenological hamiltonian .
e. o. acknowledges support and hospitality from the graduate school of science and the physics department of nagoya university in june 2009 , when part of this work was completed .
let us consider a system with @xmath200 different types of particles indexed by @xmath201 with masses @xmath202 . under a galilean boost , @xmath203 and @xmath204 .
this transformation is realized by the unitary operator : @xmath205 \end{aligned}\ ] ] one has : @xmath206 where the total mass @xmath207 and the total momentum : @xmath208 is the total momentum of the system . in second quantization ,
the operator @xmath209 takes the form : @xmath210 , \end{aligned}\ ] ] and one has : @xmath211 .
therefore , in bosonization , a galilean boost takes the form @xmath212 , i.e. , @xmath213 . from the conservation equation for particles of type @xmath214 ,
@xmath215 and the bosonization relations ( [ eq : haldane - density - bosons ] ) , we have that the current density is @xmath216 . using the equation of motion , we can rewrite @xmath217 .
therefore , under a galilean boost , we will have : @xmath218 turning to the particle current , in the rest frame , the particle current will be @xmath219 .
this can be seen for instance by calculating @xmath220 . from this result , it is clear that one must have : @xmath221 for each @xmath214 . in the case @xmath222 this gives eq .
( [ eq : galilean - constraint ] ) . an alternative derivation of eq .
( [ eq : galilean - constraint ] ) is obtained by calculating the ground state energy under twisted boundary conditions ( [ eq : definition - m ] ) using second order perturbation theory @xcite .
one finds : @xmath223 where @xmath224 is the current of particles of type @xmath214 . because of the translational invariance of the system , the eigenstates of the hamiltonian are also eigenstates of the total momentum operator @xmath225 .
since the ground state is non - degenerate , this implies that @xmath226 , i.e. , for the bose - fermi mixture : @xmath227 , leading to the following expressions for @xmath17 : @xmath228 where : @xmath229 obviously , @xmath230 . using the expressions ( [ eq : m - response ] ) ,
we obtain : @xmath231 so that @xmath232 . | we study a mixture of one - dimensional bosons and spinless fermions at incommensurate filling using phenomenological bosonization and green s functions techniques .
we derive the relation between the parameters of the microscopic hamiltonian and macroscopic observables .
galilean invariance results in extra constraints for the current current interactions .
we obtain the exact exponents for the various response functions , and show that superfluid fluctuations are enhanced by the effective boson - fermion density - density interaction and suppressed by the effective boson - fermion current - current interaction . in the case of a bosonized model with purely density - density interaction ,
when the effective boson - fermion density - density interaction is weak enough , the superfluid exponent of the bosons has a non - monotonous variation with the ratio of the fermion velocity to the boson velocity .
by contrast , density - wave exponent and the exponent for fermionic superfluidity are monotonous functions of the velocity ratio . |
a classical example of self - assembly is crystallization . at low temperatures
the crystalline phase is typically stable , and thus grows spontaneously from solution through homogeneous nucleation .
if several competing crystalline phases are allowed by microscopic interactions , the efficient production of a desired phase often requires heterogeneous nucleation from a seed of this phase , with precise annealing protocols @xcite .
more complex microscopic interactions may lead to a glassy phase with many competing structures ; however , it is generally impossible to control local compositions or microscopic interactions to obtain a particular structure .
recently , there has been a dramatic change in macromolecular and colloidal assembly techniques , made possible by the use of biopolymers , such as dna , to create a large variety of inter - component interactions .
indeed , biomolecules offer exquisite control of microscopic interactions that allows self - assembly of diverse large structures .
examples range from nanoparticle assemblies @xcite , which can also form macroscopic crystals @xcite , to structures using dna itself as a building material . in the latter case , dna origami uses short dna strands to controllably fold a long backbone strand into different well controlled structures @xcite , while short strands of dna by themselves can also build up complex three dimensional objects @xcite .
similar efforts are underway using rationally - designed proteins by creating complementary binding sites on their surfaces @xcite .
up until now , however , experimental and theoretical studies have been limited to devising interactions for the assembling a single structure .
this is to be contrasted with biological systems , where many different self - assembled structures can be formed within the cell cytoplasm .
these assembled structures can in fact share some of their components and can be dynamically induced independently from one another @xcite . here , we propose a new mechanism for the self - assembly of many different structures from one large set of shared components .
each structure is multifarious , i.e. , is made out of many different types of components .
such self - assembling systems , which we propose calling `` multifarious assembly mixtures '' , are stable and yet responsive .
this means that the mixtures do not form structures spontaneously , but can be controllably induced to assemble a specific structure .
different structures are encoded " through the choice of molecular interactions and thus stored " in the mixture , to then be retrieved " by changing only a small number of parameters . the theoretical framework introduced below allows calculation of the capacity of these systems , i.e.
, how many different independent structures can be stored and retrieved in a mixture of @xmath0 species of components . in the traditional approach of self assembly without shared components , if each structure s is composed of the same number @xmath1 of different species , only @xmath2 different structures can be self - assembled .
in contrast , in multifarious assembly mixtures , many more distinct structures can be stored .
any stored structure can be retrieved with a ( super)critical nucleation seed .
multiple seeds can induce the simultaneous assembly of multiple corresponding structures .
moreover , we show that each different structure can be retrieved by changing only a small number of chemical potentials or interspecies interactions , where the number of tuned components is the number of components in a ( super)critical nucleation seed of the desired structure .
classical nucleation theory implies that the size of this seed is only weakly dependent on the size of the structure that is built .
consider @xmath0 species of interacting components in a solution kept at a constant temperature @xmath3 . in principle
, each species can have a different chemical potential @xmath4 , @xmath5 , but for simplicity we assume for now that the chemical potentials have the same value @xmath6 .
we would like the components to be able to self - assemble into one of @xmath7 distinct , multifarious structures , @xmath8 , on demand ( fig .
[ fig : generalidea]a ) . a typical multifarious structure @xmath9 is built of @xmath1 component species . in general , each species @xmath10 in the structure @xmath9 has its own multiplicity @xmath11 .
in contrast to traditional studies of self - assembly , e.g. of crystals , where the same component species appears in many copies in the assembled structure , for multifarious assembly mixtures we are interested in the case of small values of @xmath11 . indeed , for simplicity , we assume here that all component species have a single copy in every stored structures , @xmath12 , so that the number of species @xmath1 used in the structure equals the size of the structure @xmath13 . additionally , we make a simplifying assumption that all the structure sizes @xmath14 have the same value @xmath15 , so @xmath16 .
both cellular systems and recent dna - mediated assembly experiments show that a single structure @xmath9 can be robustly assembled if each pair of neighboring components , of species @xmath10 and @xmath17 ( @xmath18 ) interact through a specific binding interaction .
our next simplifying assumption is that all these interaction energies are equal : @xmath19 and we will also set all non - specific interactions to zero .
the binding interactions between different components are mediated through a discrete number of binding sites " , with a species @xmath10 having a valence @xmath20 .
for simplicity we assume that all components have the same valence @xmath21 .
how might we choose an interaction energy matrix @xmath22 so that the components are capable of assembling different desired structures @xmath8 ( fig .
[ fig : generalidea]a ) ?
the simplest general prescription that can work for arbitrary structures is to assume that two species @xmath10 and @xmath17 bind specifically with energy @xmath23 if and only if at least one of the desired structures @xmath9 requires this binding .
such a matrix @xmath22 then has the potential for `` storing '' each structure @xmath9 as a local free energy minimum ( fig .
[ fig : generalidea]b ) .
this matrix can be written as , @xmath24 this form of energy matrix implies that component species can be promiscuous in their interactions . indeed , since a given species @xmath10 binds specifically to its partners in each of the stored structures , the total number of specific binding partners for species @xmath10 can be large .
in addition to the free energy minima corresponding to the desired structures , other undesired local minima might emerge .
these correspond to chimeric structures , or chimeras " , made of chunks of different desired structures that can bind together due to the promiscuity implied by eqn .
the stability of the stored structures is determined by the size of the free energy barriers between the different minima .
for instance , if the barriers are low , chimeras will form spontaneously , even if their local free energy minima lie higher than those of desired structures ( fig . [ fig : generalidea]a ) . similarly ,
the free energy barriers between the solution of unbound components and other minima determines the solution s characteristic time @xmath25 , beyond which stored structures nucleate spontaneously and the process of the controlled retrieval of stored structures is compromised .
thus , @xmath25 is the functional `` lifetime '' of the multifarious assembly mixture .
how many different multifarious structures , each of size @xmath15 , can one store by using @xmath0 different species of components with well - chosen interspecies interactions defined by eqn ?
if each species contributed to only a single structure , the maximum capacity would be @xmath26 . by sharing species between structures , however ,
a much larger number of structures can be stored before chimeras start to dominate .
to find this increased capacity , consider components attaching to the boundary of a growing seed .
the promiscuous interactions implied by eqn . might allow the seed to bind different sets of components , resulting in chimeras .
therefore , let us compute the number of species that can specifically bind to a given boundary site of the seed .
since each component in the bulk of a stored structure has @xmath21 nearest neighbors , for an incoming component to bind stably , it must form specific bonds with @xmath27 components on the seed s boundary . due to the promiscuous nature of eqn .
, each of these @xmath27 boundary components can bind specifically to a set of @xmath28 other species .
is randomly constituted from the @xmath0 species , a given species will occur in @xmath28 of the @xmath7 stored structures and typically have a different partner in each of them .
hence , a typical species will have @xmath28 specifically binding partners . ] for randomly constituted @xmath7 structures , each set contains a fraction @xmath29 of all the @xmath0 component species .
the intersection of these @xmath27 sets , of the size @xmath30 , determines the species that can specifically bind to all the @xmath27 boundary components .
when this number is larger than @xmath31 , many different species can attach to a given boundary site on a growing seed , resulting in a proliferation of chimeras .
hence , the largest number @xmath7 of structures that can be stored is @xmath32 for @xmath33 , the exponent @xmath34 is positive and this equation implies that the capacity @xmath35 can be much larger than the traditional estimate of the capacity @xmath26 .
it is instructive to understand why @xmath36 structures i.e. , linear chains , can not share components . binding to an end of a growing chain
requires forming a bond with just one component .
if that component is promiscuous , the seed can always grow in a non - unique chimeric manner .
hence the promiscuity of individual species , implied by eqn , must be countered by the requirement on incoming particles to form multiple ( i.e. , @xmath27 @xmath37 ) bonds .
the above argument shows that the number of structures that can be stored and stabilized with @xmath0 components is large . for this to be useful
, we need to be able to retrieve each of them easily .
the retrieval can be done in three different ways .
one can introduce a nucleation seed , i.e. , a part of a stored structure , into the solution .
alternatively , one can enhance the formation of such a seed by increasing the chemical potential of its components by an appropriate amount @xmath38 , or by strengthening the interactions @xmath39 by @xmath40 for bonds found in such a seed .
these methods enhance the nucleation of one stored structure without nucleating others , despite all stored structures being made of the same set of components .
such selective nucleation is possible only for multifarious structures ; it relies on the fact that small contiguous subsets of distinct structures typically have distinct compositions .
such subsets can be used as selective nucleating seeds , or to selectively lower the nucleation barrier for one structure using the other two methods described above .
the critical question is how many different species have to be tuned in this way to successfully retrieve a particular stored structure .
the answer follows directly from general nucleation theory , which specifies a critical nucleation radius @xmath41 in terms of the chemical potential @xmath6 and bond energy @xmath42 @xcite .
the minimal seed size @xmath43 needed to recover a structure is set by @xmath41 ; smaller seeds dissolve back into components while larger seeds are supercritical and grow into stored structures .
we can make the multifarious assembly mixture responsive to smaller seeds by lowering the critical nucleation radius , for example , by lowering the ratio @xmath44 of chemical potential to bond energy ( see si text and si figs .
s2 , s5 and s6 ) .
however , lowering the critical nucleation radius also lowers the barrier to spontaneous homogenous nucleation . as noted above , critical seeds can spontaneously assemble on a characteristic timescale @xmath25 and grow into random stored structures , without any external input .
thus , at a minimum , we need @xmath25 to be much longer than the retrieval time , i.e. , time necessary for a supercritical seed to grow into a full structure .
nucleation theory , adapted to multifarious @xmath45-dimensional structures , determines @xmath25 as @xmath46 where @xmath47 is the free energy barrier , @xmath48 is the free energy per area required for creating the critical seed and @xmath49 is a time scale connected with microscopic processes . the second term on the right hand side arises because we must account for the multiplicity @xmath50 of distinct nucleation paths leading to the @xmath7 different stored structures . for small @xmath7 , we can estimate @xmath51 to account for critical seeds from different parts of the @xmath7 stored structures of size @xmath15 each ( see si text ) .
for a fixed @xmath52 , eqn . can be solved for the nucleation radius @xmath41 , and hence the minimal number of components @xmath43 that must be tuned to retrieve a structure . if all the components have a typical size @xmath53 , this number @xmath43 is of order @xmath54 where @xmath55 depends only weakly ( logarithmically ) on @xmath7 , @xmath15 and @xmath52 .
we thus conclude , that since @xmath43 is determined by the nucleation barrier , it is essentially independent of the size of the structure , @xmath15 , that is being retrieved .
note that the above equations show an unavoidable trade - off : increasing the lifetime of the multifarious assembly mixture @xmath25 necessarily increases the nucleation radius @xmath41 and hence increases @xmath43 .
thus a more stable multifarious assembly mixture requires a larger seed for recovering stored structures . in order to study different regimes of self - assembly of multifarious structures , we have considered assembly based on eqn . , on a simple @xmath56 square lattice .
individual components are square tiles that can be one of @xmath57 species .
all @xmath7 stored structures consist of @xmath58 tiles , each tile being of different species positioned inside a @xmath59 square block .
more precisely , we assume that all the species are present in all the structures , and each species appears only once in each structure , @xmath60 for all @xmath10 and all @xmath9 , so that @xmath61 for all structures @xmath8 .
in other words , each stored structure is simply a different random permutation of the tiles inside the square block .
we assume that each tile component can bind up to @xmath62 neighbors through specific binding interactions given by eqn , and that all species of tiles have the same chemical potential @xmath6 .
we run grand canonical monte carlo simulations with different numbers @xmath7 of stored structures on a square lattice of total size @xmath63 , for different values of temperature @xmath3 and chemical potential @xmath6 ( see si text and si fig .
s3 ) . starting from a particular supercritical seed ( of linear size @xmath64 ) of one of the @xmath7 stored structures , fig .
[ fig : pd ] shows a diagram of the different outcomes of our simulations , as a function of the number of stored structures , @xmath7 , and the temperature @xmath3 ( or more precisely , @xmath65 , where @xmath42 is the specific binding energy ) , for a fixed @xmath6 .
we visualize the different stored structures with different colors , with the desired structure colored in dark red . for low @xmath7 and @xmath3
, the supercritical seed indeed grows into the desired structure . in this regime of parameter space ( regime i ) , the solution behaves as a useful multifarious assembly mixture : the mixture is stable for a long time @xmath25 and stored structures can be retrieved through heterogeneous nucleation . , is much longer than the timescale for recovery , @xmath66 .
thus , even though the monte carlo dynamics do not reflect the dynamics of a realistic self - assembly system ( see e.g.@xcite ) , they do substantiate the predictions of nucleation theory , and expose different regimes of self - assembly of multifarious structures . ] for higher number of stored structures @xmath7 ( and at higher temperatures @xmath3 ) another behavior appears ( regime ii ) .
it is characterized by the spontaneous homogeneous nucleation of all stored structures from the solution : in this regime , the multifarious assembly mixture is too short lived to allow the structure retrieval , i.e. , @xmath25 becomes comparable to the time taken for a supercritical seed to grow into a full desired structure , @xmath66 . at even higher values of @xmath7
we find yet another regime of behavior ( regime iii ) , where chimeric structures dominate .
finally , at high temperatures @xmath3 , and for all values of @xmath7 , we encounter regime iv , where any initial seed disintegrates into small clusters of individual components .
the extent of different regimes depends of course on the chosen model parameters .
in particular , the chemical potential @xmath6 influences the extent of regimes i and ii ( see si text and si figs . s7 and s9 ) .
simulations presented in fig .
[ fig : diffsa ] confirm that , in regime i , the assembly of a structure can be triggered not only with supercritical nucleating seeds , but also by enhancing the chemical potential of a small set of tile species , or by increasing the bond energies between the tile species from such a set .
numerical simulations are also a way to gauge the capacity of an multifarious assembly mixture to store structures , and to compare it with the theoretical predictions presented above .
to do this , we have introduced the entire target structure as a supercritical seed , and have examined it after a fixed simulation time chosen to be shorter than the mixture s lifetime @xmath25 .
we have assessed the quality of retrieval by measuring the error , i.e. , the fraction of the final assembled structure that differs from the initial target structure ( see si text and si fig .
[ fig : scaling]a depicts the error as a function of the number of stored structures @xmath7 , for different number of particle species @xmath0 ( structure sizes being @xmath58 ) , at fixed temperature @xmath3 and chemical potential @xmath6 .
there is a transition at critical value @xmath67 , above which the error rises rapidly .
we show that the error curves for different @xmath0 collapse onto each other when plotted against @xmath68 , fig .
[ fig : scaling]b , where @xmath35 increases with increasing @xmath0 as @xmath69 with @xmath70 .
this is in a good agreement with the prediction of eq . that the memory capacity scales as @xmath71 , for the square lattice model with @xmath62 nearest neighbors ( see si text ) .
finally , we have also assessed the trade - off , expressed in eqn , between the stability of the multifarious assembly mixture , i.e. , its lifetime @xmath25 , and the minimal size @xmath43 of a seed needed for retrieval ( fig [ fig : tplots ] ) .
the minimal seed size @xmath43 increases slowly with increasing @xmath25 , and remains a small fraction of the total number of components ( @xmath72 , in this case ) in a stored structure .
the number of stored structures @xmath7 has only a modest effect on @xmath43 , in agreement with eqn ( see si text and si fig .
s7 ) . to conclude
, we have demonstrated that it is possible to store multiple structures in a solution of components with designed interactions between them . using @xmath0 different component species
, we can store as many as @xmath73 different multifarious structures of size @xmath15 and of average coordination number @xmath21 . in an extended region of parameter values
( e.g. , temperature , chemical potentials , binding energies ) , such a multifarious assembly mixture " with many stored structures is both stable and responsive ; each of the multifarious structures can be selectively grown ( retrieved ) by modifying chemical potentials or binding energies of only a small fraction of the @xmath0 component types , or by introducing an appropriate seed .
the model that we have explored is very similar to the way associative neural networks , such as hopfield s classical networks @xcite store multiple memories in a distributed way . in these models ,
a neural network is programmed to have multiple stable states , i.e. , memories , using a prescription for neuronal connections that is very similar in spirit to the pooled energy matrix in eqn .
it has been shown @xcite that if the number of programmed memories is sufficiently small , each memory is indeed a stable state and can be recovered through initial conditions in a robust manner . however , if the number of stored memories exceeds the capacity of the network , recovery is spoiled by the presence of many `` spurious memories '' undesired stable states resulting in regimes @xcite similar to those shown in fig .
[ fig : pd ] . a distinctive feature of multifarious assembly mixtures , however , is that we require the stability of the unassembled mixture itself for a long time @xmath25 , in addition to the stability of the stored structures ( see si text ) . in our simulated lattice model , different stored structures have identical components rearranged in random permutations .
thus stored structures are assumed to be independent , or
orthogonal " , as in the case of stored memories in the original hopfield model @xcite .
an important extension of the present model would be to study stored structures with built - in correlations , such as the presence of shared modules . after all , the controlled assembly of chimeric structures could be useful .
it is also important to stress that designing specific binding interactions between different components based on superposition ( eqn ) , is not the only way to create functional multifarious assembly mixtures .
although it is arguably the simplest prescription that works for generic structures , other non - linear prescriptions can be tailored for particular structures by exploiting structural motifs ( e.g. , creating multifarious assembly mixtures with higher capacity or longer lifetimes ) .
such tailored interactions have been used to store and retrieve a particular set of structures composed of a small number of component species in recent work on dna programmed assembly @xcite . in similar vein
, the ability of a protein sequence to code for multiple stored internal structures has been studied in the context of protein folding @xcite . beyond immediate applications to artificial systems with controllable binding specificity , the present model proposes a new paradigm to understand molecular aggregates in biology .
for instance , our calculations show that instead of creating new proteins for every individual structure , it is more efficient if individual proteins are used in a multiplicity of structures , as is the case in many cellular assemblies , ranging from transcription factors @xcite to ribonucleoproteins such as spliceosomes @xcite .
our calculations also indicate that such versatility can be quite high , increasing rapidly with the number of different component species in the pool .
nonetheless , different structures can be selectively assembled by reprogramming molecular interactions , e.g. by a simple modulation of the expression levels , or of the specific binding energies via post - translation modifications , of a small number of selected components .
this is indeed what seems to happen often in cellular assembly .
we hope that the theoretical framework presented here , properly generalized to far - from - equilibrium situations , will form a basis for quantitative studies of functioning , regulation and evolution of biological assembly .
we would like to thank our colleagues for discussions and their comments on the manuscript , in particular john hopfield , david huse and olivier rivoire . z.z .
acknowledges support from the george f. carrier fellowship . m.p.b .
acknowledges funding by the national science foundation through the harvard materials research science and engineering center ( dmr-0820484 ) , the division of mathematical sciences ( dms-0907985 ) , and by grant rfp-12 - 04 from the foundational questions in evolutionary biology fund .
m.p.b . is an investigator of the simons foundation .
10 lifshitz , i & slyozov , v. ( 1961 ) the kinetics of precipitation from supersaturated solid solutions . 19:35 50 .
binder , k & stauffer , d. ( 1976 ) statistical theory of nucleation , condensation and coagulation .
25:343396 .
mirkin , c. a , letsinger , r. l , mucic , r. c , & storhoff , j. j. ( 1996 ) a dna - based method for rationally assembling nanoparticles into macroscopic materials .
382:607609 .
alivisatos , a. p , johnsson , k. p , peng , x , wilson , t. e , loweth , c. j , bruchez , m. p , & schultz , p. g. ( 1996 ) organization ofnanocrystal molecules using dna . 382:609611 .
valignat , m , theodoly , o , & crocker , j. ( 2005 ) reversible self - assembly and directed assembly of dna - linked micrometer - sized colloids .
102:42254229 .
biancaniello , p , kim , a , & crocker , j. ( 2005 ) colloidal interactions and self - assembly using dna hybridization . 94:058302 .
park , s. y , lytton - jean , a. k. r , lee , b , weigand , s , schatz , g. c , & mirkin , c. a. ( 2008 ) dna - programmable nanoparticle crystallization .
451:553556 .
nykypanchuk , d , maye , m. m , van der lelie , d , & gang , o. ( 2008 ) dna - guided crystallization of colloidal nanoparticles .
451:549552 .
kim , a , biancaniello , p , & crocker , j. ( 2006 ) engineering dna - mediated colloidal crystallization .
22:19912001 .
macfarlane , r. j , lee , b , jones , m. r , harris , n , schatz , g. c , & mirkin , c. a. ( 2011 ) nanoparticle superlattice engineering with dna . 334:204208 .
rothemund , p. w. ( 2006 ) folding dna to create nanoscale shapes and patterns . 440:297302 .
douglas , s. m , dietz , h , liedl , t , hgberg , b , graf , f , & shih , w. m. ( 2009 ) self - assembly of dna into nanoscale three - dimensional shapes .
459:414418 .
wei , b , dai , m , & yin , p. ( 2012 ) complex shapes self - assembled from single - stranded dna tiles . 485:623626 .
y , ong , l. l , shih , w. m , & yin , p. ( 2012 ) three - dimensional structures self - assembled from dna bricks .
338:11771183 .
king , n. p , sheffler , w , sawaya , m. r , vollmar , b. s , sumida , j. p , andre , i , gonen , t , yeates , t. o , & baker , d. ( 2012 ) computational design of self - assembling protein nanomaterials with atomic level accuracy .
336:11711174 .
t , king , n. p , & yeates , t. o. ( 2012 ) principles for designing ordered protein assemblies .
22:653 661 .
king , n. p , bale , j. b , sheffler , w , mcnamara , d. e , gonen , s , gonen , t , yeates , t. o , & baker , d. ( 2014 ) accurate design of co - assembling multi - component protein nanomaterials . 510:103108 .
khner , s , van noort , v , betts , m. j , leo - macias , a , batisse , c , rode , m , yamada , t , maier , t , bader , s , beltran - alvarez , p , castao - diez , d , chen , w .- h , devos , d , gell , m , norambuena , t , racke , i , rybin , v , schmidt , a , yus , e , aebersold , r , herrmann , r , bttcher , b , frangakis , a. s , russell , r. b , serrano , l , bork , p , & gavin , a .- c .
( 2009 ) proteome organization in a genome - reduced bacterium .
326:12351240 .
oxtoby , d. w. ( 1998 ) nucleation of first - order phase transitions .
31:9197 .
reinhardt , a & frenkel , d. ( 2014 ) numerical evidence for nucleated self - assembly of dna brick structures .
112:238103 .
hopfield , j. j. ( 1982 ) neural networks and physical systems with emergent collective computational abilities . 79:25542558 .
hertz , j , krogh , a , & palmer , r. g. ( 1991 ) _ introduction to the theory of neural computation ( santa fe institute series)_. ( addison - wesley longman publishing co. , inc . ,
boston , ma , usa ) .
amit , d , gutfreund , h , & sompolinsky , h. ( 1985 ) storing infinite numbers of patterns in a spin - glass model of neural networks .
55:15301533 .
barish , r , schulman , r , rothemund , p , & winfree , e. ( 2009 ) an information - bearing seed for nucleating algorithmic self - assembly .
106:6054 .
fink , t & ball , r. ( 2001 ) how many conformations can a protein remember ?
87:198103 .
ihmels , j , friedlander , g , bergmann , s , sarig , o , ziv , y , & barkai , n. ( 2002 ) revealing modular organization in the yeast transcriptional network .
31:370377 .
remenyi , a , scholer , h. r , & wilmanns , m. ( 2004 ) combinatorial control of gene expression . 11:812815 .
wahl , m. c , will , c. l , & lhrmann , r. ( 2009 ) the spliceosome : design principles of a dynamic \{rnp } machine .
136:701 718 .
in our general arguments , we assumed that species @xmath10 and @xmath17 interact specifically if they are bound together in any one of the @xmath7 stored structures : @xmath74 this superposition form of @xmath75 implies that each species has multiple specifically binding partners .
we define the promiscuity of a binding site on a component of a given species as the number of species that can specifically bind to it . in our model with @xmath0 total species of components and @xmath7 stored structures of size @xmath15 each , a given species will occur in about @xmath76 stored structures .
the component will typically be bound to a different species in each of these structures .
thus the promiscuity of a typical binding site is @xmath28 . when the number of stored structures @xmath7 is large , the species interactions are highly promiscuous . as a result
, a given seed may be able to grow non - uniquely by binding distinct combinations of components and thus form chimeric structures . here
we show that there is a sudden onset of chimeras at some @xmath77 which defines the capacity . to study this
, we can specify arbitrary components along the boundary of a seed ( black tiles in fig .
s[fig : sifig_boundary_conditions]a ) and ask if there are multiple choices of species that can fill in the positions @xmath78 shown in fig .
s[fig : sifig_boundary_conditions]a , such that @xmath79 form specific bonds with each other and with all the seed components .
for example , the number of species that can form a specific bond with the component @xmath80 from the right ( i.e. , number of choices for site @xmath81 in fig .
s[fig : sifig_boundary_conditions]b ) is given by the promiscuity @xmath28 of @xmath80 s right - side binding site .
similarly , we have another set of @xmath82 choices of species for site @xmath83 that bind specifically to the component @xmath84 from below .
the choices for site @xmath85 , i.e. species that can bind specifically to both @xmath80 and @xmath84 , is given by the intersection of these two sets .
if the @xmath7 stored structures are randomly and independently constituted from the @xmath0 species , we can assume that these sets of choices for @xmath83 and @xmath81 are two random uncorrelated sets of size @xmath76 contained in the set of all @xmath0 species .
we then estimate the probability of having at least one element in the intersection of the two sets to be , @xmath86 to estimate the probability that at least one extended chimeric structure of length @xmath87 ( like that shown in red in fig .
s[fig : sifig_boundary_conditions]a ) exists , we assume a choice of species for site @xmath85 that is stably bound and compute the probability of stably - bound chimeric choices for @xmath88 , fix a choice for @xmath88 and compute choices for @xmath89 and so on . such an estimate of the probability of existence of an extended chimeric structure of length @xmath87 is given by : @xmath90 taylor expanding for small @xmath91 gives @xmath92 for large @xmath87 , this probability sharply grows at @xmath93 , where @xmath94 a similar argument applies for a general lattice structure with coordination number @xmath21 . in this more general case , a component occupying site @xmath79 must form specific bonds with @xmath27 boundary components .
hence the choice of species for each @xmath79 is given by the intersection of @xmath27 sets of size @xmath91 each , so that the taylor expansion for @xmath95 is modified to @xmath96 .
the formula for capacity with general coordination number @xmath21 is therefore : @xmath97 we can also explicitly count the number of stably - bound chimeric structures @xmath98 of length @xmath87 , i.e. , the number of choices for the set @xmath99 shown in fig .
s[fig : sifig_boundary_conditions ] .
thus , @xmath100 estimated above is the probability that @xmath101 .
we can numerically compute @xmath102 for a square lattice through explicit enumeration using a transfer - matrix - like method @xcite .
let @xmath103 be the set of species that can occupy site @xmath104 in fig .
s[fig : sifig_boundary_conditions]a by forming a specific bond with the boundary component from the right .
( for instance , @xmath105 is the set of species that specifically bind to the component species @xmath106 from the right . )
we form a `` transfer matrix '' @xmath107 between sites @xmath53 and @xmath108 by restricting the matrix @xmath109 to rows @xmath10 which are species found in @xmath103 and to columns @xmath17 which are species found in @xmath110 .
hence @xmath107 is a @xmath111 matrix of boltzmann factors for binding between species @xmath103 and @xmath110 that can stably occupy sites @xmath53 and @xmath108 .
( note that we use the interaction matrix @xmath75 between top and bottom faces , i.e. , vertical direction in fig .
s[fig : sifig_boundary_conditions]a , in this construction of @xmath112 . )
then , the sum @xmath113 of all entries of the matrix product , @xmath114 gives the partition function summed over all chimeric structures made of components @xmath115 which , by construction , form specific bonds with the ( black ) boundary components to the left in fig .
s[fig : sifig_boundary_conditions]a .
the most stable of these chimeric structures will also contain specific ( vertical ) bonds between every pair @xmath116 , i.e. , @xmath117 specific vertical bonds of energy @xmath23 each .
hence , if we multiply the partition function @xmath113 by @xmath118 and take the limit @xmath119 , only terms corresponding to chimeric structures with @xmath87 specific vertical bonds will survive .
in fact , in the large @xmath17 limit , @xmath120 precisely gives us the number @xmath102 of such stably - bound chimeric structures . using this method
, we can compute @xmath102 explicitly for given boundary conditions . averaging this count over @xmath121 realizations of @xmath7 random structures of size @xmath58
, we obtained @xmath122 shown in fig .
s[fig : sifig_boundary_conditions]c .
here we chose @xmath123 to only count chimeric structures of length comparable to the side length of the @xmath124 structure itself .
( however , any @xmath87 that grows with @xmath0 gives similar results , as supported by the probabilistic argument for @xmath100 above . ) by varying @xmath0 between @xmath125 and @xmath126 , the inset of fig .
s[fig : sifig_boundary_conditions]c shows that @xmath102 rises rapidly at @xmath127 .
both our probabilistic arguments , using @xmath100 and explicitly counting @xmath102 of chimeric structures , agree on the scaling of capacity @xmath128 .
these results also agree with the scaling extracted from monte - carlo simulations of the lattice model , presented in fig . 4 and in section [ sec : scalingsim ] below .
we require two distinct behaviors of the multifarious assembly mixture : * * responsive : * the multifarious assembly mixture must produce structures in response to an externally introduced seed ( or equivalent perturbation ) of small size , * * stable : * the multifarious assembly mixture must be stable in the absence of external signals and must not produce any structures spontaneously .
nucleation theory , adapted to multifarious structures , dictates whether such competing requirements can be implemented .
we begin with the question of how large a seed is needed to recover a stored structure .
the change in free energy , with respect to the multifarious assembly mixture , due to the presence of an @xmath129 square seed taken from one of the @xmath7 stored structures is : @xmath130 since such a structure has @xmath131 components and @xmath132 strong bonds . here
@xmath133 is the energy of specific bonds and @xmath134 is the chemical potential of each species .
( note that we neglect the change of entropy in eqn . . )
the general shape of @xmath135 is shown in fig .
s[fig : sifig_nucleationbarriers ] ; as in conventional nucleation theory , @xmath135 has a maximum for some critical size @xmath41 . a sub - critical seed of size
@xmath136 will dissolve back into its components while supercritical seeds ( i.e. , size @xmath137 ) will grow in size and into the full stored structure . hence the minimal size of the seed we must introduce to recover structures is simply given by the size @xmath41 of the critical seed .
we can calculate @xmath41 by setting @xmath138 .
we find @xmath139 with @xmath140 .
note that this relationship is independent of the number of stored structures @xmath7 ; the minimal seed size is determined by a local condition @xmath141 on the free energy landscape and is not affected by the presence of other minima .
the excess free energy potential @xmath135 shown in fig .
s[fig : sifig_nucleationbarriers ] implies that the multifarious assembly mixture is intrinsically unstable even without the external introduction of any seed , a critical seed of size @xmath41 could spontaneously emerge on some timescale @xmath25 , leading to the nucleation of random stored structures .
hence the timescale of such spontaneous nucleation @xmath25 sets the useful lifetime of the multifarious assembly mixture . in conventional nucleation theory ,
the timescale for spontaneous nucleation @xmath25 is given by arrhenius s formula for barrier crossing @xmath142 .
however , with @xmath7 multifarious structures , we need to modify this formula to account for multiple inequivalent seeds that can spontaneously nucleate distinct stored structures .
if there are @xmath143 inequivalent barriers that can be crossed , the timescale for spontaneously crossing any one of the barriers and assembling a stable stored structure is given by , @xmath144 where @xmath49 is a timescale associated with microscopic processes . in conventional nucleation theory , there is only one ( or @xmath145 ) stable phase , while seeds can vary only in shape and not in composition . in contrast , our multifarious assembly mixture can form at least @xmath7 stable structures ( i.e. , the @xmath7 stored structures ) in addition to any stable chimeric structures that might exist .
further , seeds from different parts of these multifarious structures are inequivalent in composition . for small @xmath146
, we can ignore chimeric structures and estimate @xmath147 since the @xmath7 stored structures can each be nucleated with @xmath148 inequivalent seeds .
we do not pursue the detailed form of this correction any further here ; the correction to eqn .
[ eqn : tstarfstar ] is logarithmic and numerical simulations discussed below confirm that the @xmath149 correction is modest . using the expression for @xmath150
, we find , @xmath151 note that we have an unavoidable trade - off : increasing the lifetime of the multifarious assembly mixture @xmath25 would necessarily increase the critical seed size @xmath41 . thus a more stable multifarious assembly mixture requires a larger seed for recovering stored structures . we can also rewrite the lifetime @xmath25 in terms of the parameters @xmath6 and @xmath42 as @xmath152 and we tested this relationship in our monte carlo simulations .
we carry out monte carlo simulations of this system , with @xmath0 different species of square tiles on a square grid , see fig . [
fig : simsetup ] .
each stored structure is a @xmath153 sized square composed as a random permutation of @xmath154 tiles , with each tile being of different species .
since we work in the grand canonical ensemble , the total system was larger , with a side length @xmath155 , when @xmath156 was even , and @xmath157 otherwise .
the energy of a state of the system is specified by @xmath158 , where @xmath159 labels the stored structures , while @xmath160 labels the bond directionality of the nearest neighbor tile positions , with @xmath62 being the coordination number ( or valance ) . for our square grid ,
@xmath161 labels up , right , down and left neighbor positions , respectively . for a given structure @xmath9
we consider all nearest neighbor pairs of tiles .
when a pair of tiles of species @xmath162 is found , @xmath163 is determined as the directionality of the distance vector @xmath164 , and @xmath165 is set to @xmath23 .
the elements of the total interaction matrix for the system are : @xmath166 the maximum function caps the matrix elements , since we consider a model with all bonds of the same strength .
finally , we define the energy of states .
the square grid has @xmath167 sites , each being either empty or occupied by a tile which can be one of @xmath0 different species .
an empty site can be simply treated as a tile of zeroth species , which has no binding energy , so @xmath168 when any of @xmath162 is zero , and with chemical potential @xmath169 .
a state of the system is then described by a vector array @xmath170 , where @xmath171 if @xmath172 is the tile species at grid site @xmath173 , and @xmath174 otherwise , with @xmath175 labeling site @xmath176 void of tile .
the energy of such a state is : @xmath177 where @xmath178 is the directionality defined by nearest neighbor pair @xmath179 , and @xmath180 the vector of chemical potentials for species @xmath10 .
the monte carlo algorithm we used in the simulations chooses a random grid position and changes its species with a probability @xmath181 , where @xmath40 is the total energy cost of the change calculated using eq .
typical simulations were run for @xmath182 , where @xmath183 is one lattice sweep , i.e. , @xmath167 monte carlo moves .
4a shows the error in an assembled structure , observed at the end of simulation , as a function of the number of stored structures @xmath7 , for different structure sizes @xmath15 ( and correspondingly different system sizes @xmath0 and numbers of tile species @xmath58 ) .
each simulation starts with one selected complete structure of a given size @xmath15 ( in shape of a square ) and runs for a fixed amount of time @xmath184 .
we define the error using a three - step procedure , see fig .
s[fig : errordef ] : first , in the final state of the simulation we identify the largest contiguous area of tiles that are bonded , i.e. , the largest connected structure , which we call the `` final structure '' ; next , the union of the area of the final structure with the area of the initial structure ( which is a square ) gives a total area @xmath185 , and the number of tiles that match between the initial and final structures inside @xmath185 is divided by the total number of tiles in @xmath185 to give the overlap of structures . by definition , this overlap is between zero and one .
however , since the initial structure in the simulations never dissolves in the considered regimes ( i and iii ) , the overlap is at least @xmath186 ( for the system sizes we studied @xmath187 ) .
finally , the error is defined as one minus the overlap .
this definition of error , which uses @xmath185 , consistently takes into account the multifarious assembly mistakes that occur by changing or loosing tiles in the initial structure , as well as by attaching tiles to the boundary of the initial structure . for
any given structure size @xmath15 , the error in the simulations sharply rises from zero when a certain number of stored structures was reached , and then quickly saturates at @xmath188 .
the capacity @xmath35 is the number of structures that can be stored without having a large error , and we extract its value using a finite size scaling analysis .
our ansatz for the scaling function is @xmath189 , with @xmath190 , where @xmath7 is the number of stored structures and @xmath191 is the only fit parameter in function @xmath192 .
we find it robust for analysis to consider only the datapoints for which @xmath193 , which is consistent with our interest in the sharp rise of error from zero .
we rescale the datapoints in each error curve using @xmath194 , with @xmath191 fixed , and we fit the complete set of datapoints using a polynomial of fifth degree which represents the unknown function @xmath195 .
we quantify the quality of the fit for the given @xmath191 with @xmath196 , where @xmath197 is total of squared fit residuals , @xmath198 is total of squared mean prediction errors , and @xmath199 is total of squared datapoint errors @xcite .
we minimize @xmath200 with respect to @xmath191 .
the @xmath200 varied smoothly with @xmath191 and its minimum is easily found , giving the optimal value of @xmath191 .
4b shows a very good collapse of error curves for different number of species @xmath0 when the optimal @xmath191 is used ( the parts of the curves with @xmath201 are also plotted ) , which validates the scaling ansatz .
finally , we describe the procedure for estimating the error of the calculated optimal value of @xmath191 .
we use a simple bootstrap method @xcite . consider the error curve for some system size , and let there be @xmath202 datapoints in this curve .
we form a new dataset by randomly choosing a datapoint from the curve @xmath202 times .
the new curve therefore has the same number of points as the original , but some of the original points might be missing and some might occur multiple times .
this sampling is applied to every error curve . using the new datasets ,
the new optimal value of @xmath191 is calculated .
we repeat the entire procedure a hundred times ( always starting from the original datasets ) , giving a distribution of optimal @xmath191 values .
the mean of the distribution is the @xmath191 quoted in the main text , while the standard deviation is its error .
we check that using higher order polynomials in the fit procedure does not significantly improve the quality of fits .
in addition , instead of taking the error cutoff equal to @xmath203 , we also vary it between @xmath204 and @xmath205 , but the variation in obtained optimal @xmath191 was of the same order of magnitude as the error obtained by the bootstrapping procedure . to test the relationship between @xmath41 , @xmath6 and @xmath7 ( eqns . and ) , we ran a series of simulations starting with a square - shaped seed of different sizes and various values of @xmath6 , and measure the probability that the seed dissolves .
we consider @xmath206 and @xmath207 stored structures at fixed temperature @xmath208 , well inside the retrieval regime .
s[fig : diss ] shows the dissolving probability as a function of the number of tiles in a seed , @xmath209 , each given seed approximating a disc in shape .
each probability averages @xmath210 simulation runs . from this data
we extract the critical radius of the seed , @xmath41 , at which the probability to dissolve drops to zero .
we show the extracted @xmath41 as a function of @xmath6 in fig .
s[fig : rstarmu ] .
note that in this section we redefine the quantity @xmath41 as dimensionless and equals the critical seed radius measured in the units of tile length .
we investigate different time scales in the monte carlo simulations of the lattice model .
s[fig : tstarmum]a plots both the spontaneous nucleation time @xmath25 and the recovery time @xmath211 measured as functions of the chemical potential @xmath6 .
we define the spontaneous nucleation time @xmath25 as the time when any structure , being spontaneously assembled in the homogeneous solution , reached an area of @xmath212 tiles , which is above the critical nucleation size for the largest value of the chemical potential we considered ( see fig .
s[fig : diss ] ) .
recovery time @xmath211 identifies the moment in the simulation when we first observed the seeded structure completely assemble ( seed had @xmath213 tiles , which is supercritical for all chemical potentials considered ) . in some simulations , at the moment of completed assembly , there are a few erroneous tiles attached to the structure , which we neglect here . as @xmath214 ,
the @xmath25 diverges , while the recovery time is much smaller and essentially independent of @xmath6 .
this demonstrates that there is a parameter range of the model where structure retrieval occurs much more quickly than spontaneous nucleation from the solution , fig .
s[fig : tstarmum]a . in fig .
s[fig : tstarmum]b shows how the nucleation time @xmath25 depends on the number of stored structures . the obtained result is consistent with eqns .
[ eq : q ] and [ eq : tstar2 ] of the main text . here
we present details of the finite temperature transition between regimes iii and iv presented in fig .
2 . as presented in fig .
s[fig : transitiont ] , we pick the simulations with @xmath215 and @xmath216 stored structures as examples , and look at the size of largest structure as function of temperature . at the end of each simulation we identify the `` biggest structure '' as the largest contiguous area of tiles that are bonded specific interactions , i.e. , the largest connected structure . at each temperature
we average the size ( number of tiles ) of the biggest structure over @xmath106 independent runs .
s[fig : transitiont ] reveals that the biggest structure covered most of the entire system at low temperatures ( regime iii ) , but its size sharply dropped at a certain temperature . at temperatures above this transition ( regime iv ) ,
the system is mostly filled with tiles , however , there are hardly any specific interactions between them , i.e. , the state was a solution of fluctuating components .
as expected , the transition temperature increases slowly with increasing number of stored structures to @xmath217 , with @xmath23 the binding energy . in fig .
s[fig : regimesmu]a - b we show regimes observed in monte carlo simulations as a function of the number of stored structures @xmath7 and temperature @xmath65 , starting from a particular supercritical seed ( shown in bottom panels ) . as in fig . 2
, we use different colors to visualize different stored structures , with the seeded structure colored in dark red .
bottom panels distinguish the four regimes identified in the diagram . in regime i the desired structure is retrieved through heterogeneous nucleation since the solution remains stable in the time required for assembly .
the solution in this regime is a functional multifarious assembly mixture .
regime ii is characterized by homogeneous nucleation of all structures due to reduced stability of the solution . in regime
iii , formation of structures is dominated by chimeras .
finally , in regime iv , any initial seed is disintegrated into the solution .
these figures differ from fig . 2 in the value of the chemical potential @xmath6 , which is the same for all the species . in fig .
s[fig : regimesmu]a the lifetime @xmath25 of the multifarious assembly mixture is suppressed due to the lower value of @xmath218 .
hence , regime ii suppresses the retrieval regime i compared to the result in fig . 2 , where @xmath219 . in fig .
s[fig : regimesmu]b the chemical potential is higher @xmath220 .
consequently the characteristic lifetime of the multifarious assembly mixture is @xmath221 ( see fig .
s[fig : tstarmum ] ) , resulting in complete suppression of regime ii .
our model of multifarious assembly mixtures is closely related to models of associative memory @xcite . in these models
@xcite , multiple `` memories '' are stored as stable states of a neural network by choosing the connections between @xmath0 neurons appropriately .
just as with multifarious assembly mixtures , neural networks have a finite capacity ; with @xmath0 neurons , a limited number of memories @xmath222 can be reliably stored and retrieved . if the capacity @xmath223 is exceeded , many spurious memories undesired stable states appear and interfere with retrieval .
neural networks have also been studied in the thermodynamic limit of a large number of neurons @xmath0 @xcite . in this limit @xcite
, energy barriers separating the memories grow large and each memory becomes a stable thermodynamic phase of the system , provided the number of memories @xmath224 is less than a critical value @xmath225 .
however , there is a phase transition at @xmath226 to a spin - glass phase with an exponential number of other spurious but stable states .
the properties of such phases have been worked out for different neural networks with different models of interactions between neurons , resulting in phase diagrams that resemble fig . 2 of our paper .
in fact , a recently studied @xcite @xmath56-lattice neural network , albeit with long - ranged interactions , is closely related to the large coordination number @xmath21 limit of our model .
hence , it is natural to ask about the thermodynamic properties of multifarious assembly mixtures .
for example , we can take the size of programmed structures @xmath227 to be large , with the ratio @xmath26 held finite .
such a limit might allow for a growing number of stored structures @xmath228 to become stable phases of the system , provided @xmath229 is less than a critical @xmath230 .
however , a crucial intrinsically - kinetic feature of multifarious assembly mixtures , not found in neural networks , is the stability of the unassembled mixture itself . as we showed in the paper , the unassembled mixture has a finite lifetime @xmath25 after which random stored structures are spontaneously nucleated .
this lifetime @xmath25 is set by the ratio @xmath44 of chemical potential to bond energy and hence , in principle , is independent of structure size @xmath15 and can stay finite in the thermodynamic limit . as a result , for multifarious assembly mixtures ,
the question of practical interest is a finite - time kinetic question can one recover specific structures using seeds in a time shorter than @xmath25 ? to focus on this question , we define capacity and the transition to chimeric phase through a practical finite - time notion of error we introduce an initial seed and measured error in recovery after a finite simulation time chosen to be much smaller than the lifetime of the mixture .
a thermodynamic analysis of our model would require a modified set of quantities and parameter limits .
for example , the unassembled mixture can itself be stabilized as a phase only if its lifetime @xmath25 diverges in the thermodynamic limit , which requires tuning the chemical potential @xmath231 with growing structure size @xmath227 .
in such a limit , our finite - time error can be replaced by order parameters analogous to those used in @xcite to study the transition to the chimeric regime .
we leave a detailed study of the thermodynamic limit of multifarious assembly mixtures to future work . | self - assembly materials are traditionally designed so that molecular or meso - scale components form a single kind of large structure . here , we propose a scheme to create multifarious assembly mixtures " , which self - assemble many different large structures from a set of shared components .
we show that the number of multifarious structures stored in the solution of components increases rapidly with the number of different types of components . yet , each stored structure can be retrieved by tuning only a few parameters , the number of which is only weakly dependent on the size of the assembled structure .
implications for artificial and biological self - assembly are discussed . |
again i avoid the standard journal format .
the only prerequisites for reading the main text of this paper are some familiarity with common statistical terms or phrases such as `` bias '' and `` maximum - likelihood estimation , '' which i assume most readers will recall from a college statistics course or will look up on wikipedia .
i will relegate the statistical heavy - lifting to an appendix .
perhaps the single most important contribution a mathematical modeler can make for public health would be to invent a tool that could predict the occurrence of a deadly world - wide pandemic of an infectious disease . as is well - known , most such diseases derive from zoonotic transmission events ( meaning from an animal host in which the disease organism is endemic , or is currently causing an epizootic , to humans ) ; examples include plague , influenza , ebola , hiv , sars , and mers .
when a pathogen jumps species there is a likely requirement of certain genetic modifications before it can cause efficient transmission in the new host population .
the implication is that prediction is primarily a business for geneticists but , at present , identifying the relevant mutations in the pathogen genome prior to their appearance seems beyond geneticists capabilities .
perhaps that situation will soon change ; in the mean time , it is up to epidemiologists to use case data and contact tracing to declare the moment when the pandemic is inevitable , or has already begun . but is this even possible for the superspreader ( ss ) epidemic ? as i showed in previous papers , @xcite , @xcite , an ss epidemic is unlike an influenza or measles epidemic , for which classical , deterministic mathematical models of the sir ( susceptible - infected - recovered ) type are appropriate .
the ss epidemic has random outbursts and lulls , periodic re - introductions from the zoonotic source , and a long `` kindling '' period before the pandemic begins .
( see figure [ fig0 ] for a typical simulated epidemic trajectory . )
moreover , the case mortality for sars and mers is much higher than for influenza or measles ( 16 percent for sars and 40 percent for mers , as i write , as opposed to 2 percent for the 1918 flu ) , so by the time susceptibles drop limiting the pandemic , the world would have suffered hundreds of thousands to millions of deaths . thus the key issue is not predicting the size of the epidemic , but whether it is looming and we must quickly find a way to stop it . in paper ii of the series
, i used the information available to me ( case - incidence data ) to fit some ss models to mers data through april 2014 and asserted that a pandemic was likely .
since the data was less than optimal , i used a `` generic '' technique from pattern recognition to find models that fit .
this work is more illustrative , so i will assume available detailed contact data of just the type we would wish to have in order to estimate model parameters , including @xmath0 , the average number of secondary ( directly - infected ) cases due to one primary case ( which could be sporadic , from zoonotic transmission , or the secondary of yet another primary ) . in some instances this kind of who - infected - whom data will be readily available through screening , e.g. , for health care workers exposed to a patient ( who have comprised a large fraction of infections in the mers epidemic ) , or passengers on a plane with the primary case ; but for others it will be difficult , e.g. , for community - derived infections .
i ll indicate the ( mock ) `` data , '' which will be generated in simulations , by @xmath1 , where @xmath2 stands for the number of secondary cases caused by the i - th infected case .
of course , some contact data would be missing , resulting in misclassification of secondary cases as sporadic , causing undercounts of secondaries ; i ll discuss the impact of this problem later . for the model
i choose a rather extreme two - level ss example that appeared in the model fit of paper ii ; the parameters were : @xmath3 ( low - level infectiousness ) @xmath4 , @xmath5 ( high - level ) @xmath6 , and @xmath7 .
in addition , the noninfectious and infectiousness periods were assumed lognormal with standard deviation of twice the means .
both the span of reproductive numbers and the large variance of the infectious period generate `` extra - poisson variation '' ( terminology from @xcite ; @xmath8 , where @xmath9 is the variance of the secondaries ) and strongly affect the statistical issues , as we will see .
since @xmath0 is the parameter that will drive the prediction , how would we estimate it from the secondaries data ? since we are presented with an i.i.d .
sample , the obvious ( non - parametric ) approach is simply : = 1n_i=1^n n_i .
however this estimator is highly biased downward , as is seen in figure [ fig1 ] .
all the estimates were less than one ( meaning no pandemic ) .
what aspects of this problem might contribute to generating this peculiar bias towards low @xmath0 estimates ?
first , data points that one would normally regard as outliers , and might even drop from the analysis , namely the superspreader events , are in fact informative here .
second , there is skewing of distributions from the lognormal infectious periods . finally , there is `` right censored data , '' which is a peculiarity of trying to estimate parameters during an ongoing epidemic .
the simulations used for figure [ fig1 ] were halted at 2000 total cases , on the logic that passing some threshold case load ( hopefully smaller , but i will return to that issue below ) is likely the moment when we would prepare the statistical analysis . however , ss epidemics are particularly prone to sudden explosions .
thus many people may have been infected in the previous weeks
but then there would not have been time for them to generate as many secondary cases as would be expected ( and any investigative or reporting lag would exacerbate this problem ) .
what can we do about these problems ? as a second try
, i adopted the famous , asymptotically optimally - efficient , maximum likelihood method to fit the model .
this made some improvement , see figure [ fig2 ] .
maximum likelihood is well known to be biased in small samples , with bias of @xmath10 , where @xmath11 here means the number of primary cases in the simulation ( 2000 ) .
one might expect it to be small , but apparently something about the data still causes trouble . described in @xcite , as it is based on a likelihood . for more on this issue , see subsequent footnote .
] when faced with the problem of estimating skewed distributions , one is directed to try transforming the data .
so i replaced @xmath2 by @xmath12 and fit the three - parameter model , again assuming sufficient data , by equating model - averaged @xmath13 to empirical averages , see the appendix for details .
this worked much better ; see figure [ fig3 ] . however , clearly bias remained despite using these statistical tricks .
what about censorship ?
i simulated secondaries ( for 2000 primaries ) directly from the probability rule ) of the appendix ; thus the computer chose the infectiousness type , then a normal r.v .
, @xmath14 and a corresponding lognormal infectious period , then ran a poisson process to the end of that period , for 2000 cases , 1000 times . ] with the same parameters as the model , but without running the branching process and found figure [ fignosimulation ] .
thus , the most important factor biasing the data was not skewing or outliers , but right - censorship .
this censorship may occur in practise because : the primary is still in the asymptomatic ( non - infectious ) or infectious period , or some secondaries are still asymptomatic and were not picked up through contact - tracing and screening ( by e.g. , rt - pcr ) , when the prediction was made .
however , we pay a major cost for dropping cases , by lowering the power of the test . then . ]
although there are well - known methods for improving estimation in the presence of censorship ( by imputation of missing entries ) , perhaps a preferable approach is to get better data ! now let us assume that , in addition to secondaries data ( the @xmath15 ) , we have available the times of onset and termination of infectiousness for the primary cases .
this is not a great hardship ; for sars or mers , the infectious period may coincide with the symptomatic period , and the endpoint of infectiousness with either the patient s death or isolation .
this kind of data could be obtained from patient records and so should be easier to derive than contact data .
so i assumed ( and stored in simulations ) infectiousness periods ; the data now consists of @xmath16 , omitting cases where infectiousness onset was censored and if termination was , setting : @xmath17 final time minus onset .
( this choice automatically drops primaries still in the non - infectious period , while keeping secondaries infected before the cut - off time , even though they may take a few more weeks to locate . ) with this amplified data - set , maximum - likelihood analysis is appropriate ( see appendix ) .
moreover , we do not have to model the infectious period
. see figure [ figtimes1 ] .
table 1 compares statistical properties of the four methods mentioned above .
the log - transformed with secondaries data and the mle with both secondaries and infectious - period data roughly tied on standard error , but the latter had much smaller bias .
statistical properties of four estimators of @xmath0 for an ss epidemic.@xmath18 [ cols="^,^,^,^,^,^",options="header " , ] @xmath18 parameters as in text ; 1000 runs stopped at 2000 cases.@xmath19 some runs had too few secondary `` high - level '' ( @xmath20 ) cases , or nr did not converge.@xmath21 standard error ; square root of : ( bias squared plus variance).@xmath22 secondaries data only.@xmath23 secondaries plus infectiousness periods .
what about estimation in an epidemic with @xmath24 ? see figure [ fig5 ] .
although there is some overestimation of @xmath0 , it is very minor .
in fact , in only 220 of 1,000 runs were there enough high - level events ( with more than five secondaries ) to even carry out parameter estimation ; these runs are unlikely to be judged pandemics , leaving only about 40 out of 1,000 ( or 4 percent ) which would generate a false positive ( crying `` pandemic ! ''
when there was nt one ) .
note that all forms of statistical bias tend towards generating false negatives , i.e. , a false sense of security that we will escape the horrors of a world - wide epidemic .
( hence their `` stuttering chains '' ) , in which the skewness , outlier , and censorship problems under discussion here do not arise . using their technique to argue that @xmath24 , as in @xcite , may be circular reasoning .
nevertheless , using my models and knn fitting technique , in august 2013 i did agree with the lancet authors that @xmath0 for mers was less than one at that time .
but , looking at fig .
[ fig4 ] , i now wonder whether we were all mistaken and that period was merely the `` kindling '' phase of a pandemic . ]
the mle method with secondaries and infectious - periods data did a good job at overcoming statistical bias .
however , there is a fourth source of bias going in the same direction shown in figures [ fig4 ] and [ timesfig4 ] .
figure [ fig4 ] shows estimations of @xmath0 derived from secondaries data by the log - transformed method , and fig .
[ timesfig4 ] from secondaries plus infectious periods by the mle method , up to the number of cases shown on the x - axis , early in the epidemic .
note how the estimates rise abruptly and then fall ; what is happening is that the estimates are driven upwards by an ss event and then fall as more non - ss events accumulate .
note also how , ignoring the sawtooth pattern , the general trend is upwards .
[ timesfig4 ] looks better , but both figures have very few dots because of too few `` high - spreader events , '' e.g. , with @xmath20 , to run the nonparametric estimators , early in the epidemic . )
call it `` temporal bias , '' by which i mean bias due to estimating too early ; it points in the same direction as the statistical biases we saw in earlier diagrams , towards underestimating @xmath0 .
= something , is it probably x percent higher ? ''
i hope to address this question in a later publication . ] in addition to techniques for decreasing bias , we need some way of assessing confidence in our prediction .
after some experimentation with popular methodology ( e.g. , resampling and the `` jackknife '' ) for constructing confidence intervals suggests using ordinary , normal - distribution based cis , but recall that figure assumed 2,000 cases ; with fewer we can not be confident of the normal law . ] which worked poorly at least in my implementation , i recurred to the ancient fisherian `` reject the null hypothesis '' philosophy .
fisher s method has the advantage that it addresses our major concern : that the declaration of a pandemic might be due merely to chance .
however , where modeling is involved there is a pitfall in fisherism : one must not adopt absurd models ( e.g. , here , uniform or poisson cases ) as `` null models , '' because by that methodology any desired model can be ratified .
i adopted for the null a model with the same observed frequencies of lower / higher types , then selected @xmath3 and @xmath5 at random but restricted so that @xmath25 . by simulation
we obtain a probability that the observed @xmath0 would be equaled or exceeded in non - pandemic ss models , i.e. , a `` type - i error . ''
( see the appendix for details . ) that permits an announcement with confidence , e.g. , `` the probability of a pandemic has now exceeded ( say ) 90 percent . '' as i showed in paper i of this series , ss epidemics can have a high probability that a given infection chain started from one `` index '' case goes extinct .
for example , with the model selected here for illustrative purposes , the extinction probability , @xmath26 , was about 0.991 .
as a consequence of these high extinction rates , there may be a long period in which sporadic cases appear but the infection - chains peter out , which again biases us towards optimism that we have escaped the worst .
now consider prediction .
suppose that , as the epidemic progresses , we apply our statistical analysis to case / contact data and , at some point , obtain an estimated @xmath0 that is greater than one , with a p - value less than 0.1 but when will the pandemic start ?
here we must take into account fig .
[ fig4 ] , which suggests that our first indication of a pandemic will probably coincide with an ss event .
so we should estimate the time - to - pandemic by : = p_e^n(1 - p_e).[time_eqn ] where @xmath27 denotes the number of secondary cases of the case appearing at the instant in question ( which will admittedly take some extra time to locate ) .
the first factor takes into account that , if one of the chains begun by these secondaries goes pandemic , the pandemic is already underway ( so an omitted term : @xmath28 ) , and otherwise we wait until some future sporadic case generates the pandemic .
figure [ fig6 ] shows a histogram of @xmath27 is fig .
[ fig6 ] is due to starting the analysis of the @xmath15 s just after when total cases exceeded 200 , to avoid lots of unfittable instances .
frequently , the first analysis generated a significant @xmath29 , halting the search . since most cases are low - spreaders who generate no secondaries , the last @xmath27 was often zero .
incident cases can be zero , too , because the disease in humans had gone temporarily extinct .
the period explored was between 200 and 400 cases . ] and
[ fig7 ] the times - to - pandemic , made using the mle - with - infectiousness - periods method , for runs with p - value less than 0.1 .
note that many of the latter are large , even years . a somewhat more conservative prediction results by replacing `` @xmath27 '' in equation ( [ time_eqn ] ) by the number of cases existing at the prediction point ( counting secondaries created by the superspreader ) , producing figure [ fig8 ] .
figure [ fig9 ] shows the estimated @xmath0 s at the prediction . seems to show a right - end tail , whose origin in this complicated context eludes me . ]
two other issues will produce more bias , in the familiar direction .
as pointed out in paper i , a long `` kindling period '' when sporadic cases generate infection clusters that disappear will surely lead anyone unaware of extra - poisson variation and extinction probabilities to imitate the managers of the space shuttle program before the challenger catastrophe ( who , after each launch without a disaster , were reassured about their risk judgement ) .
finally , secondary cases missing at random will produce bias towards underestimating @xmath0 .
to summarize : i have presented a small study , using simulations from mathematical models and conventional statistical ideas , on the topic of whether we should be able to predict the occurrence of a superspreader epidemic of the sars / mers type given contact data .
the main conclusion is that care must be taken that mathematical modeling and statistical methodology do not add to a false sense of security which will arise anyway from the long `` kindling period '' and certain other , non - mathematical biases we all share .
the methods i propose in this paper can overcome the statistical bias problems , granted adequate data and some modeling assumptions .
that a pandemic can often be predicted long before it begins is perhaps not so surprising ; think of predicting when somebody will win in a repeated drawing in a lottery , given facts such as how many people are buying tickets and the number of winning balls in the urn . fitting
a model is a means of finding those facts ; then mathematics ( meaning today , computers ) allows us to make the prediction .
re those `` other '' sources of error , consider what psychologists call `` explanatory bias '' ( but which might better be called `` explanatory hubris '' ) .
i refer to the universal tendency to imagine that one has found the explanation for an unusual or shocking event , and hence to believe that it was a one - time occurrence or that it can easily be prevented in the future .
( _ viz_. that use of the aspirator in the er that produced 140 + sars cases in the prince of wales hospital in hong kong in 2003 , or the putative wave of camel births in saudi arabia in the spring of 2014 , which somehow did nt cause a bump in mers cases a year earlier . )
no one , including myself , desires to use stochastic modeling when a deterministic explanation can be found .
but sometimes we can do better by incorporating randomness into our theory as a preventive for hubris .
finally , there is the bureaucratic tendency to err on the side of not creating panic or generating reports that might displease the boss .
governments in the past have gone as far down this road as to deny the occurrence of earthquakes , nuclear meltdowns , or disease outbreaks . in regards to mers in 2014 , as i write governments have behaved more responsibly in reporting cases to who than was true for sars in 2003 , which is a step in the right direction .
epidemiologists can not be expected to give sound advice to governments if they have to work from inadequate or distorted data .
mers poses a potentially a greater risk to the world than did sars , which went extinct after the zoonotic source , an exotic wild cat sold in live - animal markets in china , was eliminated . , permit . ]
if indeed the reservoir for mers is camels , which are plentiful in the middle east and northern africa , re - introductions into the human population may continue for a long time . and
the case mortality is frightful .
we modelers should not add to the false - sense of security problem , especially when thousands or millions of lives are at risk .
the results of this paper demonstrate that , with some attention to statistical issues , modeling can make a contribution to preventing a global mers epidemic .
the model and fixed parameters ( mean and variance of non - infectious and infectious periods ) were described in previous papers .
i modified the simulation routine to record the case index of primaries that infected each secondary , then back - tracked to produce the @xmath15 .
if it were not for censoring , the statistical problem would be straighforward : given an i.i.d .
sample of nonnegative integers drawn from a distribution : estimate the model parameters , denoted by letter @xmath30 .
@xmath14 is a standard normal random variable and @xmath31 ( fixed ) is chosen to give the required variance ( i.e. , the true infection period for each case equals the mean infection period times @xmath32 ) .
@xmath33 denotes expectation over @xmath14 . here
@xmath30 stands for either @xmath34 or , equivalently , @xmath35 where these iterations either failed immediately , because @xmath38 had the wrong sign ( positive , when it should be negative ) , indicating lack of convexity of the likelihood surface , or ( in most runs ) reduced the magnitude of the score by a factor of @xmath39 .
( the expectation over @xmath14 in equations ( [ dist_eqns ] ) was performed numerically , using 200 evaluation points . ) starting at , e.g. , random choices of the @xmath40 s , usually failed by the determinant - sign , indicating that the likelihood is rarely globally convex for this problem , raising lack - of - identifiability issues . for the @xmath42 s by one - dimensional line searches .
( the sums over @xmath43 can be truncated by careful treatment of the summand exponent , using stirling . ) defining @xmath44 as usual gave fig .
[ fig3 ] . where inf stands for mean infectious period . again using the nonparametric estimates as starting points ,
the hessian determinant had the right sign , newton - raphson iterates converged quickly and reduced the score to zero .
re confidence , i first tried leaving out one of the @xmath16 , but the resulting confidence intervals were absurdly narrow .
i then tried resampling , but now they were too wide .
moreover , i do nt have a grasp of what such confidence statements are supposed to mean ; what we really want to know is how likely was the observation that @xmath29 , if in truth it was nt .
so i retreated to making p - values by simulation , as follows .
given the estimated lower - type probability , @xmath45 , i randomly chose @xmath3 and @xmath5 so that @xmath46 , starting with @xmath5 uniform on @xmath47 $ ] , 1,000 times . for each choice
, i simulated a trajectory and refit the model to the mock data , generating a new @xmath0 ; if this @xmath0 was equal to or greater than the original estimate , it counted as rejecting the null hypothesis when it was true .
then the p - value was set to : ( count)/1,000 . the method had power , with a type - i error rate of 0.1 and 1,000 cases , to reject the null when it was false ( namely for the illustrative model , with @xmath7 ) , of 0.77 . for you statisticians who might be interested : i did consider several other ideas / improvements . in the book @xcite i advertised the knn method of model fitting , which arose in computer science back in the 1950s for problems in pattern recognition .
this `` generic '' method was used in paper i for fitting ss epidemic models to some mers cluster data , and in paper ii to case data , but as it does not directly aim at estimating @xmath0 it may not be convincing to many people . and ,
as knn reduces to if you can simulate fast enough , it should suffer from the bias problem i demonstrated in fig .
[ fig1 ] . on that issue
, there is a large literature on reducing bias in ml estimation .
the idea is to modify the likelihood , perhaps by subtracting something which will cancel the bias .
however , all the suggestions i found in the literature ( which are typically tested on trivial , exactly - solvable , examples ) involve taking three or more derivatives of the loglikelihood .
my enthusiasm flagged at that point .
re identifiability for the @xmath2-only data , perhaps the power - law model from paper ii would be an improvement .
or is there some simpler model with plausible biological interpretation but also with globally identifiable parameters , granted sufficient data ?
( however , it is a mortal sin to abandon realism and mutilate a model simply in order to gain identifiability of parameters . )
i leave these problems to those interested , but remark that , noting the extreme skewness and outliers in distributions generated by ss epidemics , any such methods should be thoroughly tested by simulation from an appropriate model . | in two previous papers , i introduced superspreader ( ss ) epidemic models , offered some theoretical discussion of prevention issues , and fitted some models to data sets derived from published accounts of the ongoing mers epidemic ( concluding that a pandemic is likely ) . continuing on this theme , here i discuss prediction : whether , in a disease outbreak driven by superspreader events , a rigorous decision point meaning a declaration that a pandemic is imminent can be defined .
i show that all sources of prediction bias contribute to generating false negatives ( i.e. , discounting the chance of a pandemic when it is looming or has already started ) .
nevertheless , the statistical difficulties can be overcome by improved data gathering and use of known techniques that decrease bias .
one peculiarity of the ss epidemic is that the prediction can sometimes be made long before the actual pandemic onset , generating lead time to alert the medical community and the public .
thus modeling is useful to overcome a false sense of security arising from the long `` kindling times '' characteristic of ss epidemics and certain political / psychological factors , as well as improve the public health response . |
in many strongly correlated electronic systems , superconductivity seems to be linked to the vicinity of an electronic or magnetic instability with a characteristic phase diagram . currently , there is a growing interest on the systems where this electronic instability is a charge density wave .
superconductivity and cdw share the electron - phonon interaction in terms of microscopical mechanism .
these kind of competitions were also revived with the recent and quasi - systematic observation of a superconducting pocket at the border of the cdw in the dichalcogenides [ 1,2 ] .
these superconductors can show deviations from the weak coupling bcs theory , which raises the question of the interplay between the two orders .
the cdw opens a gap on parts of the fermi surface and can be at the origin of the modification of the superconducting gap structure .
this question was mainly asked in the dichalcogenides systems where the cdw is two dimensional ( 2d ) .
it is important to study other systems where cdw coexist with superconductivity .
here we focus on a compound , lu@xmath0ir@xmath1si@xmath2 , which offers a completely different type of cdw : 1d , 1st order , and involves a large part of the fermi surface .
lu@xmath0ir@xmath1si@xmath2 ( p4/mbm tetragonal space group ) is constituted of 1d chains of first neighbors lutetium atoms along the c - axis , with a strong interchain coupling .
when the temperature is decreased below t@xmath3 , a large jump of the electrical resistivity signs that a large part of the fermi surface is involved in the cdw . moreover , from specific heat measurements , the transition has been suggested to be first order [ 3 ] .
but , the origin of this first order transition is still unclear , and so far , a full thermal hysteresis at t@xmath3 has not observed .
the periodic lattice distortion associated with the cdw manifests itself by the formation of x - ray satellites reflections at wave vectors g+/-n.q@xmath3 where g is a reciprocal lattice vector , n is an integer and q@xmath3=(0,0 , 3/7 ) .
the temperature dependence of the integrated intensity of the satellites reflections peaks , which is related to the cdw order parameter , deviates strongly from a bcs - like dependence [ 3 ] .
recently , the origin of the cdw has been attributed to the peierls mechanism [ 4 ] . in this paper
, we present electrical resistivity measurements up to 900k on high quality single crystals .
we show a full hysteresis at the cdw transition evidencing the first order nature of the transition and the absence of any other cdw transitions at higher temperature .
we also show that the cdw and superconductivity compete for the same electrons .
finally , we present measurements of specific heat and magnetic penetration depth in the superconducting state .
a polycrystalline sample of lu@xmath0ir@xmath1si@xmath2 was prepared in an induction furnace , with a cold copper crucible and a high - purity argon atmosphere .
we used a stoichiometric amount of lu ( 99.96% ames lab . , or 98% alfa aesar ) , ir ( 99.9% alfa aesar ) and si ( 99.9999% alfa aesar ) .
single crystals of lu@xmath0ir@xmath1si@xmath2 were grown by two different techniques . on the one hand , a quick thermal quench of the melted polycrystalline sample favored the spontaneous growth of high purity `` whiskers '' oriented along the c - axis ( in the following we will use `` w '' to denote these samples ) [ 5 ] . on the other hand , large single crystals were also grown in a three - arc furnace by a czochralski technique following the procedure previously published [ 3 ] .
these latter samples were cut with a wire saw along the crystallographic axes , and subsequently annealed at 900c for 7 days under ultra - high vacuum to reduce the internal stress ( we will denote them as `` bar '' ) . for the electrical resistivity measurements , we used a traditional four wire technique , with platinum or silver paste to ensure the electrical contacts .
the high temperature resistivity was measured in a homemade furnace under vacuum , with a thermometer close to the alumina sample holder .
the specific heat measurements down to 0.4k were performed using the usual relaxation technique in a commercial ppms instrument from quantum design .
finally , the magnetic penetration depth was measured with a high stability lc oscillator , operating at 14 mhz , driven by a tunnel diode
. the superconducting sample is placed at the center of the excitation coil .
a small variation of the magnetic penetration depth induces a proportional variation of the inductance and so of the resonant frequency .
the calibration is done thanks to the dimension of the sample and by extracting the sample from the excitation coil at low temperature .
the experimental set - up is described elsewhere [ 6 ]
1 shows the resistivity up to 900k for a whisker ( where j//c ) and for a bar aligned along the a - axis . at low temperature , for the whiskers ( respectively bar ) a sharp superconducting transition
is measured : tsc = 4.05 k ( resp .
4.6k ) at 50% of the normal state resistivity .
the width of the transition is 50mk ( resp .
200mk ) , with a criterion of 10% - 90% of the normal state resistivity . above t@xmath3 77k , a sudden drop of the resistivity signs the increase of the density of states due to the destruction of the cdw . to check the order of the transition , we increased and decreased the temperature at the vicinity of the cdw at different speed .
a clear thermal hysteresis of the resistivity @xmath4 t 1k , independent of the speed rate , evidences the first order nature of the transition .
a thermal hysteresis with @xmath4 t 3k had already been reported in the iso - structural compound lu@xmath0rh@xmath1si@xmath5 at the vicinity of the cdw transition [ 7 ] .
usually , a first order transition is attributed to a lock - in transition with the lattice of a pre - existing incommensurate cdw at larger temperature , potentially above the room temperature [ 8 ] . to check the presence of this possible cdw
, we performed resistivity measurements up to 900k .
no anomaly on the resistivity is observed .
the data are well fitted with a phenomenological model with two parallel resistors [ 9 ] , one corresponding to the usual resistivity due to impurity and electron - phonon scatterings ( bloch - grneisen model ) , and a second due to saturation when the mean free path l is of the order of the lattice parameter ( ioffe - regel criterion ) : @xmath6 here @xmath7 is the debye temperature and @xmath8 is the bloch grneisen formula for an electron - phonon scattering .
an excellent fit is obtained with @xmath7=315k , and for j//a ( resp .
j//c ) @xmath9.cm ( resp .
220 ) , @xmath10.cm ( resp 0.24 ) and @xmath11.cm ( resp .
these values are close to the values for other superconducting metals [ 10 ] .
a second batch of single crystals was prepared with lutetium including 1% of tantalum ( samples denoted as `` ta '' ) .
scanning electron microscopy and energy dispersive x - ray spectrometry ( edx ) analysis show some small inclusions of pure ta in the matrix of the lu@xmath0ir@xmath1si@xmath2 .
these inclusions are homogeneously repartitioned in the sample .
it is not clear if the ta has also been substituted to a particular site and/or is in interstitial in the host material .
the temperature dependence of the resistivity is compared to pure lu@xmath0ir@xmath1si@xmath2 in fig .
2 . the jump of resistivity due to the cdw is totally suppressed . around 70k , a small
minimum of resistivity is reminiscent of the cdw in the pure compound , this suggests the possible presence of a strongly disordered cdw in some parts of the samples . here , the mechanism at the origin of the suppression of cdw ordering temperature could be a chemical pressure induced by the inclusions of ta .
further analysis should be performed to clarify this point .
the most surprising is the strong increase of the critical temperature of the superconducting transition : from 4.6k in the pure compound to 5.7k .
a similar effect has already been clearly reported by substituting si by ge [ 11 ] , or under pressure [ 12 ] .
this further suggests that both orders are in competition for the same electrons .
to get further insight regarding the superconducting properties of lu@xmath0ir@xmath1si@xmath2 , we performed specific heat measurement at 0 magnetic field . in the normal state , the specific heat
can be fitted with a sommerfeld coefficient @xmath12 of 22mj / mol / k@xmath13 and a debye temperature of 320k , in good agreement with the previous report [ 12 ] .
this sommerfeld coefficient corresponds to a density of states at the fermi level of 0.53 ev@xmath14.@xmath15 per atom . in fig .
3 we show the specific heat after subtraction of the normal state contribution . the critical temperature is found to be 3.75k with a width of 120 mk signing a good homogeneity .
the temperature dependence of the specific heat in the superconducting state is well described by a weak coupling bcs theory .
this behavior is strongly different to other system where both orders coexist .
for example , 2h - nbse@xmath16 shows the coexistence of a two dimensional ( 2d ) charge density wave ( cdw
t@xmath3 33k ) and superconductivity ( tsc 7k ) .
this superconductor shows deviations from the weak coupling bcs theory , which raises the question of the interplay between the two orders [ 13 ] . since these deviations are also present in the iso - structural system 2h - nbs@xmath16 despite the absence of cdw [ 14 ] , it turns out they could come from a peculiar electron - phonon anisotropy general to the transition metal dichalcogenide family .
finally , the temperature dependence of the variation of the magnetic penetration depth @xmath17(t ) was measured in the meissner state for an ac - magnetic field applied along the c - axis . in this geometry
the supercurrents are flowing only in the basal plane .
the superfluid density pairs is proportional to @xmath18(t))@xmath19 where i refers to the crystallographic direction , @xmath20 to the absolute value of the magnetic penetration depth at 0k and @xmath17 to its variation with temperature .
@xmath20 is related only to the fermi surface . whereas , the temperature dependence @xmath17 is also related to the superconducting gap
. the low temperature approximation for a type ii superconductor with a fully open gap is given by : @xmath21 where 0 is the superconducting gap , @xmath22 the boltzmann constant and @xmath20 the magnetic penetration depth at 0k . in fig 4 .
we show a typical temperature dependence of the superfluid density for a single crystal and a whisker , using @xmath23 of 409 nm [ 15 ] .
all the samples measured show the same temperature dependence . on a wide temperature range below tc ,
the dependence is very close to the one expected for a weak coupling bcs superconductor in the clean limit .
this is surprising , because the system is in the dirty limit according to the magnetic penetration depth and the residual resistivity values .
moreover at low temperature we observe a small , but clear deviation to the clean or dirty limit of the bcs theory .
indeed up to 0.4tc , the temperature dependence of the magnetic penetration depth is better described by a power law @xmath24 with @xmath25 close to 3 ( best fit : @xmath25=2.94(1 ) and @xmath26=350(5)nm ) .
the origin of these discrepancies needs further investigations . to conclude , we have presented high temperature electrical resistivity showing a clear thermal hysteresis at the vicinity of the cdw transition .
it is a strong evidence of the first order nature of the transition . up to 900k
, no anomalies are found .
this result suggests that the first order nature of the transition has another origin than in the case of two - dimensionnal dichalcogenides .
the superconducting properties are close to a weak coupling bcs theory at the opposite of the dichalcogenides 2h - nbse@xmath16 or 2h - nbs@xmath16 which are at the vicinity of a second order cdw transition .
+ [ 1 ] morosan et al ; nature phys . 2 ( 2006 ) 544 | we measured the electrical resistivity from 2k up to 900k on high quality single crystals of lu@xmath0ir@xmath1si@xmath2 .
a clear thermal hysteresis was found at the onset of the charge density wave ( cdw ) , evidencing the first order nature of the transition . when tantalum is included in the compound ,
the cdw is destroyed and the superconducting critical temperature is enhanced .
finally , we present specific heat and magnetic penetration depth in the meissner state . we show that the superconducting properties are very close to a weak coupling bcs superconductor . |
in the recent years it has been demonstrated that ultracold atoms loaded in optical lattices provide an ideal realization of lattice hamiltonians @xcite .
the control of the system parameters , in particular of the ratio between interactions and kinetic energy , has allowed the experimental achievement of the most famous mott insulator to superfluid phase transition in 2002 @xcite . from there on , the number of implementations of hubbard - like models using cold gases in optical lattices has undergone an incredible growth ( see , e.g. , @xcite ) : it is now possible to mimic single- and multi - species bose and fermi - hubbard models , extended hubbard models by using dipolar gases @xcite , and spin chain models @xcite .
more recently , hubbard models characterised by non - trivial topology , synthetic magnetic field , artificial gauges @xcite , or synthetic dimensions @xcite have been realized .
most of the theoretical studies on lattice hamiltonians regard many - body or single particle properties , especially in the case of non trivial topology . on the other hand ,
lattice hamiltonians show interesting features also as far as the few body physics is concerned . indeed ,
when the effect of interactions is combined with the presence of the lattice , composite objects on a lattice behave very differently with respect to their counterpart in free space @xcite .
first of all the dispersion relation , i.e. the effective mass of bound particles depends on the binding energy .
this is due to the impossibility in the lattice of separating the relative motion and the center - of - mass degrees of freedom .
moreover the spectrum of the scattering states , the so - called essential spectrum , is bounded from below and from above .
therefore bound states can exist both for attractive and for repulsive interactions @xcite .
the existence of exotic " repulsive bound pairs ( rbp ) has been directly observed for the first time a few years ago for an ultra - cold bose gas in an optical lattice @xcite . in this work ,
we consider two atoms with two coherently coupled internal levels in a one dimensional ( 1d ) optical lattice .
the two - body properties of this system can be investigated analytically , allowing for a clear understanding of the interplay between the exchange term ( phase coupling ) and the intra- and inter - species interactions ( density coupling ) .
the bound states of the system can be described in terms of a pseudo - spin wavefunction .
thanks to its spinorial structure , the bound states can , in certain regimes , show the co - presence of repulsive and attractive character or the co - presence of free ( delocalized ) and bound ( localized ) nature .
those features are absent in the single species situation .
moreover , we show that an _ effetive momentum - dependent magnetic field _ emerges for the pseudo - spin of the dressed dimers , by means of a coupling between the internal state of the bound state and its center - of - mass motion .
the emergent coupling between internal and external degrees of freedom for the bound states is due to the different dispersion relations of the bare dimers in the different coupling channels and is very easily understood in terms of dressed bound states .
the same effect also emerges in a dimer of two distinguishable particles , provided that at least one of them possesses two coherently coupled internal levels .
the paper is organised as follows : in sect .
[ model ] we describe the model and make the link to the single species repulsively bound pairs .
we introduce the concept of _ momentum - dependent _ internal wavefunction for the dimers .
we present an intuitive explanation of the physics behind this effect using an effective hamiltonian and an effective pseudo - spin picture . in sect .
[ sec : lippmann - schwinger ] , we solve analytically our model via a lippmann - schwinger treatment .
we investigate in detail the properties of the bound states and discuss the effect of bloch oscillations on the dimer pseudo - spin dynamics at the semiclassical level .
we describe in detail the case of hybridized bound and scattering states , which are not present in the analogous single species model . in sect .
[ asym ] , we generalize our results in the presence of species - dependent hopping and intra - species interaction and compare the effective model with exact diagonalization results . in sect .
[ sec : distinguishable ] , we show that analogous results are obtained for dimers of two distinguishable particles , as long as one of them has two coherently coupled internal levels .
the experimental implementation and relevance of our work in the cold - gases context is discussed in the conclusions .
we consider a two - component single - band bose - hubbard model with an exchange term between the two species , characterized by a frequency @xmath0 . in second quantization
the hamiltonian reads @xmath1 where @xmath2 the operators @xmath3 ( @xmath4 ) and @xmath5 ( @xmath6 ) are the annihilation ( creation ) operators for a particle on site @xmath7 in the internal state @xmath8 and @xmath9 , respectively . in this work
, we set the lattice constant @xmath10 . a possible realization of model ( [ full_h ] )
can be provided by bosonic atoms in a 1d deep optical lattice , with two hyperfine states resonantly coupled via an external laser field . at the many - body level a number of properties of this model in different regimes have been already studied by means of bosonization techniques or using density matrix renormalization group numerical approaches ( see , e.g. , @xcite ) . in the present work we focus instead on the two - body physics and in particular on the bound states properties .
let us first recall that in a single species 1d bose - hubbard hamiltonian an on - site interaction @xmath11 creates a dimer at energy @xmath12 where @xmath13 is the center - of - mass momentum .
depending on the sign of the interaction parameter @xmath11 , the bound state lies above or below the scattering states , whose energies define the essential spectrum @xmath14 $ ] ( see , e.g. , @xcite ) . for @xmath15 , the two - body bound state
is usually referred to as repulsively - bound pair @xcite .
the intuitive explanation behind repulsively bound pairs is provided by the fact that the interaction energy of a bound state can not be converted into kinetic energy , due to the limited energy bandwidth .
the dimer has a non - trivial dispersion relation , which becomes more and more flat for increasing interactions ( see , e.g. , @xcite and references therein ) . for @xmath16 , the curvature of the dispersion relation at @xmath17 , proportional to the inverse effective mass ,
is related to two - body hopping in second order perturbation theory @xmath18 . if we have two species @xmath8 and @xmath9 without coherent coupling , i.e. @xmath19 in eq .
( [ ham ] ) , there is a single bound state at energy @xmath20 given , for @xmath21 , by eq .
( [ ebs0 ] ) with @xmath11 replaced by @xmath22 or @xmath23 , depending on whether the two particles are in the same or in different internal states .
the bound state wavefunction can be written as @xmath24 , where @xmath25 is the relative motion wavefunction and @xmath26 describes the internal composition of the bound state , with to be the symmetric superposition for two particles in species @xmath8 and @xmath9 . the antisymmetric combination , which for bosons would lead to odd scattering wavefunctions , does not play any role in the problem under consideration in this work . ] @xmath27 the structure of the eigenfunctions is very simple because the internal state composition @xmath28 does not depend on @xmath13 . ] .
adding a finite @xmath0 leads to a mixing of the previously described bound states .
neglecting for the moment the presence of the scattering continuum and its influence on the bound states , let us define for the bound states an effective hamiltonian whose diagonal elements are given by the bound state dispersions @xmath29 in eq .
( [ ebs0 ] ) . assuming that the bound state wavefunctions @xmath30 do not depend strongly on @xmath28 ,
the exchange term provides the off - diagonal matrix elements @xmath31 .
hence for each value of @xmath13 , one obtains an hamiltonian of an effective @xmath32-system which reads @xmath33 with @xmath34 , @xmath35 , @xmath36 .
we postpone to sect .
[ asym ] the discussion of the general case @xmath37 and @xmath38 and , for the sake of simplicity , we consider in most of the paper the @xmath39 symmetric situation where @xmath21 and @xmath40 . in that case ,
@xmath41 , @xmath42 , @xmath36 and the diagonalization of the effective hamiltonian leads to two coupled bright states and a dark state not affected by the coupling @xmath0 .
the dark state , at energy @xmath43 , is the coherent superposition of the two intra - species dimers providing the bell state @xmath44 and does not depend on the center - of - mass momentum .
the bright _ upper _ ( @xmath45 ) and _ lower _
( @xmath46 ) bound states , have energies @xmath47 their internal wavefunction can be written as @xmath48 in general the internal composition of the bright states does depend on the center - of - mass momentum .
this is physically due to the fact that interactions try to preserve polarised bound states , while coherent coupling prefers to force the atoms in a balanced superposition of the two internal levels . from eq .
( [ eff - states ] ) and the reasoning above , in order for the momentum - dependence of the bound states to be pronounced , one needs relevant bare intra- and inter - species energy differences @xmath49 combined with @xmath50 and @xmath23 not much larger than the hopping parameter @xmath51 .
conversely , the condition @xmath52 , i.e. @xmath53 , defines the @xmath54 symmetric case and @xmath55 , the system hamiltonian is characterized by the @xmath39 symmetry under the exchange of @xmath8 and @xmath9 , corresponding to a rotation of @xmath56 around the @xmath0-axis ; under the condition @xmath57 , the system acquires a further @xmath54 symmetry and becomes invariant for any rotation around @xmath0 .
] , where also the bright bound states are characterized by the @xmath13-indipendent internal wavefunctions @xmath58 , corresponding to both atoms in the superpositions @xmath59 , eigenstates of @xmath60 ( see eq .
( [ bomega ] ) ) .
the effective hamiltonian in eq . ( [ effective_hamiltonian ] ) assumes bound states well separated from the continuum . for repulsive ( attractive ) interactions it can happen that the lower ( upper ) bound state energy enters the upper ( lower ) scattering continuum . even in this case , many of the conclusions provided for the upper ( lower ) bound state by the effettive model may still remain valid ( see sect.[sec : bound ] ) .
however , as we will show in sect .
[ sec : bscat ] , interesting phenomena can occur due to the hybridisation between lower ( upper ) bound states and scattering states .
in particular , the eigenfunctions present scattering character when projected on the upper ( lower ) two - body eigenstate of @xmath61 and bound state character when projected onto the orthogonal lower ( upper ) eigenstate of @xmath62 .
while we are mainly interested in atoms satisfying bose statistics , the previous results apply also for two distinguishable particles . in this case
it is enough that a rabi coupling between two different internal levels is present for one of the particles .
the effective hamiltonian , as well as the exact results obtained via lippmann - schwinger approach , can be easily extended ( see sect.[sec : distinguishable ] ) .
the advantage is that using two different atomic species or isotopes the range of available parameters could be broader , possibly allowing to obtain a larger momentum dependence of the bound states . before discussing the exact solution ,
let us describe the effect of bloch oscillations on the pseudo - spin of the dimers . as discussed in sect .
[ sec : lippmann - schwinger ] , in the @xmath39 symmetric case , the bound state hamiltonian in eq .
( [ effective_hamiltonian ] ) separates in a dark manifold and in a two - dimensional bright manifold spanned by the states @xmath58 ( see eq .
( [ bomega ] ) ) . introducing a pseudo - spin notation for the @xmath64 states ,
the dynamics is driven at each @xmath13 by the hamiltonian @xmath65 the density matrix for a pseudo - spin @xmath63 can be written as @xmath66 . from eq .
( [ h ? ] ) , the equations of motion for the pseudo - spin , describing a dimer wavepacket with center of mass momentum @xmath13 , correspond at a semiclassical level to the dynamics of a spin in an effective magnetic field @xmath67 with @xmath68 . in fig .
[ eff_spin_su2 ] , we plot the probability of finding the system in the two - body state @xmath69 , corresponding to the expectation value of @xmath70 , for two particles prepared at @xmath71 in the state @xmath69 at @xmath17 .
the pseudo - spin of a wavepacket with center of mass momentum @xmath13 rotates at a frequency @xmath72 and @xmath73 oscillates periodically in time . if the center of mass momentum @xmath13 varies in time , the consequent spin dynamics results in a spin precessing in an effective time - dependent magnetic field @xmath74 . in particular , bloch oscillations created by an external force @xmath75 induce , in the semiclassical approximation , a linear variation in time of the center of mass moment @xmath76 .
the effect of the time - dependent @xmath13 manifests itself in a chirping of the precession frequency @xmath77 and in the appearance of a beating frequency , as reported fig .
[ eff_spin_su2 ] ( red line in ( a ) and green line in ( b ) ) .
the previous treatment is useful and illustrative in the @xmath39 symmetric case and taking as initial condition a state belonging to the bright manifold . in the more general case ,
e.g. , starting with a bound state @xmath78 , or in the asymmetric case for which also the dark state is involved in the dynamics , an effective @xmath79 treatment is required . namely one can write directly the effective hamiltonian in eq .
( [ effective_hamiltonian ] ) using the gell - mann matrices @xmath80 , @xmath81 as @xmath82 where we introduced the pseudo - magnetic field @xmath83 the density matrix can now be written as @xmath84 .
the equations of motion for the spin of a wave packet of center of mass momentum @xmath13 become @xmath85 where we use the standard notation for antisymmetric tensor @xmath86 defining the @xmath87 lie algebra @xmath88=if^{ljk}\lambda_k$ ] .
in this section , we solve exactly the two - body problem for hamiltonian ( [ full_h ] ) using the lippmann - schwinger formalism .
we restrict ourselves to the @xmath39 symmetric case of @xmath89 and @xmath55 , which allows for compact analytical expressions and a clear interpretation of the results .
the more general case is treated in sect .
[ asym ] where the results are obtained by exact diagonalisation and a generalized effective model . in order to write the lippmann - schwinger equations , it is convenient to define @xmath90 to be the free hamiltonian . since we are considering @xmath89 , so that @xmath91=0 $ ] , the most suited basis is given by the 2-body internal eigenstates of @xmath92 @xmath93 the state @xmath94 corresponds to the dark state discussed above .
conversely , as it will become clear in eq.([forma_u ] ) , the states @xmath64 are coupled to each other by the interaction term . the essential spectrum @xmath95 is provided by union of the three scattering spectra @xmath96 obtained by shifting the essential spectrum of a single component bose - hubbard model by @xmath97 for @xmath98 . to describe the external degrees of freedom , we follow the standard procedure of introducing the centre - of - mass coordinate @xmath99 and the relative coordinate @xmath100 for two particles at lattice positions @xmath101 and @xmath102 . the center of mass and relative coordinates do not separate on a lattice , but
still the center of mass momentum @xmath103 is a good quantum number .
therefore , the eigenstates can be written as spinor wavefunctions @xmath104 with @xmath105 . inserting this ansatz in eq .
( [ full_h ] ) , we get the discrete schrdinger equation for the relative motion @xmath106 \psi_{k}^{\sigma}(r ) + \sum_{\sigma'}u _ { \sigma,\sigma ' } \delta_{r,0 } \psi_{k}^{\sigma'}(r ) , \end{split } \label{discr_schr_tutto}\ ] ] where @xmath107 is the kronecker - delta and the parametric dependence of the kinetic energy on @xmath13 is contained in the discrete gradient @xmath108 $ ] .
the interaction matrix reads @xmath109 as already mentioned , interactions mix only the states @xmath64 spanning the so - called bright manifold , while the dark state @xmath94 is completely decoupled . for each value of @xmath13 ,
let us call @xmath110 the free hamiltonian , with eigenstates @xmath111 satisfying @xmath112 .
the formal solution of the lippmann - schwinger equation reads @xmath113 where @xmath114 is the free retarded green s function .
bound states do exist if the homogeneous equation @xmath115 has non - zero solutions , or equivalently if there exist values @xmath116 such that @xmath117=0 .
\label{determinanteenergia}\ ] ] in our case the green s function components read @xmath118 where we define @xmath119 and @xmath120 the dark state gives rise to a bound state equivalent to the one of the single species hubbard model with @xmath121 .
the bright states are coupled by the interactions and the condition for the existence of bound states reads @xmath122 as already noticed in the previous section , for @xmath52 the solution simplifies to two bound states with @xmath13-independent internal states @xmath64 and effective interaction @xmath123 . in the general case ,
the two internal states @xmath64 are mixed giving rise to @xmath13-dependent superpositions .
as in the previous section , we name @xmath45 ( @xmath46 ) the upper ( lower ) bound state . typical bound and scattering spectra are reported in fig .
[ ex_diag1 ] .
since the dark state is completely decoupled , its spectrum is not shown in the figures . in panel
( a ) the parameters are such that there exists two well defined bound states above the essential spectrum .
the dashed lines are the approximated bound state energies given by eq .
( [ eig_eff ] ) , that are almost exact in this case . in panel
( b ) both bound states are well defined for all @xmath13 values .
however due to the vicinity of the lower bound state to the scattering continuum , the results provided by the effective model ( dashed line ) for its energy are not accurate . in panel ( c ) , the lowest bound state enters the upper scattering continuum in the brillouin zone center . in that case , scattering and bound states get hybridized , as we will discuss in detail in sect .
[ sec : bscat ] . in spite of the presence of the continuum , which is not accounted for by the effective model , the prediction for the upper bound state energy is almost perfect . for @xmath52 ,
one obtains the @xmath54 symmetric case , shown in panel ( d ) . in this case , all bound states are perfectly defined and characterized by the @xmath13-indipendent @xmath64 internal wavefunctions . in spite of the energy of the lower bound state being immersed in the upper scattering continuum , there is no bound - scattering states hybridisation , since the internal wavefunction of bound and scattering states at the same energy are orthogonal .
once the bound state energies @xmath124 are determined by solving ( [ cond - bs ] ) , the bound state wavefunctions @xmath125 , for @xmath126 , can be obtained by @xmath127 the explicit form of the green s function written in eq .
( [ green_final ] ) implies that the amplitudes in the @xmath64 states have the following spatial dependence @xmath128 with @xmath129 previously defined in eq .
( [ lambda ] ) .
eventually the ratio at @xmath130 of the spinor components can be written , e.g. , as @xmath131 from eq .
( [ bs_wf ] ) , it appears clearly that the bound states are given by two exponentially localized wavefunctions for the @xmath64 components with different @xmath13-dependent decay constants , , ensures real values of the @xmath132 s .
moreover , the choice of sign in ( [ lambda ] ) , garantees that @xmath133 , providing an exponentially decreasing wavefunction . ] as shown in figs . [ upper_wf ] and [ lower_wf ] for upper and lower bound state respectively .
the sign of @xmath134 determines the sign of @xmath129 , which is negative for bound state energies above the @xmath135-essential spectrum and positive for bound state energies below @xmath136 . assuming @xmath137 , the upper bound state is always above both essential spectra , implying negative values for both @xmath129 and a @xmath56-paired oscillating wavefunction typical of repulsively bound pairs @xcite .
instead the lower bound state can lie above or between the two essential spectra . in the latter case , @xmath138 is negative , indicating a repulsive character of the bound state in the @xmath139 internal state , but @xmath140 is positive , indicating that in the @xmath69 internal state the bound state is actually attractive ( see fig .
[ lower_wf ] ) .
we remind that the @xmath64 channels describe bound states for two atoms in the rotated internal states @xmath141 .
therefore the bound state wavefunction shows repulsive or attractive nature , or the co - presence of both , depending on which quantisation direction is chosen in the measurement of the internal state . in the previous part
, we have written the bound state wavefunction in terms of the @xmath58 states and have shown how the two components depend on the center of mass momentum @xmath13 .
the dependence of the internal wavefunction of the dimers on the center of mass momentum @xmath13 can be referred to , in a broad sense , as a _ spin - orbit coupling _ effect . from the experimental point of view , rather than considering the @xmath58 basis , it might be more straighforward to carry out the measurements using the basis @xmath142 , already introduced to develop the effective model in sect . [ model ] .
the population in the different two - body states @xmath143 , @xmath144 and @xmath145 defines a _ polarization _
vector @xmath146 $ ] for each bound state ( @xmath147 ) , where obviously @xmath148 . to quantify the emergent spin - orbit coupling
, we look at the @xmath13-dependence of the polarizazion in the upper bound state . in fig .
[ pol_symm ] , we show a typical situation for @xmath149 and @xmath150 , @xmath151 . a more detailed analysis , including also the effect of asymmetries between @xmath152 and @xmath153 will be presented in sect . [ asym ] .
+ as we have already mentioned , for repulsive interaction the lower bound state can enter the upper scattering spectrum .
in other words , for @xmath137 , there are energies in the scattering spectrum of two atoms in channel @xmath69 that are immediately close to a pole in the @xmath154-matrix for two atoms in channel @xmath139 , identified by the condition @xmath155 . in the @xmath54 symmetric case @xmath156 , the two internal states @xmath58 define two independent channels . for each channel , at the energies identified by the poles of the green s function , one finds either a scattering or a bound state , like in the single species case . instead , in the general case of different interaction parameters and in the presence of a coherent coupling between two species ,
the two atoms can be found in a superposition of scattering and bound states .
this intriguing concept will be formalized here below .
the lippmann - schwinger equations for @xmath157 read @xmath158 where we introduced the internal momentum @xmath159 satisfying @xmath160 , the green s function @xmath161 has been defined in ( [ green_final ] ) and the bare green s function for the scattering states reads @xmath162 the components of the spinor wavefunction are therefore @xmath163 , \nonumber\\ & & \psi_k^- ( r)=g_k^{-}(r,0,e ) \frac{u - u_{ab}}{2}\frac{1}{1-g_k^{-}(0,0,e)\frac{u+u_{ab}}{2}}. \nonumber\end{aligned}\ ] ] from the above expressions , it is straightforward to extract some important properties of the hybridised state : first of all , the @xmath164 component presents a delocalized wavefunction , typical of scattering states ; on the other hand , @xmath165 shows a localized exponentially decreasing wavefunction typical of bound states . unless @xmath52 , the two components feel each other and are hybrized .
the amplitude of the @xmath165 component is amplified , the closer the energy is to the bare bound state energy of two atoms in the two - body state @xmath139 .
at the same time the @xmath164 component undergoes a phase shift determined by both by the interaction coefficient @xmath123 in the @xmath166 channel and the presence of the bound component in the @xmath139 channel .
eventually , in the strongly interacting limit , where the scattering part strongly dominates over the solution of the homogeneous equation , the wavefunction is fermionized and developes a dip at @xmath130 . in the presence of the @xmath165 component
, the strongly interacting limit is reached close to the pseudo - resonance condition @xmath167 .
those features are reproduced in fig .
[ pseudo_1_wf ] where we plot the wavefunction varying @xmath13 from the brillouin zone edge towards the center for the parameters corresponding to fig .
[ ex_diag1](c ) : at the brillouin zone boundary there exists a purely bound state ; around @xmath168 the eigenstates become hybridized presenting both bound and scattering characters .
in this section , we address the most general case where @xmath37 and @xmath169 using exact diagonalization and generalizing the effective model . as a basis for exact diagonalization calculations , it is convenient to use all possible distributions in the lattice sites of two bosons in the single particle atomic states @xmath8 and @xmath9 and @xmath170 .
hence , one can take as a basis all possible distributions in the lattice sites of two bosons in the internal single particle states @xmath59 . ] .
we label all eigenstates with a well - defined value of the center of mass momentum @xmath13 and determine the spectrum for bound and scattering states . in order to compare the information about the polarization of the system provided by effective hamiltonian
, we should integrate the wavefunction provided by the exact solution over relative distance .
the effective hamiltonian , written in the basis @xmath142 , has the shape shown in eq .
( [ effective_hamiltonian ] ) presenting off - diagonal couplings equal to @xmath171 , and diagonal elements , which now explicitely read @xmath172 the most relevant difference with respect to the symmetric case is that the dark and bright manifolds are not defined anymore and in general three non degenerate bound states are to be expected .
depending on the specific choice of parameters , the @xmath13-dependence of the polarization can be strongly enhanced or strongly suppressed . to quantify the amount of spin - orbit coupling present in the system ,
we define the visibility @xmath173-\min_k[\mathcal{p}^u_\alpha]$ ] for the different components of the polarization @xmath174 .
we first consider the symmetric case . from fig . [ vis_pol_symm ]
, we see that for fixed @xmath40 , @xmath89 , the visibility of @xmath175 is largest when @xmath176 and @xmath177 largest .
the visibility for the other components follows straighforwardly from the relations @xmath178 and the normalization condition .
larger values of @xmath150 produce very similar behaviours but , as expected , an overall smaller amount of coupling between internal and external degrees of freedom .
taking as a reference the parameters used in figs .
[ pol_symm ] , @xmath179 , @xmath180 and @xmath181 , which roughly satisfy the above conditions , we study how the visibility is affected when asymmetry between the two species is introduced by varying @xmath182 and @xmath183 .
the results for the visibility of @xmath175 is shown in fig .
[ vis_pol_asymm ] . in this case
a non trivial dependence is observed , in particular as far as the effect of the different tunneling parameters is concerned . in general
, the effective model produces very reliable results for the energy and polarization of the upper bound state in a broad range of parameters , as one can for instance observe in fig .
[ pol_symm ] .
this includes the cases where the lowest bound state is actually immersed in the scattering continuum and consequently not well described by the effective model . to get a more precise idea of the regimes of validity of the effective model
, we calculated the relative difference of the upper bound state energy and polarization between effective and exact model . at the brillouin zone edge @xmath184
the effective hamiltonian always produces exact results , since there are no scattering states available to hybridize the bound states .
one obtains exact agreement for @xmath19 , which is a trivial case , and for @xmath52 as expected , due to the restored @xmath54 symmetry .
the agreement is good for @xmath185 and @xmath186 , namely the cases where the dominant coherent coupling creates bound states in the coherent @xmath64 superpositions .
however , it is important to note that even if in the case of largest spin - orbit coupling , namely @xmath187 , the error becomes of the order of several percent , the predictions of the effective model remain qualitatively reliable .
as discussed in the previous sections , the emergent momentum dependent polarisation is a consequence of the competition between breaking of galilean invariace , coherent coupling and interactions .
therefore , a @xmath13-dependence of the internal state composition of dressed dimers is quite a general feature , not unique to the previously discussed system . in particular ,
similar concepts can be applied to the case of two distinguishable particles , as long as at least one of them is characterised by two internal states , namely @xmath8 and @xmath9 coupled by an exchange term @xmath0 .
let us called @xmath188 the state of the second species or isotope atom .
interactions are characterized by the two coefficients @xmath189 and @xmath190 .
the different terms of the 2-particle hubbard - like hamiltonian in eq .
( [ full_h ] ) now read @xmath191 with @xmath192 . as in sect .
[ sec : lippmann - schwinger ] , the convenient basis is provided by the two - body internal eigenstates of @xmath92 @xmath193 with eigenvalues @xmath194 .
the eigenstates of the hamiltonian can be written as spinor wavefunctions @xmath195 with @xmath196 .
provided that @xmath197 , the relative motion for @xmath198 is described again by eq .
( [ discr_schr_tutto ] ) and interactions are now given by the @xmath199 matrix @xmath200 the analogy with the case described in detail in sect .
[ sec : lippmann - schwinger ] is striking and similar conclusions can be reached .
analogously , we can write an effective hamiltonian under the form @xmath201 where eq .
( [ ebs0 ] ) with @xmath202 provides the diagonal elements .
two different interaction couplings @xmath203 lead to @xmath13-dependent polarisation for the bound states as previously discussed . from the experimental point of view , considering two distinguishable particles could facilitate the realization of different interaction and hopping strengths .
in this work we have investigated the bound states of two coherently coupled interacting bosons .
we have shown an emerging momentum dependent polarisation due to the non separability of the centre of mass and relative coordinates .
we have derived an effective hamiltonian , which has allowed us to give a direct interpretation of the coherently coupled system in terms of effectively coherently coupled bound states , and tested its validity against exact solutions based on the lippmann - schwinger equation and exact diagonalization .
we have extended our study to the most general case where different tunneling parameters and interaction strengths for the two internal states are considered , in order to possibly access realistic experimental situations and found the conditions to optimize the correlations between internal and external degrees of freedom .
the general character of our results has been stressed by introducing a similar model for two indistinguishable particles where only one needs to be dressed by a rabi coupling . on the experimental side
, many of the ingredients necessary to study the physics of spin - momentum coupling are already available .
many experiments have been already realised for coherently coupled bose gas , starting from the seminal investigations in 1999 @xcite , to the josephson and classical bifurcation experiments @xcite , the realisation of polarisation dependent persistent currents @xcite , till the most recent experiments on spin - orbit coupling , artificial gauge fields and synthetic dimensions , where optical lattices are also present ( see e.g. @xcite ) .
the most difficult issue is the possibility of realising large enough differences between intra- and inter - species interactions .
indeed for the most commonly used @xmath204rb atoms , the difference is very small and feshbach resonances are difficult to implement .
a possible idea would be to consider two different atomic species , as discussed at the end of this work .
alternatively , one could lift the degeneracy of the hyperfine states in , e.g. the @xmath205 manifold and use spin - selective microwave pulses to couple the @xmath206 states in such a way to reduce the intra - species interaction leaving the inter - species unchanged @xcite .
another possibility could be to work with spin - dependent lattices as recently shown in @xcite . in this configuration , aside from breaking @xmath39 symmetry , by changing the overlap between @xmath8 and @xmath9 species one could also sensibly tune the inter - species interaction .
the authors thank j. catani , l. fallani and w. zwerger for useful discussions and m. di liberto for a careful reading of the manuscript .
this work has been supported by erc ( qgbe grant ) and provincia autonoma di trento .
acknowledges support from the alexander von humboldt foundation . | we study the two - body bound and scattering states of two particles in a one dimensional optical lattice in the presence of a coherent coupling between two internal atomic levels . due to the interplay between periodic potential , interactions and coherent coupling , the internal structure of the bound states depends on their center of mass momentum .
this phenomenon corresponds to an effective momentum - dependent magnetic field for the dimer pseudo - spin , which could be observed in a chirping of the precession frequency during bloch oscillations .
the essence of this effect can be easily interpreted in terms of an effective bound state hamiltonian .
moreover for indistinguishable bosons , the two - body eigenstates can present simultaneously attractive and repulsive bound - state nature or even bound and scattering properties . |
the kinetics of surface particle adsorption and of transport through interfaces play a key role in surfactant phenomena @xcite , membrane biology and cell signaling @xcite , marine layer oceanography @xcite , and other biological and chemical processes .
particle adsorption may fundamentally alter the physical and chemical properties of the interface , and it is crucial to understand both equilibrium and dynamical properties of the adsorbed layers @xcite . in the seminal work of ward and tordai @xcite ,
a bulk phase acting as a reservoir of particles is physically limited by an empty surface onto which the particles can adsorb .
particles are assumed to lower their free energy with respect to the bulk phase by irreversibly and instantaneously adsorbing onto the interface . under these conditions , the total concentration of adsorbed particles
may be estimated in relation to measurable interfacial properties , such as the dynamic surface tension .
several applications , extensions and alternate approaches to this work have been proposed @xcite . in particular
, adsorption dynamics in the ward - tordai setting can be rederived through a free energy approach @xcite , allowing for the inclusion of ionic surfactant effects and electrostatic interactions . in many biochemical systems ,
the complete adsorption of a particle arriving from the bulk requires a series of auxiliary transformations at the surface before the particle can be successfully incorporated , or ` fused ' into the surface .
these intermediate steps gives rise to a lag - time in the complete adsorption process .
for example , the incorporation of emulsifying proteins onto an air - water interface may be delayed by the unfolding of the polypeptide at the interface @xcite .
adsorption of proteins on polymer - grafted interfaces , such as the glycocalyx layer of vascular endothelial cells , is also delayed due to the progressive insertion of the protein through the polymer brush@xcite .
kinetic delays have also been observed in the adsorption of the hemagglutinin glycoprotein ( ha ) of the influenza virus as it enters target host cellular membranes @xcite .
the mechanisms underlying this delay are not known in detail but are believed to involve conformational changes of ha molecules into fusion enabling complexes , mediated by the presence of binding receptors and coreceptors on the target cell membrane @xcite .
similarly , the incorporation of an hiv particle into a t - cell or a macrophage is possible only after the gp120 glycoprotein of the hiv virus membrane recognizes and binds to the target cell surface receptor cd4 , and subsequently to other coreceptors such as ccr5 or cxcr4 . as in the case of ha and influenza
, the exact number of gp120-bound receptors and coreceptors required for hiv particle fusion is yet unknown and might depend on gp120 conformations and receptor / coreceptor binding cooperativity @xcite .
the complex nature of surface biochemistry makes quantitative kinetic measurements challenging .
recently , the binding kinetics of the cd4 cellular receptor to the gp120 hiv ligand have been measured under different experimental conditions yielding widely different dissociation rates @xcite . in this work
, we will provide a quantitative framework that can be used to better understand the experimentally observed lag - times in surface kinetics phenomena that involve multistage surface chemistry . .
particles such as viruses are spread over the cell layer in a thin supernatant film .
inset : after initial nonspecific viral adsorption on the supernatant - cell interface , cellular receptors and coreceptors bind to the virus via a certain stoichiometry , forming fusion intermediates @xmath1 .
the subsurface layer of thickness @xmath2 , the subsurface concentration @xmath3 , and the adsorbed species @xmath4 are discussed in the text.,height=201 ] in particular , we will explicitly consider intermediate , reversible steps for surface binding in the ward - tordai formalism , deriving an effective boundary condition to complement the bulk diffusion process .
chemical transitions among the surface species will introduce memory terms in the boundary conditions for the bulk concentration .
our analysis can be readily applied to the titration of replication - incompetent virus via a colony formation assay @xcite as shown in fig.[fig1 ] .
in this section , we motivate and derive the equations coupling bulk diffusion to surface layer evolution . we consider a general , linear reaction scheme to describe the multistep surface reaction dynamics .
effective boundary conditions for diffusion from the bulk are derived in sect.3 .
as we shall discuss in detail , we are able to embody the response of the adsorbing particle system to the existence of intermediate chemical steps at the surface , into a unique delay kernel regulating the boundary dynamics .
all microscopic details stemming from the surface dynamics , no matter how complicated , are contained in the derived memory kernel .
our approach includes ligand rebinding to surfaces , found to be important for analyzing surface plasmon resonance assays of biochemical systems @xcite . in sect.4
we particularize our surface reaction scheme to a specific markov process chain and evaluate all physically relevant quantities . in the continuum limit
, the density of particles @xmath5 in the bulk phase obeys the convection - diffusion equation @xmath6 + \frac 1 { k_b t}\nabla\cdot\left[{d}\,n\nabla u \right ] , \label{convectiondiffusion}\ ] ] where @xmath7 and @xmath8 are the local diffusion coefficient and potential of mean force , respectively , and @xmath9 is the thermal energy
. spatial variation of @xmath7 and @xmath8 may arise from interactions with the interface as shown in fig.[layer ] .
boundary conditions are typically applied at the mathematical surface onto which the particles adsorb or reflect . by balancing the diffusive flux just above this mathematical interface with the particle rate of insertion into the interface
, a mixed boundary condition arises @xmath10 here , @xmath11 denotes the substrate ; its normal direction is @xmath12 .
the parameter @xmath13 , which has the physical units of speed , is proportional the probability @xmath14 ( often called the accommodation coefficient @xcite or sticking probability @xcite ) that a particle is adsorbed into the mathematical interface upon collision .
we define @xmath15 such that in the limit @xmath16 and @xmath17 , eq.[bc0 ] is equivalent to @xmath18 , an absorbing boundary condition .
a reflecting boundary condition , @xmath19 , arises when @xmath20 .
equations [ convectiondiffusion ] and [ bc0 ] are commonly used to model simple diffusion - adsorption processes at surfaces . in many applications , particles at an interface
undergo chemical or physical modifications that control for example , surface reactivity , surface tension @xcite , and conductivity @xcite .
biological examples include tissue factor initiated coagulation reactions and viral entry .
coagulation factors must work their way through the glycocalyx layer before they can be enzymatically primed by the membrane - bound tissue factors @xcite .
entry of viruses , such as hiv , into cells require the binding of membrane - bound receptors and coreceptors before fusion with the target cell can occur .
all of these processes can be thought of as reactions at the membrane surface .
immediately after adsorption from the bulk , the surface particle concentration , whether of coagulation factors or of virus particles , is denoted by @xmath21 .
for example , in the case of viruses , we can identify the @xmath22 state as being that of a virus bound to @xmath23 cd4 surface receptor .
the initially adsorbed species can then kinetically evolve into the other species @xmath24 representing virus particles with @xmath25 bound receptors or coreceptors .
the kinetics among the @xmath26 surface species follows the linear rate equation @xmath27 , where @xmath28 , @xmath29 is the transition matrix among the @xmath26 surface states , and @xmath30 is the source of the first , originating source species @xmath21 coming from the bulk . .
the source species @xmath21 is supplied by the bulk surface concentration at the interface , @xmath31 . in the context of virus recognition and infection , the intermediate steps label various numbers of receptors or coreceptors associated with the surface bound virus particle .
for example , @xmath24 may denote the surface concentration of hiv particles with @xmath32 receptors and coreceptors attached .
this catenary model could also represent successive degrees of insertion of an absorbing species through a polymer brush or glycocalyx coated interface.,height=33 ] figure [ chain ] illustrates a simple example of a linear surface reaction scheme that can be described by the above linear rate equation . in this case the reaction matrix @xmath29 is tridiagonal .
general reaction matrices can also be analyzed since our results depend only on the eigenvalues and eigenvectors of @xmath29 . because the surface densities @xmath33 carry units of number per area , and the bulk densities are expressed by number per volume , any kinetic parameter linking bulk source concentrations to those at the interface must introduce a physical length scale .
diamant and andelman @xcite have introduced the sublayer thickness as a mathematical step coupling the bulk density to surface density . here , we physically motivate this `` surface layer '' and the associated transport .
let us thus introduce a thin layer of thickness @xmath2 near the surface , in which the particle density is denoted @xmath3 and is still expressed in units of number per volume .
the continuum approximation eq.[convectiondiffusion ] breaks down when resolving the transport within distances of a few mean free paths .
if we identify the sublayer thickness @xmath2 with the mean free path @xmath34 , as shown in fig.[layer ] , we must solve eq.[convectiondiffusion ] with a nonuniform @xmath8 , and possibly a nonuniform @xmath7 , to within a distance @xmath35 of the interface .
for the choice @xmath36 , the adsorption velocity @xmath37 can be approximated by the thermal velocity @xmath38 such that @xmath39 .
the value of the bulk density at the boundary @xmath40 is defined as the sublayer density : @xmath41 .
the equation for the rate of change of the number per area of molecules in the thin layer , @xmath42 can be obtained by balancing the latter with the diffusive flux into the layer @xmath43 , the adsorption into the surface concentration @xmath21 , and the spontaneous desorption from the initially adsorbed species @xmath21 occurring at rate @xmath44 .
the complete set of equations coupling the variables @xmath45 and @xmath24 is thus @xmath46 + \frac 1 { k_b t } \nabla\cdot\left[{d}\,n \ , \nabla
u \right ] , \label{convectiondiffusion2 } \\
\displaystyle \l{\dd n_{s } \over \dd t } & = & -f + d \,{\bf \hat n}\cdot\nabla n \ , \big|_{\r = \s+\l{\bf n } } + q_{1}\ , \g_{1},\label{ns } \\
{ \dd { \bf \g } \over \dd t } & = & { \bf m}{\bf \g } + { \bf f } , \quad \ , { \bf f } = ( f,0,0,\ldots,0 ) .
\label{gamma}\end{aligned}\ ] ] here , @xmath47 , and @xmath48 $ ] is the flux of the surface concentration @xmath31 into the incipiently adsorbed species @xmath21 .
this functional may depend on interactions among the adsorbed species @xmath33 , including cooperative or crowding effects , and may be modeled using free energies and chemical potential differences between the bulk and surface @xcite . , or the mean free path @xmath34 , depending upon which is larger.,height=192 ] a further simplification can be introduced by defining a different sublayer thickness @xmath49 , where @xmath50 is the typical range of the particle - surface interaction as shown in fig.[layer ] . in this case , at least at a distance @xmath50 from the interface , @xmath7 is constant , @xmath8 is zero and eq.[convectiondiffusion ] is approximated by the standard diffusion equation @xmath51 all effects of the potential of mean force @xmath8 and spatially varying @xmath7 are now subsumed into an effective source @xmath52 $ ] .
this is consistent with all previous treatments @xcite in which transport in the bulk phase was described by simple diffusion with uniform @xmath53 and @xmath54 .
provided @xmath55 $ ] is independent of @xmath24 , equations [ convectiondiffusion]-[gamma ] can be explicitly solved in simple geometries . for low surface densities @xmath24 such that additional adsorption is not hindered by steric exclusion , one can assume a form @xmath56 $ ] independent of surface concentrations @xmath33 .
for @xmath57 , @xmath13 is now interpreted as an effective adsorption coefficient allowing eq.[convectiondiffusion ] to be replaced by eq.[diffusion0 ] , and simplifying the bulk concentration equation .
following the original work of ward and tordai , subsequent studies on adsorption and dynamic surface tension measurements @xcite eliminate the bulk density at the interface in eq.[ns ] to yield two coupled integro - differential equations for @xmath31 and @xmath21 which must be numerically self - consistently solved . here , we solve the linear eqs.[gamma ] independently from the bulk densities , but with a source term @xmath56 $ ] that connects the surface concentrations @xmath24 with the bulk concentration @xmath5 . if @xmath56 $ ] is independent of @xmath24 , the explicit solution to eqs .
[ gamma ] can be found by evaluating the eigenvalues @xmath58 and corresponding eigenvectors @xmath59 of the chemical transition matrix @xmath29 . denoting the similarity
transform matrix @xmath60 such that @xmath61diag(@xmath58 ) , the surface densities are @xmath62\delta_{1 m } \dd t'\right ] \\[13pt ] \gamma_{k}(t ) & \displaystyle = \sum_{j , m = 1}^{n } v^{-1}_{kj}v_{jm } \gamma_{m}(0)e^{\lambda_{j}t } \\[13pt ] \ : & \hspace{6 mm } \displaystyle + \sum_{j=1}^{n}v^{-1}_{kj}v_{j1}\int_{0}^{t}\!e^{\lambda_{j}(t - t')}f[n_{s}(t')]\dd t ' , \label{gammasoln } \end{array}\ ] ] where @xmath63 are the intermediate surface concentrations at @xmath64 . if there are no spontaneous sources of the surface cell intermediates , all eigenvalues @xmath65 . from eq.[gammasoln ] , in the case @xmath66 , @xmath67 is proportional to @xmath68 $ ] . upon substituting @xmath69 by setting @xmath70 in eq .
[ gammasoln ] , we substitute @xmath69 into eq.[ns ] , and find a concise description of the diffusion - adsorption process : @xmath71 @xmath72 \dd t ' , \label{effectiveboundary}\ ] ] @xmath73 where @xmath74 is the kernel constructed from the eigenvalues and eigenvectors of @xmath29 .
it is composed of an instantaneous response the immediate depletion of @xmath31 due to adsorption into the @xmath21 surface species and delay terms arising from the surface kinetics of eq.[gamma ] .
the complete set of equations [ diffusion1]-[kernel ] is one of our main findings .
this result explicitly shows how multistage adsorption is modeled by a bulk diffusion equation with an nonlinear integro - differential boundary condition that incorporates the delay arising from the multistep kinetics . under this scheme ,
all effects of surface reactions are incorporated in the kernel @xmath74 .
our analysis can be carried further by specifying a linear form for the @xmath33-independent source term @xmath75 = \gamma n_{s}(t ) , \label{linearapprox}\ ] ] which simply takes the source for the surface concentration @xmath76 of the first species to be proportional to the subsurface concentration .
the surfaces densities @xmath4 can be found by substituting @xmath3 , derived from eq.[diffusion1]-[kernel ] , into the expression for @xmath68 $ ] in eq.[gammasoln ] .
note that the boundary condition eq.[effectiveboundary ] contains a singular perturbation , and that for times @xmath77 , the `` outer solution '' approximation @xmath78 yields the standard mixed boundary condition eq.[bc0 ] with an additional memory kernel .
moreover , in the linear approximation of eq .
[ linearapprox ] , the convolution of the delay term in the effective boundary condition [ effectiveboundary ] is amenable to analysis by laplace transforms . for simplicity
, we will assume a simple one - dimensional problem where all quantities vary spatially only in the direction normal to an infinite , flat interface at @xmath79 .
we nondimensionalize all quantities by using @xmath2 as the unit of length , and @xmath80 as the unit of time .
henceforth , in all equations , we make the replacements @xmath81 and @xmath82 . to render the notation less cumbersome we omit the bars from the redefined quantities . in the discussion that follows ,
the @xmath83 parameters are intended as nondimensional . upon taking the laplace
transform in time of the dimensionless forms of eqs.[diffusion1 ] , [ effectiveboundary ] , and [ kernel ] , we obtain @xmath84 @xmath85 where @xmath86 and @xmath87 is the initial , dimensionless constant bulk and sublayer concentration . the general solution to the bulk density @xmath88 from eq.[lpdiffusion1 ]
is @xmath89 once the bulk density is derived , all other quantities can be found by inverse laplace transforming @xmath90 . in the absence of spontaneous sources of the surface intermediates , @xmath91 . in this case , it is possible to show that @xmath92 only has a simple pole at @xmath93 and a branch cut on @xmath94 $ ] . performing the integral along the latter , we find the exact results @xmath95 and @xmath96 where @xmath97 equations [ nzt]-[l ] are used to numerically compute all of our results in the next section . for completeness , analytic expressions for
asymptotically short and long time limits are derived in the appendix .
we now specify a surface reaction scheme and construct its delay kernel by using its associated eigenvalues and eigenvectors . for applications such as multiple receptor binding of the adsorbed species ,
we consider a reversible sequential markov process among @xmath26 chemical intermediates @xmath98 . as shown in fig.[chain ] , formation of state @xmath99 from state @xmath24 occurs at rate @xmath100 , while the reverse step occurs at rate @xmath101 .
the final state @xmath102 is irreversibly annihilated , by transport across the membrane , or by irreversible reaction , with rate @xmath103 .
we can then explicitly write the dimensional form of eq.[gamma ] , where @xmath29 is a tridiagonal transition matrix , as @xmath104 - ( p_{1}+q_{1 } ) \g_{1}+q_{2}\g_{2 } , \\[13pt ] \displaystyle { \dd \g_{i } \over \dd t } & = & \displaystyle p_{i-1}\g_{i-1}-(q_{i}+p_{i})\g_{i}+q_{i+1}\g_{i+1 } \quad 2 \leq
i \leq n-1 \\[13pt ] \displaystyle { \dd \g_{n } \over \dd t } & = & \displaystyle -(p^{*}+q_{n})\g_{n}+p_{n-1}\g_{n-1}. \\
\nonumber \end{array}\ ] ] in the simplest case where there is only one surface intermediate before transport across the interface , @xmath105 and the dimensionless ( @xmath106 ) delay kernel is simply @xmath107 .
the sublayer concentration @xmath3 derived from eq.[nzt ] , and the surface concentration evaluated from eq.[gamma2 ] , are shown in figs.[plot1 ] for various values of @xmath13 .
let us estimate typical parameter values for viral fusion or molecular binding processes .
typical diffusion constants for viruses of diameter @xmath108 nm and in aqueous environments , are @xmath109@xmath110/s . using the typical screened electrostatic interaction potential , @xmath111 cm , we estimate the dimensionless diffusion coefficient @xmath112s@xmath113 . on the other hand , typical diameters of small ligand molecules
are of the order of 1 nm yielding a nondimensional diffusion constant @xmath114s@xmath113 .
the nondimensional @xmath13 estimated using the thermal velocity @xmath38 is now @xmath115 . for virus particles @xmath116s@xmath117 , while for molecular ligands @xmath118s@xmath117 .
the dissociation rate @xmath44 is highly variable and typically falls in the broad range @xmath119 s@xmath120 to @xmath121s@xmath120 . for the gp120-cd4 interaction ,
the dissociation has been estimated in model systems @xcite to be @xmath122s@xmath120 , while the detachment rate for mutant viral species @xcite can be as high as @xmath123s@xmath120 .
lower dissociation rates are possible in tighter binding ligand receptor pairs such as egf - receptor @xcite where @xmath124s@xmath120 . for other pairs such as p - selectin and its receptors
@xcite , @xmath125s@xmath120 to @xmath126 s@xmath120 .
the sticking probability @xmath14 is proportional to the binding probability of upon ligand - receptor contact , multiplied by the receptor area fraction at the interface .
the factor @xmath14 depends on the receptor density , but is typically of the order @xmath127 . in fig.[plot1](a )
we plot the sublayer density @xmath128 as a function of time . for @xmath129 , fig.[plot1](a ) shows that the sublayer density @xmath3 starts at its initial value @xmath87 and decreases with a nondimensional rate proportional to @xmath13 , eventually monotonically reaching @xmath130 .
if the annihilation rate @xmath103 is decreased , @xmath31 may no longer be monotonic .
the observed increase in the surface concentration is due to the slow consumption of material at the interface , allowing some of the material to desorb after being delayed at the interface , rather than irreversibly reaching the final annihilated or fused state .
for example , when @xmath131 , the surface concentration @xmath3 first decreases but recovers slightly at longer times , before ultimately decaying to zero as shown in fig.[plot1](b ) . for various @xmath132 with @xmath133 and @xmath134 .
( b ) the bulk density profiles @xmath135 as a function of position @xmath136 at times @xmath137 , and @xmath138.,height=412 ] note that the curves for @xmath3 exhibit a transient at short times determined by @xmath139 . beyond this transient , the full solution we plot in fig.[plot1](b ) reduces to the outer solution , corresponding to setting @xmath140 on the left hand side of eq.[effectiveboundary ] .
fig.[plot1](c ) explicitly shows different behaviors of the full and outer solutions during the transient time .
the effect of the @xmath141 term is to slow down the initial decrease of @xmath31 , particularly for short times within the transient defined by @xmath139 .
the effects of neglecting the boundary layer are less pronounced for larger bulk diffusivities @xmath53 .
the temporal evolution of @xmath142 is strongly dependent on @xmath143 .
in fact , the nonmonotonicity of @xmath142 for small @xmath103 shown in fig.[plot1](b ) arises from the build - up of @xmath76 indicated in fig.[plot2](a ) which can get released back into the subsurface layer . for smaller @xmath103
, @xmath76 reaches larger values .
as long as @xmath144 , both @xmath31 and @xmath76 vanish at sufficiently long times .
complete particle depletion near the surface occurs in dimensions less than two because there is no bounded steady - state solution to the diffusion equation and the depletion zone moves away from the interface for all times as shown in fig.[plot2](b ) . despite free diffusion ,
the bulk is unable to sustain a particle source near the surface as is known from classic diffusion theory @xcite .
the replenishment at small annihilation rates @xmath132 also manifests itself in the bulk . in the case shown in fig.[plot2](b ) , as time increases from @xmath145 to @xmath146 , the bulk concentration near the interface recovers before ultimately decreasing according to eq.[longtime ] . in the surface reaction scheme .
( a ) the sublayer density @xmath31 as @xmath26 is increased .
the initial rapid fall from @xmath147 is imperceptible on this scale .
( b ) the surface concentrations @xmath33 as a function of time for @xmath148.,height=412 ] for general @xmath26 , the eigenvectors and eigenvalues must be explicitly computed . in fig .
[ plot3](a ) , we plot @xmath31 as a function of @xmath26 for uniform @xmath149 and uniform @xmath150 . for small @xmath150 ,
the surface kinetics is a highly biased random walk away from @xmath151 toward @xmath76 , resulting in a larger @xmath31 . both small @xmath150 and large @xmath26 , hinder the annihilation process and impart a more reflective character to the interface .
after initial transients , both @xmath31 and the surface concentrations @xmath33 maintain a high level for a long time before dissipating .
larger @xmath26 also effectively trap surface material in the surface reservoir @xmath33 .
the relative amounts of @xmath33 for @xmath148 are shown in fig.[plot3](b ) .
for the small @xmath152 used , most of the surface density lies in the initial species @xmath76 , decreasing in the latter species .
we derived an effective , integro - differential equation for the boundary condition of a simple diffusion process .
the approach presented differs from the typical ward and tordai treatments since we use a linear model for the time rate of change of the initially adsorbed species @xmath21 , rather than eliminating the bulk diffusion equation .
the effects of intermediate chemical steps at the boundary are described by a delay kernel that can be decomposed using laplace transforms .
this kernel is an explicit function of the eigenvalues and eigenvectors of the surface reaction transition matrix .
our results suggest that measurement of a few quantities , such as fluorescence monitoring of the sublayer density @xcite , can be used to reconstruct the principal components of @xmath74 . this approach can be used to probe qualitative features of the surface kinetics important in modeling cell membrane signaling and viral infection , where a sequence of chemical steps at the surface are required before initiation of signaling or viral fusion . in hiv infection ,
the initial adsorption rate would be proportional to the surface cd4 concentration , and the subsequent rates in the reaction scheme in fig.[chain ] would depend on the coreceptor concentrations , their surface mobilities , as well as the effects of cooperative binding @xcite .
all of these physical attributes are encoded in the distribution of eigenvalues and eigenvectors of @xmath29 . for simple linear reaction schemes on a flat surface
, we find explicit dependences of the surface concentrations @xmath33 and sublayer concentrations on the eigenvalues and eigenvectors of the transition matrix . for smaller annihilation rates @xmath132 , and at least one ( @xmath153 ) surface intermediate , we find that the surface concentration persists and can replenish the bulk concentrations @xmath31 after its initial decay
. the depletion zone in the bulk can also recover .
delays that induce instabilities in dynamical systems have been well established @xcite . here ,
although the delay occurs in a boundary condition , we observe nonmonotonic behavior arising in the bulk concentrations as well .
this rebounding effect is also apparent if one differentiates eq.[effectiveboundary ] with respect to time , giving a second order , harmonic oscillator - like equation , plus a dissipative coupling to the bulk concentration .
whether the surface concentration or the bulk concentration near the surface vanishes at long times depends on the surface kinetics as well as geometry . if the combined surface kinetics towards annihilation is slow relative to bulk diffusion , the decay of the sublayer concentration @xmath31 can be extremely slow .
similarly , if the number of surface states is large , there is an effective delay to annihilation and a higher probability that a surface species can detach and replenish the sublayer concentration .
this effect is very sensitive to the annihilation rate @xmath132 and the size of the reaction @xmath26 and can keep the subsurface concentration high for essentially all times .
a number of extensions and related approaches to this and related systems can be readily investigated .
for example , in applications such as surfactant adsorption , the surface concentration @xmath154 can be appreciable and suppress additional adsorption . if surface species @xmath24 has molecular area @xmath155 , an adsorption term including steric exclusion would be @xmath156 .
the surface rate equations remain linear in @xmath157 , but with a time - dependent transition matrix @xmath29 .
the effective boundary condition eq.[effectiveboundary ] is now nonlinear in @xmath3 through @xmath56 $ ] .
however , for many biochemical applications ( such as cell signaling and virus adsorption and entry ) the total surface concentration is low such that @xmath158 and the adsorption term can be linearized . in our one - dimensional analysis ,
as long as @xmath26 is not too large and there is an appreciable annihilation process , the surface concentrations all vanish in time .
the effects of multistage adsorption can also be explored on surfaces of arbitrary shape , particularly for cylinders and spheres . for multistage processes on a sphere , the sublayer concentration approaches a positive value @xmath159 .
we also expect positive eigenvalues @xmath160 of @xmath29 to have a striking effect on the transport .
finally , although we have only considered simple linearized surface reaction schemes with negative eigenvalues , systems that support oscillations , such as those involved in surface - mediated cell signalling , could also be treated within our framework .
features of the surface reactions and the bulk concentrations near the reacting surface remain coupled through the kernel @xmath74 . under certain conditions , nonlinear surface reaction schemes may also be linearized .
one example is in the _ stochastic _ representation of the surface reactions . if we write the surface quantities in terms of the probability distribution function @xmath161 that there are @xmath162 molecules of of type 1 , @xmath163 of type 2 , etc .
, the surface reactions can be written as a linear master equation .
this allows our approach to be applied when @xmath76 in the last term of eq .
[ ns ] is interpreted as @xmath164 , the ensemble average @xmath165 . using this interpretation ,
the full problem can be solved using linear methods similar to those presented , albeit for extremely large matrix dimension .
the authors thank benhur lee for stimulating discussions .
tc acknowledges support from the nsf ( dms-0349195 ) , and the nih ( k25ai41935 ) .
part of this work was done during the cells and materials workshop at ipam ucla .
here , we derive asymptotic expression for bulk and surface densities . the trivial short time behavior of the subsurface density is @xmath166 , independent of the surface reactions since the first physical phenomenon to occur is particle adsorption from the bulk to the interface , at rate @xmath13 .
the condition @xmath170 can be interpreted as a comparison between two typical velocities . the usual diffusive velocity , @xmath171 ,
is compared to an effective reaction velocity expressed by @xmath13 modulated by surface effects through the kernel @xmath172 .
we may thus define an effective damkhler number @xmath173 , so that eqn.[largez ] is valid only at large distances and for large values of @xmath174 .
the leading term on the right - hand - side above is independent of the surface kinetics : the first information to have traveled away from the interface is the initial depletion of the @xmath31 layer into the surface and interfacial effects emerge as first order corrections . in the @xmath175 limit
, the dominant contribution to @xmath135 comes from small values of @xmath176 in eq.[nzt ] .
approximating @xmath177 with its @xmath178 limit , we find the asymptotic long time limit this expression is valid only if the surface dynamics include a net sink of material .
as long as there is some annihilation , @xmath180 and eq.[longtime ] holds . a similar consideration of the small-@xmath176 dominated integration in eq.[gamma2 ] yields for the surface concentrations where @xmath182 is the largest eigenvalue of the chemical transition matrix @xmath29
. both eq .
[ longtime ] and [ gammaasymp ] exhibit diffusion - limited @xmath183 behavior .
these general results rely only on the linearity of @xmath56 $ ] and are valid for _ any _ surface reaction scheme through the eigenvalues and eigenvectors of the transition matrix @xmath29 and the resulting function @xmath184 . | we derive the equations that describe adsorption of diffusing particles onto a surface followed by additional surface kinetic steps before being transported across the interface .
multistage surface kinetics occurs during membrane protein insertion , cell signaling , and the infection of cells by virus particles .
for example , viral entry into healthy cells is possible only after a series of receptor and coreceptor binding events occur at the cellular surface .
we couple the diffusion of particles in the bulk phase with the multistage surface kinetics and derive an effective , integro - differential boundary condition that contains a memory kernel embodying the delay induced by the surface reactions .
this boundary condition takes the form of a singular perturbation problem in the limit where particle - surface interactions are short - ranged .
moreover , depending on the surface kinetics , the delay kernel induces a nonmonotonic , transient replenishment of the bulk particle concentration near the interface .
the approach generalizes that of ward and tordai @xcite and diamant and andelman @xcite to include surface kinetics , giving rise to qualitatively new behaviors .
our analysis suggests a simple scheme by which stochastic surface reactions may be coupled to deterministic bulk diffusion . |
the problem of providing computational support for medical diagnosis has been approached from many directions including logical reasoning , fuzzy logic , set theory , rough set theory , if - then rules , bayesian networks , classical parametric and non - parametric statistics , artificial neural networks , case - based reasoning , support vector machines , perceptrons , possibility theory , and more , as well as various aggregations or combinations of methods @xcite .
this paper describes a novel approach to diagnosis based on the sp theory of computing and cognition ( described below ) .
the main attractions of this approach are : * a format for representing diseases that is simple and intuitive . * an ability to cope with errors and uncertainties in diagnostic information . * the simplicity of storing statistical information as frequencies rather than conditional probabilities . * a method for evaluating alternative diagnostic hypotheses that yields true probabilities . * a framework that should facilitate unsupervised learning of medical knowledge and the integration of medical diagnosis with other ai applications .
_ it must be stressed that the primary purpose of this paper is conceptual _ : to describe an approach to medical diagnosis that is significantly different from the main alternatives and with potential advantages compared with those alternatives .
although a prototype of the proposed new system exists , it is not yet a shrink - wrapped software system that is ready for immediate application . like any other diagnostic system , the sp system must be equipped with a body of relevant knowledge and the creation of such a body of knowledge ( by automatic learning or by ` knowledge elicitation ' from experts ) is a major undertaking in its own right .
key elements of the sp theory are first described , just sufficient for present purposes .
section [ application_section ] describes how the theory may be applied to medical diagnosis , viewed as a process of pattern recognition .
this section also discusses how the sp system relates to several aspects of the diagnostic process , including causal reasoning and the process of acquiring the knowledge that is needed for accurate diagnosis .
section [ alternatives_section ] compares this new approach to medical diagnosis with some of the alternatives .
the paper concludes with an outline of what still needs to be done in this programme of research and with a review of the main points that have been made .
the sp theory grew out of a long tradition in psychology that many aspects of brain function may be understood as information compression ( see , for example , @xcite ) .
it is based on principles of _ minimum length encoding _ pioneered by solomonoff @xcite , wallace and boulton @xcite , rissanen @xcite and others ( see also @xcite ) .
an overview of the theory is presented in @xcite and more detail may be found in other papers cited there ( see also @xcite ) .
the sp theory has been developed as an abstract model of _ any _ system for processing information , either natural or artificial . in broad terms
, the system receives ` new ' information from its environment and transfers it to a repository of ` old ' information . at the same time , it tries to compress the information as much as possible by finding patterns that match each other and merging or ` unifying ' patterns that are the same .
an important part of this process is the building of ` multiple alignments ' as described below .
the sp framework is turing - equivalent in the sense that it can model a universal turing machine @xcite but it has much more to say about the nature of ` intelligence ' than the turing model of computing ( or equivalent models such as lamda calculus @xcite or the post canonical system @xcite ) . to date , the main areas in which the sp framework has been applied are probabilistic reasoning , pattern recognition and information retrieval @xcite , parsing and production of natural language @xcite , modelling concepts in logic and mathematics @xcite , and unsupervised learning @xcite .
two computer models of the sp system have been developed : * sp62 is a partial realisation of the theory that does not transfer any information from new to old .
this model tries to compress the new information in terms of the old information by building multiple alignments of the kind that will be seen below .
sp62 also contains procedures for calculating the probabilities of inferences that may be drawn from alignments .
a slightly earlier version of this model ( sp61 ) is described quite fully in @xcite .
both versions are relatively robust and mature .
* sp70 realises all the main elements of the theory , including the transfer of information from new to old .
in addition to building multiple alignments like sp62 , the model compiles one or more alternative ` grammars ' for the information in new , using principles of minimum length encoding .
this model , and its application to unsupervised learning , is described quite fully in @xcite .
more work is required to realise the full potential of this model . in the sp system , _
all _ kinds of knowledge are stored as arrays of atomic _ symbols _ in one or two dimensions called _ patterns_. in work to date , the main focus has been on one - dimensional patterns ( i.e. , sequences of symbols ) but it is envisaged that , at some stage , the concepts will be generalised to patterns in two dimensions . for present purposes , we may define patterns and symbols as follows : * a _ pattern _ is a sequence of symbols bounded by end - of - pattern characters such as ` ( ' and ` ) ' , not shown in the examples in this paper . * a _ symbol _ is a string of non - space characters bounded by white space ( space characters , line - feed characters and the like ) . *
any symbol can be matched with any other symbol and , for any one pair of symbols , the two symbols are either ` the same ' or ` different ' .
no other result is permitted .
* symbols have no intrinsic meaning such as ` add ' for the symbol ` + ' in arithmetic or ` multiply ' for the symbol ` @xmath0 ' .
any meaning attaching to an sp symbol takes the form of one or more other symbols with which it is associated in a given set of patterns .
* each pattern has an associated integer value representing the absolute or relative frequency of occurrence of that pattern in some domain . despite the extraordinary simplicity of this format for representing knowledge ,
the way in which sp patterns are processed within the system means that they can model a wide variety of established representational schemes , including context - free and context - sensitive grammars , class - inclusion hierarchies , part - whole hierarchies , discrimination networks and trees , if - then rules , and others .
a key part of the process of matching patterns is the building of ` multiple alignments ' , described and illustrated here .
the process of building multiple alignments in the sp system provides a unified model for a variety of computational effects including fuzzy pattern recognition , best - match information retrieval , probabilistic and exact styles of reasoning , unsupervised learning , planning , problem solving and others , as described in @xcite . this subsection and the ones that follow describe the main elements of the multiple alignment concept as it has been developed in the sp theory and explains how multiple alignments are created and evaluated in the sp system . in bioinformatics ,
a multiple alignment is an arrangement of two or more dna sequences or sequences of amino acid residues so that matching symbols are aligned .
[ dna_figure ] shows a typical example .
the general idea is that , by judicious ` stretching ' of sequences , as many symbols as possible are aligned with each other .
a variety of measures of the ` goodness ' of alignments are used but they all tend to favour alignments where the number of aligned symbols is high and the gaps between them are relatively few and relatively small . ....
g g a g c a g g g a g g a t g g g g a | | | | | | | | | | | | | | | | | | | g g | g g c c c a g g g a g g a | g g c g g g a | | | | | | | | | | | | | | | | | | | | | a | g a c t g c c c a g g g | g g | g c t g g a | g a | | |
| | | | | | | | | | | | | | | g g a a | a g g g a g g a | a g g g g a | | | | | | | | | | | | | | | | |
g g c a c a g g g a g g c g g g g a .... in the sp framework , the concept of multiple alignment has been modified as follows : * one or more of the sequences ( termed _ patterns _ , as described in section [ representation_of_knowledge ] ) are classified as ` new ' and the rest are ` old ' . * a ` good ' alignment is one where the new patterns can be encoded economically in terms of the old patterns in the alignment , as will be explained . * any one pattern may appear more than once in one alignment , not as two or more _ copies _ but as two or more _ appearances _ of a _ single _ pattern .
this has implications for the way alignments are formed and for the representation of recursive structures .
these aspects of the multiple alignment concept are not relevant to the main proposals here and will not be considered further in this paper .
readers who wish to know more may consult the sources cited earlier .
normally , the sp62 model is run with relatively few new patterns and a relatively large ` database ' or ` dictionary ' of old patterns .
the system typically forms a set of _ alternative _ alignments , each one of which represents a possible encoding of the new pattern or patterns in terms of one or more of the old patterns .
[ spelling_alignment ] shows a simple example , with one new pattern ( in row 0 , representing a badly - spelled version of the word ` experimentation ' ) and one old pattern ( in row 1 , representing the correctly - spelled version of the word ) . and [ parsing_alignment ] , and the old patterns are shown in the other rows , one pattern per row and in an order that is entirely arbitrary , without special significance .
as we shall see , alignments can sometimes fit better on the page if they are rotated by @xmath1 and in this case the new pattern or patterns are shown in column 0 with the old patterns in the other columns , one pattern per column .
] this is the best alignment produced by sp62 with the new pattern as shown and a dictionary of old patterns , each one of which represents one word in its correctly - spelled form .
e p e r i m a n t a t x p i u n 0 | | | | | | | | | | | | 1
< e3 e x p e r i m e n t a t i o n > 1 .... fig .
[ parsing_alignment ] is a slightly more complicated example , the best alignment produced by sp62 when it is supplied with one new pattern ( ` j o h n r
u n s ' ) and a set of old patterns , each one of which represents a grammatical rule . this alignment shows how the sentence ` j o h n r u n s ' ( in row 0 ) may be analysed ( ` parsed ' ) into its parts .
the old patterns in rows 1 to 3 represent grammatical rules : ` @xmath2 s @xmath2 n @xmath3 @xmath2 v @xmath3 @xmath3 ' in row 3 means that a ( simple ) sentence is composed of a noun ( ` n ' ) followed by a verb ( ` v ' ) , ` @xmath2
n 0 j o h n @xmath3 ' in row 2 means that ` j o h n ' is a noun , and ` @xmath2 v 1 r u n s @xmath3 ' in row 1 means that ` r u n s ' is a verb . .... 0 j o h n r u n s 0 | | | | | | | | 1 | | | | < v 1 r u n s > 1 | | | | | | | 2 < n 0 j o h n >
| | | 2 | | | | | | 3 < s < n > < v > > 3 .... as previously mentioned , a ` good ' multiple alignment is one where the new pattern or patterns can be encoded economically in terms of the old patterns in the alignment . how is this evaluation done ? every symbol has an associated ` weight ' which is the number of bits needed to encode that symbol . and
the weight is derived from the frequency of occurrence of the symbol using the shannon - fano - elias method ( see @xcite ) ( a method that is similar to the well - known huffman method ) .
the frequency value for any symbol is derived from the frequency value of the pattern ( or patterns ) in which that symbol appears ( as described in section [ representation_of_knowledge ] ) . in the context of medical diagnoses , the frequency associated with any given pattern is the frequency of occurrence of the disease that is represented by that pattern ( see section [ diseases_as_patterns ] , below ) .
a few of the symbols ( normally one or two ) near the beginning of each old pattern are classified as _ identification _ symbols or ` id - symbols ' .
for example , the id - symbol in the pattern ` @xmath2 e3 e x p e r i m e n t a t i o n @xmath3 ' in fig .
[ spelling_alignment ] is ` e3 ' , and the id - symbols in the pattern ` @xmath2 n 0 j o h n @xmath3 ' in fig .
[ parsing_alignment ] are ` n ' and ` 0 ' .
a ` code ' for any alignment may be derived quite simply by scanning the alignment from left to right looking for columns in the alignment that contain _ one _ code symbol , not matched with any other identical symbol .
the code for the alignment is the sequence of symbols that have been found , in the given order .
for example , the code derived in this way from the alignment in fig .
[ spelling_alignment ] is ` e3 ' and the code derived from the alignment in fig .
[ parsing_alignment ] is ` s 0 1 ' .
the next step in the evaluation of a given alignment is the calculation of a ` compression score ' or ` compression difference ' as : @xmath4 where @xmath5 is the total size ( in bits ) of those symbols within the new pattern that have been matched to old symbols within the alignment , and @xmath6 is the total size ( in bits ) of the symbols in the code that has been derived as just explained .
adjustments to this score are made to take account of gaps in the alignment like those that may be seen in fig .
[ spelling_alignment ] .
the details of these calculations and adjustments are explained in @xcite . in bioinformatics
, it is generally understood that the abstract ` space ' of alternative possible alignments between two or more sequences is , with few exceptions , astronomically large which means that it can not be searched exhaustively .
all practical methods for finding ` good ' alignments amongst two or more sequences use heuristic methods such as hill climbing ( or ` descent ' ) , ` beam search ' , ` genetic algorithms ' or the like that search selectively and exclude large parts of the search space . with methods like these
, one can find good approximate solutions in a reasonable time but one can never be sure of finding the best possible solution ( unless the sequences are very short and very few ) .
finding good multiple alignments in the sp system is no different . at the heart of the sp system for building multiple alignments
is an improved version of ` dynamic programming ' for finding full matches and good partial matches between two patterns ( see , for example , @xcite ) . unlike standard versions of dynamic programming
, the procedure used in the sp models : * can find good matches between patterns without restrictions on the lengths of the patterns . * can normally find several _ alternative _ alignments between two patterns , not just one . *
it allows the ` depth ' or thoroughness of searching to be varied according to need .
given one new pattern and a database of old patterns , sp62 first builds a set of alignments , each one of which is between the new pattern and _ one _ of the old patterns . from this set
, it selects the best few alignments , using the measure described in section [ evaluation_of_alignments ] .
each of these alignments can itself be treated as if it was a single pattern .
so , in the next stage , sp62 builds larger alignments , each one of which is between one of the selected alignments and one of the old patterns or between one of the selected alignments and another of those alignments . as before , the program selects the best of the alignments that have been formed .
the process is repeated in this way until no more alignments can be found .
the process for building alignments containing two or more new patterns is a generalisation of what has been described here . at its most abstract level ( section [ sp_theory ] ) ,
the sp model is conceived as a system that learns by transferring new information to its repository of old information and compressing it at the same time .
this abstract conception has now been realised more concretely in the form of the sp70 computer model @xcite that is capable of learning simple grammars from raw data .
however , further development of the model is needed to realise its full potential .
the current model has two stages : 1 . from partial alignments between patterns ,
the model creates new patterns that are added to the repository of old patterns as explained below .
2 . amongst the patterns that are generated in this way
, some are ` good ' in terms the principles of minimum length encoding and others are ` bad ' . in the second stage of processing , the model measures the frequency with which each pattern may be recognised in the raw data and then it uses this information in a hill - climbing search amongst subsets of the old patterns to find one or more sets of patterns that are good in terms of the principles of minimum length encoding . the remaining patterns may be discarded .
it is envisaged that , when the model is more fully developed , these two stages will be repeated so that the system can progressively bootstrap a set of patterns that are good in terms of the principles of minimum length encoding , and thus represent a distillation of the patterns of redundancy in the original data .
sp70 is currently targeted at the learning of syntax in natural languages .
given a partial alignment like this : .... 0 t h e g i r l r u n s 0
| | | | | | | 1 < % 1 t h e b o y r u n s > 1 .... the program creates patterns like these : .... < % 2 t h e > < % 3 r u n s >
< % 4 0 b o y > < % 4 1 g i r l > < % 5 < % 2 > < % 4 > < % 3 > > .... the first four are derived from coherent sequences of matched symbols and coherent sequences of unmatched symbols in the alignment and they correspond to what we would normally recognise as words .
each one has a grammatical category represented by id - symbols such as ` % 2 ' , ` % 3 ' and ` % 4 ' .
notice that ` b o y ' and ` g i r l ' belong to the same disjunctive category \{`b o y ' , ` g i r l ' } because they are alternatives at the same point in the original alignment and they both share the id - symbol ` % 4 ' . the id - symbols ` 0 ' and
` 1 ' serve to distinguish the two alternatives in that category . the last pattern in this example ties everything together by listing the sequence of categories in the original alignment .
it is an ` abstract ' pattern describing the overall structure of the two original sentences . for an application like medical diagnosis ,
the style of learning just described is probably not entirely appropriate .
section [ automatic_learning ] describes how similar principles may be applied to medical data .
if the potential of these ideas can be realised , the sp system should facilitate the automatic or semi - automatic construction of knowledge bases from raw medical data .
the time complexity of the sp62 model in a serial processing environment is approximately o@xmath7 , where @xmath8 is the size of the new pattern or patterns ( in bits ) and @xmath9 is the total size of the patterns in old ( in bits ) . in a parallel processing environment
, the time complexity may approach o@xmath10 , depending on how well the parallel processing is applied .
the space complexity in serial or parallel environments is o@xmath11 .
further details may be found in @xcite . in medical diagnosis
, it seems reasonable to suppose that there will normally be a fairly small maximum for the number of signs and symptoms ( abbreviated hereinafter as ` symptoms ' ) exhibited by any one patient .
correspondingly , there should be a maximum size for the size of the set of new patterns that are used to represent the patient s symptoms .
if we take this to be a constant value for @xmath8 , then in a serial processing environment the time complexity is approximately o@xmath11 and in a parallel processing environment it may approach o(1 ) .
to a large extent , medical diagnosis may be viewed as a problem of ( fuzzy ) pattern recognition : finding the best fit between a given set of symptoms for an individual patient and the symptoms associated with one or more diseases .
however , causal reasoning also has a part to play when , for example , it is understood that a given disease is caused by a bacterial or virus infection .
this section presents an example showing how the sp system may be applied to medical diagnosis , viewed as a process of pattern recognition .
the system may also support causal reasoning about medical problems and this is discussed briefly in section [ causal_reasoning ] , below .
other aspects of the proposals are discussed in other subsections . in the sp scheme ,
knowledge about diseases may be stored as patterns in a repository of old information , and the symptoms for an individual patient may be represented as a set of one or more new patterns .
a pattern in the store of old information may represent one disease and its associated symptoms , or a combination of diseases ( see section [ disease_combinations ] , below ) , or it may represent a cluster of symptoms that tend to occur together in two or more different diseases ( see section [ dereferencing_of_pointers ] , below ) . in addition , old may include patterns that play supporting rles ( see section [ uncertainties_in_diagnosis ] , below ) .
the frequency value of each pattern may be used to represent the absolute or relative frequency with which a given disease or cluster of symptoms is found in a given population .
these figures may be derived from epidemiological surveys or they may be estimated by medical experts . by way of illustration ,
[ patterns_1 ] shows five examples of such patterns , one describing the symptoms of chicken pox , another describing the symptoms of smallpox , the third describing the cluster of symptoms which is described as ` fever ' and two more describing the class of ` high ' temperatures ( 3839@xmath12c and 40+@xmath12c ) .
the number in brackets after each pattern is a very rough estimate of the relative frequencies of occurrence of the corresponding disease or condition . ....
< disease > chicken_pox : < dname >
chicken_pox < /dname > < r2 > fever < /r2
appetite > normal < /appetite > < chest > normal < /chest > < chills > no < /chills > < cough > no < /cough > < diarrhoea > no < /diarrhoea
> < fatigue > no < /fatigue
> < lymph_nodes > normal < /lymph_nodes > <
malaise > yes < /malaise
> < muscles > normal < /muscles > < nose > normal < /nose
> < skin > rash < /skin > < throat > normal < /throat > < weight_change > no < /weight_change
> < causative_agent >
chicken_pox_virus < /causative_agent > < treatment > chicken_pox_treatment < /treatment >
< /disease > ( 2500 ) < disease > smpx : < dname > smallpox < /dname > < r1 > flu_symptoms < /r1 > <
appetite > normal < /appetite > < chest > normal < /chest > < diarrhoea > no < /diarrhoea
> < fatigue > no < /fatigue
> < lymph_nodes > normal <
/lymph_nodes > < malaise > no <
/malaise > < skin > rash with blisters < /skin > < weight_change > no < /weight_change
> < causative_agent > smallpox_virus < /causative_agent > < treatment > smallpox_treatment < /treatment >
< /disease > ( 5 ) < disease > fever < breathing > rapid < /breathing > < face > flushed < /face > < temperature > < t1 > < /t1 > < /temperature > <
/disease > ( 15000 ) <
t1 > 38 - 39 < /t1 > ( 14705 ) < t1 > 40 + < /t1 > ( 147 ) .... each of the first three patterns begins and ends with a pair of symbols ` @xmath2disease@xmath3 ... @xmath2/disease@xmath3 ' which indicate that the pattern describes a disease or a cluster of disease symptoms . within each pattern , there are similar pairs of symbols , each one marking the beginning and end of a ` field ' which describes some aspect of the disease or cluster .
for example , ` @xmath2dname@xmath3 chicken_pox @xmath2/dname@xmath3 ' provides the name of the chicken pox disease , ` @xmath2skin@xmath3 rash @xmath2/skin@xmath3 ' describes one of its symptoms , ` @xmath2causative_agent@xmath3 chicken_pox_virus
@xmath2/causative_agent@xmath3 ' describes what causes the disease , and ` @xmath2treatment@xmath3 chicken_pox_treatment @xmath2/treatment@xmath3 ' is a remarkably unhelpful description of how to treat the disease which would , of course , be much more detailed in a fully developed knowledge base . within the pattern for chicken pox , the field ` @xmath2r2@xmath3 fever @xmath2/r2@xmath3 ' indicates that ` fever ' is one of the symptoms of the disease .
however , by contrast with other fields like those just mentioned , the symbol ` fever ' is , in effect , a reference or pointer to a cluster of symptoms such as rapid breathing , flushed face and high temperature described in the third pattern in the same figure . in a similar way , ` @xmath2r1@xmath3 flu_symptoms @xmath2/r1@xmath3 ' in the pattern for smallpox is a reference or pointer to another pattern , not shown in the figure , that describes a cluster of symptoms associated with influenza and flu - like diseases .
the way in which pointers like these are dereferenced in the sp system will be seen in the next section .
readers who are familiar with xml @xcite will notice that pairs of symbols like ` @xmath2disease@xmath3 ... @xmath2/disease@xmath3 ' or ` @xmath2dname@xmath3 ... @xmath2/dname@xmath3 ' are rather like the start and end tags used to mark the elements of an xml document .
however , by contrast with xml and related languages such as html , symbols of that kind have no formal status in the sp system and the styles of symbols are not defined within the system .
any convenient style may be used such as ` disease ... # disease ' or ` disease ... % disease ' and in some applications it is not necessary to provide any distinctive markers for the beginnings and ends of patterns or fields .
the concept of ` field ' has no formal status in the sp system either .
the process of diagnosis may be modelled by the building of one or more multiple alignments .
[ alignment_1 ] shows the best alignment created by sp62 with a set of new patterns shown in fig .
[ patterns_2 ] that describe ` john smith ' and his symptoms and a set of old patterns like those shown in fig .
[ patterns_1 ] that represent diseases or aspects of diseases . and [ parsing_alignment ] , the alignment in fig .
[ alignment_1 ] has been rotated by @xmath13 to allow the alignment to fit better on the page . as previously noted ,
the new patterns are shown in column 0 and the old patterns are shown in the other columns , one pattern per column in an order that is arbitrary and without special significance . ]
.... < patient > john_smith
< /patient > <
face > flushed < /face
> < appetite > poor < /appetite
> < breathing > rapid < /breathing > < muscles > aching < /muscles >
< chills > yes < /chills > < fatigue > yes < /fatigue
> < lymph_nodes > normal <
/lymph_nodes > < malaise > no < /malaise
> < nose > runny < /nose
> < temperature > 38 - 39 < /temperature
> < throat > sore <
/throat > .... .... 0 1 2 3 4 5 < disease > ---------- < disease > ---------- < disease > ---- < disease > flu : ------------------ : < patient > ------ < patient >
john_smith < /patient > -----
< /patient > < dname > ------------ < dname > influenza < /dname > ----------- < /dname >
< r1 > --------------- < r1 > flu_symptoms ------- flu_symptoms < /r1 > -------------- < /r1 > <
r2 > ------------------------------------ < r2 > fever -------- fever < /r2 > ----------------------------------- < /r2 > < appetite > ----- < appetite > --------- < appetite > poor normal < /appetite > ---- < /appetite > -------- < /appetite > < breathing > ---- < breathing > -------------------------------------------- < breathing > rapid ------------------------------------------------------------------- rapid < /breathing > --- < /breathing > ------------------------------------------- < /breathing > < chest > ------------ < chest > normal < /chest > ----------- < /chest >
< chills > ------- < chills > --------------------------------
< chills > yes ------------------------------------------------------ yes < /chills > ------ < /chills > ------------------------------- < /chills > < cough > --------------------------------- < cough > yes < /cough > -------------------------------- < /cough > < diarrhoea > --------
< diarrhoea > no < /diarrhoea > ------- < /diarrhoea > < face > --------- < face > ------------------------------------------------- < face > flushed ----------------------------------------------------------------- flushed < /face > -------- < /face > ------------------------------------------------ < /face > < fatigue > ------ < fatigue > ---------- < fatigue > yes no < /fatigue >
----- < /fatigue > --------- < /fatigue > < headache > ------------------------------ < headache > yes < /headache > ----------------------------- < /headache > < lymph_nodes > -- < lymph_nodes > ------ < lymph_nodes > normal ------------------------------ normal < /lymph_nodes > - < /lymph_nodes > ----- < /lymph_nodes > < malaise > ------
< malaise > ----------
< malaise > no ---------------------------------- no < /malaise >
----- < /malaise > --------- < /malaise
> < muscles > ------ < muscles > ------------------------------- < muscles > aching --------------------------------------------------- aching < /muscles > ----- < /muscles > ------------------------------ < /muscles > < nose > --------- < nose > ---------------------------------- < nose > runny ---------------------------------------------------- runny < /nose > -------- < /nose > --------------------------------- < /nose
> < skin > -------------
< skin > normal < /skin > ------------ < /skin > <
temperature > -- < temperature > ------------------------------------------ < temperature >
< t1 > ----------- < t1 > 38 - 39 ------------------------------------------------------------------------------------ 38 - 39 < /t1 > ---------- < /t1 >
< /temperature > - <
/temperature > ----------------------------------------- < /temperature > < throat > ------- < throat > -------------------------------- < throat > sore ----------------------------------------------------- sore < /throat > ------ < /throat > ------------------------------- < /throat >
< weight_change > ---- < weight_change >
no < /weight_change > --- < /weight_change > < causative_agent > -- < causative_agent > flu_virus < /causative_agent > - < /causative_agent >
< treatment > -------- < treatment > flu_treatment < /treatment > -------
< /treatment > < /disease >
--------- < /disease > --------- < /disease > --- < /disease > 0 1 2 3 4 5 .... an alignment like this may be interpreted as the result of a process of recognition . in this case , the symptoms that have been recognised are those of influenza , as shown in column 2 .
the following subsections discuss aspects of the alignment and of this interpretation . in an application like this ,
it is convenient but not essential to include amongst the old patterns a ` framework ' pattern like the one shown in column 1 .
this is a generalised pattern for diseases of all kinds that lists the main categories associated with diseases such as ` @xmath2dname@xmath3 @xmath2/dname@xmath3 ' ( the name of the disease ) , ` @xmath2breathing@xmath3 @xmath2/breathing@xmath3 ' ( the state of the patient s breathing ) and ` @xmath2temperature@xmath3 @xmath2/temperature@xmath3 ' ( the patient s temperature ) , but it does not specify specific values for any category .
this framework pattern serves as an anchor point for symbols in other patterns and facilitates the formation of multiple alignments in accordance with the rules described in @xcite and earlier publications . in an application like medical diagnosis , it is not obvious that there is any intrinsic order to the symptoms of a disease or associated descriptors such as the name of the patient . in describing a patient s symptoms
, it should make no difference whether ` high temperature ' is mentioned before ` runny nose ' or the other way round . in the sp framework , each pattern that describes a disease or a cluster of symptoms necessarily imposes an order in which categories of descriptors are specified .
however , users of the system may specify the patient s symptoms in any order that is convenient .
this is because symptoms are described using a set of new patterns and there is no intrinsic order amongst the new patterns supplied to the system . in our example , new patterns were supplied to the sp62 model in the order shown in fig .
[ patterns_2 ] but in the alignment shown in fig .
[ alignment_1 ] they appear in a completely different order .
notice that this freedom in the ordering of descriptors only applies to whole patterns .
when two or more symbols in one pattern are matched to two or more symbols in another , the order of the symbols in one pattern must be the same as the order of the matching symbols in the other pattern .
as already noted , a symbol like ` fever ' or ` flu_symptoms ' in one pattern may serve as a reference or pointer to another pattern that describes a cluster of symptoms that may be found in two or more different diseases . in fig .
[ alignment_1 ] , we can see how such pointers are ` de - referenced ' in the sp system .
the symbol ` flu_symptoms ' in column 2 is matched to the same symbol in column 3 where flu - like symptoms are listed .
likewise , the symbol ` fever ' in column 3 is matched to the same symbol in column 4 where the symptoms of fever or listed .
fever is itself part of the cluster of flu - like symptoms .
the provision of named clusters like these saves the need to specify the corresponding symptoms redundantly in each of the diseases where such clusters appear .
diagnosis is not an exact process : * most diseases are ` family resemblance ' or ` polythetic ' concepts because the majority of symptoms associated with any given disease are neither necessary nor sufficient for the diagnosis of the disease : they are ` characteristic ' of the disease in the sense that any one such symptom need not be present in every case and any of them may be associated with other diseases .
* there may be and frequently are errors in the observation or recording of symptoms .
sp62 can accommodate these kinds of uncertainty in diagnosis in two distinct ways : * because it looks for a global best match amongst patterns , it does not depend on the presence or absence of any particular symptom . notice how sp62 has succeeded in constructing the alignment shown in fig .
[ alignment_1 ] despite there being no match for ` poor ' in the new pattern ` @xmath2appetite@xmath3 poor @xmath2/appetite@xmath3 ' and ` yes ' in the new pattern ` @xmath2fatigue@xmath3 yes @xmath2/fatigue@xmath3 ' and no match for many of the symbols in the old patterns .
+ although the system does not depend on the presence or absence of any one symptom , particular symptoms can have a major impact on diagnosis , as described in section [ explaining_away ] , below . * within the sp framework , it is not necessary for every symptom of a disease to be recorded as a specific value .
for example , in column 4 of the alignment in fig .
[ alignment_1 ] , the pair of symbols ` @xmath2t1@xmath3 @xmath2/t1@xmath3 ' represents a set of alternative values for the temperature associated with fever . in this case
, there are just two values , represented in old by the patterns ` @xmath2t1@xmath3 38 - 39 @xmath2/t1@xmath3 ' ( high temperature ) and ` @xmath2t1@xmath3 40 + @xmath2/t1@xmath3 ' ( very high temperature ) , as shown at the bottom of fig .
[ patterns_1 ] .
the first of these patterns is shown in column 5 of the alignment , matched to the temperature of the patient shown in column 0 .
+ being able to specify symptoms as sets of alternative values allows the system to accommodate the kind of variability which is so prominent in many diseases . in medical diagnosis ,
it is quite usual for the physician to consider alternative hypotheses about what disease or diseases the patient may be suffering from .
the sp framework provides a model for this process in the way the system builds alternative alignments for any given pattern or set of patterns in new .
alignments and the corresponding diagnoses may be evaluated as follows .
as previously noted ( section [ evaluation_of_alignments ] ) , a ` compression difference ' is calculated for each alignment as shown in equation [ equation_1 ] .
the value @xmath6 that is used in that equation may be translated into an _ absolute _
probability for the given alignment :
@xmath14 for any one alignment ( the @xmath15th alignment ) in a set of alternative alignments , @xmath16 , that encode the same symbols from new , a _ relative _ probability may be calculated as : @xmath17 a fuller account of the way probabilities are calculated may be found in @xcite .
given that the new patterns represent the symptoms of one patient at a particular time and given that each pattern in old describes a single disease or a single cluster of symptoms that may form part of the description of one or more diseases , then each alignment formed by sp62 represents an hypothesis about any _
one _ disease that the patient may have . where alternative alignments encode different subsets of the symbols in new , it is possible that the patient may be suffering from two or more diseases at the same time .
this possibility is discussed in section [ disease_combinations ] , below . however , where two or more of the best alignments encode exactly the same symbols from new , then they represent _ alternative _ diagnostic hypotheses and they may be compared using values for relative probability ( @xmath18 ) . when sp62 formed the alignment shown in fig .
[ alignment_1 ] , it also formed a similar alignment , matching exactly the same symbols in new , in which column 2 contained a pattern representing the symptoms of smallpox , instead of the pattern for influenza .
the relative probability values calculated in this case were 0.99950 for influenza and 0.00049 for smallpox , reflecting the prevalence of those two diseases in the world today .
the symptoms of influenza and smallpox are quite similar , except for the very distinctive rash and blisters that occur in smallpox . the example shown in fig .
[ alignment_1 ] is silent about whether john smith had a rash and blisters or not .
if a rash and blisters had been seen to be absent , this would have been represented as ` @xmath2skin@xmath3 normal @xmath2/skin@xmath3 ' .
given this lack of information about the state of the patient s skin , he may have either influenza or smallpox but he is very much more likely to have the former than the latter , as indicated by the calculated probabilities .
if ` @xmath2skin@xmath3
rash_with_blisters @xmath2/skin@xmath3 ' is added to the symptoms recorded in new , and if sp62 is run again with the augmented set of symptoms , the best alignment found by the system is similar to that shown in fig .
[ alignment_1 ] but with the pattern for smallpox ( the second pattern in fig .
[ patterns_1 ] ) instead of the pattern for influenza in column 2 and with a match shown between ` @xmath2skin@xmath3 rash_with_blisters @xmath2/skin@xmath3 ' in the set of new patterns and the same symbols in the pattern that describes smallpox .
however , in this case _ there is no other alignment that matches the same symbols in new_. consequently , the relative probability of the best alignment is 1.0 . in short , the addition of one distinctive symptom to the list of symptoms has a dramatic effect on the relative probabilities calculated by the system . instead of a vanishingly small probability for smallpox ( 0.00049 ) ,
the system now assigns it a probability of 1.0 , in accordance with our intuitions . from this result
, we may conclude that the patient certainly has smallpox and that his aching muscles and runny nose are due to smallpox , not influenza .
this is the phenomenon of ` explaining away ' : `` if a implies b , c implies b , and b is true , then finding that c is true makes a _ less _ credible . in other words , finding a second explanation for an item of data
makes the first explanation less credible . ''
( @xcite , with the emphasis as in the original ) .
as noted above , it is possible for a patient to suffer from two or more diseases at the same time . given that the old patterns in the system describe single diseases , then the system would create two or more ` good ' alignments , each one corresponding to one of the diseases that the patient is suffering from .
if we want the system to calculate probabilities for combinations of diseases , then the repository of old patterns must contain patterns that represent combinations of that kind .
each such pattern may be constructed economically using references to the component diseases , in much the same way that clusters of symptoms may be referenced , as described in section [ dereferencing_of_pointers ] . as with single diseases , frequency values for a combinations of diseases
may be obtained from population surveys or by the judgement of medical experts . in the absence of any direct evidence of a statistical association between two or more diseases , it seems reasonable to assume that they are statistically independent . in such cases , frequency values
may be derived straightforwardly via normalised values for the frequencies of occurrence of individual diseases .
whether the frequency values for combinations of diseases are measured , estimated or derived , they can be used for the calculation of @xmath19 values and probabilities in exactly the same way as for single diseases .
of course , there are so many possible combinations of diseases that it would be impossible to store information about them all .
a more practical option may be to store information in old about individual diseases and combinations of diseases that are known to have a statistical association with each other .
one may assume that all other combinations of diseases are statistically independent . in a multiple alignment like the one shown in fig .
[ alignment_1 ] , any symbol within an old pattern that is _ not _ matched to a symbol in new represents an inference that may be drawn from the alignment . in this example
, we may infer from the alignment _
inter alia _ that the patient is likely to have a cough and a headache and that the standard treatment for influenza is required .
probabilities of these inferences can be calculated as described in @xcite .
if a ` good ' alignment makes a prediction about some marker that may be found in the patient s blood or something that may be observed in an x - ray , this may be interpreted as a suggestion to the physician that he or she should order an appropriate blood test or x - ray .
if tests of that kind or other kinds of investigation are instigated as a result of the inferences drawn from preliminary alignments , the results of those investigations , together with the original symptoms , may be fed back into the system as new information .
the system may then be run again and the alignments that are created may suggest a final diagnosis or the need for further investigation and so on .
apart from the kinds of inference just described , medical diagnosis often seems to involve a ` deeper ' kind of reasoning about the causes of symptoms and diseases , using knowledge of bacteria , viruses , anatomy , physiology and so on .
the sp framework supports a variety of styles of reasoning , including probabilistic ` deductive ' reasoning , abductive reasoning , nonmonotonic reasoning and ( as we saw in section [ explaining_away ] ) ` explaining away ' ( see @xcite ) .
so there are reasons to believe that , within the sp framework , it may be possible to extend the pattern recognition analysis described above to incorporate causal styles of reasoning .
recent investigation has confirmed this expectation .
the input - output relations of each subsystem within a larger system can be modelled in the sp framework as a set of patterns , and causal connections can be established by matching outputs to inputs . as with the analysis described above , a ` framework ' pattern
is also needed to ensure that alignments can be formed in an appropriate manner .
these potential applications of the system need further exploration and development .
one of the attractions of the sp system is that it allows concepts to be represented at multiple levels of abstraction ( e.g. , ` cat ' , ` mammal ' , ` vertebrate ' , ` animal ' ) in the manner of object - oriented design and , via the building of multiple alignments , it allows a specific entity ( such as `` my cat tibs '' ) to be recognised at several different levels of abstraction @xcite . to some extent
, this idea is already illustrated by the example shown in fig .
[ alignment_1 ] .
the concept of ` fever ' , represented by the pattern in column 4 of the figure , may be seen as a superclass comprising all the diseases where the patient may be feverish .
likewise , the pattern for flu symptoms ( column 3 in the figure ) may be seen as a superclass of the diseases in which such symptoms may be seen .
by contrast with the classification of animals and plants , the hierarchy of diseases tends to be relatively flat . however , there is scope for the recognition of classes and subclasses in the variants of diseases such as influenza and diabetes . with the sp system ,
each variant of a given disease may be recorded as a pattern that specifies the symptoms that are characteristic of the variant .
provided that pattern contains a symbolic link to another pattern describing the main symptoms of the disease , there is no need to repeat those symptoms redundantly in each of the variants . broadly speaking , the knowledge that is required in any artificial system for medical diagnosis
can be obtained ` manually ' from experts or written sources , or it may be obtained by the automatic or semi - automatic abstraction of knowledge from raw medical data , or some combination of the two . the sp system has potential to facilitate any or all of these processes .
it should be apparent from the example described above that the sp system provides a means of representing medical knowledge in a form that is simple and intuitive .
the simplicity of representing all knowledge as patterns is , perhaps , less important than the fact that this system allows computer - based knowledge to be expressed in a form that apparently reflects the natural structure of the original concepts .
this feature of the system should facilitate traditional kinds of knowledge elicitation from experts or written sources .
medical experts should have little difficulty in expressing their knowledge directly in the form of sp patterns . given that such experts are often busy and their time is , in any case , expensive , there are advantages if at least some of the process of building computer - based knowledge bases can be undertaken by knowledge engineers without specialised medical training .
it should be possible for such people to derive a good deal of the necessary knowledge from medical text books and other written sources .
section [ unsupervised_learning ] presented an outline description of how the sp70 model learns the kinds of structures found in the syntax of natural languages . as was indicated in that section ,
that style of learning is probably not entirely appropriate for medical data but the same general principles should apply .
this subsection describes in outline how the sp system may be applied to the learning of medical knowledge .
the simplest kind of ` learning ' is simply to keep a record of the symptoms of each specific patient and the corresponding diagnosis .
this is the principle of ` case - based learning ' and , in conjunction with a system that kind find good partial matches between the description of a new patient and stored knowledge of old patients , it can be quite practical and useful in making new diagnoses .
since the sp system is capable of finding good partial matches between patterns , it could well be used in this way , as suggested in section [ case_based_reasoning ] , below
. however , medical practitioners normally recognise a degree of abstraction in medical knowledge as described in section [ class_hierarchy ] : a concept like ` fever ' describes a cluster of symptoms that is found in two or more different diseases ; diseases like influenza often come in several variants or subclasses ; and symptoms may be described in terms of ranges rather than specific values as in our example of temperature ( section [ uncertainties_in_diagnosis ] ) . as an illustration of the way in which the sp system may create these kinds of abstraction , consider two imaginary patients with two different diseases and symptoms represented by the patterns ` x a b y c d e z f ' and ` a p q b c r d s e f t ' . with these two patterns ,
the sp system would create an alignment like this : .... 0 x a b y c d e z f 0 | | | | | | 1 a p q b c r d s e f t 1 , .... and from the matched symbols in this alignment it may derive the pattern ` a b c d e f ' . with the addition of some appropriate id - symbols ,
this pattern may serve like the pattern for ` fever ' in figs .
[ patterns_1 ] and [ alignment_1 ] : it represents a cluster of symptoms that appears in two or more different diseases . alternatively , this pattern may represent the symptoms of a general class of diseases with two variants represented by the two original patterns , modified so that the shared cluster of symptoms are replaced by a pointer to the general class .
with regard to categories like the set of two alternative temperatures represented by the patterns ` @xmath2t1@xmath3 38 - 39 @xmath2/t1@xmath3 ' and ` @xmath2t1@xmath3 40 + @xmath2/t1@xmath3 ' in fig .
[ patterns_1 ] , the sp system may derive this kind of disjunctive category in much the same manner as the disjunctive class \{`b o y ' , ` g i r l ' } in the example in section [ unsupervised_learning ] . the very simple format for representing knowledge described in section [ representation_of_knowledge ]
is intended to be as nearly ` universal ' as possible in the sense that it is designed to represent a wide range of different kinds of knowledge .
this should be much more nearly true when the concept of _ pattern _ has been generalised to two dimensions . in a similar way
, the concept of multiple alignment described in section [ processing_knowledge ] is intended to be a ` universal ' model for a wide range of different kinds of processing : pattern recognition , information retrieval , various kinds of reasoning , and so on . to the extent that these two objectives can be realised
, they should facilitate the seamless integration of different kinds of application , including medical diagnosis .
it should , for example , be relatively easy to apply a natural language interface to an sp system for medical diagnosis or to integrate the kind of visual pattern recognition needed in the diagnosis of different kinds of skin cancer with other kinds of medical expertise .
as mentioned in the introduction , a wide variety of philosophies and systems have been applied to the problem of medical diagnosis . in this section ,
i briefly review some of the more prominent of these approaches and compare them with the sp approach , as described in this paper .
rule - based systems ( like the well - known mycin system @xcite ) contain if - then rules where the ` if ' side of any rule is a collection of one or more conditions for the rule to fire connected by logical operators such as ` and ' , ` or ' ( which may be inclusive or exclusive ) and ` not ' .
by contrast , the sp system expresses all knowledge in the form of patterns . at first sight ,
sp patterns lack the expressive power of if - then rules .
but the effect of such rules can be modelled within the sp system if that is required @xcite . and
if medical diagnosis is viewed as a process of pattern recognition ( as in this paper ) , then sp patterns and the sp framework are , arguably , a more natural and flexible medium for the representation and processing of knowledge than are if - then rules .
to illustrate this last point , we can express the distinctive features of influenza by a rule such as : .... if chills and cough and headache and aching muscles and runny nose and sore throat then influenza ( probability = 0.9 ) .... although there may be a probability associated with the rule ( as shown ) , the rule has an intrinsic logic which , if strictly applied , means that the rule will _ only _ fire if all the conditions are satisfied .
by contrast , a pattern like the one shown in column 3 of fig .
[ alignment_1 ] may appear in the best alignment when any reasonably large subset of its symbols have been matched .
if one attempted to achieve this kind of flexibility with an if - then rule using combinations of and , or and not , the rule would become very complex .
alternatively , one might split up the rule into a number of smaller rules , one for each symptom or combination of two or three symptoms but again the result would be relatively complex . in systems like mycin and some of its successors , the ` probabilities ' that the system calculates are really measures of confidence without the theoretical underpinnings of probability theory . in other systems , `` ... formal approaches based on probability theory are precise but can be awkward and non - intuitive to use . ''
by contrast , the sp framework allows true probabilities to be calculated quite simply ( see section [ calculation_of_probabilities ] and @xcite ) and strictly in accordance with established theory ( as described in sources such as @xcite ) .
one of the attractions of artificial neural networks for the support of medical diagnosis is that they can be trained with appropriate data , thus by - passing the need for the manual compilation of knowledge by medical experts or knowledge engineers .
however , `` a major drawback is that ` knowledge ' embedded [ in the neural network ] is cryptically coded as a large number of weights and activation values . as a consequence ,
the lack of neural network validation tools is often one of the reasons limiting their use in practice , especially in the context of medical diagnosis where physicians can not trust a system without explanation of its decisions . ''
@xcite . while there may be scope for extracting rules from a trained neural network ( _ ibid .
_ ) , this adds complexity and uncertainty to the technology and defeats the other main attraction of a neural network : as a classifier of specific cases in terms of the learned knowledge . as a system for unsupervised learning of knowledge structures from raw data ,
the sp system is not yet a rival to existing neural network systems .
however , the system has clear potential for unsupervised learning and , if that potential can be realised , the system has the advantage that its knowledge is stored in a form that can be read and understood by people . meanwhile , if it is supplied with knowledge about diseases derived from experts or text books , it can be used for diagnostic classification of individual patients .
given the variability of diseases and other uncertainties associated with medical diagnosis ( section [ uncertainties_in_diagnosis ] ) , the field of fuzzy logic ( and fuzzy set theory ) has the obvious attraction that it has been designed with the explicit intention of providing a model for ` fuzzy ' concepts and ` fuzzy ' operations on them ( see , for example , @xcite ) .
in purely theoretical terms , the field of fuzzy logic may be criticised because it introduces a fairly elaborate conceptual framework to accommodate the undoubtedly fuzzy nature of many human concepts but this conceptual framework is poorly integrated with other ideas about the nature of human cognition .
by contrast , the sp theory grew out of research in psychology and it provides a unified model for several aspects of human perception and cognition @xcite .
considerations of that kind may be discounted as not relevant to the practicalities of medical diagnosis .
but in that connection fuzzy logic has the drawback that it introduces another layer of complexity to the already difficult process of eliciting knowledge from medical experts @xcite . there seems to be some scope for ameliorating this problem by the provision of appropriate tools ( _ ibid . _ ) but the basic problem remains .
by contrast , the sp system allows concepts to be expressed in a simple , intuitive manner and , at the same time , it accommodates much , perhaps all , of the fuzziness of medical diagnosis .
two of the main differences between bayesian networks ( see , for example , @xcite ) and the sp system are : * bayesian networks focus on the binary relationship between any given node in the network and each of its parent nodes ( if any ) . in this respect
, they inherit some of the thinking behind if - then rules .
by contrast , the sp system is oriented towards the representation and processing of associations ( expressed as patterns ) that may contain arbitrarily many elements .
* correspondingly , any given bayesian network stores its statistical knowledge in the form of tables of conditional probabilities , one for each node in the network .
by contrast , the sp system stores its statistical knowledge in the form of integers , one for each pattern , representing the absolute or relative frequency of that pattern in some domain . these and related differences seem to underlie some of the apparent drawbacks of bayesian networks compared with the sp framework : * the directional nature of bayesian networks does not sit easily with the non - directional nature of medical syndromes . *
the process of calculating probabilities of inferences in a bayesian network is substantially more complicated than the calculation of probabilities for alignments and inferences in the sp framework . *
the tables of conditional probabilities required for bayesian networks are significantly more complex than simple measures of frequency that are used in the sp system . notwithstanding
the development of special methods for eliciting conditional probabilities from experts @xcite , the process of building up the necessary tables of conditional probabilities is likely to be much harder than measuring or estimating an integer value for each disease , reflecting its absolute or relative frequency in a given domain .
a major attraction of case - based reasoning in medical diagnosis ( see , for example , @xcite ) is that , compared with many of the alternatives , it can considerably simplify the process of acquiring the necessary knowledge . in its simplest form ,
a case - based system merely requires a description of one or more specific examples of each disease and a search algorithm that can find exact matches or good partial matches between the symptoms of a given patient and one or more of the stored records . in some respects , the sp system is like a case - based system and it could indeed be used like a case - based system . to use it in this way ,
each of the old patterns should represent a specific case ( including its diagnosis ) and the new pattern or patterns should represent the symptoms of a patient for whom a diagnosis is required .
the capabilities of the system for finding exact matches and good partial matches between patterns will allow it to retrieve patterns for previously - diagnosed cases that are similar to any given current case .
the main advantages of the sp system compared with the case - based approach to diagnosis are : * it facilitates the description of diseases in generalised terms without the need to specify exact values for every category of symptom . in our main example , we saw how the temperature of a patient with a disease like influenza may be specified as a range of alternative values ( section [ uncertainties_in_diagnosis ] ) .
any other category of symptom may be treated in the same way .
* it allows one to specify clusters of symptoms that are found in two or more different diseases and it allows one to describe diseases at two or more levels of abstraction ( section [ class_hierarchy ] ) .
both of these things facilitate the description of diseases without the need to repeat information unnecessarily where similar patterns are found in different diseases or varieties of disease .
as we have seen , the sp system accommodates the main elements of medical diagnosis , viewed as a problem of pattern recognition , and there are reasons to believe that it may also provide support for causal reasoning in medical diagnosis .
however , the sp62 model is only a prototype that serves for research and demonstration .
it is not yet a system with ` industrial strength ' .
the main developments that are needed to reach that goal are : * the provision of a well - designed graphical user interface .
* there is probably scope for improvements in the search methods that are used within the system .
* there is scope for the application of parallel processing both to improve the scaling properties of the system ( section [ computational_complexity ] ) and to increase absolute speeds of processing . *
naturally , the system needs to be provided with appropriate knowledge . for each area of application ,
a set of patterns needs to be developed that describes the diseases and symptom clusters in that domain .
* at some stage after the development of a realistic knowledge base , the performance of the system must be validated against the judgement of human medical experts .
the potential payoff from these developments is a system that allows knowledge about diseases to be expressed in a simple , intuitive manner , that can cope with errors and uncertainties in knowledge about diseases and knowledge about individual patients , that simplifies the acquisition and storage of statistical information , that calculates true probabilities of diagnoses , that smooths the path to the automatic or semi - automatic abstraction of medical knowledge in the future , and should facilitate the integration of medical diagnosis with other kinds of application .
j. g. wolff , learning syntax and meanings through optimization and distributional analysis , in : y. levy , i. m. schlesinger and m. d. s. braine , eds . , categories and processes in language acquisition ( lawrence erlbaum , hillsdale , nj , 1988 ) , pp .
copy : www.cognitionresearch.org.uk/lang_learn.html#wolff_1988 .
j. g. wolff , information compression by multiple alignment , unification and search as a unifying principle in computing and cognition , artificial intelligence review 19 , no . 3 ( 2003 ) 193230 .
copy : http://arxiv.org/abs/cs.ai/0307025 .
j. g. wolff , information compression by multiple alignment , unification and search as a framework for human - like reasoning , logic journal of the igpl 9 , no
. 1 ( 2001 ) 205222 , first published in the _ proceedings of the international conference on formal and applied practical reasoning ( fapr 2000 ) _
, september 2000 , issn 14694166 .
copy : www.cognitionresearch.org.uk/papers/pr/pr.htm .
j. g. wolff , ` computing ' as information compression by multiple alignment , unification and search , journal of universal computer science 5 , no .
11 ( 1999 ) 777815 .
copy : http://arxiv.org/abs/cs.ai/0307013 .
j. g. wolff , probabilistic reasoning as information compression by multiple alignment , unification and search : an introduction and overview , journal of universal computer science 5 , no .
7 ( 1999 ) 418462 .
copy : http://arxiv.org/abs/cs.ai/0307010 .
j. g. wolff , syntax , parsing and production of natural language in a framework of information compression by multiple alignment , unification and search , journal of universal computer science 6 , no .
8 ( 2000 ) 781829 .
copy : http://arxiv.org/abs/cs.ai/0307014 .
j. g. wolff , mathematics and logic as information compression by multiple alignment , unification and search , technical report , cognitionresearch.org.uk ( 2002 ) .
copy : http://arxiv.org/abs/math.gm/0308153 .
j. g. wolff , unsupervised grammar induction in a framework of information compression by multiple alignment , unification and search , in : c. de la higuera , p. adriaans , m. van zaanen and j. oncina , eds . ,
proceedings of the workshop and tutorial on learning context - free grammars , 2003 , pp .
this workshop was held in association with the 14th european conference on machine learning and the 7th european conference on principles and practice of knowledge discovery in databases ( ecml / pkdd 2003 ) , september 2003 , cavtat - dubrovnik , croata .
copy : http://arxiv.org/abs/cs.ai/0311045 .
j. g. wolff , unsupervised learning in a framework of information compression by multiple alignment , unification and search , technical report , cognitionresearch.org.uk ( 2002 ) .
copy : http://arxiv.org/abs/cs.ai/0302015 .
t. bray , j. paoli , c. m. sperberg - mcqueen , e. maler , extensible markup language ( xml ) 1.0 ( second edition ) , technical report , world wide web consortium , w3c recommendation , 6 october 2000 .
copy : www.w3.org/tr/rec-xml ( 2000 ) .
j. fox , d. glasspool and j. bury , quantitative and qualitative approaches to reasoning under uncertainty in medical decision making , in : s. quaglini , p. barahona and s. andreassen , eds . ,
proceedings of the 8th conference on artificial intelligence in medicine in europe , aime 2001 , vol .
2101 of lecture notes in computer science ( springer , heidelberg , 2001 ) pp . 272282 .
k. boegl , k .-
adlassnig , y. hayashi , t. e. rothenfluh and h. leitich , knowledge acquisition in the fuzzy knowledge representation framework of a medical consultation system , artificial intelligence in medicine 30 ( 2004 ) 126 .
l. c. van der gaag , s. renooij , c. witteman , b. aleman and b. taal , probabilities for a probabilistic network : a case study in oesophageal cancer , artificial intelligence in medicine 25 ( 2002 ) 123148 .
althoff , r. bergmann , s. wess , m. manago , e. auriol , o. i. larichev , a. bolotov , y. i. zhuravlev and s. i. gurov , case - based reasoning for medical decision support tasks : the inreca approach , artificial intelligence in medicine 12 ( 1998 ) 2541 .
gerry wolff is director of cognitionresearch.org.uk .
previously , he has held academic posts in the school of informatics , university of wales , bangor , and the department of psychology , university of dundee , a research fellowship with ibm in winchester , uk , and he has worked for four years as a software engineer with praxis systems in bath ,
his first degree is in natural sciences ( psychology ) from cambridge university and his phd is in cognitive science from the university of wales .
he is a chartered engineer and member of the british computer society . since 1987
his research has focussed on the development of the sp theory .
previously his main research interests were in developing computer models of language learning .
he has numerous publications in a wide range of journals , collected papers and conference proceedings . | this paper describes a novel approach to medical diagnosis based on the sp theory of computing and cognition .
the main attractions of this approach are : a format for representing diseases that is simple and intuitive ; an ability to cope with errors and uncertainties in diagnostic information ; the simplicity of storing statistical information as frequencies of occurrence of diseases ; a method for evaluating alternative diagnostic hypotheses that yields true probabilities ; and a framework that should facilitate unsupervised learning of medical knowledge and the integration of medical diagnosis with other ai applications .
_ keywords : _ medical diagnosis , information compression , multiple alignment , sp theory , pattern recognition , causal reasoning . |
magnetic reconnection is thought to be a process , being responsible for many eruptive plasma phenomena in space plasmas and astrophysical plasmas , like geomagnetical substorms or eruptive flares .
although magnetic reconnection in two dimensions ( 2d ) is fairly well understood , e.g. , see the comments in , it would be interesting to have more detailed informations about the topological and geometrical structure of flow and field lines in the vicinity of the singular points of plasma flow and magnetic field .
classical reconnection scenarios , see or sweet - parker model ( see , e.g. , ) propose a magnetic null point and a stagnation point flow into the diffusion region , i.e. the stagnation point is inside this diffusion region . where the first to analyse the case of incompressible 2d mhd with constant resistivity .
it turned out that either the magnetic field must be of higher order in the spatial variables @xmath0 and @xmath1 or that there is no stagnation point flow , but a shear flow .
therefore their result is , that the classical hyperbolic stagnation point flow needs higher order terms , concerning the spatial variables ( @xmath0 and @xmath1 ) .
solutions locally containing only higher order terms do not allow for topologically / structurally stable magnetic fields .
therefore either no reconnection can take place , as in the case of higher order terms no hyperbolic magnetic field can exist , or the stagnation point flow is not of hyperbolic type .
annihilation solutions have been studied , where chose a special ansatz for the solution of the resistive mhd to get reconnection solutions .
they emphasize the first order momentum equation and neglect the energy transport ( equation ) or rather the entropy conservation ( equation ) , starting with a nonlinear perturbation of magnetic annihilation solutions .
this lead them to so called reconnective annihilation solutions , where only one of the two separatrix - lines are crossed , and the other is only tangent to the converging streamlines .
the current sheet has a one - dimensional structure ( straight line ) .
the results found by @xcite , and confirm the results found earlier by , who found more shear - like flows instead of typical stagnation point flows . later on and extended the method to curvelinear current sheets .
it was shown by and later on in extended form by that under certain circumstances ( like constant resistivity or current depending / anomalous resistivity and sub - alfvnic flow etc . )
reconnection is impossible , the so called anti - reconnection theorems . in 3d for constant resistivity
a careful analysis of topologically different solutions has been discussed in . in 2d
such an analysis is missing and should be done here , but without taking the restriction to constant resistivity into account .
in contrast to the aforementioned models we do not search for special solutions , but the most _ general _ solution without constant or special non - constant resistivity or a specific non - idealness . thus our analysis covers all forms of non - ideal terms . in the sense of topological fluid dynamics
we want to get insight how the field and streamlines are rooted in the null point(s ) of flow and field , i.e. which geometrical shapes of field and streamlines are possible and which not .
we define and investigate the influence of an effective non - constant resistivity and use hereby the full energy equation of resistive mhd instead of using only the assumption of incompressibility .
the problems concerning ( exact ) analytical models very often are : * the physical quantity resistivity is often only a constant smallness parameters * if really gradients of the resistivity @xmath2 and maybe also the amplitude / absolute value are recognized as important for the non - ideal process , what are the shapes of such resistivities enabling magnetic reconnection ? * it is not clear which _ flow _ topologies are allowed to generate reconnection , reconnective annihilation solutions , or other solutions for general non - ideal or resistive terms thus our aims are : * to show not every non - ideal or resistive process in the vicinity of a null point is a magnetic reconnection process * it is not enough to have a localized resistivity to get a reconnective solution , finding parameters that mark the boundary between reconnection and non - reconnection solutions * why should there be no reconnection process close to the magnetic null point and what happens if we do not restrict ourselves to complete incompressible dynamics without energy transport ? * finding analytical _ and _ exact solutions close to the null point * detailed investigation of topological skeleton of flow and magnetic field in the frame of resistive mhd ; it is an analysis of the resistive system close to singular points of flow and magnetic field * performing linearization , necessary to get information about the skeleton of magnetic reconnection we will concentrate on the topology and geometrical properties of flow and field and discuss different cases with respect to their implications for magnetic reconnection . to summarize : the main goal is to get the topological and geometrical skeleton of the resistive mhd equations , i.e. , if the jacobian of the magnetic field at the magnetic null point has only real eigenvalues ( hyperbolic null point ) , which eigenvalues can be found for the jacobian of the plasma velocity at the stagnation point ? in our analysis the null point of the velocity field ( = stagnation point ) , should have an almost identical position and a vanishingly small offset to the magnetic null point .
the resistvity @xmath2 should be a positive quadratic form in @xmath0 and @xmath1 close to the magnetic null point , indicating that the resistivity indicates real dissipation in the form of ohmic heating and annihilation of magnetic flux , see also the explanations and calculations in the section [ sws ] .
how does the geometrical structure of the velocity field look like , and which kind of eigenvalues are allowed by the resistive or in general non - ideal mhd equations ?
the topological classification of 2d vector fields in the vicinity of their null points is described in the textbooks ( with connection about phase portraits of dynamical systems , i.e. systems of ordinary differential equations ) e.g. of @xmath3@xmath4 or @xmath5 .
the topological structure of magnetic fields in the vicinity of null points is described , e.g. , by@xmath5 and concerning the construction of ideal mhd flows , e.g. in@xmath5 . our interest is to ask which topology and geometry of the _ macroscopic _ flow correspond to which magnetic topology and geometry in the frame of mhd .
in contrast to the analyses mentioned in the previous paragraph we here have to investigate the topological and geometrical properties of _ both _ vector fields , i.e. that of the plasma flow and that of the magnetic field . to perform this analysis we solve the resistive mhd system with a linear ansatz for both vector fields , i.e. , we perform a taylor expansion of both vector fields in the vicinity of their null points , neglecting derivatives higher than the first ones .
that implies that we allow the stagnation point (= null point of the velocity field ) of the velocity field @xmath6 to have a small offset concerning the magnetic null point , but neglecting here , the influence of the second derivatives of the velocity field
. an influence would be noticeable for significant offsets , but is negligible for very small offsets , justifying our assumptions .
we will justify this assumptions afterwards at the end of section [ clf ] , estimating the influence of an offset caused by the gravitational force in the case of completely linear velocity and magnetic fields , where the second derivatives of the vector fields vanish automatically .
now let @xmath7 be the offset in @xmath0-direction and @xmath8 the offset in @xmath1-direction .
then we can express the class of linear velocity fields with the help of their first derivatives , i.e. jacobians : @xmath9 in analogy to the magnetic field
@xmath10 @xmath11 according to@xmath5 , every magnetic field of the form of eq.([2b ] ) can be rotated such , that it is represented by a magnetic flux function @xmath12 , so that a standard null point of the magnetic field appears with a constant current density @xmath13 const around the origin , where the magnetic field is given by @xmath14 .
the corresponding scenario of a flare loop , with its symmetry axis being almost perpendicular to the photosphere , is given in fig.[bild0 ] .
the current density is @xmath15 such that @xmath16 the different topologies of 2d vector fields are represented by two independent parameters , the threshold current @xmath17 , and the eigenvalues of @xmath18 , the @xmath19 s , or @xmath20 , the current in @xmath21direction @xmath22 implying a bijective relation between the variables @xmath23 and @xmath24 and the threshold @xmath17 and the actual current @xmath20 , to be precise @xmath25 we will use the four variables @xmath26 if it is convenient , i.e. combinations of the four to simplify the corresponding terms in the equations .
the eigenvalue @xmath19 determines the topological structure of the field and the geometrical shape of the field lines . for a divergence free case
there are only three main types of such fields : * the case that the eigenvalue is zero ( @xmath27 ) corresponds to the one - dimensional current sheet ( degenerated case ) .
* the case that @xmath28 corresponds to field lines being topological circles ( geometrical ellipses ) , if @xmath29 ( @xmath30 ) , then geometrical circles . all the cases mentioned in the last sentence
are so called o - points .
* the case of the so called x - points , where @xmath31 .
only for the case in the last item a magnetic separatrices exist .
such separating field lines must exist to enable magnetic reconnection .
a necessary condition for magnetic reconnection in 2d is that the plasma flow crosses magnetic separatrices , see , e.g.,@xmath5,@xmath5,@xmath5 , @xmath5 , or@xmath5 .
the current free case is given by @xmath32 , i.e. @xmath33 and should be excluded , as dissipation in such a case would have nothing to do with the electric current .
the following analysis is restricted to _
pure _ resistive dynamics , i.e. ohmic heating / dissipation without any other loss terms like viscosity or heat conduction .
we choose the coordinate system in such a way , that the gravity is directed in negative @xmath1-direction ( the unit vector in @xmath1-direction is @xmath34 ) .
the basic resistive stationary mhd equations in 2d are given by ( following , e.g. , ) @xmath35 due to stationary maxwell equations the electric field in the 2d case has to be constant , i.e. @xmath36const see , e.g. , .
the same shape of the equations occur , if we introduce a generalized non - ideal term on the right - hand side of ohm s law . as the current density is constant close to the x - point ( see eq.[1b ] )
, the shape of an effective resistivity is that of the non - idealness @xmath37 ( where only the @xmath21-component is non - zero ) , writing @xmath38 the implication is that @xmath39 , as @xmath40 is constant close to the null point .
the non - idealness resulting from two - fluid approach can be written for the one fluid approach , e.g. , following@xmath41 , as @xmath42 where @xmath43 is the particle density , @xmath44 the absolute value of the electron charge , @xmath45 the hall term , @xmath46 the anisotropic electron pressure tensor , @xmath47 the electron acceleration term ( in our case averaged over a fitting characteristic time scale , as our approach is quasi - stationary ) , and the tensor divergence @xmath48 describes the influence of electron inertia .
but even this two - fluid model does not include all terms which may be ( additionally ) responsible for anomalous resistivity or non - idealness , like , e.g. , electron scattering by wave turbulence generated due to the buneman instability ( often considered , e.g. , and ) , ionization - recombination terms , non - linear electron - ion momentum transfer terms , electron and ion viscosities , radiative losses / gains , and so on .
as we do not specify the non - idealness , having introduced the effective resistivity , our analysis can be used for all physical models with different types of non - ideal terms . for a more specific view but for non - local models of reconnection with a generalized ohm s law , see , e.g.,@xmath5 . the term @xmath49 is therefore only a _
necessary _ trigger enabling reconnection , not a sufficient one , see , e.g. , @xmath5 .
we will calculate the ( almost ) complete solution space of the resistive ( non - ideal ) mhd close to the null points of flow field and magnetic field .
the aim is to find the general correlation between the jacobians of the plasma velocity and the magnetic field , and the shape of the unspecified non - idealness . at first both vector fields
are treated as completely linear , i.e. unbounded fields in an unbounded domain , see section [ clf ] .
this does not exclude the possibility that in a bounded domain the fields at the boundary are maybe not analytic , but the extrapolation of the field may produce reasonable finite and regular solutions .
in section [ coli ] we concentrate on the linearized fields and take only first order terms of the spatial variables into account . a similar method to determine the structure of the non - ideal term and
the flow and the magnetic field has been proposed and done , but for the case @xmath50 and for global fields , in the frame of a toy model in@xmath5 . with this linear or linearized fields we can draw conclusions with respect to the other mhd quantities , like pressure , density and resistivity :
the lowest order of the magnetic field is linear , the current density is constant , and therefore the lorentz force @xmath45 is linear in @xmath0 and @xmath1 .
this is the reason that all other terms should be also at ( the ) most of first order in the spatial variables .
this leads to the conclusion that the plasma pressure @xmath51 is at highest order quadratic in @xmath0 and @xmath1 , to allow for a linear pressure force , i.e. we can express @xmath51 by @xmath52 where @xmath53 to @xmath54 are constant coefficients . as the velocity close to the stagnation point can be represented as a linear term in @xmath0 and @xmath1 , the @xmath55 term is also linear in @xmath0 and @xmath1 , or constant .
the mass continuity equation , eq.([4 ] ) implies therefore that the mass density , @xmath56 , has to be constant in lowest order close to the stagnation point . of course @xmath56 could be a simple linear function close to the null point but
that would imply that the mass density has to vanish on a surface , or rather straight line in 2d , depending on the gradient of the density .
this null - line _ could _ be somewhere inside the linear region . in one direction the density
could then become negative .
we assume that the density @xmath56 is constant in the vicinity of the null points of magnetic and flow field to prevent @xmath56 to have a negative or zero value within the domain of interest .
in addition it is a reasonable assumption that close to a stagnation point there is a stagnation region , i.e. the density has a maximum or minimum close to the stagnation point . in the case of completely linear fields
it is now evident that the mass continuity equation requires a divergence free velocity field , if we assume all terms to have the same or comparable orders in @xmath0 and @xmath1 .
one has also to take into consideration that at the magnetic null point @xmath57 equals @xmath58 and if we assume that the resistivity is constant , then @xmath59 everywhere in the vicinity of the neutral point .
the same holds for special cases of non - constant resistivity , e.g. , if @xmath2 is a function of the current density @xmath20 only , one can conclude that in the case of a linear field or in the vicinity of the null point the current density is constant and therefore also the function @xmath60const , as @xmath61const .
the electric field @xmath57 is constant also outside the linear region , but the resistivity and current density are , of course , not constant there .
this implies that the term @xmath62 equals to zero within the linear region and therefore the flow is field - aligned everywhere _ within this region _ in lowest order of the spatial variables . to get not such trivial solutions , which are non - reconnective in lowest order
, we have to regard the resistive ohm s law as a _ definition equation _ for the spatially dependent resistivity , i.e. @xmath63 .
therefore , to enable reconnection in the vicinity of the null point , the resistivity can not be a function of the current density only .
we designate and regard this coefficient as an effective resistivity or short resistivity , even if this coefficient originates from collisional theory . as the current density is constant inside within the region of the linear field approximation , the ( effective ) resistivity
is a substitute expression for a general non - ideal term , determining , but vice versa also determined by , the flow and magnetic field line structure !
this relation between the velocity or flow field @xmath6 , the magnetic field @xmath10 and the resistivity @xmath2 will be analysed in this article .
one can also recognize that eq.([7 ] ) ( together with the incompressibility ) at the stagnation point would require a vanishing resistivity in the absence of additional dissipation terms . written with all coefficients and comparing with all orders of @xmath0 and @xmath1 we get the following system of equations , first from the euler or momentum equation @xmath64 and the equations from the energy equation , first neglecting source or heating terms , only including ohmic heating , @xmath65 where @xmath66 .
the equations above are ordered after their physical meaning , the first five equations eqs.([9 ] ) - ( [ 13 ] ) correspond to the first order momentum equations , while eqs.([14 ] ) - ( [ 19 ] ) represent the terms of the energy equation .
the eqs.([12 ] ) , ( [ 13 ] ) and ( [ 14 ] ) are of zeroth order in @xmath0 and @xmath1 .
the eqs.([9 ] ) , ( [ 10 ] ) , ( [ 15 ] ) and ( [ 16 ] ) are of first order in @xmath0 and @xmath1 , the eq.([17 ] ) is of the order @xmath67 and eq.([18 ] ) of second order in @xmath0 , respectively eqs.([19 ] ) is of second order in @xmath1 .
in the first subsection we concentrate on linear fields , i.e. fields that are unbounded in an unbounded domain . in the second subsection
we take only zero and first order terms of @xmath0 and @xmath1 into account , i.e. we regard a taylor expansion of maximum order one . if @xmath68 then from eqs.([18 ] ) and ( [ 19 ] ) it follows @xmath69 and with two terms of the momentum equation , namely eq.([9 ] ) and eq.([10 ] ) , excluding the current free case @xmath32 , i.e. @xmath33 and the case with o - points , i.e. @xmath30 , we infer that @xmath70 thus @xmath71 should be valid , leading to @xmath72 . as @xmath72 is not representing a realistic equation of state
, @xmath73 must be zero , _ taking terms of every order into acount_. this is unaffected by the neglecting other source terms , not mentioned here , as the interesting heat conduction term is of the order zero in @xmath0 and @xmath1 , and therefore affecting the eq.([14 ] ) only .
now we use the assumption that @xmath74 .
then we rewrite the mhd equations : @xmath75 with eqs.([6 ] ) and ( [ 7 ] ) it can clearly be recognized that at the stagnation point the resistivity must be zero . as the electric field @xmath57 is constant and @xmath59 at the magnetic null point and the stagnation point ,
we conclude that the electric field is zero ( everywhere ) .
the effect of a vanishing compressibility is in resistive mhd also connected with a vanishing electric field .
in addition , as the @xmath0-offset @xmath7 in the fourth equation ( a zeroth order term of the force ) of the system ( [ 24 ] ) couples an additional pressure gradient in @xmath0-direction to this offset only , this offset can be set to zero without loss of generality .
therefore , from @xmath76 and @xmath77 we infer @xmath78 , see the fourth and fifth equation of the system ( [ 24 ] ) ( part of the energy / entropy equation ) .
if we would further allow @xmath79 in the case of non - vanishing gravity , then there would exist only static solutions either ( as in this case @xmath80 must be zero it implies that @xmath81 ) , or the other possible solution branch would need an adiabatic exponent @xmath72 . for the non - degenerated case ( @xmath82 ) the general solution for the system ( [ 24 ] )
has two branches : @xmath83 where the discriminant @xmath84 must fulfill @xmath85 to guarantee that @xmath80 is real .
this leads to @xmath86 ^ 2 - 4 b^2 \gamma^2 j_{z}^2 \geq 0 \ , , \label{28}\end{aligned}\ ] ] where @xmath87 is the discriminant of the quadratic equation eq.([27 ] ) .
the inequality eq.([28 ] ) guarantees a real value of @xmath88 .
that leads together with the convention @xmath89 @xmath90 this implies that either the actual current density @xmath91 , then the field lines are hyperbolas or topological circles , or @xmath92 , then only hyperbolic field lines or ( anti-)parallel are possible .
let us now estimate the influence of the offset in @xmath1-direction on the solution , e.g. , for the situation in the corona .
we can express the offset in @xmath1-direction by using the equation for the offset @xmath8 in eq.([25 ] ) @xmath93 assuming that @xmath94 .
we demand that the configuration should really be an x - point , but not a one - dimensional sheet , therefore we assume that the relation between the two characteristic currents is given by a value not much larger than @xmath95 ( correponds to an opening angle of the smaller separatrix angle of only about 2 degrees ) . with the electron charge @xmath44 , the number density @xmath43 and the drift velocity @xmath96 and @xmath97 , it follows for @xmath8 @xmath98 the value @xmath99 must now be compared with the typical lengthscale and values of a magnetic field close to a null point , i.e. the coronal magnetic field is about @xmath100 ( the index c stands for coronal ) and the scale on which the magnetic field varies linearly should be larger than a debye length ( for coronal parameters in the range of mm to cm ) m ] , but much smaller than the active region loops ( @xmath101 m ) to get an upper limit for @xmath99 .
we therefore use the approximate current sheet thickness @xmath102 . to estimate the current sheet thickness of a finite current sheet , we assume again that the current density @xmath103 , where @xmath96 is the drift velocity , which should not exceed the thermal velocity ( of the electrons ) , i.e. @xmath104 . let @xmath105jk@xmath106 the boltzmann constant , @xmath107a@xmath108s@xmath109kg@xmath106m@xmath110 the dielectric constant , @xmath111 kg the mass of the proton , @xmath112 kg the mass of the electron , @xmath113m@xmath114 the number density , and @xmath115k ( as the thermal temperature is about @xmath116k , we consider an effective temperature @xmath117 , thus that the corresponding drift velocity @xmath96 is definitely below the critical drift velocity given by temperature @xmath118 to guarantee a stable stationary current , instead of an kinetic instability see , e.g. , ) the approximate temperature ( for the coronal parameters see , e.g. , ) .
we choose a small current density to get the upper limit of @xmath8 .
the offset is then given by @xmath119 the lower bound of the current sheet thickness is approximately given by @xmath120 the fraction of both values should be @xmath121and is then determined by @xmath122 thus for typical coronal parameters the offset in @xmath1-direction has a value of about mm or even some orders of magnitude less , in contrast to the value of the typical lengthscale of the linear x - point region which is at least of the order of one meter or some tenth of meters or even more .
this is an argument or rough justification to neglect the term of the interaction between the shift and the second derivative of @xmath6 .
even if we assume that the drift velocity @xmath96 is , let us say six orders , smaller than the thermal velocity , the fraction @xmath123 would be of the order of some percent .
therefore , even in this extreme regime of the drift velocity , the stagnation point is located within the current sheet , assuming that the sheet is really two - dimensional ( non - singular ) and its extensions in both coordinate directions are about of the same order .
the complete mentioned discussion is valid exactly only in the case of unbounded fields , but it shows that in general the offset due to gravity can be neglected .
the discriminant @xmath87 in eq.([28 ] ) , being larger than zero is only a _
necessary _ criterion .
that @xmath124 has also to be guaranteed . to fullfil the relation @xmath125 we define two values , namely @xmath126 to represent the relation eq.([27 ] ) @xmath127 which implies for ci @xmath128 and for cii @xmath129 .
the values @xmath130 are explicit functions of the current densities @xmath17 and @xmath20 .
this gives us restrictions for @xmath131 : for @xmath91 the lower bound for @xmath131 is given by the left relation of eq.([31 ] ) otherwise the lower bound is zero , writing @xmath132 such that the non - admissible region is then given by @xmath133\ , , \end{aligned}\ ] ] and @xmath134 is given by @xmath135 for @xmath136 all values of @xmath131 are allowed .
there are no restrictions for @xmath131 , as for @xmath32 the critical value @xmath137 and for @xmath138 all values of @xmath134 are imaginary . for cii
it is clear that the criterion is given by @xmath139 for @xmath138 no value of @xmath131 can satisfy the relation eq.([35 ] ) . only for @xmath91 solutions
exist in the region beyond the curve fig.[bild ] and the @xmath20-axis .
thus we can summarize : for @xmath91 the allowed values for @xmath140 are given by the relation @xmath141 in the case of a vanishing @xmath131 or @xmath80 the equations again lead to the statement that either @xmath142 in both cases or that again @xmath72 .
this implies that the allowed flows are of higher orders with respect to the spatial variables , like indicated in @xmath143 or need special , maybe unphysical thermodynamical constraints .
also the case of general shear flows leads to @xmath72 .
such solutions will not be discussed here .
the resistivity is given by the @xmath21-component of @xmath144 and results in the quadric @xmath145 neglecting the term with @xmath8 , the above equation shows that the isocontours of the resistivity have four branches . in two quadrants
@xmath2 has a positive sign and the other two quadrants represent an effective negative resistivity .
this example of unbounded fields shows an interesting behaviour because the solutions are exact analytical ones and they exhibit the fact that there is a special relation between the derivatives of @xmath6 and of @xmath10 connected about algebraic varieties which generate the solutions .
we stop these investigations here due to the problem of negative resistivity . to interpret this negative effective resistivity ( or negative non - idealness ) and
search for a reasonable physical interpretation or correponding physical mechanism needs further investigation .
here @xmath68 in general , as the equation which contradicts the thermodynamical constraints , the last two equations of eqs.([18 ] ) and ( [ 19 ] ) , being part of the energy equation , are neglected due to their higher order in @xmath146 . in this case
we only need to regard equations , being of first order in @xmath0 and @xmath1 , namely @xmath147 @xmath148 which contain only terms of first order .
this is a nonlinear system consisting of eight , but , of course , effectively seven equations for nine unknowns . in this case
the stagnation point is identical with the magnetic null point .
the general solutions is now represented by @xmath149 with free parameters @xmath73 and @xmath131 ( and , of course , the magnetic parameters the threshold current @xmath17 and the actual current @xmath20 ) . the parameter @xmath80 is not free , as it is determined by the fact that the quadric ( surface ) , representing the resistivity , must be an elliptic paraboloid .
we have to introduce another parameter @xmath150 to express @xmath80 as function of @xmath131 and @xmath150 .
the elliptic paraboloid is the only quadric that allows a positive resistivity with one zero ( the vertex of the paraboloid ) at the magnetic null point .
all other quadrics can be excluded , with the exception of the degenerated case of a cylindrical paraboloid , see , e.g. , or .
the cylindrical paraboloids are limiting cases with @xmath151 as is shown as in the example in fig.[cross4 ] ( here the specific value is @xmath152 ) . the general solution is then parameterized by @xmath153 @xmath154 with the restriction @xmath155 $ ] .
multiplication of eq.([cl16 ] ) with @xmath131 , inserting this expression into the expression for the eigenvalue @xmath156 , and completing the square gives @xmath157 if @xmath158 and @xmath159 as assumed , the magnetic field is of hyperbolic type and it is guaranteed that the stagnation point is also of hyperbolic type , i.e. has two purely real eigenvalues , namely one positive value and a negative counterpart . although the solutions shown here are no reconnection solutions , as the electric field has to be zero , because only resistive dissipation is included in our investigation , they represent almost reconnection solutions(=reconnection - like solutions ) where the necessary condition for reconnection is fulfilled .
let us investigate now the geometry , respectively the slopes of the separatrices of the flow and the magnetic field .
defining @xmath160 , the magnetic separatrix is given by @xmath161 and @xmath162 $ ] ( we restrict this to @xmath163 ) and can therefore be expressed by @xmath164 where @xmath165 is the slope of the both magnetic separatrix lines or to say the both asymptotes .
as @xmath80 is given by the general solution , eq.([cl16 ] ) , by defining a stream function @xmath166 via @xmath167 we can integrate the above equations and get for the stream function @xmath168 the fluid separatrix is here given by @xmath169 , geometrically this are asymtotical lines ( asymptotes ) .
we will briefly discuss the problem that the magnetic separatrix is partially identical with the hydrodynamic separatrix . in this case
the plasma flow can only take place across one part of the separatrix , or both separatrix lines are identical and no reconnection can take place
. one can clearly recognize that in the case @xmath74 and thus @xmath170 both corresponding asymtotical branches(@xmath171separatrix lines ) have the same slope .
thus the hydrodynamical separatrix and the magnetic separatrix are identical , the plasma can not cross the magnetic separatrix and therefore no reconnection can take place . writing @xmath172 ( if @xmath173 ) , where @xmath174 is the slope of the hydrodynamic separatrix , and inserting this into @xmath175 with the parametric expression for @xmath80 in eq.([cl16 ] ) , we get the slopes @xmath176 of the two asymptotes / separatrix lines @xmath177 where @xmath178 . for @xmath179
the asymptotes are given by @xmath180 and @xmath181 , and for @xmath142 we get @xmath180 and @xmath182 as separatrices .
let @xmath183 , then @xmath184 there is _ one _ corresponding slope of the flow separatrix with respect to one of the magnetic separatrix lines , if @xmath151 .
therefore we can formulate the following theorem ( in analogy to known anti - reconnection theorems ) , which is now restricted to almost reconnective solutions with vanishingly small electric fields and resistive disspation only , i.e. vanishing non - resistive dissipation mechanisms : [ [ theorem ] ] theorem + + + + + + + _ if the flow close to the null point is in good approximation incompressible , the electric field is negligible and other kinds of dissipation mechanisms than the resistive dissipation are also negligible , then the plasma flow can not cross the magnetic separatrices if for the both slopes of the both magnetic separatrix lines @xmath185 with @xmath186 and the following is valid : @xmath187 if only i.a ) or i.b ) or ii.a ) or ii.b ) is valid then only one magnetic separatrix can be crossed and as the other stream lines converge to the second magnetic separatrix line without crossings , one could call this reconnective annihilation , see , e.g. , priest & forbes ( 2000 ) . the necessary condition for a complete non - crossing is @xmath74 for @xmath188 and for only partly crossing flows @xmath151 . _ for almost all values of @xmath150 , here @xmath189 , the flow in fig.[cross4 ] ( left column ) crosses all four separatrix branches ( all two separatrix lines ) , the resistivity is positive ( only zero at the null point , middle panel ) , and their isocontours are ellipses ( bottom panel ) . for the special parameter @xmath152
it can be seen in fig.[cross4 ] ( right column ) that the flow , like for the aforementioned magnetic reconnective annihilation solutions , crosses only the separatrix line with the positive slope , while it converges to the other magnetic separatrix , being almost field aligned ( top panel ) , the resistivity being also postive ( middle panel ) , and the isocontours of the resistivity are straight lines ( bottom panel ) .
we now focus on the effective resistivity and therefore on the nature of ohmic heating / disspation .
the resistivity is given by @xmath190 .
\label{resi1}\end{aligned}\ ] ] to ensure that the dissipation is positive we prove that the term @xmath191 is larger than zero ( or zero ) for all @xmath150 and @xmath146 .
let @xmath192 and @xmath193 as required for an x - point .
then @xmath194 this implies that the sign of the resistivity @xmath2 depends on the sign of @xmath195 . as the dissipation should be positive , i.e. @xmath196
, thus @xmath73 and @xmath20 must have the same sign . for @xmath151
it can be recognized from eq.([resi1 ] ) and from the bottom panel fig.[cross4 ] ( for the case @xmath152 ) that the resistivity contour lines are straight lines given by @xmath197 , and therefore the resistivity has the structure of a one - dimensional sheet .
the case @xmath151 therefore reflects the 1d character of the non - idealness , which can also be found in the papers of .
we analyse the solution space of the mhd system with a generalized non - idealness in ohm s law close to an x - type magnetic null point .
this procedure is done to get non - ideal / resistive but formally non - reconnective solutions that can be regarded as reconnection solutions with a vanishing electric field .
such solutions show the parameters and their relation amongst each others , to get non - ideal non - reconnective , i.e. either almost reconnective annihilation or completely almost reconnective (= reconnection - like ) solutions .
we assume that the flow close to the magnetic null point is governed strictly by the pure non - ideal / resistive system of mhd equations , and that the stagnation point is close to the null point ( or that both singular points in the plane are identical ) .
the focus is on the region close to the magnetic null point to investigate the possible type of topology of the corresponding stagnation point ( flow ) . to use a closure for the equations and to analyse the resistive mhd system we use the corresponding energy equation .
we further assume that the density is constant close to the stagnation point .
this assumption is not a necessary , but a plausible condition concerning the mass distribution in the vicinity of stagnation points .
this leads to a vanishing divergence of the plasma velocity and as a logical consequence to a vanishing resistivity close to the stagnation point .
inserting this into resistive ohm s law implies that the electric field has to vanish at the stagnation point .
we conclude that the electric field has to be zero , and as a consequence of the 2d stationary approach the electric field has to be zero everywhere .
this case of vanishing divergence of the velocity field and thus of a vanishing electric field can be regarded as a physical approximation or limit .
obviously a classical reconnection process requires a non - vanishing divergence of the velocity field to produce a non - zero reconnection rate , determined by the constant electric field @xmath57 , see , e.g. , .
from resistive ohm s law one can clearly recognize that close to the magnetic null point the electric field is identical to the convective electric field generated by the product from resistivity and current density . as the electric field is zero
, the resistivity inside the finite current sheet must be zero at the magnetic neutral point .
the effect is that the resistivity can not be constant close to the magnetic neutral point .
therefore the resistive ohm s law is an equation determining the topological type of the magnetic field and the velocity field and connecting / relating it to the shape of the resistivity , which is a quadratic function of the spatial coordinates .
we call this spatially varying resistivity _ effective resistivity _ or for short resistivity as usual .
as the current density is constant close to the magnetic null point , the right hand side of ohm s law , eq.([6 ] ) could be regarded as a generalized non - ideal term .
a _ necessary _ condition for reconnection is that the plasma flow can cross magnetic separatrices .
we found that crossing of magnetic separatrices requires also an x - type stagnation point flow if the effective resistivity and thus the dissipation should be positive .
the sufficient criterion for reconnection is the crossing of the separatrix and the non - vanishing electric field .
this reconnection process can take place for / in the incompressible case if there is not only a resistive , but an additional non - resistive dissipation ( term ) that occurs only in the energy equation .
therefore one possibility could be heat conduction .
another possibility is the viscous case with constant viscosity . here
, the laplacian of the velocity field vanishes in the first order momentum equation of the ions ( navier - stokes ) as the velocity field is linear .
these cases have to be investigated in more detail but have been discussed under the assumption of constant resistivity , e.g. , by .
the analyses started by us here will be extended to configurations that have density gradients close to the stagnation point and therefore allow for a non - zero electric field .
the extension to energy equations allowing for a deviation from the classical energy equation of resistive mhd will provide us with a relation between the deviation from pure resistive dissipation and deviation from the classical ( x - type ) stagnation point flow .
nickeler acknowledges financial support from gaavr grant no .
iaa300030804 and m. karlick from gar grant no .
nickeler is grateful to dr .
michaela kraus for the help in preparing fig.2 and careful reading of the manuscript .
amann , h. : 1995 , gewhnliche differentialgleichungen walter de gruyter , berlin .
arnold , v. : 1992 , ordinary differential equations , springer , berlin .
bartsch , e. : 1984 , mathematische formeln , veb fachbuchverlag , leipzig .
baty , h. , forbes , t.g . and priest , e.r . : 2009 , _ phys . of plasmas
_ * 16 * , 012102 .
braginsky , s.i .
: 1965 , _ rev .
plasma phys . _
* 1 * , 205 .
bronstein , i.n . , and
semendjajew , k.a . : 1987 , taschenbuch der mathematik , verlag harri deutsch , frankfurt / main .
buechner , j. and elkina , n. : 2006 , _ space sci .
_ * 121 * , 237 .
cowley , s.w.h . : 1976 , _ j. geophys .
_ * 81 * , 3455 .
craig , i.j.d . and henton , s.m .
: 1995 , _ astrophys . j. _ * 450 * , 280 .
craig , i.j.d . and rickard , g.j .
: 1994 , _ astron .
_ * 287 * , 261 .
craig , i.j.d . and watson , p.g . : 2003 , _ solar phys .
_ * 214 * , 131 .
goedbloed , j.p . and poedts , s. : 2004 , principles of magnetohydrodynamics , cambridge university press , cambridge .
karlick , m. and brta , m .- j .
: 2005 , _ solar phys .
_ * 247 * , 335 .
nickeler , d. and fahr , h .- j .
: 2005 , _ adv .
space res .
_ * 35 * , 2067 .
nickeler , d. , goedbloed , h. , and fahr , h .- j . : 2006 , _ astron .
_ * 454 * , 797 .
papadopoulos , k. : , 1977 , _ rev . of geophysics _ * 15 * , 113 .
parnell , c. , neukirch , t. , and priest , e. : 1996 , _ phys . of plasmas
_ * 3 * , 759 .
petschek , h.e . : 1964 , in w.n .
hess ( ed . ) , the physics of solar flares , aas - nasa symp . ,
priest , e. and cowley , s.w.h . : 1975 , _
j. of plasma phys .
_ * 14(2 ) * , 271 . priest ,
e. and forbes , t. : 2000 , magnetic reconnection cambridge university press , cambridge .
priest , e. , titov , v.s .
, vekstein , g.e . , and rikard , g.j .
: 1994 , _ j. geophys .
_ * 99(a11 ) * , 21467 .
reitmann , v. : 1996 , regulre und chaotische dynamik , b.g .
teubner verlagsgesellschaft , leipzig .
schindler , k. , hesse , m. , and birn , j. : 1988 , _ j. geophys .
res . _ * 93 * , 5547 .
sonnerup , b.u. . : 1979 , in l.j .
lanzerotti , c.f .
kennel , and e.n .
parker ( eds . ) solar systems plasma physics , vol.3 , north holland , amsterdam , p. 45 .
sonnerup , b.u. . ,
baum , p.j . ,
birn , j. , et al . : 1984 , in d.m . butler and k.
papadopoulos ( eds . ) , solar terrestrial physics : present and future , nasa , ref .
publ . vol .
1120 , p. 1 .
stix , m. : 2002 , the sun , springer , berlin / heidelberg / new york .
sweet , p.a .
nuovo cimento _ * 8 * , 188 .
tassi , e. , titov , v.s . , and hornig , g. : 2002 , _ phys .
letters a _ * 302 * , 313 .
titov , v.s . and hornig , g. : 2000 , _ phys . of plasmas _ * 7 * , 3542 .
titov , v.s . , tassi , e. , and hornig , g. : 2004 , _ phys . of plasmas _ * 11 * , 4662 .
watson , p.g . , and craig , i.j.d .
: 1998 , _ astrophys . j. _
* 505 * , 363 .
vasyliunas , v.m . : 1975 , _ rev .
geophys . _
* 13 * , 303 . | the appearance of eruptive space plasma processes , e.g. , in eruptive flares as observed in the solar atmosphere , is usually assumed to be caused by magnetic reconnection .
the process of magnetic reconnection is often connected with singular points of the magnetic field .
we therefore analyse the system of stationary resistive / non - ideal magnetohydrodynamics ( mhd ) in the vicinity of singular points of flow and field to determine the boundary between reconnection solutions and non - reconnective solutions .
we find conditions to enable the plasma to cross the magnetic separatrices also inside the current sheet , close to the current maximum .
the results provide us with the topological and geometrical skeleton of the resistive mhd fields .
we therefore have to perform a local analysis of almost all non - ideal mhd solutions without a specific non - idealness .
we use taylor expansions of the magnetic field , the velocity field and all other physical quantities , including the non - idealness , and with the method of a comparison of the coefficients , the non - linear resistive mhd system is solved analytically . in the vicinity of a stagnation point , it is reasonable to assume that the density is constant .
we find that the electric field has to be zero and that the non - ideal term / resistivity has to depend on the spatial coordinates and can not be constant , otherwise it has to be zero everywhere .
it turns out that not every non - ideal flow is a reconnective flow and that pure resistive / non - ideal mhd only allows for such reconnection - like solutions , even if the non - idealness is localized to the region around the magnetic null point .
it is necessary that the flow close to the magnetic x - point is also of x - point type to guarantee positive dissipation of energy and annihilation of magnetic flux .
if the non - idealness has only a one - dimensional , sheet - like structure , only one separatrix line can be crossed by the plasma flow , similar to reconnective annihilation solutions . |
let @xmath0 be a two - sided standard brownian motion in @xmath6 having @xmath8 .
if @xmath7 is a stopping time with respect to the filtration @xmath9 , then the shifted process @xmath10 is a one - sided brownian motion independent of @xmath4 .
however , the two - sided shifted process @xmath11 need not be a two - sided brownian motion .
moreover , the example of a fixed time shows that even if it is , it need not be independent of @xmath4 .
we call a random time @xmath2 an _ unbiased shift _ ( of a two - sided brownian motion ) if @xmath2 is a measurable function of @xmath1 and @xmath11 is a two - sided brownian motion , independent of @xmath4 .
we say that a random time @xmath2 _ embeds _ a given probability measure @xmath5 on @xmath6 , often called the _ target distribution _ , if @xmath4 has distribution @xmath5 . in this paper
we discuss several examples of nonnegative unbiased shifts that are stopping times .
however , we wish to stress that nonnegative unbiased shifts are not assumed to have the stopping time property , see for instance example [ exnonstop ] .
the paper has three main aims .
the first aim is to characterise all unbiased shifts that embed a given distribution @xmath5 .
the second aim is to construct such unbiased shifts . in particular
, we solve the skorokhod embedding problem for unbiased shifts : given any target distribution we find an unbiased shift which embeds this target distribution ( and is also a stopping time ) .
the third and final aim is to discuss some properties of unbiased shifts . in particular , we discuss optimality of our solution of the skorokhod embedding problem for unbiased shifts .
the case when the embedded distribution is concentrated at zero is of special interest .
let @xmath12 be the local time at zero .
its right - continuous ( generalised ) inverse is defined by @xmath13 = r\ } , \quad & r \geq 0,\\ \sup\{t < 0 : \ell^0 [ t,0 ] = -r\ } , \quad & r < 0.\end{cases } \end{aligned}\ ] ] note that @xmath14 and @xmath15 if @xmath16 .
we prove the following theorem .
[ littletwin ] let @xmath17 .
then @xmath18 is an unbiased shift embedding @xmath19 .
this result formalises the intuitive idea that two - sided brownian motion looks globally the same from all its ( appropriately chosen ) zeros , thus resolving an issue raised by mandelbrot in ( * ? ? ?
* , p. 385 ) and reinforced in @xcite .
another way of thinking about this result is that if we travel in time according to the clock of local time we always see a two - sided brownian motion .
this is analogous to a well - known property of the two - sided stationary poisson process with an extra point at the origin : the lengths of the intervals between points are i.i.d .
( exponential ) and therefore shifting the origin to the @xmath20th point on the right ( or on the left ) gives us back a two - sided poisson process with an extra point at the origin .
in fact , much of the work behind the present paper was inspired by recent developments for spatial point processes and random measures @xcite . in the terminology of @xcite ,
brownian motion is _ mass - stationary _ with respect to local time , see section [ secmassstat ] . theorem [ littletwin ] is relatively elementary . to state the further main results of this paper ,
we need to introduce some notation and terminology .
it is convenient to define @xmath1 as the identity on the canonical probability space @xmath21 , where @xmath22 is the set of all continuous functions @xmath23 , @xmath24 is the kolmogorov product @xmath25-algebra , and @xmath26 is the distribution of @xmath1 . define @xmath27 , @xmath28 , and the @xmath25-finite and stationary measure ( see section [ secpalm ] ) @xmath29 expectations ( resp
. integrals ) with respect to @xmath30 and @xmath31 are denoted by @xmath32 and @xmath33 , respectively . for any @xmath34
the shift @xmath35 is defined by @xmath36 an _ allocation rule _
@xcite is a measurable mapping @xmath37 that is _ equivariant _ in the sense that @xmath38 a _ random measure _
@xmath39 on @xmath6 is a kernel from @xmath22 to @xmath6 such that @xmath40 for @xmath31-a.e .
@xmath41 and all compact @xmath42 .
if @xmath39 and @xmath43 are random measures , and @xmath44 is an allocation rule such that the image measure of @xmath39 under @xmath44 is @xmath43 , that is , @xmath45 then we say that @xmath44 _ balances _ @xmath39 and @xmath43 . if @xmath44 balances @xmath39 and @xmath43 and @xmath25 is an allocation rule that balances @xmath43 and another random measure @xmath46 , then the allocation rule @xmath47 balances @xmath43 and @xmath46 .
let @xmath48 be the random measure associated with the _ local time _ of @xmath1 at @xmath28 ( under @xmath26 )
, see also section [ secpalm ] . if @xmath5 is a probability measure on @xmath6 , then below implies that @xmath49 defines a random measure @xmath50 .
this is an _ additive functional _ of brownian motion , that is , it has the invariance property , see section [ secpalm ] for more details . in fact , any diffuse additive functional must be of this form with a @xmath25-finite _ revuz measure _ @xmath5 , see e.g. ( * ? ? ?
* chapter 22 ) . for any random time
@xmath2 we define an allocation rule @xmath51 by @xmath52 since @xmath53 , there is a one - to - one correspondence between @xmath2 and @xmath51 .
our key characterisation theorem is based on a result in @xcite , which will be recalled as theorem [ th2 ] below .
[ main1 ] let @xmath2 be a random time and @xmath5 be a probability measure on @xmath6 .
then @xmath2 is an unbiased shift embedding @xmath5 if and only if @xmath51 balances @xmath12 and @xmath50 . for any probability measure @xmath5 on @xmath6
we denote by @xmath54 the distribution of a two - sided brownian motion with a random starting value @xmath55 with law @xmath5 .
we show in section [ secmassstat ] that all these distributions coincide on the invariant @xmath25-algebra . a general result in @xcite
( see also ( * ? ? ?
* theorem 10.28 ) ) then implies that there is a random time @xmath2 ( possibly defined on an extension of @xmath56 ) such that @xmath57 has distribution @xmath58 ( under @xmath26 ) .
the next two theorems yield a much stronger result .
they show that @xmath2 can be chosen as a _ factor _ of @xmath1 , that is , as a measurable function of @xmath1 , see @xcite for a similar result for poisson processes .
moreover , this factor is explicitly known .
the proof is based on theorem [ main1 ] and on a general result on the existence of allocation rules balancing jointly stationary orthogonal diffuse random measures on @xmath6 with equal conditional intensities , see theorem [ main3 ] .
[ main2 ] let @xmath5 be a probability measure on @xmath6 with @xmath59 . then the stopping time @xmath60=\ell^\nu[0,t]\big\}\end{aligned}\ ]
] embeds @xmath5 and is an unbiased shift .
the stopping time was introduced in @xcite as a solution of the _ skorokhod embedding problem_. this problem requires finding a stopping time @xmath7 embedding a given distribution @xmath5 , see @xcite for a survey .
it has apparently not been noticed before that is an unbiased shift .
if @xmath5 is of the form @xmath61 where @xmath62 and @xmath63 , then theorem [ main2 ] does not apply .
in fact , if @xmath64 then is an unbiased shift embedding @xmath65 . still we can use theorem [ main2 ] to construct unbiased shifts without any assumptions on @xmath5 : [ main2a ] let @xmath5 be a probability measure on @xmath6 . then there exists a nonnegative stopping time that is an unbiased shift embedding @xmath5 .
in theorem [ littletwin ] we have @xmath66 and @xmath67 if @xmath16 .
it is interesting to note that unbiased shifts @xmath2 ( even if they are not stopping times ) are almost surely non - zero as long as the condition @xmath68 is fulfilled : [ prop3 ] let @xmath5 be a probability measure on @xmath6 such that @xmath64 .
then any unbiased shift @xmath2 embedding @xmath5 satisfies @xmath69 in contrast to the previous theorem , if @xmath2 is an unbiased shift with @xmath70 , then the probability @xmath71 may take any value : [ p11 ] for any @xmath72 $ ] there is an unbiased shift @xmath7 embedding @xmath19 and such that @xmath73 . a solution @xmath2 of the skorokhod embedding problem is usually required to have good moment properties , but some restrictions apply .
for instance , if the target distribution @xmath5 is not centered , by ( * ? ? ?
* theorem 2.50 ) , we must have @xmath74 .
if the embedding stopping time is also an unbiased shift the situation is worse , even when @xmath5 is centred .
[ t14complete ] suppose @xmath5 is a target distribution with @xmath59 , and @xmath75 , and the stopping time @xmath76 is an unbiased shift embedding @xmath5 .
then @xmath77 under the same assumptions on @xmath5 the unbiased shift constructed in satisfies @xmath78 dropping the stopping time assumption we show in theorem [ pmoment ] that @xmath79 for any unbiased shift @xmath2 embedding a target distribution @xmath5 with @xmath64 .
if the target distribution is concentrated at zero and @xmath2 is nonnegative but not identically zero , we show in theorem [ posmoment ] that @xmath80 .
nonnegativity is important in this result , example [ exexpon ] provides an unbiased shift with @xmath81 that has exponential moments .
theorem [ pminimal ] further shows that , in addition to the nearly optimal moment properties stated above , the stopping times @xmath2 defined in are also _ minimal _ in a sense analogous to the definition in @xcite ( see also @xcite , or @xcite for a survey ) .
this means that if @xmath82 is another unbiased shift embedding @xmath5 such that @xmath83 , then @xmath84 .
our discussion of minimality is based on a notion of stability of allocation rules , which is similar to the one studied in @xcite .
the structure of the paper is as follows .
section [ secpalm ] provides some background on palm measures and local time .
section [ secmassstat ] presents a general result on mass - stationarity for diffuse random measures on the line , implying theorem [ littletwin ] .
section [ sexinv ] contains the proof of a more general version of theorem [ main1 ] .
section [ secexistence ] presents the key general result on balancing diffuse jointly stationary random measures , theorem [ main3 ] , implying theorem [ main2 ] .
section 6 contains the proofs of theorems [ main2a ] , [ prop3 ] and [ p11 ] . in sections [ secstability ] and [ secproperties ] we discuss minimality and moment properties of unbiased shifts , including the proof of theorem [ t14complete ]
section [ secremarks ] concludes with some remarks .
recall the definition of the measure @xmath31 on @xmath85 . since @xmath86 for any compact @xmath42
, @xmath31 is @xmath25-finite .
we also note the invariance property @xmath87 the proof of is only based on the stationary increments of @xmath1 , see @xcite .
corollary [ timep ] provides an alternative definition of @xmath31 . a random measure @xmath39
is called _ invariant _ if @xmath88 where @xmath89 is the borel @xmath25-algebra on @xmath6 . in this case the _ palm measure _
@xmath90 of @xmath39 ( with respect to @xmath31 ) is defined by @xmath91}(s){{\bf 1}}_a(\theta_sb)\ , \xi(ds ) , \quad a\in\mathcal{a}.\end{aligned}\ ] ] this is a @xmath25-finite measure on @xmath85 .
if the _ intensity _ @xmath92 of @xmath39 is positive and finite , @xmath90 can be normalised to yield the palm probability measure of @xmath39 .
even though @xmath90 is generally not a probability measure , we denote integration with respect to @xmath90 by @xmath93 .
the invariance property implies the _ refined campbell theorem _
@xmath94 for any measurable @xmath95 .
we recall the following result from @xcite .
[ th2 ] consider two invariant random measures @xmath39 and @xmath43 on @xmath6 and an allocation rule @xmath44
. then @xmath44 balances @xmath39 and @xmath43 if and only if @xmath96 where @xmath97 is defined by @xmath98 , @xmath99 .
recall that @xmath48 is the random measure associated with the _ local time _ of @xmath1 at @xmath28 ( under @xmath26 ) .
this means that @xmath100 for all measurable @xmath101 .
we assume that @xmath12 is @xmath30-a.e .
diffuse for any @xmath28 and that @xmath102 the global construction in @xcite guarantees the existence of a version of @xmath12 with these properties , see also ( * ? ? ?
* proposition 22.12 ) or ( * ? ? ? * theorem 6.43 ) .
this construction implies , moreover , that it is no restriction of generality to assume that @xmath103 and @xmath104 equation implies that @xmath105 is @xmath30-a.e .
diffuse for any @xmath28 and has the invariance property .
the following result is essentially from @xcite , see also @xcite .
[ palmlocal ] let @xmath106 .
then @xmath107 is the palm measure of @xmath105 .
_ let @xmath95 be measurable .
by definition of @xmath31 we have @xmath108 by this equals @xmath109 where the equality comes from a change of variables and fubini s theorem . by this
equals @xmath110 since @xmath111 , we obtain @xmath112 and hence the assertion .
equation is the refined campbell theorem in the case @xmath113 . in particular
, it implies that @xmath105 has intensity @xmath114 : @xmath115)=1.\end{aligned}\ ] ]
in this section we prove theorem [ littletwin ] . moreover ,
we show that the invariance property in theorem [ littletwin ] characterises mass - stationarity ( defined below ) of general diffuse random measures on the line .
let @xmath39 be a diffuse random measure on @xmath6 and @xmath116 be a random element in a space on which the additive group @xmath6 acts measurably .
let @xmath117 denote the shift of @xmath118 by @xmath119 .
let the pair @xmath120 be defined as the identity on a canonical measurable space equipped with a @xmath25-finite measure @xmath121 .
let @xmath122 denote lebesgue measure on @xmath6 .
the pair @xmath118 is called _ mass - stationary _ if , for all bounded borel subsets @xmath123 of @xmath6 with @xmath124 and @xmath125 and all nonnegative measurable functions @xmath126 , @xmath127 here we set @xmath128 whenever @xmath129 .
mass - stationarity is a formalisation of the intuitive idea that the origin is a typical location in the mass of a random measure .
the invariance property can be interpreted probabilistically as saying that if the set @xmath123 is placed uniformly at random around the origin and the origin shifted to a location chosen according to the mass distribution of @xmath39 in that randomly placed set then the distribution of the pair @xmath118 does not change .
the invariance property in the following theorem is a new characterisation of mass - stationarity .
it is similar to the well - known characterisation in the simple point process case ( see e.g. ( * ? ? ?
* theorem 11.4 ) ) and is certainly more transparent than .
it is however restricted to the diffuse case on the line while works for general random measures in a group setting .
the result below is also new , but the equivalence of mass - stationarity and palm measures was established in @xcite for abelian groups and in @xcite for general locally compact groups .
[ mass - stat ] assume that @xmath130 and let @xmath131 , @xmath132 , be the generalised inverse of the diffuse random measure @xmath39 defined as in .
then @xmath133 if and only if @xmath118 is mass - stationary and if and only if the distribution @xmath121 of @xmath120 is the palm measure of @xmath39 with respect to a @xmath25-finite stationary measure @xmath134 .
the measure @xmath134 is uniquely determined by @xmath121 as follows : for each @xmath135 and each bounded nonnegative measurable function @xmath126 , @xmath136 first assume .
then , @xmath137=0\}=\bq\{\xi[s_1,s_1+\varepsilon]=0\}=0 $ ] , for any @xmath138 , where the second identity comes from @xmath139 @xmath121-a.e.and the definition of @xmath140 .
this easily implies that @xmath141 let @xmath142 be a bounded borel with @xmath124 and @xmath125 .
changing variables and noting that , for any @xmath143 in the support of @xmath39 , we have @xmath144 for @xmath122-a.e .
@xmath145 , we obtain that the left - hand side of equals @xmath146 where we have changed variables to get the equality . the key observation and assumption yield that the above equals @xmath147 thus holds ,
that is , @xmath118 is mass - stationary . by ( * ?
* theorem 6.3 ) equation is equivalent to the existence of a stationary measure @xmath134 such that @xmath121 is the palm measure of @xmath39 with respect to @xmath121 .
@xcite inversion formula ( see also ( * ? ? ?
* section 2 ) ) implies that @xmath134 is uniquely determined by @xmath121 and that , moreover , @xmath148 .
fix @xmath135 .
for the claim that @xmath134 defined by is stationary when holds , see lemma [ secondq * ] below . to show that @xmath121 is then the palm measure of @xmath39 with respect to this @xmath134 let @xmath126 be nonnegative measurable and use for the first step in the following calculation , @xmath149}(s ) f(\theta_s(x,\xi))\,\xi(ds ) = \be_{\bq}\iint { { \bf 1}}_{[0,1]}(s ) { { \bf 1}}_{[0,s_w]}(t ) f(\theta_s\theta_t(x , \xi ) ) \,\theta_t\xi(ds)\,dt\\ & = \be_{\bq}\iint { { \bf 1}}_{[0,1]}(v - t ) { { \bf 1}}_{[0,s_w]}(t ) f(\theta_v(x , \xi))\,\xi(dv)\,dt\\ & = \be_{\bq}\iint { { \bf 1}}_{[0,1]}(s_r - t ) { { \bf 1}}_{[0,s_w]}(t ) f(\theta_{s_r}(x,\xi))\,dr\,dt\\ & = \be_{\bq}\iint { { \bf 1}}_{[0,1]}(-s_{-r}-t ) { { \bf 1}}_{[0 , s_{w}(\theta_{s_{-r}}(x , \xi))]}(t ) f(x,\xi)\,dr\,dt\\ & = \be_{\bq}f(x , \xi)\iint { { \bf 1}}_{[-1,0]}(u ) { { \bf 1}}_{[s_{-r},\ , s_{-r+w } ] } ( u)\ , dr\,du\\ & = w\,\be_{\bq}f(x , \xi),\end{aligned}\ ] ] where we have used and for the fourth identity and the final identity holds since the double integral equals @xmath150 .
finally , if @xmath121 is the palm measure of @xmath39 with respect to a @xmath25-finite stationary measure @xmath134 , then theorem [ th2 ] implies once we have shown for any @xmath17 that the allocation rule @xmath151 defined by @xmath152 balances @xmath39 with itself , that is , @xmath153 asssume @xmath154 .
then , outside the @xmath134-null set @xmath155 we obtain for any @xmath156 ( interpreting @xmath157 $ ] as @xmath158 $ ] for @xmath159 ) that @xmath160\le r , \xi[s , b ] > r\ } \ , \xi(ds)\\ & = \int { { \bf 1}}\{s\le b , r < \xi[s , b]\le r+\xi[a , b ] \ } \ , \xi(ds)= \xi[a , b],\end{aligned}\ ] ] which implies the desired balancing property .
the case @xmath161 can be treated similarly .
[ secondq * ] let @xmath82 be a random time and @xmath134 be the measure defined by setting , for each bounded nonnegative measurable function @xmath126 , @xmath162 if @xmath163 has the same distribution as @xmath118 under @xmath121 then @xmath118 is stationary under @xmath134 . for each @xmath126 as above and @xmath34 , @xmath164 where the third identity follows from the assumption that @xmath163 has the same distribution as @xmath118 under @xmath121 . as a corollary
we obtain an alternative construction of the stationary measure by integrating over time rather than space .
[ timep ] let @xmath135 and consider @xmath165 defined by
. then @xmath166 , where @xmath167 _ proof of theorem [ littletwin ] . _
the result follows from theorem [ mass - stat ] and lemma [ palmlocal ] .
the _ invariant @xmath25-algebra _ is defined by @xmath168 we now apply theorem [ littletwin ] to prove the following result which we need in the proof of theorem [ main2 ] in section [ secexistence ] .
[ h1 ] let @xmath169 .
then either @xmath170 for any @xmath28 ( in which case @xmath171 ) or @xmath172 for any @xmath28 ( in which case @xmath173 ) .
we first show that @xmath174 we use here the random times @xmath175 ( see ) for integers @xmath20 . by theorem [ littletwin ] , for any integer @xmath20 , the processes @xmath176 and @xmath177 are independent one - sided brownian motions .
this implies that the processes @xmath178 are independent under @xmath26 . since , by , @xmath179)=1\ } = \inf\{t\ge 0 \colon \ell^0(b,[t_n , t_n+t])=1\ } = t_{n+1}-t_n\end{aligned}\ ] ] holds @xmath26-a.s .
for any @xmath180 , the @xmath181 have the distribution of a one - sided brownian motion stopped at the time its local time at @xmath182 reaches the value @xmath114 .
clearly we have that @xmath183 for a suitably defined measurable function @xmath184 . by invariance of @xmath185 and definition of the family @xmath186 , @xmath187 where the final equation holds @xmath26-a.s .
since iid - sequences are ergodic ( by the law of large numbers ) , we obtain .
the refined campbell theorem implies ( with @xmath122 denoting lebesgue measure ) @xmath188 provided that @xmath189 .
assume now that @xmath190 .
then implies that @xmath191 for all compact @xmath42 .
letting @xmath192 , we obtain @xmath193 , that is , @xmath194 on the other hand , by , @xmath195 for @xmath122-a.e .
@xmath196 so that @xmath170 for @xmath122-a.e .
@xmath196 . therefore @xmath171 .
by this implies @xmath170 for all @xmath28 .
in this section we prove a result that is more general than theorem [ main1 ] .
recall that @xmath197 for a probability measure @xmath65 on @xmath6 .
[ main1a ] let @xmath2 be a random time and @xmath198 be probability measures on @xmath6 .
then the following two assertions are equivalent . 1 .
it is true that @xmath199 @xmath200 , and that @xmath201 and @xmath4 are independent under @xmath202 .
the allocation rule @xmath51 defined by balances @xmath203 and @xmath50 .
we start by noting that the random measures @xmath203 and @xmath50 are invariant in the sense of .
this follows from the invariance of local time , and fubini s theorem .
let us first assume that ( i ) holds .
then we have for any @xmath204 that @xmath205 lemma [ palmlocal ] and fubini s theorem imply that @xmath58 is the palm measure of @xmath50 .
therefore we obtain from theorem [ th2 ] that @xmath51 balances @xmath203 and @xmath50 .
assume now that ( ii ) holds . by theorem [ th2 ]
we obtain for any @xmath204 that @xmath206 this implies @xmath207 for any @xmath208 and any @xmath209 .
this yields ( i ) .
[ remsub]an _ extended _ allocation rule is a mapping @xmath210 $ ] that has the equivariance property .
the balancing property can then be defined as before . using these concepts , theorem [ main1a ]
can be proved for a subprobability measure @xmath211 .
the conditions in ( i ) have to be replaced with @xmath212 , @xmath213 and the independence of @xmath201 and @xmath4 under @xmath214 .
in this section we prove theorem [ main2 ] . the proof is based on the following new balancing result for general random measures on the line , which is inspired by @xcite . [ main3 ]
let @xmath39 and @xmath43 be jointly stationary orthogonal diffuse random measures on @xmath6 with finite intensities .
assume further that @xmath215 \big| \mathcal i \big ] = \be\big [ \eta[0,1 ] \big| \mathcal
i \big ] \quad \bp\text{-a.e.}\ ] ] then the mapping @xmath37 , defined by @xmath216=\eta[s , t]\},\quad s\in\r,\end{aligned}\ ] ] is an allocation rule balancing @xmath39 and @xmath43
. for convenience , theorem [ main3 ] has been formulated for random measures defined on our canonical @xmath217 . however , the special structure of that space is of no importance here .
we start the proof of theorem [ main3 ] with an analytic lemma . here and later
it is convenient to work with the continuous function @xmath218 , defined by @xmath219-\eta[0,t],&\text{if $ t\ge 0$,}\\ \eta[t,0]-\xi[t,0],&\text{if $ t < 0$}. \end{cases}\end{aligned}\ ] ] [ analytic ] suppose @xmath39 and @xmath43 are orthogonal diffuse measures
. then @xmath220$},\ ] ] provided that @xmath221 for all @xmath222 .
the proof of lemma [ analytic ] rests on three further lemmas .
[ des ] + * for @xmath39-almost every @xmath143 there exists @xmath223 with @xmath224 . * for @xmath43-almost every @xmath143 there exists @xmath223 with @xmath225 .
it suffices to prove @xmath226 , as @xmath227 follows by reversing the roles of @xmath39 and @xmath43 .
recall that @xmath39 and @xmath43 are orthogonal and hence there exists a borel set @xmath185 with @xmath228 and @xmath229 .
we need to show that , for each @xmath230 , @xmath231 given any @xmath232 we may choose an open set @xmath233 with @xmath234 .
we can cover @xmath235 by a countable collection @xmath236 of nonoverlapping intervals @xmath237 $ ] , @xmath238 , @xmath239 , such that @xmath240 . indeed , suppose that @xmath241 is a connected component of @xmath242 , which intersects @xmath235 .
if there is a minimal element @xmath143 in @xmath243 let @xmath244 be the minimum of @xmath245 and the distance of @xmath143 to the right endpoint of @xmath241 .
add the interval @xmath237 $ ] to the collection @xmath236 and remove it from @xmath235 and @xmath242 .
if no such minimum exists we can pick a strictly decreasing sequence @xmath246 , @xmath247 , converging to the infimum . let @xmath248 be the minimum of @xmath245 and the distance of @xmath249 to the right endpoint of @xmath241 , and , for @xmath250 , let @xmath251 be the minimum of @xmath245 and @xmath252 .
add all intervals @xmath253 $ ] to the collection @xmath236 and remove their union from @xmath235 and @xmath242 .
note that after one such step ( performed in every connected component ) all of @xmath235 in connected components of length at most @xmath245 will be removed , and the lower bound of the intersection of all other connected components with @xmath235 , if finite , is increased by at least @xmath245 .
also , after one step , the intersection of any connected component with @xmath235 is either empty or bounded from below .
therefore , every set of the form @xmath254\cap a_\epsilon$ ] will be completely covered after finitely many steps by nonoverlapping intervals , as required .
observe that @xmath255 for every interval in the collection , and hence @xmath256 the result follows as @xmath232 was arbitrary .
we now fix @xmath257 and decompose @xmath126 on @xmath258 $ ] according to its backwards running minimum @xmath259 given by @xmath260 see figure 1 for illustration .
the nonegative function @xmath261 can be decomposed on @xmath258 $ ] into a family @xmath262 of excursions @xmath263 with starting times @xmath264 $ ] .
note that an excursion @xmath263 is a function such that there exists a number @xmath265 , called the lifetime of the excursion , such that @xmath266 , @xmath267 for all @xmath268 , and @xmath269 for all @xmath270 . formally putting @xmath269 for all @xmath271
the decomposition can be written as @xmath272 note that the intervals @xmath273 , @xmath274 , are disjoint .
we denote by @xmath123 the complement of their union in @xmath258 $ ] , i.e.@xmath275 \colon f(t)=m(t)\}.$ ] for every @xmath274 we have @xmath276 we only have to show that @xmath277 for @xmath39-almost every @xmath278 . by lemma [ des ] ( a ) , for @xmath39-almost every @xmath278 , there exists @xmath223 such that @xmath224 . as @xmath279 , by continuity of @xmath126
, we infer that there exists @xmath280 such that @xmath281 .
therefore @xmath282 as required .
[ rest ] we have @xmath283 first observe that if @xmath284 , then @xmath285 for all @xmath286 $ ] .
if @xmath287 for all @xmath286 $ ] then @xmath288 .
otherwise there exists a maximal @xmath286 $ ] with @xmath289 .
then @xmath290 is a true local minimum of @xmath126 in the sense that there exists @xmath291 with @xmath292 for all @xmath293 and @xmath294 for all @xmath295 .
in particular there are at most countably many levels @xmath296 where this can happen .
fixing such a level @xmath297 we note that @xmath298 .
summing over all these levels we see that @xmath299 .
we conclude the proof by showing that @xmath300 .
lemma [ des ] ( b ) ensures that , for @xmath43-almost every @xmath301 $ ] there exists @xmath223 such that @xmath225 , which implies that @xmath302 .
hence the stated equality follows .
_ proof of lemma [ analytic ] .
_ taking the sum over the equations in the previous two lemmas we obtain @xmath303.$ ] this implies @xmath304 , \quad a\geq 0,\ ] ] as any @xmath271 with @xmath305 satisfies @xmath306 $ ] , and lemma [ des ] ( a ) implies that @xmath39-almost every @xmath271 with @xmath307 satisfies @xmath308 , and so @xmath309 .
_ proof of theorem [ main3 ] .
_ define @xmath310 recall that by lemma [ analytic ] we have @xmath311 on @xmath312 provided @xmath313\ge \eta[s , t]$ ] for all @xmath314 . by lemma [ des ] this holds for @xmath315-a.e .
moreover , by stationarity of @xmath315 we have that @xmath316=0,\xi^\infty\ne 0\}=0 $ ] for all @xmath317 .
we infer that @xmath318 using the refined campbell theorem twice , we obtain @xmath319\ } \ , ds\\ & = \be \int { { \bf 1}}\{\xi^\infty\not=0 , \tau(s)\in[0,1]\ } \ , \xi(ds)\\ & = \be { { \bf 1}}\{\xi^\infty\not=0\ } \ , \eta^*[0,1 ] .
\end{aligned}\ ] ] using first and then our assumption gives @xmath320 = \be { { \bf 1}}\{\xi^\infty\not=0\ } \eta[0,1 ] = \be { { \bf 1}}\{\xi^\infty\not=0\ } \xi[0,1],\ ] ] and together with we infer that @xmath321,\ ] ] and therefore @xmath322 for @xmath39-a.e .
@xmath143 , @xmath31-a.e .
in particular , this implies that @xmath44 is a well - defined allocation rule .
an analogous argument implies that @xmath323 where @xmath324=\eta[t , s]\}\ ] ] is the inverse of @xmath44 .
we now use this to show that @xmath44 balances @xmath39 and @xmath43 .
fixing @xmath156 we aim to show that @xmath325=\eta[a , b]$ ] .
if @xmath326 for all @xmath327 $ ] this holds by lemma [ analytic ] .
otherwise we apply this lemma to suitably chosen alternative intervals . to this end
let @xmath328 \colon \text{$f(s)\leq f(t)$ for all $ a\le t \le b$ } \}\ ] ] be the leftmost minimiser of @xmath126 on @xmath329 $ ] . as @xmath330\geq f(a^*-\frac1n)-f(a^
* ) > 0 $ ] for all sufficiently large @xmath247 , we find a decreasing sequence @xmath331 with @xmath332 and hence @xmath333
. then @xmath334 $ ] and @xmath335 if @xmath336 . assuming first that @xmath336 we obtain from lemma [ analytic ] that @xmath337$},\ ] ] which implies the statement .
now assume that @xmath338 . in this case
we get @xmath311 on @xmath339 $ ] and on @xmath340 $ ] for every @xmath20 , and the result follows as @xmath341 . _ proof of theorem [ main2 ] .
_ by theorem [ h1 ] , @xmath342| \mathcal i ] = \be [ \ell^\nu[0,1]| \mathcal i]$ ] almost surely .
since @xmath12 and @xmath50 are orthogonal , we can combine theorems [ main3 ] and [ main1 ] to obtain the result .
assume in theorem [ main2 ] that @xmath5 is a subprobability measure .
then @xmath2 takes the value @xmath343 with positive @xmath26-probability . indeed , by remark [ remsub ] , defining the extended allocation rule @xmath44 by @xmath344
we get that @xmath44 balances the restriction of @xmath12 to @xmath345 and @xmath50 .
assertion ( i ) of theorem [ main1a ] remains valid in the sense explained in remark [ remsub ] and gives @xmath346 .
now assume in theorem [ main2 ] that @xmath5 is a locally finite measure with @xmath347 .
then @xmath348 and @xmath44 balances @xmath12 and @xmath349 .
the proof of theorem [ main3 ] still yields the inequality @xmath350 .
in particular @xmath43 is a diffuse ( and invariant ) random measure and must therefore be of the form @xmath351 for some measure @xmath352 , see e.g. ( * ? ?
* theorem 22.25 ) .
in fact , @xmath352 is a probability measure smaller than @xmath5 . hence @xmath2 is an unbiased shift embedding @xmath352 .
some properties of @xmath352 can be found in @xcite .
the nonnegative unbiased shifts in theorem [ main2 ] , theorem [ main2a ] and in theorem [ littletwin ] are all stopping times . in the next example
we construct a nonnegative unbiased shift embedding a distribution not concentrated at zero , which is not a stopping time .
[ exnonstop]let @xmath353 .
we define an allocation rule @xmath44 that balances @xmath12 and @xmath48 and such that @xmath354 is nonnegative but not a stopping time .
the mapping @xmath44 is the composition of the following five allocation rules .
let @xmath355 balance @xmath12 and @xmath48 according to theorem [ main2 ] .
let @xmath356 balance @xmath48 and @xmath48 by shifting forward one mass - unit , that is , let @xmath357 be defined by with @xmath358 and with @xmath12 replaced with @xmath48 .
let @xmath359 balance @xmath48 and @xmath12 according to theorem [ main2 ] .
finally define @xmath360 by shifting _ backward _
one mass - unit in the local time at @xmath196 , that is , let @xmath360 be defined by with @xmath361 and @xmath12 replaced with @xmath48 . the composition @xmath44 of these allocation rules balances @xmath12 and @xmath48
moreover , @xmath362 .
however , @xmath2 is not a stopping time .
this example can be extended to a general target distribution @xmath5 .
in this section we prove theorems [ main2a ] , [ prop3 ] , and [ p11 ] .
in contrast to the previous section we allow here for an atom at @xmath182 .
_ proof of theorem [ main2a ] .
_ let @xmath363 such that @xmath364 and define @xmath365 theorems [ main1 ] and [ main2 ] imply that the allocation rule @xmath366=\ell^\mu[s , t]\big\},\quad s\in\r,\end{aligned}\ ] ] balances @xmath12 and @xmath203 .
the same theorems imply that there is an allocation rule @xmath367 that balances @xmath105 and @xmath12 .
define @xmath368 then we have for any borel set @xmath42 outside a fixed @xmath31-null set that @xmath369 where we have used ( and @xmath364 ) in the penultimate equation . hence @xmath44 balances @xmath12 and @xmath50 . theorem [ main1 ]
now implies that @xmath354 is an unbiased shift embedding @xmath5 .
_ proof of theorem [ prop3 ] .
_ let @xmath2 be any unbiased shift embedding @xmath5 and define @xmath370 . outside a fixed @xmath31-null
set we obtain for any borel set @xmath142 that @xmath371 where we have used to obtain the final identity .
this implies that @xmath372 assuming now that @xmath64 we obtain @xmath373 for @xmath12-a.e .
@xmath143 , @xmath31-almost everywhere .
lemma [ palmlocal ] now implies .
_ proof of theorem [ p11 ] .
_ let @xmath374 , where @xmath375 is given by with @xmath376 .
define a stationary random measure @xmath39 by @xmath377 .
the allocation rule @xmath378=1\big\}\ ] ] balances @xmath39 with itself .
define @xmath379 it is easy to see that @xmath44 balances @xmath12 with itself .
lemma [ palmlocal ] and theorem [ main1 ] ( or a direct calculation ) implies that @xmath354 satisfies @xmath380 since @xmath2 is an unbiased shift , the proof is complete .
the following definition is a one - sided version of the notion of stability introduced in @xcite for point processes .
we call an allocation rule @xmath37 balancing @xmath39 and @xmath43 _ right - stable _ if @xmath381 for all @xmath317 and @xmath382 roughly speaking this means that the mass of pairs @xmath383 such that @xmath143 would prefer the partner of @xmath384 over its own partner , while @xmath385 would prefer @xmath143 over @xmath384 as a partner , vanishes . [ rstable ] let @xmath39 and @xmath43 be invariant random measures satisfying the conditions of theorem [ main3 ] , and suppose @xmath37 is the allocation rule constructed in the theorem
. then @xmath44 is right - stable . by lemma [ des ] ( a ) and continuity of @xmath126 , we have for @xmath39-a - e .
@xmath143 that @xmath386 for all @xmath387 .
hence @xmath388-almost every pair @xmath383 with @xmath389 satisfies @xmath390 contradicting the definition of @xmath44 .
right - stable allocation rules have a useful minimality property . [ mini ]
any right - stable allocation rule @xmath44 balancing two measures @xmath39 and @xmath43 is minimal in the sense that if @xmath25 is another allocation rule balancing @xmath39 and @xmath43 such that @xmath391 for @xmath39-almost every @xmath317 , then @xmath392 by right - stability of @xmath44 we have , _ for @xmath39-almost every @xmath393 , _
@xmath394 \longleftrightarrow \tau(s)\in[a,\tau(a ) ] \quad \xi\text{-a.e.\ $ s$}.\end{aligned}\ ] ] from the assumption @xmath391 and we obtain for any @xmath395 $ ] that @xmath396 $ ] implies @xmath397 $ ] for @xmath39-almost every @xmath143
. therefore @xmath398=\int { { \bf 1}}\{\tau(s)\in[a , t]\}\ , \xi(ds ) \le \int { { \bf 1}}\{\sigma(s)\in[a , t]\}\ , \xi(ds ) = \eta[a , t].\end{aligned}\ ] ] this implies @xmath399\}= { { \bf 1}}\{\sigma(s)\in[a , t]\}\quad \xi\text{-a.e.\ $ s\in\r$}.\ ] ] therefore @xmath44 and @xmath25 coincide @xmath39-almost everywhere on @xmath400)$ ] .
now fix some @xmath401 and recall the definition of the backwards running minimum @xmath402 and the set @xmath403 .
we have seen that the complement of @xmath123 consists of countably many intervals @xmath404 as above and therefore @xmath44 and @xmath25 coincide @xmath39-almost everywhere on @xmath405\setminus c)$ ] . on the other hand , by lemma [ rest ] we have @xmath406 , as required to finish the argument .
one could define an allocation rule @xmath44 to be _ stable _ if latexmath:[\[\xi\otimes\xi\big\ { ( s , t ) \colon @xmath44 of theorem [ main3 ] does not satisfy this .
we do not know if stable allocation rules in the above sense exist , or if they are unique .
an unbiased shift @xmath2 is called _ minimal unbiased shift _ if @xmath408 and if any other unbiased shift @xmath409 such that @xmath410 and @xmath411 satisfies @xmath84 .
the following theorem provides more insight into the set of all minimal unbiased shifts . the result and its proof
are motivated by proposition 2 in @xcite .
[ pminexist ] let @xmath2 be an unbiased shift embedding the probability measure @xmath5 and such that @xmath408
. then there exists a minimal unbiased shift @xmath412 embedding @xmath5 and such that @xmath413 .
let @xmath414 denote the set of all unbiased shifts @xmath409 embedding @xmath5 and such that @xmath410 .
this is a partially ordered set , where we do not distinguish between elements that coincide @xmath26-a.s . by the hausdorff maximal principle ( see , e.g. ( * ? ? ?
* section 1.5 ) ) there is a _ maximal chain _ @xmath415 .
this is a totally ordered set that is not contained in a strictly bigger totally ordered set .
let @xmath416 then there is a sequence @xmath417 , @xmath247 , such that @xmath418 as @xmath341 .
since @xmath419 is totally ordered it is no restriction of generality to assume that the @xmath417 are decreasing @xmath26-a.s .
define @xmath420 . by construction and monotone convergence @xmath421
we also note that @xmath422 .
we claim that @xmath412 is a minimal unbiased shift embedding @xmath5 and first show that @xmath412 is an unbiased shift .
let @xmath423 , and consider continuous and bounded functions @xmath424 and @xmath425 .
let @xmath426 .
since @xmath427 for any @xmath247 we have that @xmath428 by bounded convergence the above left - hand side converges towards @xmath429 as @xmath341 .
the monotone class theorem implies that @xmath412 is an unbiased shift embedding @xmath5 .
it remains to show the minimality property of @xmath412 .
assume on the contrary that there is some unbiased shift @xmath409 embedding @xmath5 such that @xmath430 and @xmath431 .
the last two relations imply that @xmath432 by this means that @xmath433 . on the other hand , since @xmath434 , we have that @xmath435 , contradicting the maximality property of @xmath419 .
as announced in the introduction the stopping time is a minimal unbiased shift : [ pminimal ] let @xmath5 be a probability measure on @xmath6 with @xmath59 .
then @xmath2 defined by is a minimal unbiased shift .
let @xmath409 be an unbiased shift embedding @xmath5 and such that @xmath410 .
theorem [ main1 ] implies that the allocation rules @xmath436 and @xmath51 balance @xmath12 and @xmath50 . by theorem [ rstable ]
, @xmath51 is right - stable @xmath31-a.e .
the assumptions yield @xmath437 @xmath31-a.e . by theorem [ mini ]
we therefore have @xmath438 @xmath31-a.e .
this readily implies that @xmath84 .
in this section we discuss moment properties of unbiased shifts . the following two theorems together
were stated as theorem [ t14complete ] in the introduction .
[ t14lower ] suppose @xmath5 is a target distribution with @xmath59 , and @xmath75 , and the stopping time @xmath76 is an unbiased shift embedding @xmath5 .
then @xmath77 we start the proof with a reminder of the barlow - yor inequality @xcite , which states that , for any @xmath439 there exist constants @xmath440 such that , for all stopping times @xmath2 , @xmath441^{p } \le c\ , \be_0 t^{p/2}.\ ] ] hence it suffices to verify that @xmath442^{1/2}=\infty$ ] . the proof of this fact uses an argument similar to that in the proof of theorem 2 in @xcite .
let @xmath443 be the allocation rule associated with @xmath2 and set @xmath444=t\}$ ] .
then , on the one hand , @xmath445\}\,\ell^0(ds ) = \be_0\int^{t_t}_0{{\bf 1}}\{\tau(s)-s > t_t - s\}\ , \ell^0(ds)\\ & = \int^t_0\bp_0\{\tau(t_s)-t_s > t_t - t_s\}\,ds = \int^t_0\bp_0\{t\circ\theta_{t_s}>t_{t - s}\circ\theta_{t_s}\}\,ds\\ & = \int^t_0\bp_0\{t > t_s\}\,ds=\be_0[\ell^0[0,t]\wedge t],\end{aligned}\ ] ] where we have used the strong markov property at @xmath446 ( or theorem [ littletwin ] ) for the fourth step and change of variable for the second and fifth steps . on the other hand , the fact that @xmath44 balances @xmath12 and @xmath50 easily implies that @xmath447\}\,\ell^0(ds)\ge ( \ell^0[0,t_t]-\ell^\nu[0,t_t])_+.\ ] ] hence , combining these two facts , @xmath448\wedge t]\ge \be_0 ( \ell^0[0,t_t]-\ell^\nu[0,t_t])_+.\end{aligned}\ ] ] observe that @xmath449=t=\be\ell^\nu[0,t_t]$ ] , where the second identity is a trivial consequence of the second ray - knight theorem ( see theorem 2.3 in chapter xi of @xcite ) and in fact a consequence of general palm theory .
jensen s inequality and again the second ray - knight theorem imply that @xmath450)^2\le \be_0 \int ( \ell^x[0,t_1])^2\,\nu(dx ) = \int ( 1+|x|)\,\nu(dx)\end{aligned}\ ] ] which is finite by assumption .
hence , by the central limit theorem , @xmath451-\ell^\nu[0,t_t])_+=:c_1>0.\ ] ] assume for contradiction that @xmath452^{1/2}]<\infty$ ] .
since @xmath453\wedge t)\le \ell^0[0,t]^{1/2}$ ] , dominated convergence implies that @xmath454\wedge t]\to 0 $ ] as @xmath455 , contradicting the positivity of @xmath456 .
note that the unbiased shifts defined in satisfy the conditions of theorem [ t14lower ] if @xmath5 has finite mean .
the next result shows that they have nearly optimal moment properties .
[ t14 ] let @xmath5 satisfy @xmath75 , and let @xmath2 be the stopping time constructed in .
then , for all @xmath457 , @xmath458 the proof of theorem [ t14 ] uses a result similar to theorem4 ( ii ) in @xcite and theorem 2 in @xcite , which is of independent interest and may also serve as another example for theorem [ main3 ] .
we consider the ` clock ' @xmath459+\ell^\nu[0,t]=r\big\}\ ] ] and random measures @xmath39 and @xmath43 on the positive reals given by @xmath460:=\ell^0[0,u_r],\qquad \eta[0,r]:=\ell^\nu[0,u_r ] , \quad r\geq 0.\ ] ] [ t41 ] let @xmath39 and @xmath43 be defined as above and let @xmath461=\eta[0,t]\}$ ]
. then @xmath462 , but for some @xmath463 we have @xmath464 for all @xmath34 .
the proof of @xmath462 is very similar to theorem 2 in @xcite and is therefore omitted .
we prove here the upper bound for the tail asymptotics ( only this part is needed ) .
this result is similar to theorem 6 ( ii ) in @xcite , but due to the specific form of @xmath409 we can use a more direct argument . for any @xmath465 let @xmath466\leq i+1\}$ ] . using the second ray - knight theorem and the assumption on the first moment of @xmath5 as in
, we see that the sequence @xmath467 is an i.i.d .
sequence of random variables with mean one and finite variance .
define , for @xmath247 , @xmath468 let @xmath469 and fix @xmath470 .
then , for any @xmath471 , @xmath472 by a classical result of spitzer @xcite , see also ( * ? ? ?
* theorem 1a in section xii.7 ) , the first term on the above right - hand side is bounded by a constant multiple of @xmath473 . by chebyshev
s inequality we have @xmath474 = \frac{\lfloor at\rfloor}{(t-2\lfloor at\rfloor)^2 } \,\be_0[(1-y_1)^2],\ ] ] which is bounded by a constant multiple of @xmath475 .
this completes the proof . _
proof of theorem [ t14 ] . _
the variable @xmath409 , defined in proposition [ t41 ] , satisfies @xmath476+\ell^\nu[0,t]=2\ell^0[0,t].\ ] ] it remains to relate the tail behaviour of @xmath477 $ ] ( which we know ) to that of @xmath2 ( which we require ) . to this end
we observe that for @xmath478 and @xmath479 , using ( * ? ? ?
* theorem 6.10 ) , @xmath480 < 1/\theta\big\ } & = \bp_0\big\{\inf_{s > t } \frac1{\sqrt{s/\log s } } \max_{0\leq
r\leq s } |b_r| < 1/\theta\big\ } \\ & \leq \sum_{k=0}^\infty \bp_0\big\ { \frac1{\sqrt{t+k } } \max_{0\leq r\leq t+k } |b_r| < \frac{2}{\theta\sqrt{\log(t+k)}}\big\}.\end{aligned}\ ] ] by a step in the proof of chung s law of the iterated logarithm , see
e.g. ( * ? ? ?
* ( 2.1 ) ) , @xmath481 and hence we have @xmath482 < 1/\theta\big\ } \le t^{-1/4},\ ] ] for a sufficiently large constant @xmath483 . for sufficiently large @xmath384
we have @xmath484>\sqrt{t}\ } + \bp_0\big\{\inf_{s > t } \frac1{\sqrt{s/\log s}}\ell^0[0,s ] < 1/\theta \big\},\ ] ] and the right hand side in this inequality is bounded by a constant multiple of @xmath485 .
the result follows directly by integration .
next we turn to unbiased shifts @xmath2 embedding a measure @xmath486 , which need neither be stopping times , nor nonnegative .
we conjecture that any such shift satisfies @xmath487 . at the moment
we can only prove the following weaker result .
[ pmoment ] if @xmath2 is an unbiased shift embedding a probability measure @xmath486 , then @xmath488 the idea of this proof is due to alex cox .
we work under the probability measure @xmath26 . by definition of an unbiased shift @xmath489 and @xmath490
are independent brownian motions .
moreover , the pair @xmath491 is independent of @xmath4 .
assume that @xmath492 , where @xmath493 is chosen such that @xmath494 .
( if there is no such @xmath493 we find an @xmath495 such that @xmath496>0 $ ] and assume @xmath497 . ) if @xmath498 , then @xmath499 , so that @xmath500 if @xmath501 , then @xmath502 , so that @xmath503 hence @xmath504 .
it is well - known that @xmath505 and @xmath506 . since @xmath507 and @xmath508 are independent , this property transfers to @xmath409 .
it follows that @xmath509 unbiased shifts embedding @xmath19 also have bad moment properties if they are nonnegative ( or , by time - reversal , nonpositive ) but not identically zero .
the result can be compared with theorem 3(i ) in @xcite . however , the proofs are very different .
[ posmoment ] if @xmath7 is an unbiased shift such that @xmath70 and @xmath510 , then @xmath511 we assume for contradiction that @xmath512 .
define a probability measure @xmath513 on @xmath22 by setting @xmath514 for each bounded nonnegative measurable function @xmath126 . by lemma [ secondq * ]
, @xmath513 is stationary . to show that , on the invariant @xmath25-algebra @xmath236 , the process @xmath1 has the same distribution under @xmath513 as under @xmath26 , take @xmath169 and recall from theorem [ h1 ] that @xmath515 . but
@xmath516 or @xmath114 according as @xmath517 or @xmath114 , as required . by ( * ? ? ?
* theorem 2 ) we infer from this that @xmath518 with respect to the total variation norm . on the other hand , for every @xmath291 , @xmath519 implying @xmath520 for all @xmath291 , which is a contradiction .
in contrast to the two theorems above , we shall see below that unbiased shifts can have _ good _ moment properties if they can assume both signs .
[ exexpon ] we construct a nonzero unbiased shift @xmath2 embedding @xmath19 , which has @xmath521 for some @xmath522 .
let @xmath523 be the countable collection of maximal nonempty intervals @xmath524 with the property that @xmath525 for all @xmath526 and @xmath527 for some @xmath528 .
we assume that the collection is ordered such that @xmath529 for all @xmath530 .
we define an allocation rule @xmath44 by the requirement that , for @xmath531 , @xmath532 it is easy to see that @xmath44 balances @xmath12 with itself , and hence by theorem [ main1 ] , we have that @xmath533 is an unbiased shift embedding @xmath19 . moreover , we have @xmath534 where @xmath535 and @xmath536 . @xmath140 and @xmath537 are obviously independent and identically distributed , and it is easy to see that they , and hence @xmath538 , have the required moment property .
if @xmath76 is an unbiased shift such that @xmath70 and @xmath510 , then we conjecture that @xmath74 ( strengthening theorem [ posmoment ] ) , but we can not prove this without additional assumptions .
one such assumption ( covering @xmath18 defined in for @xmath291 ) is that @xmath539 for some @xmath540 such that @xmath541 is @xmath26-almost surely in the @xmath25-algebra generated by @xmath542 .
indeed , in this case we have @xmath543 where @xmath544 . by the markov property @xmath545 since @xmath546 for all @xmath547 and @xmath548 .
note that this argument does not use that @xmath2 is unbiased .
in this section we make some comments on our results and their possible extensions .
we begin with the following counterpart of theorem [ main3 ] for simple point processes .
[ tpp ] let @xmath39 and @xmath43 be invariant simple point processes on @xmath6 defined on some probability space equipped with a flow and an invariant @xmath25-finite measure .
assume that @xmath39 and @xmath43 have finite intensities and that @xmath549\mid \mathcal{i}]=\be[\eta[0,1]\mid \mathcal{i}].\end{aligned}\ ] ] then @xmath550=\eta[s , t]\},\quad s\in\r,\end{aligned}\ ] ] is an allocation rule balancing @xmath39 and @xmath43 .
theorem [ tpp ] is a one - sided ( and one - dimensional ) version of the stable matching procedure described in @xcite .
it can be proved by adapting the ideas of theorem [ main3 ] to a discrete and therefore much simpler setup . in the point process case
the allocation rule is right - stable in the sense of section [ secstability ] and it is not difficult to show that it is the unique right - stable allocation balancing @xmath39 and @xmath43 .
we conjecture that this uniqueness property also holds in the general case and therefore can be added to theorem [ rstable ] .
[ statincr ] let @xmath551 be a right - continuous real - valued stochastic process with left - hand limits and @xmath552 , defined on its canonical probability space @xmath21 .
assume that @xmath116 has stationary increments .
defining @xmath30 , @xmath28 , as before , the definition still yields a stationary measure .
assume that there is a ( measurable ) family @xmath48 , @xmath28 , of local times satisfying .
under technical assumptions on this family ( weaker than and ) theorem [ main1 ] still holds . one
does not need to assume that the local times are diffuse .
however , one can not expect the existence results ( e.g. theorem [ main2a ] ) to hold in such a general setting .
consider the setting of example [ statincr ] and assume in addition that @xmath116 is a lvy process .
there are assumptions on @xmath116 that guarantee the existence of diffuse local time measures that are perfect in the sense of and , see @xcite . under these assumptions theorems [ main2 ] , [ main2a ] ,
[ littletwin ] , [ h1 ] , and [ timep ] all hold .
also the minimality properties stated in theorems [ pminexist ] and [ pminimal ] as well as the stability assertion of theorem [ rstable ] remain true in this more general setting .
* acknowledgements : * this research started during the oberwolfach workshop ` new perspectives in stochastic geometry ' and was supported by a grant from the _
royal society_. all this support is gratefully acknowledged .
we wish to thank sergey foss and takis konstantopoulos for helpful discussions of theorem [ tpp ] and alex cox for providing the idea of the proof of theorem [ pmoment ] .
kagan , y.a . and
vere - jones , d. ( 1996 ) .
problems in the modelling and statistical analysis of earthquakes . in athens conference on
applied probability and time series , vol 1 : applied probability ( springer lecture notes in statistics 114 ) , eds c.c .
heyde , y .. v .
prohorov , r. pyke , s.t .
rachev , springer , berlin , 398425 . | let @xmath0 be a two - sided standard brownian motion . an _ unbiased shift
_ of @xmath1 is a random time @xmath2 , which is a measurable function of @xmath1 , such that @xmath3 is a brownian motion independent of @xmath4 .
we characterise unbiased shifts in terms of allocation rules balancing additive functionals of @xmath1 . for any probability distribution @xmath5 on @xmath6
we construct a stopping time @xmath7 with the above properties such that @xmath4 has distribution @xmath5 .
in particular , we show that if we travel in time according to the clock of local time we always see a two - sided brownian motion
. a crucial ingredient of our approach is a new theorem on the existence of allocation rules balancing jointly stationary diffuse random measures on @xmath6 .
we also study moment and minimality properties of unbiased shifts . _ 2000 mathematics subject classification . _ 60j65 ; 60g57 ; 60g55 . _
key words and phrases .
_ brownian motion , local time , unbiased shift , allocation rule , palm measure , random measure , skorokhod embedding . |
many gaps exist in the understanding of the post - main sequence evolution of the most massive ( @xmath640 m@xmath5 ) stars , due in large part to the small number of such stars identified as being in this evolutionary phase .
it is known that during part of the post - main sequence phase , identified as the luminous blue variable ( lbv ) stage ( conti 1984 ) , these stars lose a large amount of mass in a short time interval ( e.g. chiosi & maeder 1986 ) .
the identifying characteristics of an lbv in addition to its blue colors are ( 1 ) a mass loss rate of ( @xmath7 m@xmath5yr@xmath1 ) , ( 2 ) a low wind velocity of no more than a few hundred kms@xmath1 , ( 3 ) photometric variations of up to 2 magnitudes on time - scales ranging from months to decades , and ( 4 ) the presence of a circumstellar ( and sometimes bipolar ) nebula ( nota & clampin 1997 ) . less than a dozen confirmed lbvs are known in our galaxy . however , in recent years a number of lbv candidates ( lbvc s ) have been found .
candidates generally have both of the first two characteristics listed above and one of the other two ( for reviews see nota & lamers 1997 ) .
one recently discovered lbvc is the central star of the radio shell g79.29 + 0.46 ( higgs et al . 1994 , hereafter hwl ; and waters et al .
1996 , hereafter paper i ) .
the principal evidence for its candidacy are its high luminosity and mass loss rate ( estimated as @xmath31 @xmath8 10@xmath9 m@xmath5yr@xmath1 in paper i ) and the presence of a slightly bipolar ring nebula .
few spectra have been reported and the only spectral features in them thought to be associated with the star are several recombination lines of hydrogen and helium , most notably h@xmath2 ( hwl ) and br@xmath2 ( paper i ) .
hwl also reported the presence of diffuse interstellar lines and bands .
in this paper we present and discuss additional optical and near - infrared spectra of the central star of g79.29 + 0.46 ( hereafter g79 * ) .
the resolution of the optical spectrum is considerably higher than the spectrum published by hwl and the infrared spectra provide considerably more wavelength coverage than the spectrum in paper i. the new spectra allow us to estimate the excitation , luminosity and current activity of the star .
we conclude that although g79 * may have gone through a more intense mass loss episode in the recent past , it probably still is in the luminous blue variable stage of evolution .
a log of observations is provided in table [ obslog ] .
high resolution optical spectra of g79 * were obtained in 1995 with the utrecht echelle spectrograph ( ues ) mounted on the william herschel telescope on la palma .
the tek 5 detector ( 1024@xmath81024 pixels ) was used .
the slit length was 15 arcsec , allowing a portion of the nebula to be observed simultaneously .
insufficient time was available to obtain the blue part of the spectrum , which is very faint due to the high extinction .
two overlapping settings were chosen to cover the red part of the spectrum , with three exposures obtained at each setting .
a spectrum of a thar lamp combined with telluric lines in the spectrum of the g79 * allowed the wavelength calibration to be determined to an accuracy of 1 kms@xmath1 and provided an estimate of @xmath10 for the resolving power .
data reduction used the midas echelle package . because of the faintness of the object ( @xmath11 16 mag is about the limiting magnitude for the ues ) and because the position of the orders on the ccd was not constant during the night , considerable care had to be taken ( for details see voors 1999 ) .
the signal - to - noise ratio of the continuum in the final spectrum ( not shown ) increases from @xmath125 around 6000 to @xmath1250 around 8000 . .observing log [ cols="^,^,^",options="header " , ] the strength of the dibs are a good measure of the interstellar extinction due to diffuse matter . in several ob associations in cygnus a tight relation exists between the strength of the @xmath135797 dib and @xmath14 , ( chlewicki et al . , 1986 ) . assuming this relation for g79.29 + 0.46 , which is very close in the sky to cyg ob2
, we obtain @xmath14 = 2.1 @xmath15 0.3 .
this is much less than the values derived above .
the discrepancy may be resolved if additional reddening is present in the form of molecular clouds .
g79.29 + 0.46 is only 9 distant from the center of the obscured star - forming region dr 15 , which is thought to contain a few hot stars ( odenwald et al . ,
an ir spectrum obtained with iso - sws morris et al .
( in preparation ) shows bands of h@xmath16o and co@xmath16 ice .
it remains to be determined whether the extinction derived from the iso spectrum accounts for the difference between the two values of @xmath14 , but it seems safe to conclude that the extinction to g79.29 + 0.46 contains both diffuse and molecular contributions .
from the radial velocity of g79 * and after allowing for peculiar motions of up to 10 km s@xmath1 , we derive a kinematic distance of 2 kpc from the galactic rotation curve with an uncertainty of approximately a factor of two ( see , e.g. , hwl fig .
trams et al .
use their best fitting model atmosphere , with a luminosity of 3@xmath810@xmath17 l@xmath5 and mass loss rate of 1.3@xmath810@xmath18m@xmath5 ( considerably higher than the estimate of paper i ) to derive a distance of 1.8 kpc .
it is virtually certain that the distance exceeds 0.7 kpc since no interstellar clouds closer than that distance are known which could cause the large observed extinction .
the dr 15 complex , which we have suggested is responsible for some of the extinction to g79.29 + 0.46 , is 1 kpc distant .
we conclude that the distance to g79.29 + 0.46 is in the range 1 - 4 kpc , with a most likely value of @xmath32 kpc .
= 8.8 cm = 8.0 cm
line profiles , averaged along the 15@xmath19 slit , excluding the regions influenced by stellar radiation , are shown in fig .
the profile of h@xmath2 was averaged over a slightly smaller region along the slit than [ nii]@xmath136583 , because of contamination by the bright central peak of stellar h@xmath2 .
the widths and the shapes of the broad components of the [ nii ] and the h@xmath2 nebular profiles agree very well , h@xmath2 being 2.3 times stronger .
however , the narrow nebular component seen in h@xmath2 is missing in [ nii ] .
clearly the narrow and broad components originate from different regions .
the broad profiles of both lines are well fitted by a gaussian with fwhm 45@xmath153 kms@xmath1 .
the fwzi is about 70 kms@xmath1 .
the central wavelength of the gaussian is @xmath20 2 kms@xmath1 and agrees very well with the radial velocity of the central star , @xmath21 kms@xmath1 .
more recent observations ( voors 1999 ) show that both the broad and narrow components extend well beyond the ring nebula , implying that they are interstellar in origin and probably associated with emission from the cygnus - x region .
even at the minimum distance of 1 kpc the luminosity of g79 * is 1@xmath810@xmath17 l@xmath5 ; at the maximum distance of 4 kpc it approaches that of some of the most luminous stars known .
its temperature and spectral type are not well defined , but the presence of neutral species in emission and absorption as well as n ii lines in absorption suggests it is similar to a mid b supergiant .
the high luminosity , high mass loss rate , moderate wind velocity of 110 km s@xmath1 , and presence of a circumstellar nebula all are hallmarks of an lbv .
the massive dusty ring nebula surrounding g79 * indicates a period of much higher mass loss in the recent past ( paper i ) , possibly during a red supergiant phase .
the star may evolve into a wolf - rayet star .
variability has not been demonstrated , but given the paucity of observations this is not surprising .
more detailed spectra , such as those recently obtained for the pistol star ( najarro et al .
1998 ) and fmm362 ( geballe et al . 2000 ) , are needed for g79 * in order to allow quantitative analysis , including a direct estimate of the mass loss rate , and more careful comparison with other lbvc s and lbv s .
we thank p. w. morris for permission to reproduce the @xmath22 band spectra of ag car and wra 751 , m. van kerkwijk for the spectrum of sgr 1806 - 20 , and b. wolf for helpful comments .
we also thank the @xmath23 group in utrecht for useful discussions and the referee for helpful comments .
lbfmw acknowledges financial support from a nwo `` pionier '' grant .
f. n. acknowledges dgycit grants pb96 - 0883 and esp98 - 1351 ukirt is operated by the joint astronomy centre on behalf of the u.k . particle physics and astronomy research council . we have made use of the simbad database , operated at the centre de donnes astronomiques , strasbourg , france .
bernat a.p . ,
lambert d.l . , 1978 ,
pasp 90 , 520 bowen i.s . , 1947 ,
pasp 59 , 196 chiosi c. , maeder a. , 1986 , ara&a 24 , 329 chlewicki g. , van der zwet g.p . , van ijzendoorn l.j . , et al . , 1986 ,
apj 305 , 455 conti p.s . , 1984 ,
_ observational test of stellar evolution theory _ , iau symposium 105 , eds .
a. meader , a renzini , p. 233 .
fuchs y. , mirabel f. , chaty s. , et al .
, 1999 , a&a 350 , 891 geballe t.r .
, najarro f. , figer d.f 2000 , apj in press grandi s.a .
1975 , apj 196 , 465 hamann f. , simon m. , 1987 , apj 327 , 876 higgs l.a .
, wendker h.j . ,
landecker t.l . , 1994 , a&a 291 , 295 ( hwl ) kurucz r.l . , 1979 ,
apjs 40 , 1 lopes d.f .
, damineli neto a. , de freitas pacheco j.a . , 1992 ,
a&a 261 , 482 mcgregor p.j . , hyland a.r .
, hillier d.j . , 1988 , apj 324 , 1071 morris p.w . , eenens p.r.j . , hanson m.h . , et al . , 1996 ,
apj 470 , 597 mountain , c.m . , robertson , d. , lee , t.j . ,
wade , r. 1990 , proc .
spie 1235 , 25 mnch g. , 1948 , apj 108 , 116 najarro f. , hillier d.j .
, kudritzki r.p . ,
1994 , a&a 285 , 573 najarro f. , hillier d.j .
, figer d.f .
, geballe t.r .
1998 , in _ the central parsecs , galactic center workshop 1998 _ , eds .
h. falcke , a. cotera , w. huschl , f. melia , and m rieke , in press .
nota a. , clampin m. , 1997 , in : _ luminous blue variables : massive stars in transition _ , asp conference series , vol .
120 , eds .
a. nota and h. lamers nota a. , lamers h. 1997 , _ luminous blue variables : massive stars in transition _ , asp conference series , vol . 120 , eds .
a. nota and h. lamers odenwald s.f . ,
campbell m.f . , shivanandan k. , et al . , 1990 , aj 99 , 288 thompson r.i . , boroson t.a . , 1977 ,
apj 216 , l75 torres - dodgen a.v . , tapia m. , carroll m. , 1991 , mnras 249 , 1 trams n.r .
, van tuyll c.i . ,
voors r.h.m .
, et al . , 1999 , _
variable and non - spherical stellar winds in luminous hot stars _ , iau colloquium 169 , eds .
b. wolf , a. fullerton and o. stahl van kerkwijk m.h . ,
kulkarni s.r . ,
matthews k. , neugebauer g. , 1995 , apj 444 , l33 voors r.h.m . 1999 , thesis , university of utrecht waters l.b.f.m .
, izumiura h. , zaal p. , et al .
, a&a 313 , 866 ( paper i ) | we report optical and near - infrared spectra of the central star of the radio source g79.29 + 0.46 , a candidate luminous blue variable . the spectra contain numerous narrow ( fwhm @xmath0 100 kms@xmath1 ) emission lines of which the low - lying hydrogen lines are the strongest , and resemble spectra of other lbvc s and b[e ] supergiants .
a few prominent infrared lines are unidentified .
the terminal wind speed is determined from h@xmath2 to be 110 kms@xmath1 .
the strength of h@xmath2 implies the presence of a very dense wind .
extended emission from h@xmath2 and [ nii ] was detected but appears to be associated with the cygnus x region rather than the radio source .
both diffuse interstellar bands and interstellar absorption lines are present in the optical spectrum of the central star , suggesting that there are both diffuse and molecular cloud components to the extinction and implying a minimum distance of 1 kpc and minimum luminosity of @xmath310@xmath4l@xmath5 for the star .
the new spectra and their analysis indicate a low excitation , dense , and slowly expanding wind and support the lbvc classification . |
the quest for a natural way to break electroweak symmetry has long been the central motivation for constructing theories for physics beyond the standard model ( sm ) at the tev scale .
however , the same logic applied to the even more severe fine - tuning problem associated with the cosmological constant would have predicted new physics near @xmath6 ev , which we have no evidence for .
this suggests the possibility that our notions of naturalness are misleading , and that other fine - tuning mechanisms may be at work in nature . eliminating the use of naturalness as a guiding principle for model - building allows for some drastic changes to particle physics lore .
arkani - hamed and dimopoulos have recently argued for a theory with `` split '' supersymmetry @xcite ( also @xcite ) . in this model
the `` structure '' @xcite and `` atomic '' @xcite principles were used to explain the smallness of uv sensitive parameters ; namely the cosmological constant and the higgs vacuum expectation value .
chiral symmetries keep all fermions light and a single fine tuning does the same for one higgs scalar , while all other scalars are at the high susy breaking scale .
this framework , while preserving the successes of the mssm such as gauge coupling unification @xcite , also salvages some of its difficulties , giving a simple explanation for the absence of fcncs and cp violation ; the non - discovery of superpartners , and a light higgs ( see also @xcite ) , while simultaneously solving a variety of cosmological difficulties associated with the gravitino and moduli .
an important quantitative prediction of the minimal model is the mass of the higgs , which has been computed to lie between 120 and 170 gev @xcite . as pointed out in @xcite , however , this prediction is sensitive to physics above @xmath0 and the presence of new matter beneath @xmath0 . in this paper
we explore the bounds on the higgs mass in these more general versions of split susy which continue to conform to the essential philosophy of the model , in order to provide a falsifiable test of theories that are built on these principles .
we analyse these models at one loop and examine the limits on the boundary value of the higgs quartic coupling at the susy breaking scale . in section [ sec : minmass ] we show that there exist mechanisms by which this can be made negative . how negative is determined by requiring stability of the sm vacuum . next , in analogy with the triviality bound in the sm , we use the requirement of perturbativity to the cutoff to put an upper limit on this boundary value .
we take a totally agnostic viewpoint , assuming that there is some unknown physics in the region between @xmath0 and the unification scale , @xmath7 , that effectively decouples the higgs quartic from the electroweak gauge couplings .
this could include new d - terms or f - terms such as in the nmssm .
we then see how far this bound can be pushed in the large @xmath8 limit by : * varying @xmath0 and adding n su(5 ) @xmath9s at the weak scale , maximizing n at each @xmath0 in order to maintain perturbative gauge coupling unification * adding yukawa couplings between the higgs and new fermions that come in complete multiplets of su(5 ) we consider each of these in turn in section [ sec : maxmass ] , rge evolving the couplings down to the weak scale , where we can calculate the physical higgs mass .
this gives us a firm prediction of this class of models , based on a minimal number of reasonable assumptions .
we conclude in section [ sec : conclusion ] , with some discussion of possible interesting extensions of this work .
we begin by discussing how we can _ decrease _ the higgs quartic coupling at @xmath0 .
suppose that in the theory above this scale , we have an additional gauge singlet scalar field @xmath10 that picks up a mass term @xmath11 from susy breaking .
like the higgses , it has an @xmath12-charge of zero so that the superpotential term @xmath13 is forbidden .
however , the following @xmath14-term is permitted @xmath15 integrating out @xmath16 induces a term of the form @xmath17 , and our effective theory beneath @xmath0 now contains a negative contribution to the higgs quartic coupling of @xmath18 this can exceed the usual gauge d - term contribution proportional to @xmath19 , giving rise to the possibility that the sm vacuum state is not the true vacuum of the theory and allowing for the eventual decay of our vacuum to the true one by bubble nucleation .
we use the methods in @xcite to calculate the decay rate per unit volume by approximating this to a pure @xmath20 theory ( see e.g @xcite ) : @xmath21 where @xmath12 is the size of the bubble by which this process takes place . in practice
this integral is just dominated by the scale at which the integrand is maximized .
as we will see , @xmath22 will turn out to be largest for @xmath23 , so the rate is dominated by @xmath24 .
we musy have @xmath25 for this decay not to have occurred already , and solving this equation allows us to bound @xmath1 : @xmath26 saturating this bound we find the higgs mass shown in figure [ fig : minmass ] for two different values of @xmath8 . note the consistency of this lower limit on the higgs mass for @xmath27 gev with the lep - ii bound . we do not wish to lower @xmath0 any further since this will bring back the problems associated with the mssm that we were trying to alleviate .
adding matter at the weak scale does not disrupt one - loop gauge coupling unification as long as this matter can be grouped into complete multiplets of su(5 ) .
it does , however , increase the value of the couplings at @xmath7 since it contributes ( the same ) positive quantity to each rge .
hence we need to ensure that we do not add so many particles that the couplings become non - perturbative before unification takes place .
this limits n@xmath286 for susy breaking scale around @xmath29 gev , for example .
we use these values to show in figure [ fig : higgsmassvslambda ] how the low energy physical higgs mass changes with @xmath1 .
note that there is very little gain in mass for @xmath30 at high energies .
next we take the higgs quartic close to its perturbative limit at the cutoff and vary @xmath0 , maximizing n at each scale to find an upper bound on the higgs mass .
we expect the bound to increase substantially with n ; however as can be seen in figure [ fig : higgsmassnplusnbar ] this is not exactly the case ; at least for low cutoff there is no significant difference between the higgs mass in the theory with and without extra @xmath9s .
this seems rather counterintuitive since increasing all gauge couplings increases the boundary values of the higgs - gaugino yukawa couplings ( @xmath31s ) at the susy breaking scale , which in turn should feed into the higgs mass .
the reality of the story for @xmath32 is rather more complicated , however , and intimately involves three other couplings , @xmath33 , @xmath34 and @xmath35 , in the terms @xmath36 as well as @xmath37 .
figure [ fig : contrib ] contains a graph of each of these contributions to the @xmath32 rge .
notice that for low cutoff , the running of @xmath32 is dominated by itself and , since we have decoupled its boundary condition from the electroweak gauge couplings , it relies on none of the quantities that change on adding @xmath9s . even if @xmath32 was not the dominant coupling , increasing @xmath34 would actually decrease the higgs mass since it contributes via the positive @xmath38 term , partly undoing the effects of increasing @xmath35 and decreasing @xmath33 . for high cutoff on the other hand ,
the quartic runs down enough so that not only do the weak gauge coupling and gaugino yukawa start playing a much bigger part in determining its running ( although @xmath33 still does not ) , but the @xmath39 term actually becomes the dominant one .
increasing n therefore increases the higgs mass through larger @xmath34 as well as @xmath35 . due to this property of the rges in this theory
our method of adding @xmath9s will not increase the higgs mass much beyond 250 gev . as a check we can see in figure [ fig : higgsmassnplusnbarlowbc ]
similar results using the boundary conditions for the mssm and the new fat higgs @xcite respectively .
since these are much smaller than the perturbative limit used in the previous example , the results with and without new gut multiplets start to differ at a lower energy , confirming that @xmath32 now does not need to run down as much before the other couplings start becoming important . ,
although in reality the latter two need to be separated slightly in order for us to legitimately use the weak limit bound described in the paper . recall that this model already contains 4 @xmath9s of its own above the confinement scale - we need to take these into account in our perturbative unification constraint . ]
returning to the effect of the top yukawa , we saw that this was negligible for all cutoffs since its value at low energies is fixed , and so it never becomes comparable in size to the other terms in the @xmath32 rge .
this observation inspires an alternative approach in which n vector - like gut multiplets of quarks and leptons ( @xmath40s ) are added and coupled to the higgs in the usual fashion .
these have vector - like masses , and the new top - like yukawa coupling , @xmath41 , plays exactly the same role as @xmath33 , except that it is not fixed at low energies and therefore can be more instrumental in determining the higgs mass .
first we examine how the low energy value of this additional yukawa changes with boundary condition at the cutoff ( see figure [ fig : ytvsyt ] ) . as with @xmath32 , it is relatively insensitive to its boundary condition for @xmath422 .
for such a large yukawa the higgs mass is also independent of the boundary value of @xmath32 , suggesting that its rge is now controlled by the new yukawa as intended .
now we can analyse the higgs mass with changing cutoff in a similar manner to the ( @xmath43 ) case from earlier , maximizing n at each energy and comparing with our previous results in figure [ fig : maxhiggsmassxtend ] .
if we examine the different contributions to the running of @xmath32 as before , we see that the larger @xmath44 indeed plays a leading role along the entire cutoff spectrum , giving rise to an overall increase in higgs mass with n , to a maximum of about 400 gev .
it is possible that , as it stands , this modification with a dominant top - type yukawa coupling will give rise to undesirably large oblique parameters , especially a large positive contribution to t. this issue can be resolved in two ways , neither of which significantly affect our results .
firstly we could increase the masses of these new fermions , which suppresses the higher - dimensional operators contributing to precision electroweak measurements like the t parameter .
alternatively , we could impose an approximate custodial @xmath45 symmetry in the new matter sector , fixing the same boundary conditions on the bottom - type yukawa as the top - type .
either way , these models can be made consistent with current experimental data .
we see that it is possible to give limits on the physical higgs mass of @xmath46 in arbitrary extensions of split susy , including additional gut matter multiplets / gut singlets that are otherwise decoupled from the sm . adding new yukawa couplings
pushes the upper bound to around @xmath47 gev .
it is interesting that the _ lower _ limit on the higgs mass has already been excluded by lepii .
however , we see that the possible hint for a 115 gev higgs can be accomodated in split susy models with @xmath48 gev and a slightly negative higgs quartic coupling near this scale .
further analysis is required to determine how robust these predictions are to adding two - loop contributions to the running ( see @xcite for rges ) . from the corresponding sm results @xcite
, we expect that this will decrease our limits by of the order of @xmath49 gev
. it would also be interesting to see how nmssm - like boundary conditions for the higgs quartic , or even different uv completions of the nmssm , such as @xcite , affect these bounds .
many thanks to nima arkani - hamed for suggesting the possibility of a negative quartic and also for numerous invaluable discussions on the vacuum stability bound among other interesting things . n. arkani - hamed and s. dimopoulos , arxiv : hep - th/0405159 . g. f. giudice and a. romanino , arxiv : hep - ph/0406088 . s. weinberg , phys .
lett . * 59 * , 2607 ( 1987 ) .
v. agrawal , s. m. barr , j. f. donoghue and d. seckel , phys .
d * 57 * , 5480 ( 1998 ) [ arxiv : hep - ph/9707380 ] .
s. dimopoulos and h. georgi , nucl .
b * 193 * , 150 ( 1981 ) .
s. dimopoulos , s. raby and f. wilczek , phys .
d * 24 * , 1681 ( 1981 ) .
w. j. marciano and g. senjanovic , phys .
d * 25 * , 3092 ( 1982 ) .
m. b. einhorn and d. r. t. jones , nucl .
b * 196 * , 475 ( 1982 ) .
l. e. ibanez and g. g. ross , phys .
b * 105 * , 439 ( 1981 ) .
n. sakai , z. phys .
c * 11 * , 153 ( 1981 ) .
p. langacker and n. polonsky , phys
. rev .
d * 52 * , 3081 ( 1995 ) [ arxiv : hep - ph/9503214 ] .
j. d. wells , arxiv : hep - ph/0306127 .
a. arvanitaki , c. davis , p. w. graham and j. g. wacker , arxiv : hep - ph/0406034 . s. coleman , `` aspects of symmetry : selected erice lectures '' .
k. m. lee and e. j. weinberg , nucl .
b * 267 * , 181 ( 1986 ) .
g. isidori , g. ridolfi and a. strumia , nucl .
b * 609 * , 387 ( 2001 ) [ arxiv : hep - ph/0104016 ] .
s. chang , c. kilic and r. mahbubani , arxiv : hep - ph/0405267 .
n. k. falck , z. phys .
c * 30 * , 247 ( 1986 ) .
m. e. machacek and m. t. vaughn , nucl .
b * 249 * , 70 ( 1985 ) .
m. e. machacek and m. t. vaughn , nucl .
b * 236 * , 221 ( 1984 ) .
m. e. machacek and m. t. vaughn , nucl .
b * 222 * , 83 ( 1983 ) .
g. altarelli and g. isidori , phys .
b * 337 * , 141 ( 1994 ) .
r. harnik , g. d. kribs , d. t. larson and h. murayama , arxiv : hep - ph/0311349 . | we investigate the limits on the higgs mass in variations of split supersymmetry , where the boundary value of the higgs quartic coupling at the susy breaking scale ( @xmath0 ) is allowed to deviate from its value in the minimal model of arkani - hamed and dimopoulos .
we show that it is possible for @xmath1 to be negative and use vacuum stability to put a lower bound on this coupling , and hence on the mass of the physical higgs .
we also use the requirement of perturbativity of all couplings up to the cutoff to determine an upper limit for the higgs mass in models which are further modified by additional matter content . for @xmath2 gev
we find @xmath3 gev @xmath4 gev if the new matter is not coupled to any standard model field ; and @xmath3 gev @xmath5 gev if it has yukawa couplings to the higgs . |
the small magellanic cloud ( smc ) and the large magellanic cloud ( lmc ) are gas rich irregular galaxies and satellites of the milky way ( mw ) . given the proximity to the galaxy and their high galactic latitude , the magellanic clouds ( mcs ) are perfect laboratories for extragalactic studies , since the light emitted from them is not much attenuated by the dust present in the galactic disk .
these objects are highly studied , motivated by the fact that , through understanding them , we can also better comprehend , for instance , galaxy evolution from high redshifts to now ; the structure of irregular dwarf galaxies ; the interstellar medium ( ism ) content at the external parts of huge spiral galaxies ; star formation at low metalicity environments ; dwarf galaxies in interaction ; and the relation of satellite galaxies with their hosts .
the mcs are low metalicity galaxies @xmath12 and @xmath13 for the lmc and smc , respectively @xcite .
this peculiarity makes their ism particularly different from the one of our galaxy . among the major differences , we can point out the high gas - to - dust ratio @xcite and the submm excess emission @xcite .
the smc consists of two main features : bar and wing .
the bar component is the main body of the smc and corresponds to its northern part , which has a position angle ( pa ) of @xmath14 .
the wind is located @xmath15 east of the bar and corresponds to its southern part @xcite . the dynamical evolution of the smc lmc
mw system seems to have generated two interesting structures the magellanic stream and the magellanic bridge .
these structures are well illustrated in the hi column density map ( figure 2 ) presented by @xcite .
the magellanic stream is a gas tail covering over @xmath16 , also going through the galactic south pole .
one scenario says that this gas might have been pulled out of the smc and lmc due to tidal interaction between the mcs and the mw @xcite . on the other side ,
the magellanic bridge seems to have been generated due to the same effect , but between the smc and the lmc @xcite .
more recently , high precision proper motions from the _ hubble space telescope _ showed that the mcs are moving @xmath17 km / s faster than earlier estimates @xcite .
these high velocities favor a scenario where the mcs are in their first infall toward the mw @xcite . in this scenario
, @xcite found that the magellanic stream may have been formed by lmc tides on the smc before the system was accreted by the mw .
@xcite studied several orbits consistent with the recent proper motions .
they assumed an orbital history where the mcs became a strongly interacting pair just recently . in this picture ,
their first close encounter provided enough tidal forces to disrupt the smc s disk and create the magellanic stream .
in contrast to the magellanic stream , the magellanic bridge has a young stellar population @xcite , besides being composed of gas and dust @xcite .
the study of gas - to - dust ratio of the magellanic bridge confirms the hypothesis that the magellanic bridge is formed of material that was stripped from the smc @xcite .
moreover it was found that at the magellanic bridge the gas - to - dust ratio is about @xmath18 times higher than in the galaxy and about two times higher than the value expected by its metalicity , indicating that some of the dust in that region must have been destroyed .
it is not well known whether the mcs were formed as a binary system or were dynamically coupled about @xmath19 gyr ago @xcite .
the kinematic history reconstruction of the mcs suggests that their last closest approach occurred about @xmath20 gyr ago , when they came within 27 kpc distance of each other , a likely progenitor event for the construction of the bridge .
another possible result of this interaction is the stretching in northeastern and wing sections toward the lmc @xcite .
the last interaction was also likely responsible for the bimodality of the smc along its depth @xcite , a feature that is corroborated by the velocity components in hi data , which are @xmath21 km s@xmath22 in difference @xcite .
distances to cepheids show that the high velocity component is located further in distance , while the low velocity component is closer .
both components appear to be expanding at around @xmath23 km s@xmath22 .
the low velocity component itself is further subdivided into two other groups , one extending from the center to the northeast ( ne ) at a location of @xmath24 kpc , the other , extending from the bar to the southwest ( sw ) at a location of @xmath25 kpc .
the bar is the most extended region ( @xmath26@xmath27 kpc ) and therefore the hi profiles in that direction are very complex @xcite .
the study of the hi component at the smc shows that it rotates @xcite , indicating that the smc contains a disk component .
distances to cepheids also show that the two gas components are about @xmath28 kpc apart and have a depth of @xmath29 kpc . a double component , relating to the distances , is also observed in cepheids located in the bridge , consistent with the two gaseous components . besides the bridge s bifurcation
, it is also , on average , closer to us than the smc @xcite .
recently , red clump ( rc ) and rr lyrae stars ( rrls ) were used to estimate the smc s line of sight depth .
@xcite used these stars to study the three - dimensional structure of smc and found that smc has an average depth of @xmath30 kpc .
they concluded also that the ne region is closer to us and an elongation along the ne
sw direction is seen .
@xcite used rc stars to study the eastern and western regions of smc and found that the first has a larger depth ( @xmath31 kpc ) , while the second has a much shallower depth ( @xmath10 kpc ) .
some of their eastern fields showed a bimodality , with one component located at @xmath32 kpc and the other at @xmath33 kpc away from us , leading them to conclude that the closer stellar component was stripped from smc during the past interaction .
the study of the stellar distribution of smc shows that it has a spheroidal or slightly ellipsoidal structure @xcite .
a stellar disk structure is also observed ; however , there is a discrepancy between the gaseous and stellar disk parameters @xcite .
the overall picture shows that smc possesses a stellar and a gaseous disk component , as well as a spheroidal stellar component .
the eastern region and magellanic bridge are closer in distance to us and show a bimodal structure .
these regions are also the ones that were probably most affected by the past interaction , which explains why they may be approaching lmc , which is located at @xmath34 kpc from us .
the line of sight depth estimations show a relative high dispersion from region to region , with the eastern side being deeper than the western side .
the magnetic field of smc was previously studied using several techniques optical interstellar polarization , faraday rotation , and synchrotron emission .
the interstellar polarization is caused by the alignment of the dust grains angular momentum with a local magnetic field .
the unpolarized light emitted by the stars become polarized due to the dichroic extinction by dust grains . therefore ,
optical polarization vectors trace the magnetic field projected in the plane of the sky , and a polarization map can be interpreted as a magnetic field map in that direction .
the phenomenon known as faraday rotation happens , when radiation travels through an ionized medium with a local magnetic field . the light plane of polarization rotates , if the local magnetic field has a parallel component to the direction of propagation of the light .
the incident light is decomposed in to two parts circularly polarized with opposite rotations .
the synchrotron emission is produced when charged particles are accelerated by a magnetic field .
the radiation emitted by these particles is polarized , because the acceleration is not isotropic .
@xcite used optical polarization data from @xcite and obtained an ordered magnetic field projected in the plane of the sky of @xmath35 g and a turbulent component of @xmath36 g . @xcite
analyzed optical polarization data from @xcite and estimated an ordered sky - projected magnetic field of @xmath37 g and a turbulent field of @xmath38 g .
@xcite also estimated the line of sight ordered magnetic field , through faraday rotation , using data from the australia telescope compact array ( atca ) .
they obtained @xmath39 g , with the field pointing away from us .
in addition , there are several studies based on the radio continuum synchrotron emission @xcite . using synchrotron emission and assuming energy equipartition between the cosmic rays and magnetic fields , @xcite estimated the smc s total magnetic field to be about @xmath40 g .
photoelectric polarization data from @xcite , @xcite , and @xcite suggested that smc has an ordered magnetic field aligned with the magellanic bridge .
polarized synchrotron emission , despite its weak nature in the smc , also confirmed such an alignment @xcite .
besides the alignment with the bridge , @xcite observed an alignment of the ordered magnetic field with the smc s bar .
the magnetic field may have played an important role in shaping the bridge , since the interstellar magnetic field of galaxies can be strong enough to influence the dynamics of their gas @xcite .
therefore , this work aims to study the structure of the magnetic field using ccd data , which is more precise than photoelectric .
we concentrated our study on the ne and wing regions of the smc , since these regions may have been more affected by the last interaction with lmc .
we also have some data of the magellanic bridge , from the part closer to the smc , a structure probably formed due the interaction of the smc lmc system .
the paper is structured as follows . in section [ sec : data ] , we describe the observational data , the reduction process , and how we built the catalog .
the estimates for the foreground polarization are presented in section [ sec : foreground ] .
we analyze the polarization patterns for each field observed in section [ sec : ppatterns ] .
the magnetic field geometry is discussed in section [ sec : bgeometry ] .
the alignment of the ordered magnetic field with respect to the magellanic bridge is discussed in section [ sec : alignment ] .
we estimate the magnetic field strength in section [ sec : bintensity ] .
finally , in section [ sec : conclusion ] , we summarize our results and the main conclusions .
the observational data were taken at cerro tololo inter - american observatory ( ctio ) over the course of five nights in 1992 november 1317 , applying the optical polarization technique and using the @xmath41 m telescope .
the instrumental setup consisted of a half - wave retarder plate followed by a calcite , @xmath42 filter , and the detector .
this setup gives us , for each star , two images with orthogonal polarizations .
at least two sets of images , taken with the half - wave retarder plate positioned at different angles , are necessary for each object / field , to obtain the linear stokes parameters @xmath43 and @xmath44 @xcite .
some standard stars polarization were measured using 16 images ( see table [ tab : standards ] ) .
further details about the instrumental technique can be found in @xcite .
a tek1k-1 1k x 1k ccd , with a plate scale of @xmath45/pixel , read noise of @xmath46 e@xmath47 rms , and gain of @xmath48 e@xmath47/adu was used . during the observational run , bias and flat - field images
were taken to correct instrumental noise .
one non - polarized standard star was observed , in order to check whether there was any instrumental polarization or not . also , three polarized standard stars were observed , in order to get the conversion for the polarization angles ( @xmath49 ) into the equatorial system .
twenty - eight fields of 8 x 8 arcmin size , situated at the ne and wing sections of the smc as well as the magellanic bridge , were observed .
we decided to concentrate our data efforts on these regions for the reasons mentioned previously , relating to the last interaction with the lmc , about @xmath20 gyr ago @xcite .
the date of the observation , number of positions the retarder plate was rotated , and integration time for the standard stars are presented in table [ tab : standards ] .
the coordinates and date of the observation for the 28 fields are presented in table [ tab : fields ] . for the fields ,
the half - wave retarder plate was rotated four times and the integration time was @xmath50 s per frame .
figure [ fig : fields ] shows the positions of these fields .
, c. m. , & dufour , r. j. 1998 , in revista mexicana de astronomia y astrofisica , vol .
27 , vol . 7 , revista mexicana de astronomia y astrofisica conference series , ed .
r. j. dufour & s. torres - peimbert ( mexico , d.f . : instituto de astronomia , universidad nacional autonoma de mexico ) , 202 , a. m. , de oliveira , c. m. , carciofi , a. , & et al , .
2012 , in american institute of physics conference series , vol .
1429 , american institute of physics conference series , ed . j. l. hoffman , j. bjorkman , & b. whitney ( stellar polarimetry : from birth to death .
aip conference proceedings ) , 244247 , a. m. , loiseau , n. , rodrigues , c. v. , & piirola , v. 1990 , in iau symposium , vol .
140 , galactic and intergalactic magnetic fields , ed .
r. beck , r. wielebinski , & p. p. kronberg ( international astronomical union ) , 255 , a. m. , rodrigues , c. v. , margoniner , v. e. , pereyra , a. , & heathcote , s. 1996 , in astronomical society of the pacific conference series , vol .
97 , polarimetry of the interstellar medium , ed . w. g. roberge & d. c. b. whittet ( san francisco : astronomical society of the pacific ) , 118 | we present a new optical polarimetric catalog for the small magellanic cloud ( smc ) .
it contains a total of 7207 stars , located in the northeast ( ne ) and wing sections of the smc and part of the magellanic bridge .
this new catalog is a significant improvement compared to previous polarimetric catalogs for the smc .
we used it to study the sky - projected interstellar magnetic field structure of the smc .
three trends were observed for the ordered magnetic field direction at position angles ( pas ) of ( @xmath0 ) , ( @xmath1 ) , and ( @xmath2 ) .
our results suggest the existence of an ordered magnetic field aligned with the magellanic bridge direction and smc s bar in the ne region , which have pas roughly at @xmath3 and @xmath4 , respectively . however , the overall magnetic field structure is fairly complex .
the trends at @xmath5 and @xmath6 may be correlated with the smc s bimodal structure , observed in cepheids distances and hi velocities .
we derived a value of @xmath7 g for the ordered sky - projected magnetic field , and @xmath8 g for the turbulent magnetic field .
this estimate of @xmath9 is significantly larger ( by a factor of @xmath10 ) than the line of sight field derived from faraday rotation observations , suggesting that most of the ordered field component is on the plane of the sky . a turbulent magnetic field stronger than the ordered field agrees with observed estimates for other irregular and spiral galaxies . for the smc
the @xmath11 ratio is closer to what is observed for our galaxy than other irregular dwarf galaxies . |
recent advances in quantitative measurements of gene expression at the single - cell level @xcite have brought new insight on the importance of stochastic fluctuations in genetic circuits @xcite .
the role of fluctuations is enhanced due to the discrete nature of the transcription factors and their binding sites , which may appear in low copy numbers @xcite . as a result
, populations of genetically identical cells may show significant variability .
stochastic behavior may invoke oscillations @xcite and spatio - temporal patterns @xcite , which are unaccounted for by macroscopic chemical rate equations .
genetic circuits with feedback mechanisms may exhibit bistability , namely , two distinct stable states which can be switched by an external signal @xcite .
a low rate of spontaneous switching events may also take place . to qualify as a switch
, this rate must be much lower than the rates of the relevant processes in the cell , namely transcription , translation , and degradation of transcription factors .
genetic switches , such as the phage @xmath0 switch , give rise to different cell fates @xcite . in this switch , @xmath0 phages infect _ e. coli _ bacteria and can exist in two exclusive states , one called lysogeny and the other called lysis . when the phage enters its host , it integrates itself into the host s dna and is duplicated by cell division .
it codes for proteins that can identify stress in the host cell . in case of stress ,
the phage transforms into the lysis state . in this state
, it kills the host cell , using its dna to produce many copies of the phage , which are released and later infect other cells .
other switch circuits exist in the metabolic systems of cells .
these switches determine which type of sugar the cell will digest @xcite .
the genetic switch may also serve as a memory unit of the cell , and help determine its fate during cell differentiation .
recent advances enable the construction of genetic circuits with desired properties , that are determined by the network architecture .
these networks are constructed from available components , namely genes and promoters .
they do not require the manipulation of the structure of proteins and other regulatory elements at the molecular level .
these genes and promoters are often inserted into plasmids rather than on the chromosome . a synthetic toggle switch , that consists of two repressible promoters with mutual negative regulation , was constructed in _
e. coli _ and the conditions for bistability were examined @xcite .
the switching between its two states was demonstrated using chemical and thermal induction .
more recently , such circuit was found to exist in a natural system in which two mutual repressors regulate the differentiation of myeloid progenitors into either macrophages or neutrophils @xcite . in this paper
we analyze the genetic toggle switch using deterministic and stochastic methods . in this simple genetic circuit , two proteins , @xmath1 and @xmath2 ,
negatively regulate each other s synthesis .
the regulation is performed at the transcription level , namely the production of protein @xmath1 is negatively regulated by protein @xmath2 , through binding of @xmath3 copies of @xmath2 to the @xmath1 promoter ( and vice versa ) .
this process can be modeled by a hill function , which reduces the production rate of @xmath1 by a factor of @xmath4^n$ ] , where @xmath5 $ ] is the concentration of @xmath2 proteins in the cell , @xmath6 is a parameter and @xmath3 is the hill coefficient @xcite . in case that @xmath7 , the binding of a single protein is sufficient in order to perform the negative regulation , while for @xmath8 the _ cooperative binding _ of two or more proteins is required .
one may expect this circuit to function as a switch , with two stable states , one dominated by @xmath1 proteins and the other dominated by @xmath2 proteins .
when the population of @xmath1 proteins is larger than the population of @xmath2 proteins , the @xmath1 proteins suppress the production of @xmath2 proteins . under these conditions , the production of @xmath1 proteins will not be suppressed much by the small @xmath2 population .
therefore , the system approaches a state rich in @xmath1 proteins and poor in @xmath2 proteins .
similarly , the system may approach a state rich in @xmath2 proteins and poor in @xmath1 proteins . to qualify as a switch
, the system should be bistable . in the deterministic description ,
bistability is defined as the existence of two stable steady state solutions of the rate equations .
this description does not account for the possibility of spontaneous transitions between the two states . in the stochastic description
, spontaneous transitions do take place .
therefore , the condition for bistability is that the rate of spontaneous switching events ( due to random fluctuations rather than an external signal ) is much lower than the rates of all other relevant processes in the system .
rate equations provide the average concentrations of @xmath1 and @xmath2 proteins in a population of cells . in these equations
, bistability emerges at a bifurcation point , where two stable states emerge .
rate equations do not include fluctuations and do not account for the possibility of spontaneous transitions between the two states .
the master equation provides the probability distribution of the populations of @xmath1 and @xmath2 proteins .
the two bistable states appear as two distinct peaks in this distribution .
monte carlo simulations enable to follow the fluctuations in a single cell and to evaluate the rate of spontaneous switching events .
we examine the conditions for the system to become a switch , and calculate the rate of spontaneous transitions between its two states .
this is done for several variants of the toggle switch .
in particular , we focus on switch systems in which the repression in done without cooperative binding ( namely , @xmath7 ) .
numerous studies have concluded , using rate equations , that cooperative binding is a necessary condition for the emergence of bistability @xcite .
below we show , using a combination of deterministic and stochastic simulation methods , that this is not the case , namely a bistable switch can exist even in the absence of cooperative binding .
in particular , we show that bound - repressor degradation ( brd ) and protein - protein interactions ( ppi ) give rise to bistability , without cooperative binding , even at the level of rate equations .
these results are confirmed by stochastic simulations using the master equation and monte carlo methods .
we also consider the exclusive switch , in which the @xmath1 and @xmath2 repressors can not be bound simultaneously due to overlap between their promoter sites .
this system exhibits bistability only when stochastic fluctuations are taken into account .
the rate of spontaneous transitions between the two states is calculated as a function of the biological parameters .
the paper is organized as follows . in sec .
[ sec : general ] we consider the basic version called the general switch .
several variants of this circuit are considered in the sections that follow .
the exclusive switch is studied in sec .
[ sec : exclusive ] , the brd switch is considered in sec . [ sec : brd ] and the ppi switch is analyzed in sec . [
sec : ppi ] .
the effects of cooperative binding are studied in sec .
[ sec : cooperative ] .
the response of toggle switch systems to external signals is examined in sec .
[ sec : response ] .
the results are discussed in sec .
[ sec : discussion ] and summarized in sec .
[ sec : summary ] .
the general switch consists of two transcription factors , @xmath1 and @xmath2 , that negatively regulate each other s synthesis @xcite . a schematic description of this circuit is given in fig .
[ fig1](a ) .
the regulation is done by the binding of a protein to the promoter site of the other gene , blocking the access of the rna polymerase and suppressing the transcription process . in this circuit
there is no cooperative binding , namely the regulation process is performed by a single bound protein .
the concentrations of free @xmath1 and @xmath2 proteins in the cell are denoted by @xmath9 $ ] and @xmath5 $ ] , respectively ( by concentration we mean the average copy number of proteins per cell ) .
the copy numbers of the bound proteins , are denoted by @xmath10 $ ] and @xmath11 $ ] , where @xmath12 is a bound @xmath1 protein that monitors the production of @xmath2 , while @xmath13 is a bound @xmath2 protein that monitors the production of @xmath1 .
note that there is at most one bound repressor of each type at any given time , and thus @xmath14 .
for simplicity , we ignore the mrna level and combine the processes of transcription and translation as a single step of synthesis @xcite . the maximal production rate of protein @xmath15 is denoted by @xmath16 ( s@xmath17 ) , @xmath18 .
the degradation rate of protein @xmath15 is given by @xmath19 ( s@xmath17 ) .
while the structure of the circuits studied here is symmetric , the rate constants can be different for @xmath1 and @xmath2 .
however , for simplicity we use symmetric parameters , i.e. @xmath20 and @xmath21
. the binding rate of proteins to the promoter is denoted by @xmath22 ( s@xmath17 ) and the dissociation rate by @xmath23 ( s@xmath17 ) .
the dynamics of the general switch circuit is described by the rate equations @xcite @xmath24 } & = & g_a ( 1-[r_b ] ) - d_a [ a]- \alpha_0 [ a ] \left ( 1-[r_a ] \right ) + \alpha_1[r_a ] \nonumber \\
\dot{[b ] } & = & g_b(1-[r_a])-d_b[b]- \alpha_0[b]\left(1-[r_b]\right)+\alpha_1[r_b ] \nonumber \\ \dot{[r_a ] } & = & \alpha_0[a]\left(1-[r_a]\right)-\alpha_1[r_a ] \nonumber \\ \dot{[r_b ] } & = & \alpha_0[b]\left(1-[r_b]\right)-\alpha_1[r_b ] .
\label{eq : general_switch_rate}\end{aligned}\ ] ] it is commonly assumed that the binding - unbinding processes are much faster than other processes in the circuit , namely @xmath25 .
this means that the relaxation times of @xmath26 $ ] are much shorter than other relaxation times in the circuit . under this assumption
, one can take the time derivatives of @xmath26 $ ] to zero , even if the system is away from steady state .
this brings the rate equations to the standard michaelis - menten form @xmath24 } & = & \frac{g}{1+k[b ] } - d [ a ] \nonumber \\ \dot{[b ] } & = & \frac{g}{1+k[a ] } - d [ b ] , \label{eq : general_switch_michaelis_menten}\end{aligned}\ ] ] where symmetric parameters are used , and @xmath27 is the repression strength . for a given population of free @xmath15 repressors , the parameter @xmath6 controls the value of @xmath26 $ ] .
the limit of weak repression , @xmath26 \ll 1 $ ] , is obtained when @xmath28 \ll 1 $ ] , while the limit of strong repression , @xmath26 \simeq 1 $ ] , is obtained for @xmath28 \gg 1 $ ] .
the meaning of bistability at the level of rate equations is that at steady state the equations exhibit two distinct positive solutions . in this particular class of circuits ,
one solution is dominated by @xmath1 proteins and the other is dominated by @xmath2 proteins .
starting from any initial state , the system will converge to one of these solutions .
the solutions are stable , so the possibility of spontaneous transitions , induced by stochastic fluctuations , is not included in the rate equation description .
the steady state solutions of eqs .
( [ eq : general_switch_rate ] ) and ( [ eq : general_switch_michaelis_menten ] ) are identical
. we will now show that these equations have only one positive steady - state solution . to this end , we first take @xmath29}=\dot{[b]}=0 $ ] in eq .
( [ eq : general_switch_michaelis_menten ] ) .
we multiply each equation by the denominator of the hill function that appears in it .
we obtain : @xmath30 - kd[a][b ] & = & 0 \nonumber \\ g - d[b ] - kd[a][b ] & = & 0 .
\label{eq : general_switch_hill1}\end{aligned}\ ] ] subtracting one equation from the other we get @xmath31-[b])=0 $ ] and therefore @xmath9 $ ] must be equal to @xmath5 $ ] at steady state . the steady state values of @xmath9 $ ] and @xmath5 $ ] can be easily found . inserting @xmath9=[b]$ ] into eq .
( [ eq : general_switch_hill1 ] ) we obtain a quadratic equation whose only positive solution is @xmath32 = [ b ] = \frac { -1+\sqrt{1 + 4kg / d}}{2k}.\ ] ] standard linear stability analysis shows that this solution is always stable . as a result
, we conclude that at the level of rate equations the general switch , without cooperative binding , does not exhibit bistability . in sec .
[ sec : cooperative ] we consider the case of cooperative binding , where the rate equations do exhibit bistability .
in order to account for stochastic effects and to obtain insight on the reason that this system is not bistable , the master equation approach is applied @xcite . in this case , we consider the probability distribution function @xmath33 .
it is the probability for a cell to include @xmath34 copies of free protein @xmath15 and @xmath35 copies of the bound @xmath15 repressor , where @xmath36 , and @xmath37 .
the master equation for the general switch takes the form @xmath38 \nonumber\\ & & + g_{\rm b } \delta_{r_a,0 } [ p(n_a , n_b-1,r_a , r_b ) - p(n_a , n_b , r_a , r_b ) ] \nonumber\\ & & + d_{\rm a } [ ( n_a+1 ) p(n_a+1,n_b , r_a , r_b ) - n_a p(n_a , n_b , r_a , r_b ) ] \nonumber\\ & & + d_{\rm b } [ ( n_b+1 ) p(n_a , n_b+1,r_a , r_b ) - n_b p(n_a , n_b , r_a , r_b ) ] \nonumber\\ & & + \alpha_0 [ ( n_a+1 ) \delta_{r_a,1 } p(n_a+1,n_b , r_a-1,r_b ) - n_a \delta_{r_a,0 } p(n_a , n_b , r_a , r_b ) ] \nonumber\\ & & + \alpha_0 [ ( n_b+1 ) \delta_{r_b,1 } p(n_a , n_b+1,r_a , r_b-1 ) - n_b \delta_{r_b,0 } p(n_a , n_b , r_a , r_b ) ] \nonumber\\ & & + \alpha_1 [ \delta_{r_a,0 } p(n_a-1,n_b , r_a+1,r_b ) - \delta_{r_a,1 } p(n_a , n_b , r_a , r_b ) ] \nonumber\\ & & + \alpha_1 [ \delta_{r_b,0 } p(n_a , n_b-1,r_a , r_b+1 ) - \delta_{r_b,1 } p(n_a , n_b , r_a , r_b ) ] , \label{eq : general_switch_master}\end{aligned}\ ] ] where @xmath39 for @xmath40 and @xmath41 otherwise .
the @xmath16 terms account for the production of proteins .
the @xmath19 terms account for the degradation of free proteins , while the @xmath22 ( @xmath23 ) terms describe the binding ( unbinding ) of proteins to ( from ) the promoter site . in numerical integration
, the master equation must be truncated in order to keep the number of equations finite .
this is done by setting suitable upper cutoffs , @xmath42 and @xmath43 , on the populations sizes of free proteins .
in order to maintain the accuracy of the calculations , the probability of population sizes beyond the cutoffs must be sufficiently small .
the master equation has a single steady state solution , which is always stable @xcite .
the criterion for bistability is that the steady state solution @xmath33 exhibits two distinct regions ( peaks ) of high probabilities , separated by a gap in which the probabilities are very small .
these two regions correspond to the two states in which the system is likely to be .
if the transition rate between the peaks is small enough , the system is indeed a bistable switch .
note , that in this case , averages of the form @xmath44 where @xmath45 , @xmath2 , do not reflect the complex structure of the probability distribution .
these can be considered as averages over many cells , some dominated by @xmath1 and others dominated by @xmath2 proteins , such that the total populations of the two species are about the same . to examine the existence of bistability we consider the marginal probability distribution
@xmath46 this probability distribution was calculated for a broad range of parameters .
two representative examples are shown in fig .
[ fig2 ] . under conditions of weak repression ( small k )
, @xmath47 exhibits a single peak for which @xmath48 [ fig . [ fig2](a ) ] , in agreement with the rate equations .
this is due to the fact that the repression is weak , and the @xmath1 and @xmath2 populations are almost uncorrelated . in this case
, the cell will contain roughly the same amount of @xmath1 and @xmath2 proteins . for strong repression ,
the distribution @xmath47 exhibits a peak dominated by @xmath1 proteins and a peak dominated by @xmath2 proteins , as expected for a bistable system .
however , a third peak appears near the origin , in which both populations of free proteins are suppressed [ fig . [ fig2](b ) ] .
this peak represents a dead - lock situation , caused by the fact that both @xmath1 and @xmath2 repressors can be bound simultaneously , each bringing to a halt the production of the other specie .
the third peak provides a corridor through which the probability can flow between the other two peaks . as a result
, the system can quickly switch between the @xmath1-dominated and the @xmath2-dominated states .
in addition to the solution of the master equation , monte carlo simulations have been performed . in these simulations
one can follow the time evolution of the populations of free and bound proteins in a single cell . in fig .
[ fig3](a ) we present the population sizes of free proteins vs. time for the general switch .
it is clear that the cell can indeed be in one of three states : a state rich in @xmath1 , a state rich in @xmath2 and a state in which both proteins are in very low copy numbers .
we conclude that a necessary condition for the system to become a switch is to prevent this dead - lock situation in which both protein populations are suppressed simultaneously .
below we present several variants of the circuit in which the third peak is suppressed , giving rise to a bistable switch .
the first variant we consider is the exclusive switch , depicted in fig . [
fig1](b ) . in this circuit
there is an overlap between the promoters of @xmath1 and @xmath2 . as a result
, there is no room for both @xmath1 and @xmath2 proteins to be bound simultaneously .
exclusive binding is encountered in nature , for example , in the lysis - lysogeny switch of phage @xmath0 @xcite .
it was shown that in presence of cooperative binding , the exclusive switch is more stable than the general switch @xcite .
this is because in the exclusive switch the access of the minority proteins to the promoter site is blocked by the dominant proteins .
here we show that in the exclusive switch , stochastic effects give rise to bistability even without cooperativity between the transcription factors .
the dead - lock situation in prevented in this case , since @xmath1 and @xmath2 repressors can not be bound simultaneously . to model the exclusive switch ,
recall that the variable @xmath10 $ ] ( @xmath11 $ ] ) is actually the fraction of time in which the promoter is occupied by a bound @xmath1 ( @xmath2 ) protein @xcite .
the fraction of time in which the promoter is vacant is thus @xmath49 - [ r_b]$ ] . incorporating this into eq .
( [ eq : general_switch_rate ] ) gives rise to the following modification : in the @xmath22 terms , each appearance of @xmath10 $ ] or @xmath11 $ ] should be replaced by @xmath10+[r_b]$ ] . with this modification ,
the rate equations take the form @xmath24 } & = & g ( 1-[r_b ] ) - d [ a]- \alpha_0 [ a ] \left ( 1-[r_a]-[r_b ] \right ) + \alpha_1 [ r_a ] \nonumber \\ \dot{[b ] } & = & g ( 1-[r_a ] ) - d [ b ] - \alpha_0 [ b]\left(1-[r_a]-[r_b ] \right ) + \alpha_1 [ r_b ] \nonumber \\ \dot{[r_a ] } & = & \alpha_0 [ a ] \left(1- [ r_a ] - [ r_b ] \right)-\alpha_1 [ r_a ] \nonumber \\
\dot{[r_b ] } & = & \alpha_0 [ b ] \left(1- [ r_a ] - [ r_b ] \right)-\alpha_1 [ r_b ] .
\label{eq : exclusive_rate}\end{aligned}\ ] ] under steady state conditions , the rate equations can be reduced to the michaelis - menten form @xmath24 } & = & \frac{g}{1+{k[b]/(1+k[a ] ) } } - d[a ] \nonumber \\ \dot{[b ] } & = & \frac{g}{1+{k[a]/(1+k[b ] ) } } - d[b ] , \label{eq : exclusive_michaelis_menten}\end{aligned}\ ] ] where , as before , @xmath27 .
we will now show that even for the case of the exclusive switch , the rate equations still exhibit a single solution , thus there is no bistability .
this is done by taking @xmath29}=\dot{[b]}=0 $ ] and getting rid of the denominators , by repeated multiplications .
the resulting equations are @xmath50 -kd[a ] ( [ a]+[b ] ) & = & 0 \nonumber \\ g
+ ( kg- d ) [ b ] -kd[b ] ( [ a]+[b ] ) & = & 0 . \label{brep2}\end{aligned}\ ] ] by subtraction of one equation from the other , we find that @xmath51+[b])\}([a]-[b ] ) = 0 .
\label{brep3}\ ] ] the positive , symmetric solution , @xmath9=[b]$ ] , is given by @xmath32 = \frac{(kg - d ) + \sqrt{(kg+d)^2 + 4kgd}}{4kd}.\ ] ] the other , non - symmetric solution , given by @xmath52+[b ] ) = 0 \label{eq : nonsymmsol}\ ] ] is inconsistent with eq .
( [ brep2 ] ) unless @xmath53 , namely there is no production of @xmath1 and @xmath2 proteins , which immediately leads to @xmath9=[b]=0 $ ] . under these conditions
the solution of eq .
( [ eq : nonsymmsol ] ) is @xmath9+[b]=-1/k$ ] , which requires a negative population size and thus makes no physical sense .
therefore , the only solution for @xmath54 is the symmetric solution , @xmath9=[b]$ ] .
thus , the rate equations do not support a bistable solution for the exclusive switch for any choice of the parameters . to account for the effects of fluctuations ,
we now describe the exclusive switch using the master equation .
it is similar to to master equation for the general switch given by eq .
( [ eq : general_switch_master ] ) , except for the following modifications : ( a ) in the @xmath22 and @xmath23 terms , each time @xmath55 ( @xmath56 ) , @xmath57 , appears it should be multiplied by @xmath58 , ( @xmath59 ) ; ( b ) the constraint @xmath60 should be imposed . implementing these changes we obtain the following equation : @xmath61 \nonumber\\ & & + g_{\rm b } \delta_{r_a,0 } [ p(n_a , n_b-1,r_a , r_b ) - p(n_a , n_b , r_a , r_b ) ] \nonumber\\ & & + d_{\rm a } [ ( n_a+1 ) p(n_a+1,n_b , r_a , r_b ) - n_a p(n_a , n_b , r_a , r_b ) ] \nonumber\\ & & + d_{\rm b } [ ( n_b+1 ) p(n_a , n_b+1,r_a , r_b ) - n_b p(n_a , n_b , r_a , r_b ) ] \nonumber\\ & & + \alpha_0 \delta_{r_b,0 } [ ( n_a+1 ) \delta_{r_a,1 } p(n_a+1,n_b , r_a-1,r_b ) - n_a \delta_{r_a,0 } p(n_a , n_b , r_a , r_b ) ] \nonumber\\ & & + \alpha_0 \delta_{r_a,0 } [ ( n_b+1 ) \delta_{r_b,1 } p(n_a , n_b+1,r_a , r_b-1 ) - n_b \delta_{r_b,0 } p(n_a , n_b , r_a , r_b ) ] \nonumber\\ & & + \alpha_1 [ \delta_{r_a,0 } p(n_a-1,n_b , r_a+1,r_b ) - \delta_{r_a,1 } p(n_a , n_b , r_a , r_b ) ] \nonumber\\ & & + \alpha_1 [ \delta_{r_b,0 } p(n_a , n_b-1,r_a , r_b+1 ) - \delta_{r_b,1 } p(n_a , n_b , r_a , r_b ) ]
. \end{aligned}\ ] ] for the exclusive switch , as for the general switch , under conditions of weak repression , @xmath47 exhibits a single peak [ fig . [
fig4](a ) ] that satisfies @xmath48 .
however , as the repression strength increases two distinct peaks begin to form . for intermediate values of @xmath6 these peaks are still connected , by a corridor of non - vanishing probabilities [ fig . [
fig4](b ) ] .
monte carlo simulations show that for intermediate values of @xmath6 , the system indeed exhibits two states , one rich in @xmath1 and the other rich in @xmath2 , but rapid transitions occur between them . for strong repression ,
the distribution @xmath47 exhibits two peaks which are separated by a region with vanishing probabilities [ fig . [
fig4](c ) ] . in one peak
the @xmath1 population is suppressed , while in the other peak the @xmath2 population is suppressed , as expected for a bistable system .
the average population of the dominant protein specie in each peak is @xmath62 , while the population of the suppressed specie is @xmath63 .
monte carlo simulations show that in this case the average time between spontaneous transitions is much longer .
the typical switching time for the case shown in fig .
[ fig3](b ) is around @xmath64 seconds .
it is much longer than the time - scales of the transcription , translation and degradation processes .
it is also longer than the time between cell divisions which is of the order of @xmath65 seconds .
the monte carlo results clearly show a large number of failed attempts in which a protein of the minority specie binds to the promoter and then unbinds again , without causing the system to flip . to evaluate the switching times we performed the following procedure .
we initialized the master equation in a state which is completely dominated by a proteins , namely , @xmath66 ( where @xmath67 represents the integer part ) , and all other probabilities vanish .
the master equation was then integrated numerically and @xmath47 was calculated as a function of time .
the function @xmath68 was found to decay exponentially from its initial value , @xmath69 , to zero , according to @xmath70 .
the time constant @xmath71 is defined as the switching time @xcite .
its inverse , @xmath72 , is referred to as the switching rate .
using this procedure , we examined the dependence of the switching time , @xmath71 on the protein synthesis rate , @xmath73 [ fig . [ fig5](a ) ] , the degradation rate , @xmath74 [ fig . [ fig5](b ) ] , and the repression strength , @xmath6 [ fig . [ fig5](c ) ] . in the parameter range in which bistability takes place , we obtain that ( a ) @xmath75 , ( b ) @xmath76 and ( c ) @xmath77 . concerning fig .
[ fig5](c ) , note that system exhibits bistability only in the regime in which @xmath6 is large @xcite . for @xmath78 , @xmath79 ( s ) , which is the typical time - scale of other processes in the cell . only for @xmath80
, @xmath71 becomes significantly larger than the time scales of other processes , and the system can function as a stable switch .
the scaling properties of the switching time can be summarized by @xmath81 this result can be reproduced by a simple argument .
consider an initial state in which the system is dominated by @xmath1 proteins , while the population of @xmath2 proteins is suppressed , namely @xmath9 \gg [ b]$ ] .
in this situation the promoter site is occupied by an @xmath1 protein during most of the time . in order that the switch will flip ,
the bound a protein must unbind ( at rate @xmath23 ) .
then , a @xmath2 protein ( rather than an @xmath1 protein ) should bind to the promoter .
the probability for this to happen is @xmath82/[a]$ ] .
this @xmath2 protein should remain bound long enough in order to build up a sufficiently large population of @xmath2 proteins .
on average , the @xmath2 protein stays bound @xmath83 ( s ) , during which @xmath84 proteins of type @xmath2 are produced . after the @xmath2 repressor will unbind , the probability that the next protein that binds will be of type @xmath2 rather than @xmath1 , is thus @xmath85 $ ] ( neglecting the degradation of @xmath1 proteins , because @xmath86 ) . following this argument ,
the switching rate is given by @xmath87 \over [ a ] } \times { { g } \over { \alpha_1 [ a ] } } = g { [ b ] \over [ a]^2 } .
\label{eq : exclusive_scaling2}\ ] ] from the michaelis - menten equations we obtain that @xmath88}{[a ] } = \frac{1}{1 + k [ a ] } \approx \frac{1}{k[a]},\ ] ] since for strong repression @xmath89 \gg 1 $ ] . inserting this result into eq .
( [ eq : exclusive_scaling2 ] ) and using and @xmath9 \approx g / d$ ] we find that @xmath90 this result can be considered as the leading term in the expansion of @xmath71 in powers of @xmath73 , @xmath74 and @xmath6 .
this leading term turns out to provide a very good approximation to simulation results .
for example for @xmath91 , @xmath92 , @xmath93 , and @xmath94 ( s@xmath17 ) we get @xmath95 ( s ) , which agrees perfectly with the results of monte carlo simulations . from eq .
( [ eq : exclusive_scaling2 ] ) , and from the fact that the average copy number of the dominant specie is @xmath9 \approx g / d$ ] , we find that when the production rate , @xmath73 , is varied while keeping all other parameters fixed , @xmath96 $ ] .
otherwise , when the degradation rate , @xmath74 , is varied while all other parameters are fixed , @xmath96 ^ 2 $ ] . in general
, the switching time is @xmath97 , while the population size , @xmath9 $ ] , of the dominant specie depends on both @xmath73 and @xmath74 .
thus , by a suitable variation of the rate constants , any desired dependence of @xmath71 on @xmath9 $ ] can be obtained . in particular , by increasing @xmath6 , @xmath71 can be increased with no effect on @xmath9 $ ] .
a similar result is obtained when @xmath73 and @xmath74 are decreased by the same factor .
we thus conclude that the population size is only one of several factors that affect the switching time .
a complete description of the switching time should include all the relevant rate constants . in monte carlo simulations of a switch system with cooperative binding , the switching time was found to depend exponentialy on the copy number @xcite .
this is consistent with the discussion above , but requires a well defined protocol according to which the rate constants are varied .
consider a different variant of the general switch , in which not only free repressors , but also bound repressors are affected by degradation .
the bound - repressor degradation ( brd ) tends to prevent the dead - lock situation in which both @xmath1 and @xmath2 repressors are bound simultaneously .
this is due to the fact that degradation removes the bound repressor from the system , unlike unbinding , where the resulting free repressor may quickly bind again .
it turns out that degradation of bound repressors induces bistability not only at the level of the master equation but even at the level of rate equations .
the rate equations that describe the brd switch take the form @xmath24 } & = & g ( 1-[r_b ] ) - d [ a]- \alpha_0 [ a ] \left ( 1-[r_a ] \right ) + \alpha_1[r_a ] \nonumber \\ \dot{[b ] } & = & g ( 1-[r_a])-d [ b]- \alpha_0[b]\left(1-[r_b]\right)+\alpha_1[r_b ] \nonumber \\ \dot{[r_a ] } & = & \alpha_0[a]\left(1-[r_a]\right)-\alpha_1[r_a ] - d_r [ r_a ] \nonumber \\ \dot{[r_b ] } & = & \alpha_0[b]\left(1-[r_b]\right)-\alpha_1[r_b ] - d_r [ r_b ] , \label{eq : brd_switch_rate}\end{aligned}\ ] ] where @xmath98 is the degradation rate of the bound repressors . assuming quasi - steady state for the binding - unbinding processes we obtain the michaelis - menten equations @xmath99}&=&{g \over { 1+k[b ] } } - \left(d + { { d_rk}\over{1+k[a ] } } \right ) [ a ] \nonumber \\
% \label{eq : bdeg } \dot{[b]}&=&{g \over { 1+k[a ] } } - \left ( d + { { d_rk}\over{1+k[b ] } } \right ) [ b ] , \label{eq : dbr_switch_michaelis_menten}\end{aligned}\ ] ] where now @xmath100 .
note that the coefficients of @xmath9 $ ] and @xmath5 $ ] in the second terms in eq .
( [ eq : dbr_switch_michaelis_menten ] ) can be considered as effective degradation rate constants . for steady state conditins , eq .
( [ eq : dbr_switch_michaelis_menten ] ) exhibits the symmetric solution @xmath32 = [ b ] = \frac { [ ( d+d_rk)^2 + 4dkg]^{1/2 } - d - d_rk } { 2dk}. \label{eq : symmbrd}\ ] ] this solution exists for any choice of the parameters . in addition , in some parameter range , two non - symmetric solutions exist .
these solutions can be expressed as the solutions of the quadratic equation @xmath101 ^ 2 + ( gdk+dd_rk+d_r^2k^2-gd_rk^2)[a]+gd=0 .
\label{eq : asymmetrics}\ ] ] the condition for the existence of two different solutions of this equation is @xmath102>0 .
\label{eq : asymmetrics2}\ ] ] in order for them to be positive the condition @xmath103 must be satisfied .
thus , the bifurcation takes place at @xmath104 and the non - symmetric solutions exist for @xmath105 .
linear stability analysis shows that whenever the non - symmetric solutions exist they are stable , while the symmetric solution is stable only for @xmath106 .
the steady state populations of free @xmath1 and @xmath2 repressors vs. @xmath6 , for the brd switch , are shown in fig .
the results of numerical integration of the rate equations ( @xmath107 ) are in perfect agreement with the analytical results derived above ( solid line ) .
we conclude that the degradation of bound repressors induces bistability , even at the level of rate equations .
the emergence of bistability can be attributed to the fact that the effective degradation rate for the minority specie in eq .
( [ eq : dbr_switch_michaelis_menten ] ) is larger than the effective degradation rate for the dominant specie . this tends to enhance the difference between the population sizes and to destabilize the symmetric solution for @xmath105 .
the master equation for the brd switch can be obtained from eq .
( [ eq : general_switch_master ] ) by adding the term @xmath108 + \nonumber\\ & & d_r [ \delta_{r_b,0 } p(n_a , n_b , r_a , r_b+1 ) - \delta_{r_b,1 } p(n_a , n_b , r_a , r_b ) ] . \end{aligned}\ ] ] for steady state conditions we find that brd tends to suppress the peak near the origin of @xmath47 . for a suitable range of parameters , two separate peaks appear , which qualitatively resemble those obtained for the exclusive switch .
however , unlike the exclusive switch , there is a narrow corridor with small but non - vanishing probabilities that connects the two peaks via the origin . as a result , the switching time for the brd switch tend to be somewhat shorter than for the exclusive switch with the same parameters .
the switching times , @xmath71 , vs. the repression strength , @xmath6 , are shown in fig .
[ fig7 ] .
we now examine in what range of parameters this circuit is indeed a switch according to the master equation . unlike the rate equation where the condition for bistability is clear [ eq . ( [ eq : asymmetrics3 ] ) ] , in the case of the master equation
the notion of bistability is more subtle .
thus , in the analysis below we use the following operational criterion .
first we define the two states of the switch .
the @xmath1-dominated state is defined as the set of all states in which @xmath109 and @xmath110 .
similarly , the @xmath2-dominated state is defined by @xmath111 and @xmath112 .
the system is considered as a switch if , under steady state conditions , the total probability to be in either of these states is larger than @xmath113 .
this leaves a probability of only 0.01 for all the intermediate states , which the system must visit in order to switch between the @xmath1-dominated and the @xmath2-dominated states . as a result
, the switching rate is low .
we used this criterion in order to find the region in the @xmath114 plane of the parameter space in which the brd circuit exhibits bistability .
it was found that the brd switch exhibits bistability for large enough values of @xmath6 , as long as the value of @xmath98 is not too different from @xmath74 . if @xmath115 , the process of bound - repressor degradation is negligible and can not eliminate the dead - lock situation .
if @xmath116 , proteins bind and quickly degrade . as a result , the population of the dominant specie is reduced and bistability is suppressed . within the parameter range in which the system exhibits bistability , we examined the dependence of the switching time @xmath71 of the brd switch on the parameters @xmath73 , @xmath74 , @xmath22 and @xmath98 .
it was found that @xmath71 exhibits linear dependence on the production rate @xmath73 and on the repression strength @xmath6 ( here , @xmath6 was varied by changing @xmath22 , keeping @xmath23 and @xmath98 fixed ) .
the dependence of @xmath71 on the degradation rate @xmath74 was found to be approximately @xmath117 .
note that as @xmath74 was veried , we kept @xmath118 in order that the system remains bistable .
since @xmath6 depends on @xmath98 , it slightly varied as well . unlike the exclusive switch , where we managed to obtain the scaling properties of @xmath71 by a simple argument
, the brd switch turns out to be more complicated .
this is due to the fact that there are several processes that may lead to the flipping of the switch , such as the unbinding or the degradation of the bound repressor .
a further complication is that the two repressors can be bound simultaneously . as a result ,
we have not managed to obtain an expression for @xmath71 in the brd switch .
consider a switch circuit which in addition to the mutual repression , exhibits protein - protein interactions ( ppi ) , namely an a protein and a b protein may form a complex , ab .
the ab complex is not active as a transcription factor .
the ppi switch can be described by the following rate equations @xmath24 } & = & g ( 1-[r_b ] ) - d [ a]- \alpha_0 [ a ] \left ( 1-[r_a ] \right ) + \alpha_1[r_a ] -\gamma ab \nonumber \\
\dot{[b ] } & = & g ( 1-[r_a])-d [ b]- \alpha_0[b]\left(1-[r_b]\right)+\alpha_1[r_b ] -\gamma ab \nonumber \\ \dot{[r_a ] } & = & \alpha_0[a]\left(1-[r_a]\right)-\alpha_1[r_a ] \nonumber \\ \dot{[r_b ] } & = & \alpha_0[b]\left(1-[r_b]\right)-\alpha_1[r_b ] .
\label{eq : ppi_switch_rate}\end{aligned}\ ] ] the parameter @xmath119 is the rate constant for the binding of a pair of @xmath1 and @xmath2 proteins .
the michaelis - menten equations take the form @xmath24 } & = & \frac{g}{1+k[b ] } - d [ a ] -\gamma [ a][b]\nonumber \\ \dot{[b ] } & = & \frac{g}{1+k[a ] } - d [ b ] -\gamma [ a][b ] .
\label{eq : ppi_switch_michaelis_menten}\end{aligned}\ ] ] for steady state conditions , these equations exhibit a symmetric solution , @xmath9=[b]$ ] , for any choice of the parameters .
it is the solution of @xmath120 ^ 3 + ( \gamma+dk)[a]^2 + d[a ] - g = 0.\ ] ] since all the coefficients of powers of [ a ] are positive , this equation has only one positive solution .
also , within some range of parameters there exist non - symmetric solutions , given by the solutions of @xmath121 ^ 2 + ( d\gamma + d^2 k - g\gamma k)[a]+d^2 = 0 .
\label{eq : ppnonsym}\ ] ] the non - symmetric solutions exist only for the range of parameters in which eq .
( [ eq : ppnonsym ] ) has two positive solutions .
the condition for this can be easily expressed in terms of the coefficients in eq .
( [ eq : ppnonsym ] ) . as in the case of the brd switch
, bistability is observed even at the level of rate equations .
again , the emergence of bistability can be attributed to the fact that the effective degradation rate constant for the minority specie is larger than for the dominant specie , thus enhancing the difference between the population sizes [ note that the effective degradation rate constant for @xmath1 is @xmath122)$ ] , while for @xmath2 it is @xmath123)$ ] ] . the master equation for the ppi switch can be obtained from eq .
( [ eq : general_switch_master ] ) by adding the term @xmath124 .
\label{eq : ppi_switch_master}\ ] ] for a suitable range of parameters the steady state solution of the master equation exhibits two separate peaks . to draw the range of parameters in which bistability takes place we apply the operational criterion used above for the brd switch .
we fix @xmath73 , @xmath74 and @xmath23 and examined the system in the @xmath125 plane .
the results are plotted in fig .
[ fig8 ] ( solid line ) . for small values of @xmath119 ( weak pp interaction )
, the circuit does not exhibit bistability .
as the interaction strength increases the circuit behaves as a switch for a certain range of repression strength @xmath6 .
this range broadens as @xmath119 is increased . unlike the switch systems discussed above , in which the bistability gets stronger as @xmath6 is increased ,
the ppi switch is bistable for intermediate values of @xmath6 .
this can be understood as follows .
recall that the key to the formation of a switch is the elimination of the dead - lock situation .
the exclusive and the brd switches deal with this situation directly at the bound repressor level
. however the pp interaction does not directly affect the bound repressor . to prevent the possibility of two proteins bound simultaneously , one of them should unbind and form a complex with a protein of the other specie . in order for this to happen
, the repressors must not be bound too strongly .
therefore , the ppi switch works at intermediate repression strength .
as the ppi becomes more effective ( larger @xmath119 ) this mechanism applies at larger values of @xmath6 .
enhanced switching properties can be obtained by considering a hybrid system that combines ppi and exclusive binding .
the resulting switch exhibits bistability in a broader range of parameters than the exclusive or ppi switches alone . the master equation for the exclusive - ppi switch
is obtained from eq .
( [ eq : exclusive_switch_master ] ) by adding the term @xmath124,\ ] ] which accounts for the pp interaction . numerical results , shown in fig
. [ fig8 ] , indicate that indeed as expected the exclusive - ppi switch is a better switch than either the ppi or the exclusive switch .
the parameter range in which it exhibits bistability is broader .
thus , it is more robust to variations in the parameters than the exclusive or ppi switches .
cooperative binding is found in genetic switch systems such as the phage @xmath0 switch @xcite . in this case ,
transcription regulation is obtained only when several copies of the repressor are bound simultaneously . this situation can be achieved in two ways .
one possibility is that repressors bind to each other and form a complex , which then binds to the promoter .
the other possibility is that the repressors bind separately , but those already bound assist the other ones to bind more effectively . in the case of cooperative binding
, bistability turns out to appear even at the level of rate equations @xcite .
switch systems with cooperative binding are commonly described by @xmath24 } & = & \frac{g}{1+k[b]^n } - d [ a ] \nonumber \\ \dot{[b ] } & = & \frac{g}{1+k[a]^n } - d [ b ] , \label{eq : general_switch_hilln}\end{aligned}\ ] ] where @xmath3 is the hill coefficient .
it corresponds to the number of copies of the transcription factor which are required in order to perform the repression process .
here we focus on the case @xmath126 , and show that these equations exhibit two stable steady state solutions for some range of parameters . imposing @xmath29}=\dot{[b]}=0 $ ] in eq .
( [ eq : general_switch_hilln ] ) , we obtain @xmath30 - kd[a][b]^2 & = & 0 \nonumber \\ g - d[b ] - kd[b][a]^2 & = & 0 .
\label{eq : general_switch_hilln1}\end{aligned}\ ] ] subtracting one of these equations from the other we find that @xmath127-[b])-kd[a][b]([b]-[a])=0 .
\label{eq : amb}\ ] ] looking for a non - symmetric solution for which @xmath9 \neq [ b]$ ] , we divide eq .
( [ eq : amb ] ) by @xmath9-[b]$ ] .
we find that @xmath89[b]=1 $ ] , or @xmath5=1/k[a]$ ] .
inserting this into eq .
( [ eq : general_switch_hilln1 ] ) we get an equation for @xmath9 $ ] : @xmath128 ^ 2-gk[a]+d=0.\ ] ] this equation exhibits two distinct stable solutions @xmath32=\frac{gk \pm \sqrt{g^2 k^2 - 4 d^2 k}}{2dk},\ ] ] for @xmath129 .
this means that the system becomes bistable at the bifurcation point , @xmath130 .
in addition to these solutions , the symmetric solution @xmath9=[b]$ ] exists for any choice of the parameters .
this symmetric solution is stable for @xmath131 and becomes unstable at the bifurcation point .
consider a switch system with cooperative binding with @xmath126 , in which two proteins of the same specie bind together to form a complex or dimer .
the repression of @xmath1 synthesis is done by dimers composed of two @xmath2 proteins and vice versa .
for example , consider an exclusive switch , in which the dimers of @xmath1 and @xmath2 can not be bound simultaneously . to account for stochastic effects
, we have studied this system using monte carlo simulations . the rate constant for the formation of dimers is denoted by @xmath132 .
it is assumed that dimers can not dissociate into single proteins , but they can degrade .
the degradation rate of dimers is denoted by @xmath133 .
we examined the dependence of the switching time @xmath71 on all the parameters .
we found the following properties .
the dependence of @xmath71 on @xmath73 was found to be linear as for the exclusive and brd switch .
the dependence on @xmath74 is very weak , except for the limit in which @xmath74 is very large .
this is because the proteins tend to form dimers before they have a chance to degrade .
the dependence of @xmath71 on @xmath27 is found to be well fitted by a quadratic polynomial .
this means that for sufficiently strong repression , @xmath134
. the dependence of @xmath71 on the dimerization rate @xmath132 [ fig .
[ fig9](a ) ] exhibits interesting behavior . for small values of @xmath132
the system is not really a switch , because almost no dimers are formed .
therefore the switching time is short . for larger values of @xmath132
the dimer population increases and the system starts to function as a switch .
the switching time @xmath71 increases as the switch becomes more stable . but
from some point , increasing @xmath132 causes @xmath71 to decrease .
this is because , very fast dimerization helps the minority specie to form dimers , making it more likely to flip the switch . the dependence of @xmath71 on @xmath133 [ fig .
[ fig9](b ) ] was found to be well fitted by a cubic polynomial in @xmath135 .
this means that in the limit of slowly degrading dimers , @xmath136 . in the limit of fast dimer - degradation
the system is not bistable , because the population of dimers is too small to make the repression effective .
the switching time for this system was also studied in ref .
@xcite , where @xmath71 was presented as a function of the average copy number of the dominant specie .
however , the copy number depends in a non - trivial way on the parameters and can not be directly controlled .
therefore , we believe that in a systematic study of the switching times , it is more practical to examine the dependence of @xmath71 on the parameters themselves .
note that there is another important realization of cooperative binding in which the promoter consists of two binding sites .
when a protein binds to one of them it facilitates the binding of another protein to the second site .
the effect of this mechanism is qualitatively similar to the one shown above for dimers . in general cooperative binding induces bistability becuase it forces the minority specie to recruit at least two proteins in order to flip the switch . as a result ,
cooperative binding helps to remove the dead - lock situation in which both species are suppressed simultaneously .
until now our discussion considered only spontaneous transitions between the two states of the switch . here
we demonstrate how an external signal may lead to the flipping of the switch . in case of the @xmath0 switch ,
such an external signal may be , for example , the exposure of _
e. coli _ infected by phage @xmath0 to uv light . in the _ lac _ circuit , the external signal indicates the presence of lactose .
we assume that the effect of the external signal is that one of the proteins undergoes a conformal change that prevents its binding to the promoter .
when the signal affects the dominant specie , this may lead to the flipping of the switch .
we assume that the conformal change is fast and that it lasts for a period of time determined by the duration of the external signal .
we have performed monte carlo simulations , where the binding rate @xmath22 of the dominant specie was set to zero for some period of time ( the length of external signal ) .
we calculated the probability for a flipping of the switch during @xmath137 ( s ) , which is roughly the time between divisions of _ e. coli _ , as a function of the signal length .
the results are shown in fig .
[ fig10 ] . for short duration of the signal
, the switch has a small chance to flip .
as the duration increases the probability to flip increases too , and so for a long enough signal , the switch will eventually flip as expected ( the actual switching time depends on the parameters of the switch , like the production rate @xmath73 or the unbinding rate @xmath23 .
here we just demonstrated that in principle the switch will flip states in response to an external signal ) .
in the rate equations , the meaning of bistability is clear .
it typically appears as a result of a bifurcation . below the bifurcation
there is a single , stable solution , which becomes unstable at the bifurcation point , where two stable solutions emerge . in case of the toggle switch ,
one of these solutions is dominated by @xmath1 proteins and the other is dominated by @xmath2 proteins . since both solutions are stable , the possibility of spontaneous transitions between them due to stochastic fluctuations is not included in the rate equation model .
the objects that participate in regulatory processes in cells , namely genes , mrnas , proteins and promoter sites are discrete objects , and some of them often appear in low copy numbers .
this , together with the fact that many of the relevant processes such as diffusion , degradation as well as binding and unbinding of transcription factors are of stochastic nature , requires to consider the role of stochastic fluctuations in these regulatory processes . this can be done by using the master equation or monte carlo simulations . in the master equation ,
bistability is characterized by two separate peaks in the probability distribution .
these peaks should be sufficiently far from each other , with low probabilities in the domain between them . as a result ,
the flow of probability between the two peaks is low and the time between spontaneous switching events is long . in order to qualify as a switch
, the average time between spontaneous switching events must be much longer than the time constants of the transcription , translation and degradation processes in the cell .
for the systems studied here it was found that the general switch without cooperative binding does not exhibit bistability bistability either with the rate equations or with the master equation .
two other variants , the brd and the ppi switch systems , were found to exhibit bistability both with the rate equations and with the master equation .
however , the exclusive switch , which is not bistable at the rate equation level , was found to exhibit bistability with the master equation .
thus , in case of the exclusive switch it is clear that stochastic fluctuations play a crucial role in making the system bistable . for this system
we also found an exact phenomenological expression for the switching time in terms of the rate constants of the relevant processes .
stochastic analysis of genetic networks can be done either by direct integration of the master equation or by monte carlo simulations .
the master equation provides the probability distribution of the population sizes of all the mrna s and proteins in the simulated circuit .
it can be considered as a distribution over a large number of genetically identical cells .
the average population sizes and the rates of processes are expressed in terms of moments of this distribution .
to obtain such distributions from monte carlo simulations , one needs to repeat the simulations a large number of times and average over them .
this may be inefficient in terms of computer time , and the statistical errors may be significant . on the other hand , unlike the master equation , monte carlo simulations enable to follow the time evolution of a single cell and directly evaluate quantities such as switching times and oscillation periods . the number of equations in the master equation set increases exponentially with the number of proteins and mrnas included in the simulated circuit . as a result ,
the master equation becomes infeasible for complex networks .
recently , we have shown that for reaction networks described by sparse graphs , one can use suitable approximations and dramatically reduce the number of equations @xcite .
a related circuit , the mixed feedback loop , in which @xmath1 is a repressor to @xmath2 and the @xmath1 and @xmath2 proteins bind to form a complex was recently studied using rate equations @xcite .
it was found to exhibit bistability within a range of parameters .
genetic switch systems with mutual repression of two transcription factors , have been studied using a combination of deterministic and stochastic methods .
these system exhibit bistability , namely two stable states such that spontaneous transitions between them are rare . induced transitions take place as a result of an external stimulus .
we have studied several variants of the genetic switch , which exhibit cooperative binding , exclusive binding , protein - protein interactions and degradation of bound repressors . for each variant
we examined the range of parameters in which bistability takes place .
numerous studies have concluded that cooperative binding is a necessary condition for the emergence of bistability in these systems .
we have shown that a suitable combination of network structure and stochastic effects gives rise to bistability even without cooperative binding .
the average time @xmath71 between spontaneous transitions was evaluated as a function of the biological parameters .
hill - function models are simplifications of rate - law equations . when derived directly from rate laws , @xmath3 is expected to take only integer values , which represents the number of transcription factors required to perform the regulation .
however , when these models are used for fitting empirical data , @xmath3 is a fitting parameter which may take non - integer values
. extended circuits that include the mrna level have also been studied .
it was found that there is a perfect agreement between the results of the res for the simplified and the extended circuits , and excellent agreement with only a slight difference in peak shapes in the me .
results of @xmath71 calculated in this procedure , agree with results obtained from averaging a large number of monte carlo simulations , up to a numerical factor of 2 .
this factor rises from the fact the in the procedure we described the system can also return from the @xmath2 dominated peak to the @xmath1 dominated peak . | genetic switch systems with mutual repression of two transcription factors are studied using deterministic methods ( rate equations ) and stochastic methods ( the master equation and monte carlo simulations ) .
these systems exhibit bistability , namely two stable states such that spontaneous transitions between them are rare . induced transitions may take place as a result of an external stimulus .
we study several variants of the genetic switch and examine the effects of cooperative binding , exclusive binding , protein - protein interactions and degradation of bound repressors .
we identify the range of parameters in which bistability takes place , enabling the system to function as a switch .
numerous studies have concluded that cooperative binding is a necessary condition for the emergence of bistability in these systems .
we show that a suitable combination of network structure and stochastic effects gives rise to bistability even without cooperative binding .
the average time between spontaneous transitions is evaluated as a function of the biological parameters . |
little is known about the stellar properties , extinction , and the expected intrinsic ly@xmath2 emission of distant , high redshift galaxies .
indeed , although it has in the recent past become possible through various techniques to detect already sizeable numbers of galaxies at @xmath10 ( see e.g. the reviews of taniguchi et al .
2003 and spinrad 2004 ) the information available on these objects remains generally scant .
for example , in many cases the galaxies are just detected in two photometric bands and ly@xmath2 line emission , when present , serves to determine the spectroscopic redshift ( e.g. bremer 2004 , dickinson 2004 , bunker 2004 ) .
then the photometry is basically used to estimate the star formation rate ( sfr ) assuming standard conversion factors between the uv restframe light and the sfr , and nothing is known about the extinction , and the properties of the stellar population ( such as age , detailed star formation histories etc . ) at higher redshift ( @xmath11 ) even less information is generally available ( but see a recent study of eyles 2005 on two @xmath12 galaxies observed with hst and spitzer ) .
many objects are found by ly@xmath2 emission , but remain weak or sometimes even undetected in the continuum ( e.g. rhoads & malhotra 2001 , kodaira 2003 , cuby 2003 , ajiki 2003 , taniguchi et al .
2004 ) . in these cases the ly@xmath2 luminosity can be determined and used to estimate a sfr using again standard conversion factors .
also the ly@xmath2 equivalent width is estimated , providing some possible clue on the nature of these sources . however , this has lead to puzzling results e.g. for the sources from the lala survey which seem to show unusually large ly@xmath2 equivalent widths that are difficult to understand without invoking exceptional conditions ( popiii stars ? ; malhotra & rhoads 2002 , rhoads 2003 ) .
given the few data available for the lala sources it is fair to say that the nature of these objects , their stellar populations , extinction etc .
remain currently largely unknown ( cf .
dawson 2004 ) .
when possible , a simple comparison between the uv and ly@xmath2 sfr is undertaken providing possibly information on the ly@xmath2 transmission , i.e. the partial absorption of ly@xmath2 photons on their sight line through the intergalactic medium ( e.g. haiman 2002 , santos 2004 ) and/or on partial ly@xmath2 `` destruction '' processes close to the source ( e.g. due to dust or ism geometry ; charlot & fall 1993 , valls - gabaud 1993 , tenorio - tagle 1999 , mas - hesse 2003 ) .
notable exceptions of @xmath13 samples for which some estimate of extinction is available from multi - band photometry include work on the subaru deep survey ( ouchi 2004 ) and _ goods _ data ( e.g. papovich 2004 ) .
lehnert & bremer ( 2004 ) also discuss some preliminary information on little extinction in their @xmath14 sources .
interestingly , in their study of a @xmath15 galaxy discovered by dey ( 1998 ) , armus ( 1998 ) find indications for significant reddening ( @xmath16 mag ) from analysis of the observed sed and from the presence of ly@xmath2 emission . in a similar manner we will here present a consistent study of the stellar population properties , extinction , and ly@xmath2 emission for two galaxies at redshift @xmath11 . for this aim we use two distant ( @xmath11 ) gravitationally lensed galaxies for which multi - band photometry is available ( detection in at least 34 bands ) . through a quantitative analysis of their sed , using a vast library of empirical and theoretical template spectra
, we aim to constrain properties of the stellar populations , such as age and star formation ( hereafter sf ) history ( burst or constant sf ? ) and their extinction . furthermore by comparing the ly@xmath2 emission expected from the stellar population constraint with the observed ly@xmath2 flux we estimate consistently the ly@xmath2 `` transmission '' for the individual sources
. the ly@xmath2 transmission and sf properties derived here can in principle be used to infer the ionisation fraction of hydrogen in the igm at a given redshift ( cf .
haiman 2002 , santos 2004 ) , a key quantity of interest for the study of the reionisation history of the universe ( cf .
review from barkana & loeb 2001 ) .
obviously the present `` exploratory '' work will have to be extended to larger galaxy samples , and sophisticated tools will probably be needed to interpret such results in terms of igm properties ( cf .
gnedin & prada 2004 )
. however , this approach should be complementary to other methods probing the reionisation history by measuring the gunn - peterson optical depth observed in quasar spectra as a function of redshift ( e.g. becker 2001 , fan 2003 ) , or by comparing ly@xmath2 luminosity functions at different redshifts ( e.g. malhotra & rhoads 2004 ) .
the remainder of the paper is structured as follows . in sect .
[ s_obs ] we summarise the adopted observational constraints from the literature . our modeling technique is described in sect .
[ s_models ] .
the detailed results for each galaxy are presented in sects .
[ s_370 ] and [ s_kesr ] .
our main conclusion are summarised in sect .
[ s_conclude ] .
the two galaxies studied here are : 1 ) the probable @xmath1 galaxy recently discovered by kneib ( 2004 , hereafter kesr ) , which presently lacks of a spectroscopic redshift but for which rather accurate multi - band hst observations are available , allowing us in particular also to derive a fairly reliable photometric redshift .
2 ) the @xmath0 ly@xmath2 emitter hcm 6a behind the lensing cluster abell 370 .
we now summarise the observational data , taken from the literature .
the adopted redshift and gravitational magnification factors are listed in table [ tab_props ] . before proceeding let us mention for clarity that these two objects are generally considered to be star forming galaxies ( starbursts ) , not agn ( narrow line - type ii - or others ) , as no contradicting information is available so far .
however , one must bear in mind that some of the interpretations presented below ( and in the literature ) may need to be revised , should this assumption be incorrect . * triple arc in abell 2218 : * the observational data for this object , named abell 2218 kesr hereafter , is taken from kneib et al .
( 2004 , hereafter kesr ) and from egami ( 2005 ) .
the photometry from kesr includes observations with hst ( wfpc2 , acs , nicmos ) in ( undetected ) , , , and , and with nirc / keck in @xmath17 .
subsequently , additional photometry was obtained with nicmos / hst in the @xmath17 band ( f110w ) , and with irac / spitzer at 3.6 and 4.5 ( see egami et al . ) for our computations ( see below ) we have used the appropriate filter transmission curves . in particular , updated transmission curves were used for the acs and nicmos filters ( m. sirianni 2003 , private communication ; sirianni et al .
2004 ; r. thompson 2003 , private communication ) .
few brief comments concerning the photometry are needed here .
first , kesr present photometry for two multiple images ( a and b ) . apparently sources a and b differ in the flux ( with quoted errors of @xmath18 0.05 mag ) by 3.2 @xmath6 , whereas the fluxes in the other filters agree well within 1 @xmath6 .
differential lensing across the images together with sampling effects could be responsible for this small discrepancy .
as we are interested in a global representative sed for this source , we have chosen to use the averaged photometric sed , the magnification factors being the same for the two images . finally , we have also noted some apparent discrepancies between the measurements reported in kesr and egami , the most important one being the flux , which is @xmath5 1520 % ( 34 @xmath6 ) higher in the latter publication .
these differences are mostly due to the use of different apertures on different repixeled / rescaled images ( j. richard , 2004 , private communication ) .
again , this illustrates the difficulty in deriving reliable colors for extended arcs . to account for these small discrepancies and for the possible error
underestimate we therefore adopt a minimum photometric error of 0.15 mag in all filters .
it is worth noting that photometric errors translate into absolute flux calibration errors for fitting purposes . as we will see below , adopting the latter minimum errorbars significantly improves the sed fits .
in addition to the photometry , the non - detection of the source with keck lris spectroscopy provides an upper limit on the continuum flux between 9000 and 9300 ( kesr ) .
this upper limit will be used as an additional constraint in our sed modeling .
kesr also indicate a possible drop of the continuum below @xmath5 9800 from their keck ii nirspec spectrum . for various reasons
the reality of this spectral break is questionable .
first a true neutral hydrogen break ( `` ly@xmath2 break '' ) at @xmath5 9800 , far in the red wing of the filter , seems incompatible with the relatively strong flux measured in this filter .
furthermore , test computations show that such a break is difficult if not impossible to reconcile with our spectral modeling . in any case
the significance of this finding appears questionable as the detected continuum is extremely faint and noisy .
the reality of this spectral feature is now also questioned by egami ( 2005 ) . for these reasons
this information is discarded from our spectral fitting . in practice
we have retained the following two variants to describe the observed sed of this source : _
sed1 ) _ the average fluxes ( , , , ) of images a and b from kesr plus the irac / spitzer data of image b from egami et al . _ sed2 ) _ all fluxes from image b from egami et al .
these are treated as seds from two different objects .
furthermore , for each of these `` objects '' we have computed two cases in our sed fitting described below : _
i ) _ the observed seds in , , , , 3.6 , and 4.5 .
_ ii ) _ same as ( i ) plus the flux limits from the and keck lris non - detections .
no emission line has so far been detected for abell 2218 kesr .
its spectroscopic redshift remains therefore presently unknown but the well - constrained mass model for the cluster strongly suggests a redshift @xmath19 6 for this source ( kesr , egami 2005 ) .
the magnification factors of both images a and b is @xmath20 , according to kesr . *
abell 370 hcm6a : * the observational data of this @xmath0 galaxy is taken from hu et al . (
the photometry includes @xmath21 from keck i and ii ( lris and echellette spectrograph and imager ) and from subaru ( cisco / ohs ) . the gravitational magnification of the source is @xmath22 according to hu et al .
( 2002 ) .
photometric fluxes and errors were adopted from their fig .
3 . where possible the appropriate filter transmission curves were used .
the `` @xmath23 '' band filter transmission is somewhat uncertain , as these observations were undertaken using an rg850 filter , which together with the lris optics and the ccd response , yields a transmission similar to a @xmath23 band filter ( hu et al .
our approximate filter curve shows a blueward shift of @xmath24 by @xmath5 200 compared to the information given by hu et al .
however , since the redshift is known for this source and since we adjust the observed flux ( not magnitude ) in this band , this should not affect our conclusions .
before proceeding to the fits of the individual seds a brief comment on the available sed features seems appropriate . for obvious reasons the available sed ( basically from broad - band photometry ) of high - z ( @xmath11 ) galaxies
is primarily limited to the rest - frame uv ( when observed from the ground ) or optical spectrum ( when available e.g. with spitzer and future satellite missions ) .
the main information `` encoded '' in this sed is therefore : _ 1 ) _ the neutral hi break shortward of ly@xmath2 ( hereafter the `` ly@xmath2 '' break ) due to the strong or complete gunn - peterson trough , _ 2 ) _ the slope of the uv spectrum , and _ 3 ) _ possibly a 4000 break ( hereafter denoted balmer break ) , if present and covered by the observations .
in addition the presence of the ly@xmath2 line , mostly used to determine spectroscopically the redshift , provides clear evidence for ongoing massive star formation ( hereafter sf ) .
the position of the ly@xmath2 break depends essentially on redshift .
the uv slope depends on the intrinsic spectrum in turn depending mostly on age and sf history and on the extinction , i.e. the extinction law and the amount of reddening .
the balmer break becomes visible ( in absorption ) in the continuum of stellar populations after @xmath25 1030 myr .
ly@xmath2 emission , if due to stellar photoionisation and not agn activity , indicates the presence of young ( @xmath26 10 myr ) massive ionizing stars .
concerning the uv slope , it is useful to recall that this quantity , are defined as the power - law index of the sed in @xmath27 versus @xmath28 over a certain wavelength interval . ]
does not lend itself to determine the metallicity of a star forming galaxy from a theoretical point of view and in terms of individual objects .
the reasons are that intrinsically the uv slope shows only small variations with metallicity , and that the slope depends strongly on the exact sf history ( see e.g. leitherer & heckman 1995 , meurer 1995 ) .
this is illustrated in fig .
[ fig_beta ] where @xmath29 ( measured over the interval 1300 - 1800 ) is plotted as a function of age for populations of metallicities between solar and zero ( popiii ) and for the limiting cases of bursts and sfr = const . furthermore
, as pointed out in schaerer ( 2002 , 2003 ) and also shown in this figure , for very low metallicities ( @xmath30 ) nebular continuous emission becomes dominant even down to uv wavelengths ( longward of ly@xmath2 ) , such that the observed integrated ( stellar+nebular ) spectrum has even a flatter uv slope than high metallicity starbursts .
in other words , even for bursts , there are strong intrinsic degeneracies of @xmath29 between age and metallicity , to which the additional effect of reddening must be added , including the uncertainties on the a priori unknown extinction law .
it is therefore evident that on an individual object basis there are in general degeneracies between age , metallicity , sf history , and extinction .
however , this does not preclude the possible existence of statistical correlations between quantities such as e.g. @xmath29 and metallicity in large samples of galaxies , as known to hold e.g. for local uv selected starbursts ( cf .
heckman et al . 1998 ) .
also , as we will see below , there are cases where the uv slope and the mere fact of the presence of an emission line allow us nevertheless to lift some degeneracies and therefore to determine interesting constraints on the stellar population and on extinction .
the behaviour of the 4000 break and its use as an age indicator has extensively been discussed in the literature ( e.g. bruzual 1983 , and recently kauffmann 2003 ) . as in simple stellar populations
its amplitude is basically a monotonically increasing function of age , an estimate of the break , e.g. obtained from 3.8 - 4.5 photometry with irac / spitzer and jhk photometry in @xmath31 galaxies , provides information on the age of the light emitting stellar population . since the exact sf history can not be determined in these cases ( in contrast to studies at low @xmath32 , cf .
kauffmann ) the amplitude of the break provides a range of ages , the minimum age being given by instantaneous bursts , the maximum age from models with constant sf . for obvious reasons this maximum
`` luminosity weighted '' age derived from a measure of the balmer break is also unaffected by considerations of possible multiple stellar populations .
the same is not true for the minimum age , which is , however , of less cosmological interest . for the spectral fitting we use a slightly adapted version of the photometric redshift code of bolzonella et al .
( 2000 ) .
does standard sed fitting using a number of modeling parameters .
the free parameters for the sed modeling are : * the spectral template , * extinction and the reddening law , * a parameter describing possible deviations from the average lyman forest attenuation from madau ( 1995 ) . for abell 2218 kesr
the source redshift is also a free parameter . for the spectral templates we use a large compilation of empirical and theoretical sed , including starbursts , qso , and galaxies of all hubble types , and covering various star formation histories ( bursts , exponentially decreasing , constant sf ) and various metallicities . for most applications we group the templates in the following way : * * starbursts and qsos ( hereafter sb+qso ) : * this group includes the starburst templates with @xmath33 from @xmath34 to 0.7 from the calzetti et al .
( 1994 ) and kinney et al .
( 1996 ) atlas , the hst qso template of zheng et al .
( 1997 ) , as well as uv - optical spectrum of the metal - poor galaxy sbs 0335 - 052 with numerous strong optical emission lines ( and an extinction of @xmath35 , izotov & thuan 1998 ) kindly communicated to us by yuri izotov ( 2002 , private communication ) * * bccww+ : * bruzual & charlot ( 1998 , private communication ; cf . bruzual & charlot 1993 ) evolving synthesis models assuming bursts , constant star formation , and exponentially decaying star formation histories reproducing present day spectra of galaxies of various types ( e , s0 , sa , sb , sc , sd , and i m ) plus the empirical e , sbc , scd , and i m templates from coleman et al .
( 1980 ) , as included in the public version .
* * s03 + : * theoretical templates of starburst galaxies from schaerer ( 2003 ) covering metallicities of @xmath36 ( solar ) , 0.008 , 0.004 , 0.001 , 1/50 , @xmath37 , @xmath38 , and zero metallicity ( popiii ) . for low metallicities ( @xmath39 ) these templates have been computed for 3 different assumptions on the imf .
the spectral library includes burst models and models with a constant star formation rate ( sfr ) . for more details
see schaerer ( 2003 ) . for the present work these computations
were extended to cover ages of up to 1 gyr . these seds are available on request from the first author and on the web .
the standard extinction law adopted here is the one from calzetti et al .
( 2000 ) determined empirically from nearby starbursts .
we also explore the possible implications of other laws , such as the galactic law of seaton ( 1979 ) including the 2200 bump , and the smc law from prvot ( 1984 ) and bouchet ( 1985 ) showing no uv bump , but a steeper increase of the extinction in the uv compared to calzetti
. for the lyman forest attenuation , follows madau ( 1995 ) .
however , we allow for possible deviations from the mean attenuation by varying the lyman forest optical depths @xmath40 by a multiplicative factor taking the values of ( , 1 . , and @xmath41 ) .
typically we adopted @xmath42 2 or 3 . here
@xmath40 stands for the optical depths corresponding to the absorption between ly@xmath2 and ly@xmath29 , and between ly@xmath29 and the lyman limit respectively .
the following other minor changes have been made in our version ( 1.3ds ) of .
the calculation of the synthetic photometry deals correctly with templates including strong spectral lines ( emission or absorption ) .
furthermore we make sure to use the proper filter transmission curves usually given in photon units .
earlier versions of and other codes ( e.g. evolutionary synthesis codes ) assume sometimes ( for `` historical '' reasons ) that transmission curves be given in flux units . in case of wide filters , e.g. such as some acs / hst filters , this may lead to small differences .
other modifications concern essentially features related to the user interface ( additional outputs etc . ) . for given choices of the above parameters , the code performs a standard minimisation fit to the observed seds and determines , for each point in the parameter space , the corresponding @xmath43 value .
using these @xmath43 values , it is possible to quantify the probabilities for the main free parameters , namely extinction , age of the spectral template , sf history , etc .
when the sed fitting is based on theoretical templates , the sfr value is easily obtained and allows us to compare the expected values for the ly@xmath2 flux to the actual ones . to convert the observed / adjusted quantities to absolute values we adopt the following cosmological parameters : @xmath44 , @xmath45 , and @xmath46 km s@xmath47 mpc@xmath47 .
as a spectroscopic redshift has not been obtained ( yet ) for this galaxy we here examine its photometric redshift estimate . in fig .
[ fig_pz_7 ] we show the photometric redshift probability distributions for the two seds ( sed1 , sed2 ) of abell 2218 kesr described above using the three spectral template groups and adopting a minimum photometric error of 0.15 mag . for each redshift , quantifies the quality of the best fit model obtained varying all other parameters ( i.e. extinction , , spectral template among template group ) .
given the excellent hst ( wfpc2 , acs and nicmos ) photometry , is quite well defined : the photometric redshift ranges typically between @xmath48 5.5 and 7.3 .
outside of the plotted redshift range is essentially zero . if we assume the ( smaller ) quoted formal photometric errors ( but note the discrepancies discussed in sect .
[ s_obs ] ) becomes more peaked , i.e. the photometric redshift better defined .
this is driven by the error on the flux , which determines the red side of the `` ly@xmath2 '' break .
however , the resulting best fit value does not change much .
furthermore , the fit quality is considerably decreased .
this demonstrates the interest of such high accuracy measurements and the need for reliable error estimates .
the predicted redshift distribution is found to be quite insensitive to the exact template ( as shown in fig .
[ fig_pz_7 ] ) , to the exact value of , and to the adopted extinction law ( variations of the latter two are not shown ) .
however , we note that for this object the fits ( and ) are improved when allowing for deviations from the average madau ( 1995 ) attenuation law .
the curves shown here have been computed for @xmath49 .
more important in determining is the exact sed . as seen from fig .
[ fig_pz_7 ] the use of sed1 or sed2 lead to somewhat different distributions .
sed2 ( cf .
[ s_obs ] ) yields a somewhat larger redshifts , albeit with a somewhat yreduced fit quality .
these differences illustrate how uncertainties and difficulties in the photometric measurements of such faint sources , whose origin are briefly discussed in sect .
[ s_obs ] , propagate to the photometric redshift estimate .
all our best - fit solutions have redshift @xmath50 6.256.63 , lower than the redshift range estimated by kesr , but compatible with the more recent quantitative analysis of egami et al .
given the various free parameters , uncertainties on the intrinsic sed , etc .
we conclude that the redshift of abell 2218 kesr is likely @xmath3 6.0 7.2 . taking into account both our photometric determination and the lensing considerations of kesr .
a large number of models have been computed using the different variants of the seds describing this object ( sed1 - 2 ) , the different filter combinations ( non - detections + spectroscopic constraint ) discussed in sect .
[ s_obs ] , and varying the various model parameters .
we first discuss briefly the main salient results with the help of some illustrations . a more general discussion of the results and their dependence on various assumptions follows .
figure [ fig_7rev_all ] shows the best fit models to the sed2 including the upper limits from the and lris spectroscopy for the s03 + and sb+qso template groups .
the best fit redshifts are @xmath51 and 6.54 respectively .
these fits show in particular that the spectroscopic constraint can be accommodated simultaneously with the observed flux ; the resulting fits are within the 1 @xmath6 errors in all bands .
the best fit from the s03 + group corresponds to a burst with an age of 15 myr at solar metallicity and no extinction ( solid line ) .
similarly good fits are also obtained for lower metallicity .
the best fit with empirical starburst and qso templates ( sb+qso ) is obtained with the spectrum of the metal - poor h ii galaxy sbs 0335 - 052 ( dotted line in fig . [ fig_7rev_all ] ) . in this case
the apparent balmer break observed between the nicmos / hst and irac / spitzer domain is simply explained by the presence of strong emission lines in the 3.6 and 4.5 filters , h@xmath29 and in the 3.6 filter and he i @xmath285876 in the 4.5 filter .
e.g. for the emission lines between h@xmath52 and the total observed equivalent width ( boosted by the @xmath53 factor ) is @xmath5 9130 , as estimated from the data of izotov & thuan ( 1998 ) , compared to a filter surface of @xmath5 6600 . ] and some additional extinction to reduce the restframe uv flux .
the extinction needed is @xmath54 for the calzetti law , or @xmath55 for the prvot extinction law . in terms of age
the restframe uv to optical spectrum ( continuum and lines ) of sbs 0335 - 052 corresponds to a young population of @xmath5 35 myr according to the analysis of papaderos ( 1998 ) and vanzi ( 2000 ) .
of course , the presence of an older population in addition to the starburst can not be excluded on the present grounds . in short ,
the observed sed of abell 2218 kesr can be explained by a young population withour or with emission lines .
spectroscopy in the 34 range would be needed to distinguish the latter solution from others .
alternatively , good sed fits are also obtained with relatively `` old '' populations .
the oldest ages are obtained when invoking the longest sf timescale , i.e.constant sf . in this case
the uv restframe flux remains high ( due to the continuous formation of massive stars ) and older ages need to be attained to build up a sufficient population of evolved stars with strong balmer breaks , in order to reproduce the observed break .
this case is shown in fig .
[ fig_7rev_old ] with fits with the s03 + templates to the observed sed1 and sed2 .
the best fits solutions obtained here correspond to ages of 500 and 400 myr , no extinction , and redshifts @xmath5 6.40 and 6.57 respectively . similar , but somewhat older ages are obtained for metallicities @xmath56 , below the ones shown here . a quantitative examination of the `` maximum age '' allowed by the observations ( here sed2 )
is presented in fig .
[ fig_7rev_contour ] , showing @xmath43 contours in the extinction age plane for a given set of spectral templates ( s03 + group with constant sfr ) , the calzetti extinction law , and a fixed redshift of @xmath57 .
for these conditions the best fit is corresponds to 400 myr , zero extinction , and @xmath58 ( see fig . [ fig_7rev_old ] ) .
this figure shows that a maximum age of @xmath5 ( 250650 ) myr ( 1 @xmath6 interval ) is obtained , in good agreement with the modeling of egami ( 2005 ) .
if true , this would correspond to a formation redshift of @xmath59 8.720 for our adopted cosmological parameters . as clear from fig .
[ fig_7rev_contour ] , even with constant sf models , younger populations with some extinction can also fit , although less well , the present observations .
however , solutions with low or zero extinction are generally preferred .
when varying the star formation history between these extreme cases ( burst or sfr@xmath60const ) , i.e. considering e.g. exponentially declining sf histories , any intermediate age can be found for obvious reasons .
such cases are e.g. obtained when fitting templates from the bruzual & charlot models ( not shown here ) with exponentially decreasing sf histories and can be found in egami et al . as discussed in sect .
[ s_uv ] , considering multiple stellar populations ( cf .
eyles 2005 ) does not alter the above estimate of the maximum age determined from constant sfr models . in any case
, the data available here does not allow us to constrain the sf history and age further .
the observations obtained so far have not revealed any emission line from this object ( kesr ) . in particular ly@xmath2
emission is lacking , which could be puzzling for a source with intense star formation . as we have just seen a variety of star formation histories and ages
are possible for abell 2218 kesr .
therefore one may or may not expect intrinsic ly@xmath2 emission . a simple explanation for the apparent absence of ly@xmath2 emission could be to invoke an advanced age ( in the post starburst phase ) .
however , even with a young age it is not necessary that ly@xmath2 emission be observed .
e.g. the spectrum of the metal - poor h ii galaxy sbs 03335 - 052 which provides an excellent fit to the observed sed and shows strong emission lines ( cf .
[ fig_7rev_all ] ) does not show ly@xmath2 emission ( thuan 1997 , kunth 2003 ) .
alternatively , if intrinsically present , the ly@xmath2 non - detection could be due a variety of factors : a redshift @xmath61 placing ly@xmath2 below the spectral range discussed in detail by kesr , a flux below their strongly varying detection threshold , or other factors depressing the ly@xmath2 emission within the host galaxy ( dust , ism+hi geometry ) and in the intervening igm . in conclusion , from the available data
the apparent lack of ly@xmath2 emission from this source is not puzzling .
however , it is not completely excluded that the galaxy truely shows ly@xmath2 emission , which has so far eluded detection . after these main findings
we shall now quickly mention more `` technical '' results about the influence of various fit parameters .
quite generally , the results on the age , sf history , extinction etc . depend little on the different variants of the observed seds ( sed1 - 2 ) , on the inclusion or not of the non - detections in the fits , and on the use of the published formal errors or a minimum error of 0.15 mag ( cf .
[ s_obs ] ) .
the results discussed above are therefore quite robust with respect to these assumptions .
small differences in the best fit values can , however , be obtained .
e.g. the best fit photometric redshift can vary by up to @xmath26 0.2 depending on adopting sed1 or sed2 . in all cases
we note that sed1 allows better fits ( smaller @xmath43 ) than sed2 . adopting @xmath62 mag
also significantly increases the fit quality .
finally , considering variations around the mean lyman forest attenuation improves the fits ( especially as the hst photometry determining the `` ly@xmath2 break '' is quite accurate ) . in practice
all best fits are found with an increased lyman - forest opacity ( @xmath49 ) . to summarise ,
given the absence of a spectroscopic redshift , a fair number of good fits is found to the observations of abell 2218 kesr when considering all the free parameters .
the main conclusions from these `` best fits '' are : * generally the determined extinction is negligible or zero quite independently of the adopted extinction law .
the best fit with the empirical starburst spectrum of sbs 0335 - 052 represents an exception to this case , requiring an additional @xmath8 0.20.6 mag , depending on the adopted extinction law . *
although generally burst models fit somewhat better than those with constant star formation among the theoretical templates ( bc , s03 + ) , the data does not strongly constrain the star formation history . * typical ages between @xmath5 15 and 400 myr are obtained . a reasonable 1-@xmath6 upper bound on the age of @xmath5
650 myr can be obtained assuming constant star formation .
however , the data can also be well fit with a very young ( @xmath5 35 myr ) stellar population with strong emission lines ( using e.g. the spectrum of the metal - poor galaxy sbs 0335 - 052 ) . in this case
the apparent balmer break observed between the hst and spitzer broad - band photometry is simply due to the presence of strong emission lines affecting the red 3.6 and 4.5 filters . * given degeneracies of the restframe uv spectra between age and metallicity ( cf .
above ) no clear indication on the galaxian metallicity can be derived , in contrast to the claim of kesr .
good fits to the available data can even be found with solar metallicity starburst templates .
* depending on the star formation history and age one may or may not expect intrinsic ly@xmath2 emission , i.e. an important h ii region around the object .
the apparent absence of observed ly@xmath2 emission does therefore not provide much insight .
a more complete error analysis beyond the level presented here is difficult to achieve for a variety of reasons and clearly beyond the scope of this publication .
the theoretical templates can also be used to estimate the stellar mass involved in the starburst or the star formation rate when constant star formation is assumed . for this aim
we use all the best fits to the three seds ( sed1 - 3 ) with the s03 + templates , we assume a typical redshift of @xmath63 , and the magnification @xmath64 determined by kesr . for the adopted cosmology the distance luminosity is then @xmath6564457.8 mpc .
when constant sf is assumed one obtains the following star formation rate : @xmath66 ( for a salpeter imf from 1 to 100 ) . for the best fit ages of
@xmath5 400570 myr the total mass of stars formed would then correspond to @xmath67 .
the mass estimated from best fit burst models ( of ages @xmath5 620 myr ) is slightly smaller , @xmath68 .
if we assume a salpeter imf with @xmath69 the mass and sfr estimates would be higher by a factor 2.55 , and in good agreement with the values derived by kesr and egami ( 2005 ) .
in all the above cases the total luminosity ( unlensed ) is typically @xmath70 .
llllllllll object & @xmath71 & redshift & sf history & age & @xmath72 & sfr & stellar mass & @xmath73 & ly@xmath2 + & & & & [ myr ] & [ mag ] & [ yr@xmath47 ] & [ ] & [ ] & transmission + abell 2218 & 25 . &
@xmath5 6.07.2 & ? & 3400 & negligible ? & @xmath5 1 & ( 0.36 ) @xmath75 10@xmath76 & @xmath77 + kesr + + abell 370 & 4.5 & 6.56 & csfr/ & ? & @xmath5 1 .
& 1141 & ( 14 ) @xmath78 & ( 14 ) @xmath79 & 2390 % + hcm6a & & & young burst + + & & & composite & young+``old '' & negligible & @xmath80 0.4 - 0.8 & @xmath81 & @xmath82 & @xmath25 40 % + [ tab_props ]
overall the published sed of hcm 6a ( see fig . [ fig_sed_6a ] ) is `` reddish '' , showing an increase of the flux from @xmath23 to @xmath83 and even to @xmath84 and @xmath85 seems weak , and difficult to understand . ] . from this simple fact and the above explanations it is already clear qualitatively that one is driven towards solutions with a ) `` advanced '' age and little extinction or b ) constant or young star formation plus extinction .
however , a ) can be excluded as no ly@xmath2 emission would be expected in this case .
quantitatively , the best solutions obtained for each spectral template group is shown in the left panel of figure [ fig_sed_6a ] . indeed ,
the solutions shown correspond to bursts of ages @xmath5 50130 myr ( bccww+ , s03 templates ) and little or no extinction .
however , as just mentioned , solutions lacking young ( @xmath26 10 myr ) massive stars can be excluded since ly@xmath2 emission is observed .
the best fit empirical sb+qso template shown corresponds to the spectrum of the h ii galaxy sbs 0335 - 052 with an additional extinction of @xmath86 to reconcile the observed sed with ly@xmath2 , a young population , e.g. such as sbs 0335 - 052 , or constant sf is required . in any of these cases fitting the `` reddish '' sed requires a non negligible amount of reddening . to illustrate the typical range of possible results we show in fig .
[ fig_contour_6a ] @xmath43 contour maps and corresponding confidence intervals for solar metallicity models ( s03 + template group ) and reddened with the calzetti law .
the left panel ( burst models ) illustrates in particular the need for progressively higher extinctions the younger the bursts .
all ages @xmath25 10 myr are excluded for the absence of ly@xmath2 emission . from the constant sf models ( right panel )
we see that for a given age @xmath72 is typically @xmath5 0.51.8 mag at the 68 % confidence level . for obvious reasons
, no constraint can be set on the age since the onset of ( constant ) sf .
hence , from the photometry of hcm 6a and from the presence of ly@xmath2 we are led to conclude that this object must suffer from reddening with typical values of @xmath87 for a calzetti attenuation law . a somewhat smaller extinction ( @xmath88 ) can be obtained if the steeper smc extinction law of prvot et al .
( 1984 ) is adopted . from the present data it is not possible to distinguish the different extinction / attenuation laws . also , it is not possible to draw any constraints on the metallicity of hcm 6a from the available data ( cf .
[ s_uv ] ) .
what if we are dealing with composite stellar populations ?
indeed it is conceivable that the ly@xmath2 emission originates from a population of young stars and the `` reddish '' restframe uv flux be due to another , older population .
assuming constant sf , no loss of ly@xmath2 and standard sfr conversion factors , the observed ly@xmath2 emission implies a maximum uv flux of the order of 0.1 @xmath71jy ( and approximately constant in @xmath89 over the observed wavelength range : @xmath90 ) for an unreddened population
. the bulk of the observed flux could then be from an older population . in this case the rising spectrum from the over the jh ( and presumably k ) bands could even be due to an unreddened population ; a strongly increasing flux and probably a significant `` balmer '' break is then expected , similar to the aged burst shown in the right panel of fig .
[ fig_sed_6a ] .
this explanation should in principle be testable with spitzer observations as discussed below .
how does our possible indication for a high extinction fit in with other studies ?
at redshift @xmath91 the extinction of lyman break galaxies ( lbgs ) has been estimated by various authors ( e.g. sawicky & yee 1998 , meurer et al . 1999 ,
adelberger & steidel 2000 , shapley 2001 , ouchi 2004 )
. given their mean / median values ( typically @xmath92 0.150.2 ) and the @xmath33 , our finding of a `` high '' extinction is not exceptional .
furthermore , armus ( 1998 ) find indications for @xmath16 mag from their analysis of a @xmath15 galaxy .
however , this implies that dust extinction is likely present in starburst galaxies with redshifts above 6 .
large amounts of dust have already been observed in qsos up to similar redshift ( bertoldi 2003 , walter 2003 ) . to estimate properties such as the mass , sfr , and the intrinsic ly@xmath2 emission from hcm 6a we simply examine the predictions from the best fit models , scale them appropriately to the luminosity distance one has @xmath65 64005.7 mpc . ] , and correct for the gravitational magnification ( here @xmath22 ) .
the derived quantities are summarised in table [ tab_props ] .
first we consider single ( non - composite ) stellar populations . from the best fit constant sf models ( with variable ages ) we deduce an extinction corrected star formation rate of the order of sfr(uv ) @xmath5 11 41 for a salpeter imf from 1 to 100 . for a
commonly adopted , although unjustified , salpeter imf down to 0.1 this would increase by a factor 2.55 .
actually this estimate is not very different than the one obtained from standard sfr calibrations provided the same assumptions on the imf . indeed the observed restframe uv luminosity , e.g. derived from the average j , h , and k@xmath93 flux of @xmath94 hz@xmath47 is @xmath95 hz@xmath47 ( hu et al .
2002 ) , translates to @xmath96 , where @xmath97 is the usual sfr conversion coefficient , and @xmath98 the uv extinction . for the standard value @xmath99 from kennicutt ( 1998 ) , assuming a salpeter imf down to 0.1 , and @xmath100 2.3 .
, one has @xmath101 3588 ; for the imf used in this work ( salpeter from 1 - 100 ) this becomes @xmath101 1434 .
this assumption , and the absence of extinction correction also explains the difference with @xmath102 estimate of hu et al .
( 2002 ) . from @xmath103 , whereas the classical kennicutt ( 1998 ) calibration would yield @xmath104 without extinction correction . ] for continuous sf over timescales @xmath105 longer than @xmath5 10 myr , the total ( bolometric ) luminosity output is typically @xmath106 per unit sfr ( in ) for a salpeter imf from 1 - 100 , quite independently of metallicity .
the total luminosity associated with the observed sf is therefore @xmath107 , close to or just above the limit to possibly qualify as a luminous infrared galaxy ( @xmath108 ; cf .
sanders & mirabel 1996 ) if a significant fraction of its bolometric flux emerges in the ( restframe ) ir . for @xmath109 10
myr the estimated stellar mass is @xmath110 . from the data given by hu et al .
( 2002 ) , the observed ly@xmath2 flux is @xmath111 , with the magnification factor @xmath71 .
the ly@xmath2 luminosity per unit sfr from the same s03 models used above is @xmath112 ( ) @xmath47 for metallicities between solar and 1/50 .
the sfr deduced from ly@xmath2 would then be sfr@xmath113 0.40.8 for hcm 6a .
taking an extinction of @xmath86 ( for the calzetti law ) into account implies a reddening corrected sfr@xmath113 712 .
the ratio between sfr@xmath114/sfr(uv ) presumably reflects the incomplete ly@xmath2 transmission @xmath115 , which can be estimated in various ways .
the most consistent estimate is obtained from the comparison of the predicted ly@xmath2 luminosity of each best fit model ( obtained from fitting the broad - band sed ) to the observed ly@xmath2 luminosity . from the best fit models with @xmath116 and the calzetti law we obtain @xmath117 2354 % ; in the case of the fit with the prvot et al .
extinction law we find a higher transmission @xmath117 90 % . for comparison haiman ( 2002 ) assumed a ly@xmath2 transmission of 20 % from the data of hu et al . per definition
this ly@xmath2 `` transmission '' corresponds to the ratio of the expected / intrinsic ly@xmath2 emission from the starburst over the observed one .
the physical causes for a partial ( i.e. @xmath118 100 % ) transmission are of course open to various interpretations ( e.g. physical processes destroying ly@xmath2 photons in the host galaxy , absorption in the intervening igm etc . ) .
in fact the relatively high ly@xmath2 transmission estimated here could be somewhat surprising , given the gunn - peterson trough observations in @xmath11 quasars ( cf .
becker 2001 , fan 2003 ) and the possible presence of dust ( this work ) .
consider now the case of composite stellar populations . in this case
we retain as a rough estimate in table [ tab_props ] the @xmath119 ( from the young population ) as a lower limit , and the mass and total luminosity is derived from the best fit burst model of age @xmath5 100 myr assuming that this `` older '' population dominates the observed continuum flux .
formally we then have no handle on the ly@xmath2 transmission , except that it can not be very low ( say @xmath26 40 % @xmath120 ) since otherwise the associated uv flux from the young population @xmath121 would dominate the observed continuum flux @xmath122 .
it is interesting to examine the seds predicted by the various models at longer wavelengths , including the rest - frame optical domain , which is potentially observable with the sensitive irac camera onboard the spitzer observatory and other future missions .
in the right panel of fig .
[ fig_sed_6a ] we plot the 3 best fits to the observed data for the bccww+ and s03 + template groups ( `` burst '' solutions with no extinction ) and the sbs 03352 - 052 template ( with additional @xmath123 ) showing strong optical emission lines .
we see that these solutions have fluxes comparable to or above the detection limit of irac / spitzer .
point - source sensitivity for low and medium backgrounds for frame times of 200s and described by fazio ( 2004 ) .
these values do not include `` confusion noise '' .
] on the other hand the strongly reddened constant sf or young burst solutions do not exhibit a balmer break and are hence expected to show fluxes just below the irac sensitivity at 3.6 and significantly lower at longer wavelengths . as ly@xmath2 emission is expected only for the reddened seds the latter solutions ( low 3.6 - 4.5 flux ) are predicted to apply to hcm 6a , except if composite stellar populations are considered .
indeed , a high 3.6 and 4.5 flux could be a good indication for a composite stellar population , as discussed above .
in addition it is important to secure higher accuracy photometry especially in the near - ir ( jhk ) to assess the accuracy of the redward increasing shape of the spectrum ( in @xmath89 ) , which drives one towards solutions with non - negligible extinction .
using sed fitting techniques considering a large number of parameters ( mainly a vast library of empirical and theoretical template spectra , variable extinction and extinction laws ) we have attempted to constrain the properties of the stellar populations and ly@xmath2 emission of two strongly lensed galaxies with redshifts @xmath19 6 from their observed sed including various ground - based observations , hst , and spitzer observations .
the following main results have been obtained for these objects ( see sects .
[ s_370 ] , [ s_kesr ] ) , and summary in table [ tab_props ] ) : * * triple arc in abell 2218 * discovered by kneib ( 2004 , kesr ) .
the most likely redshift of this source is @xmath3 6.07.2 taking into account both our photometric determination and lensing considerations .
+ sed fits indicate generally a low extinction ( @xmath4 ) but do not strongly constrain the sf history .
best fits have typical ages of @xmath5 3 to 400 myr . a reasonable maximum age of ( 250650 )
myr ( 1 @xmath6 interval ) can be estimated .
however , the apparent 4000 break observed recently from combination of irac / spitzer and hst observations , can also equally well be reproduced with the template of a young ( @xmath5 35 myr ) burst where strong restframe optical emission lines enhance the 3.6 and 4.5 fluxes .
+ the estimated sfr is typically @xmath5 1 for a salpeter imf from 1 - 100 , in agreement with previous estimates .
+ given the poor constraint on age and sf history , we conclude that intrinsic ly@xmath2 emission may or may not be present in this galaxy .
the apparent non - detection of ly@xmath2 by kesr can therefore even be understood without invoking ly@xmath2 destruction . * * abell 370 hcm 6a * discovered by hu ( 2002 ) . the relatively red sed and
the presence of ly@xmath2 emission indicate basically two possible solutions : 1 ) a young burst or ongoing constant sf with non - negligible extinction ( @xmath8 0.51.8 at a 1 @xmath6 level ) or 2 ) a composite young + `` old '' stellar population . + for the first case , best fits are obtained for constant sf with @xmath7 . in consequence
previous sfr estimates for this source must likely be revised upward .
if correct , the bolometric luminosity of this galaxy is estimated to be @xmath107 , comparable to the luminosity of infrared luminous galaxies .
furthermore a ly@xmath2 transmission of @xmath5 2390 % is estimated from our best fit models .
+ alternatively the observed 0.9 - 2.2 sed could also be fit without extinction by a composite `` young '' and `` old '' stellar population , where the former would be responsible for the ly@xmath2 emission and a fraction of the restframe uv flux .
the sfr , stellar mass , and total luminosity are then lower than in case 1 .
the two scenarios may be distinguishable with irac / spitzer observations at 3.6 and 4.5 .
+ given the limited observed spectral range , the present data does not allow to draw any firm constraints on the maximum age of the stellar population . in general it should also be noted that broad - band sed fits or measurements of the uv slope do not allow one to determine the metallicity of a star forming galaxy from a theoretical point of view and in terms of individual objects given important degeneracies ( cf .
[ s_uv ] ) .
the estimates of the ly@xmath2 transmissions presented here can in principle be used to constrain the intervening igm properties , and therefore probe the reionisation of the universe . although the results obtained here from this exploratory study of just two lensed galaxies , the highest known redshift galaxies with photometric detections in at least 34 filters , can not provide a general view on the sf and igm properties at @xmath11 , there is good hope that the sample of such objects will considerably increase in the near future with the availability of large ground - based telescopes and sensitive space borne observatories such as spitzer and even more so with the planned jwst .
we thank an anonymous referee for critical comments which helped to improve the paper .
we thank eichi egami and johan richard for comments on the hst and spitzer photometry of the arc in abell 2218 kesr , jean - paul kneib for information on keck spectroscopy of this object , and yuri izotov for communicating spectra of metal - poor galaxies .
part of this work was supported by the swiss national science foundation and the cnrs .
adelberger , k.l . ,
steidel , c.c .
2000 , , 544 , 218 ajiki , m. , et al . , 2003 , , 126 , 2091 armus , l. , et al . , 1998 , , 506 , l89 barkana , r. , loeb , a. , 2001 , physics report 349 , 125 becker , r.h . , et al . , 2001 , , 122 , 2850 bertoldi , f. , et al . , 2003 , , 409 , l47 bolzonella , m. , miralles , j .- m . , pell , r. , 2000 , , 363 , 476 bouchet , p. , et al . , 1985 , , 149 , 330 bremer , m.n , et al . , 2004 , , 347 , l7 bruzual , g. , charlot , s. , 1993 , , 405 , 538 bunker , a. , et al . , 2004 , , 355 , 374 calzetti , d. , kinney , a.l
storchi - bergmann , t. 1994 , , 429 , 582 calzetti , d. , et al .
, 2000 , , 533 , 682 charlot , s. , fall , m.s . , 1993 , , 415 , 580 coleman , d.g .
, wu , c.c . , weedman , d.w .
1980 , , 43 , 393 cuby , j .-
, et al . , 2003 , , 405 , l19 dawson , s. , et al . , 2004 , , 617 , 707 dickinson , m. , et al .
, 2004 , , 600 , l99 egami , e. , et al . , 2005 , , 618 , l5 eyles , l.p . , bunker , a.j . , stanway , e.r . ,
lacy , m. , ellis , r.s .
doherty , m. , 2005 , , submitted [ astro - ph/0502385 ] fan , x. et al .
, 2003 , , 125 , 1649 fazio , g.g .
, et al . , 2004 , , 154 , 10 gnedin , n.y . , prada , f. , 2004 , , 608 , l77 haiman , z. , 2002 , , 576 , l1 heckman , t.m . , robert , c. , leitherer , c. , garnett , d.r .
, van der rydt , f. , 1998 , , 503 , 646 hu , e.m . ,
mcmahon , r.g .
, cowie , l .. l . , 1999 , , 522 , l9 hu , e.m . ,
et al . , 2002 , , 568 , l75 ; erratum : , 576 , l99 izotov , y.i . ,
thuan , t.x . , 1998 , , 500 , 188 kauffmann , g. , et al .
, 2003 , 341 , 33 kennicutt , r.c .
, 1998 , araa , 36 , 182 kinney , a. , calzetti , d. , bohlin , r. c. , mcquade , k. , storchi - bergmann , t. , & schmitt , h. r. , 1996 , , 467 , 38 kneib , j.p . , et al . , 2004 , , 607 , 697 ( kesr ) kodaira , k. , et al . , 2003 .
pasj , 55 , l17 kunth , d. , et al . , 1998 , , 334 , 11 kunth , d. , et al . , 2003 , , 597 , 263 lehnert , m.d . , bremer , m. , 2004 , the messenger , 115 , p. 27
leitherer , c. , heckman , t.m . , 1996 , , 96 , 9 madau , p. , 1995
, , 441 , 18 malhotra , s. , rhoads , j.e , 2002 , , 565 , l71 malhotra , s. , rhoads , j.e , 2004 , , 617 , l5 mas - hesse , j. m. , kunth , d. , tenorio - tagle , g. , leitherer , c. , terlevich , r. j. , & terlevich , e. 2003 , , 598 , 858 meurer , g.r . ,
et al . , 1995 , , 110 , 2665 meurer , g.r . , heckman , t.m .
, calzetti , d. , 1999 , , 521 , 64 ouchi , m. , et al . , 2004 , , 611 , 660 papaderos , p. , izotov , y.i . , fricke , k.j . ,
thuan , t.x . ,
guseva , n.g . , 1998 , , 338 , 45 papovich , c. , et al . , 2004 , , 600 , l111 prvot , m.l . , et al , 1984 , , 132 , 398 rhoads , j.e . ,
malhotra , s. , 2001 , , 563 , l5 rhoads , j.e . ,
et al . , 2003 , , 125 , 1006 sanders , d.b . , mirabel , i.f . , 1996 , araa , 34 , 749 santos , m.r . , 2004 , , 349 , 1137 sawicki , m. , yee , h.k.c . , 1998 , , 115 , 1329 schaerer , d. , 2002 , , 382 , 28 schaerer , d. , 2003 , , 397 , 527 schaerer , d. , pell , r. , 2005 , in `` the dusty and molecular universe : a prelude to herschel and alma '' , eds .
: a. wilson .
esa - sp 577 , p. 121
[ astro - ph/0412246 ] seaton , m. , 1979 , , 187 , 73 shapley , a.e , et al . , 2001 , , 562 , 95 sirianni , m. , et al . , 2004 ,
pasp , submitted spinrad , h. , 2003 astrophysics update , topical and timely reviews on astrophysics .
springer - praxis books in astrophysics and astronomy .
john w. mason .
berlin : springer - verlag and chichester , uk : praxis publishing , 2004 . , p.155 , [ astro - ph/0308411 ] taniguchi , y. , shioya , y. , fujita , s. s. , nagao , t. , murayama , t. , ajiki , m. 2003 , jkas , 36 , 123 [ astro - ph/0306409 ] ; erratum : jkas , 36 , 283 taniguchi , y. , et al .
, 2005 , pasj , 57 , 165 tenorio - tagle , g. , et al . , 1999 , , 309 , 332 thuan , t.x . ,
izotov , y. i. , lipovetsky , v. a. 1997 , , 477 , 661 valls - gabaud , d. , 1993 , , 419 , 7 vanzi , l. , hunt , l.k . ,
thuan , t.x . ,
izotov , y.i . , 2000 , , 363 , 493 walter , f. , et al , 2003 , nature , 424 , 406 zheng , w. , et al . , 1997 , , 475 , 469 | we present an analysis of two strongly lensed galaxies at @xmath0 and @xmath1 for which multi - band photometric and spectroscopic observations are available . for one source
the data include recent hst and spitzer observations .
using an sed fitting technique considering a number of parameters ( various libraries of empirical and theoretical template spectra , variable extinction and extinction laws ) we attempt to constrain the properties of their stellar populations ( age , star formation history , mass ) and their intrinsic ly@xmath2 emission .
the following main results are obtained for the individual galaxies : * * triple arc in abell 2218 , probable @xmath1 galaxy : * the most likely redshift of this source is @xmath3 6.07.2 taking into account both our photometric determination and lensing considerations .
sed fits indicate generally a low extinction ( @xmath4 ) but do not strongly constrain the star formation ( sf ) history .
best fits have typical ages of @xmath5 3 to 400 myr . a reasonable maximum age of ( 250650 ) myr ( 1 @xmath6 interval ) can be estimated .
however , the apparent 4000 break observed recently from combination of irac / spitzer and hst observations , can also equally well be reproduced with the template of a young ( @xmath5 35 myr ) burst where strong restframe optical emission lines enhance the 3.6 and 4.5 fluxes .
the estimated sfr is typically @xmath5 1 for a salpeter imf from 1 - 100 , in agreement with previous estimates .
the unknowns on the age and star formation history could easily explain the apparent absence of ly@xmath2 in this galaxy . *
* abell 370 hcm 6a : * the available ly@xmath2 and continuum observations indicate basically two possible solutions : 1 ) a young burst or ongoing constant sf with non - negligible extinction or 2 ) a composite young + `` old '' stellar population . in the first case one obtains a best fit @xmath7 , or @xmath8 0.51.8 at a 1 @xmath6 level . in consequence
we obtain sfr @xmath5 1141 , higher than earlier estimates , and we estimate a fairly high total luminosity ( @xmath9 ) for this galaxy , in the range of luminous infrared galaxies . a ly@xmath2 transmission of @xmath5 2390 % is estimated from our best fit models .
other properties ( age , sf history ) remain largely unconstrained . in case of composite stellar populations
the sfr , mass , and luminosity estimate is lower .
the two scenarios may be distinguishable with irac / spitzer observations at 3.6 and 4.5 .
galaxies : high - redshift galaxies : evolution galaxies : starburst cosmology : early universe infrared : galaxies |
core - collapse supernovae ( ccsne ) are the signposts of recent massive star formation .
since much of the massive star formation is embedded in dust , a substantial fraction of the ccsne in the universe will remain hidden in optical searches @xcite .
this is particularly true in the case of star formation taking place in the dusty environments of luminous ( @xmath7)@xmath8 $ ] . ] , and ultra - luminous ( @xmath9 ) infrared ( ir ) galaxies ( lirgs and ulirgs , respectively ) , which dominates the star formation rate density at @xmath10 ( e.g. , * ? ? ?
in many lirgs , the bulk of star formation occurs within their circumnuclear regions ( @xmath11kpc ) , so that the need for high - resolution observations becomes crucial in the detection and study of ccsne therein .
the synergy between high resolution radio and near - infrared ( nir ) _ k_-band ( where the extinction is 10 times lower than in the optical ) observations is currently being used to help build a complete picture of the supernova ( sn ) activity in dusty starbursts and lirgs .
outstanding examples are sne 2000 ft ( in ngc7469 ; * ? ? ?
* and references therein ) , 2004ip ( in iras18293@xmath123413 ; * ? ? ?
* ; * ? ? ?
* ) , 2008cs ( in iras17138@xmath121017 ; * ? ? ?
* ) , and 2008iz ( in m82 ; * ? ? ?
* ; * ? ? ?
* ) where both high resolution ( to disentangle the emission of the sn from its host ) and reduced extinction measurements ( radio and nir ) , were essential in their detection and characterisation .
arp299 is a lirg with an ir luminosity ( @xmath13l@xmath14 @xcite , at an adopted luminosity distance of 44.8mpc , assuming @xmath15kms@xmath16mpc@xmath16 .
the system is composed of two interacting galaxies , whose major nuclei ( a and b1 ) and core components in the interacting region ( c@xmath17 and c ) are bright radio and nir emitters @xcite .
arp299 is a very prolific sn factory as proved by the detection of several radio sne and sn remnants ( snrs ) in the innermost nuclear regions of arp299a and arp299b @xcite , and the detection within the last 20 years of seven optical / nir sne in the circumnuclear regions of the system ( see * ? ? ?
* ; * ? ? ?
many attempts have been made to detect at radio wavelengths the sne occurring in the circumnuclear regions of arp299 , mostly under observing programmes with the very large array ( vla ) : as333 carried out from 1990 may to 1993 december ; as525 on 1994 february to detect sn1993 g ; as568 on 1999 january , february , april and october to detect sn1999d ; aw641 on 2005 february , june and august to detect sn2005u .
we are currently monitoring arp299 at radio wavelengths under programmes ap592 and ap614 , `` unveiling the hidden population of sne in local luminous infrared galaxies '' ( pi : m. a. prez - torres ) , as part of a combined radio and nir sn search in a sample of local lirgs . in this paper
we study the nature of the most recently detected sne in the system , 2010o and 2010p , by means of their late - time radio emission .
sn2010o was discovered at optical wavelengths on 2010 january 24 @xcite .
the object exploded on 2010 january 7 and lies on a location with intermediate extinction , @xmath18mag ( kankare et al .
it was classified as a type ib sn by @xcite based on a low - resolution optical spectrum obtained with the nordic optical telescope ( not ) .
@xcite reported an x - ray transient at the position of sn2010o prior to its explosion , and suggested that the progenitor of this sn was part of a wolf - rayet x - ray binary system , similar to those found in our galaxy .
sn2010p was discovered at nir wavelengths with the not on 2010 january 18 @xcite , a few days after explosion ( 2010 january 10 , kankare et al .
the spectrum of sn2010p obtained with the gemini - north telescope revealed a deficiency of hydrogen and matched with spectra of type ib / iib sne @xcite , and the absolute magnitude and colours from optical and nir observations with the not and the gemini - north telescope yielded a likely high host galaxy extinction of @xmath19 ( kankare et al .
2013 ) .
this paper accompanies @xcite by kankare et al .
( 2013 ) which presents a study of the early optical and nir emission of sne 2010o and 2010p .
here we analyse long - term follow - up radio observations of arp299 carried out by us and obtained from archives , with particular emphasis on characterising sne 2010o and 2010p . in section [ sec :
radio ] we give details on the data reduction and analysis .
we describe our results in section [ sec : results ] , which we then discuss in section [ sec : discussion ] . in section [ sec : concl ] we summarize the conclusions of our work .
we have collected observations of arp299 obtained with the multi - element radio linked interferometry network ( merlin ) , the electronic multi - element remotely linked interferometer network ( e - merlin ) , the vla of the national radio astronomy observatory ( nrao ) , and the european very long baseline interferometry ( vlbi ) network ( evn ) . in table
[ tab : radobs ] we show basic information for these observations ( all of which have an assigned label ) including the range of frequencies observed , the total time on target ( @xmath20 ) , and the peak intensity in each epoch of j1128@xmath215925 , which was used as the phase calibrator in all the observations we report here . | we report radio observations of two stripped - envelope supernovae ( sne ) , 2010o and 2010p , which exploded within a few days of each other in the luminous infrared galaxy arp299 .
whilst sn2010o remains undetected at radio frequencies , sn2010p was detected ( with an astrometric accuracy better than 1milli arcsec in position ) in its optically thin phase in epochs ranging from @xmath0 to @xmath1yr after its explosion date , indicating a very slow radio evolution and a strong interaction of the sn ejecta with the circumstellar medium .
our late - time radio observations toward sn2010p probe the dense circumstellar envelope of this sn , and imply @xmath2 / v_{\rmn{wind } } [ 10\,{\,\mbox{\ensuremath{\rmn{km\,s^{-1 } } } } } { } ] = ( 3.0$]5.1)@xmath3 , with a 5ghz peak luminosity of @xmath4@xmath5 on day @xmath6464 after explosion .
this is consistent with a type iib classification for sn2010p , making it the most distant and most slowly evolving type iib radio sn detected to date .
[ firstpage ] supernovae : general supernovae : individual : sn2010o supernovae : individual : sn2010p |
anthracnose is a phytopathology which attacks several commercial tropical crops such as coffee .
the anthracnose of coffee is known under the name coffee berry disease ( cbd ) and its pathogen is the _ colletotrichum _ _ kahawae _ , an ascomycete fungus .
the literature on anthracnose pathosystem is extensive @xcite .
there have been several attempts to model the spread of cbd and to identify efficient control strategies @xcite .
possible control methods include genetic methods @xcite , biological control @xcite , chemical control @xcite and cultivational practices @xcite .
chemical methods appear to be the most effective , but present ecological risks .
moreover , inadequate application of chemical treatments can induce resistance in the pathogen @xcite .
a dynamical spatial model of anthracnose infection that includes chemical control was proposed and analysed in @xcite ; this paper also showed how to optimize the use of the chemical control with respect to a given cost functional .
the disease dynamics were represented by an inhibition rate that satisfies a reaction - diffusion partial differential equation with coefficients that depend on space and time .
the present paper adds to the above model the possibility of a pulse control strategy that represents cultivational practices such as pruning old , infected twigs and removing mummified fruits .
such actions are commonly performed at discrete times at regular intervals .
an additional enhancement to the model results from our relaxing the regularity conditions on the model parameters that were imposed in @xcite .
the enhanced model is able to take into account the fact that in realistic situations the application of antifungal compounds is typically not continuous in time ( although the action of these compounds once applied is continuous ) .
the remainder of the paper has the following structure . in section [ system ] , we present the system model and explain the significance of the model parameters . in section [ wellposedness ]
we establish the well - posedness of the model @xmath1 ( under certain conditions ) and its spatially - averaged version .
section [ sectioncov ] proves existence of an optimal control strategy based only on the pulse strategy , for both the spatially - averaged model and the general model .
some properties of the optimal control strategy are proven , and an algorithm for finding the optimal pulse strategy is given , which applies both spatially - averaged and general models .
section [ sectioncouv]proves the existence of an optimal control strategy using simultaneously pulse and continuous strategies , for both the spatially - averaged and the general model .
some extremal properties of these strategies are also established . in section [ sectionmodsim ]
we present system simulations of both the spatially - averaged and general models that demonstrate properties of the optimal pulse - only control .
finally , in section [ discussion ] we summarize our conclusions .
the model of anthracnose infection discussed in @xcite expressed the disease dynamics in terms of an inhibition rate @xmath2 that satisfies the following equations : @xmath3 where @xmath4 is a positive real - valued function defined on @xmath5 ; @xmath6 is a real parameter satisfying @xmath7 ; @xmath8 is a real - valued function with @xmath9 , defined on the same domain as @xmath10 ; @xmath11 is a @xmath12 matrix function which is positive definite for all @xmath13 , defined on the same domain as @xmath10 ; @xmath14 is an open , bounded subset of @xmath15 ; @xmath16 is the boundary of @xmath14 , which is assumed to satisfy @xmath17 ; @xmath18 denotes the normal vector to the boundary at @xmath19 ; @xmath20 is a real - valued function satisfying @xmath21 for @xmath22 .
the various terms in the model equations @xmath23 have practical interpretations as follows .
( see reference @xcite for a more detailed description . )
the function @xmath10 represents the inhibition pressure , which depends on climatic and environmental conditions @xcite .
it is appropriate to model @xmath10 as an almost - periodic function , taking into account yearly seasonal changes .
the term @xmath24 accounts for the ( possibly anisotropic ) diffusive spatial spreading of inhibition rate in the open domain @xmath25 , where the matrix @xmath26 contains the space- and time - dependent diffusion coefficients .
the boundary condition @xmath27 guarantees that there is no net flux of inhibition rate between @xmath14 and its exterior .
the function @xmath8 expresses the influence of chemical control ( that is , the application of fungicides ) on the inhibition rate .
@xmath28 is the inhibition rate corresponding to epidermis penetration .
once the epidermis has been penetrated , the inhibition rate can not fall below this value , even under maximum control effort . in the current paper , we propose the following modified model , that includes the possibility of impulsive control : @xmath29 where @xmath10 , @xmath30 , @xmath6 , @xmath26 , @xmath14 , @xmath18 , and @xmath20 are as above ( except that @xmath10 and @xmath26 no longer depend on @xmath2 ) and @xmath31 is an increasing sequence of nonnegative reals such that @xmath32 ; @xmath33 ; @xmath34 , where @xmath35 denotes the volume @xmath36 ; @xmath37 is a threshold value such that the inhibition rate is not measurable under it and is observable for values greater than @xmath38 ; @xmath39 is a @xmath40$]-valued function defined on @xmath41 . in @xmath42 , the inhibition rate @xmath2
is assumed to be left continuous with respect to time : @xmath43 , where @xmath44 denotes @xmath45 .
( we shall also use the notation @xmath46 to denote @xmath47 ) cultivational practices are included in the model @xmath42 as follows .
the sequence @xmath48 represents times at which cultivational interventions are possible . for example , in coffee cultivation it would be reasonable to take the @xmath49 as regularly spaced with an interval of one week .
the intervals between intervention times reflect the fact that continuous exertion of cultivational interventions such as pruning is neither practical nor efficient . at each potential intervention time @xmath50
, intervention only takes place if the infection is sufficiently serious , as determined by the threshold condition @xmath51 .
the degree of intervention at time @xmath52 at each point @xmath41 is given by @xmath53 : @xmath54 corresponds to no change in the inhibition rate at @xmath55 , while @xmath56 reduces the inhibition rate @xmath57 to @xmath58 .
in this section we verify that the model @xmath42 is well - posed ( under certain conditions ) and that the inhibition rate @xmath59 always remains between @xmath58 and @xmath60 .
we first establish well - posedness of a spatially - averaged version of the model , and then use similar techniques to prove well - posedness of the spatially - dependent model .
a spatially - averaged version of the model @xmath42 may be obtained by taking spatial averages of the equations .
the averaged model is much simpler to work with than the general model ; however , the same tools used to prove well - posedness for the averaged model can be generalized to apply to the general model .
define @xmath61 if we suppose that @xmath62 @xmath30 and @xmath63 are functions of @xmath64 only and not of @xmath55 , then @xmath65 satisfies the following impulsive differential equation : @xmath66 , \label{modelpulse3ide}\ ] ] where @xmath67 note that the divergence term in ( [ modelpulse1 ] ) vanishes in the averaged model , due to the no - flux boundary conditions ( [ modelpulse3 ] ) .
a solution of @xmath68 is a piecewise absolutely continuous real - valued function that satisfies the following equation in each interval @xmath69 \tau_{k},\tau_{k+1}\right ] $ ] : @xmath70 we impose the following conditions on @xmath10 and @xmath30 to ensure the solution s existence and uniqueness : * ( h1 ) : * @xmath71 . *
( h2 ) : * @xmath72 \right ) $ ] .
[ boundednesside]if @xmath73 is a maximal solution of @xmath74 , then @xmath73 is @xmath0 -$]valued .
let @xmath73 be a solution of @xmath68 .
since @xmath75 $ ] it suffices to establish that the restriction of @xmath73 on @xmath69 t_{0},t_{1}\right ] $ ] is @xmath0 -$]valued .
let @xmath76 and @xmath77 .
we first prove that @xmath78 .
let @xmath79 t_{0},t_{1}\right ] $ ] be the set where @xmath80 is positive .
since @xmath80 is continuous , it follows that @xmath81 is open in @xmath69 t_{0},t_{1}\right ] $ ] .
suppose that @xmath81 is nonempty ; then @xmath81 is the disjoint union of open subintervals of @xmath69 t_{0},t_{1}\right ] $ ] .
let @xmath82 be one of these intervals . then from ( [ thetaintegral ] ) and the definition of @xmath80 , for @xmath83 we have @xmath84
but we know @xmath85 , since @xmath86 ; this contradiction implies that @xmath81 is empty .
it follows that @xmath87 on @xmath69 t_{0},t_{1}\right ] $ ] , which implies that @xmath88 on @xmath69 t_{0},t_{1}\right ] $ ] . to prove that @xmath89 on @xmath69 t_{0},t_{1}\right ] $ ]
, we may use an almost identical argument , with @xmath90 replacing @xmath91 .
[ existenceglobalsolutionide]the problem @xmath92 has a unique global solution . for existence , it suffices to establish existence of a local solution and use proposition [ boundednesside ] to conclude the result based on theorem 5.7 of @xcite .
it also suffices to restrict ourselves to the set @xmath93 $ ] .
the function @xmath94 \times\left [ 0,1\right ] \mapsto\alpha\left ( t\right ) \left ( 1-x/\left ( 1-\sigma u\left ( t\right ) \right ) \right ) $ ] is integrable with respect to @xmath64 , lipschitz continuous with respect to @xmath55 , and upper bounded by @xmath10 which is also integrable with respect to @xmath64 .
then by the carathodory theorem ( see @xcite ) there is a local @xmath0 -$]valued solution .
we now prove uniqueness of the solution . given that @xmath55 and @xmath95 are solutions on @xmath96 $ ] , we have that @xmath97 using the gronwall lemma , we get @xmath98 and more generally if @xmath99 @xmath100 $ ] @xmath101 it follows that the solution is unique and depends continuously on initial conditions .
it is important to notice that ( [ contin ] ) implies that the solution of @xmath92 is continuous with respect to control strategies @xmath30 and @xmath102 .
a solution @xmath2 of @xmath103 is a piecewise absolutely continuous function of time , such that @xmath104 the function @xmath105 and satisfies @xmath106@xmath107 for all @xmath108\tau_{k},\tau_{k+1}]$ ] and @xmath109 .
we also have @xmath110 , \text { } \forall x\in\overline{\omega}\subseteq\mathbb{r}^{3}. \label{weak_3}\ ] ] we further define a _ weak solution _
@xmath2 of @xmath111 to be a piecewise absolutely continuous function with respect to time which satisfies ( [ weak_3 ] ) and @xmath112 the function @xmath113 satisfies ( [ weak2 ] ) and the following weak form of ( [ weak_1 ] ) : @xmath114 where @xmath108\tau_{k},\tau_{k+1}]$ ] and @xmath115 .
we make the following additional assumptions in order to guarantee existence and uniqueness of the weak solution : * ( h3 ) : * @xmath116 ; * ( h4 ) : * @xmath117 @xmath118 ; * ( h5 ) : * @xmath119 such that @xmath120 @xmath121 * ( h6 ) : * @xmath122 \right ) \right ) $ ] ; * ( h7 ) : * @xmath123 \right ) $ ] .
as preliminary to proving existence and uniqueness of weak solutions , we first establish the boundedness of solutions .
the proof is similar to that of proposition [ boundednesside ] .
[ boundednessipde]if @xmath2 is a maximal solution of @xmath111 then @xmath2 is @xmath0 $ ] -valued .
let @xmath2 be a solution of @xmath103 .
since @xmath124 \right ) $ ] it suffices to establish the result for the restriction of @xmath2 on @xmath69 t_{0},t_{1}\right ] $ ] .
let @xmath125 and @xmath126 .
we first prove that @xmath127 .
let @xmath128 t_{0},t_{1}\right ] \times\omega$ ] be the set where @xmath80 is positive .
let @xmath129 , and let @xmath130 .
as in the proof of proposition [ boundednesside ] , we may choose an open subinterval @xmath131 which is a connected component of @xmath81 . for almost every time @xmath83 , the function @xmath132 and @xmath133 since @xmath134 , it follows that @xmath135 for all @xmath86 , which implies @xmath136 in @xmath137 .
it follows immediately that @xmath138 in @xmath139 , so that @xmath127 on @xmath69 t_{0},t_{1}\right ] $ ] .
a similar computation to the above can be used to show that @xmath140 in @xmath141 , from which @xmath142 follows immediately . from assumptions * ( h4)-(h5 ) * , it follows that the following problem has a unique solution in @xmath143 for an arbitrary but fixed time @xmath144.@xmath145{l}\operatorname{div}\left ( a\left ( t , x\right ) \nabla w\left ( x\right ) \right ) = f\left ( x\right ) , \text { } \forall x\in\omega\\ \left\langle a\left ( t , x\right ) \nabla w\left ( x\right ) , n\left ( x\right ) \right\rangle = 0,\text { } \forall x\in\partial\omega \end{array } \right.\ ] ] where @xmath146 .
theorems 3.6.1 and 3.6.2 of @xcite imply that there is a complete orthonormal system @xmath147 of eigenfunctions and eigenvalues @xmath148 such that @xmath149 @xmath145{l}\operatorname{div}\left ( a\left ( t , x\right ) \nabla\varphi_{n}\left ( t , x\right ) \right ) = \lambda_{n}\left ( t\right ) \varphi_{n}\left ( t , x\right ) , \text { } \forall x\in\omega\\ \left\langle a\left ( t , x\right ) \nabla\varphi_{n}\left ( t , x\right ) , n\left ( x\right ) \right\rangle = 0,\text { } \forall x\in\partial\omega \end{array } \right.\ ] ] moreover , the sequence @xmath150 is @xmath151valued , and if @xmath16 is of class @xmath152 then @xmath150 is @xmath153valued .
now we make the following assumption * ( h8 ) : * the sequence @xmath154 is independent of time ( that is , @xmath155 @xmath156 ) . in particular , * ( h8 ) * holds if @xmath157 has the form @xmath158 with @xmath159 , @xmath104 . in the special case where @xmath160 ( @xmath161 is the @xmath162 identity matrix ) , we have @xmath163 .
when * ( h8 ) * is satisfied , a weak solution @xmath2 of @xmath74 can be written as @xmath164 , where each @xmath165 is an absolutely continuous function of time that for @xmath166 \tau_{k},\tau_{k+1}\right ] , $ ] satisfies@xmath167 we are now ready to prove existence and uniqueness of weak solutions , given the additional conditions we have imposed .
the proof parallels that of proposition [ existenceglobalsolutionide ] .
[ existenceglobalsolutionipde]under conditions ( * h3*)(*h8 * ) , the problem @xmath103 has a unique global weak solution @xmath2 . moreover ,
if @xmath16 is of class @xmath152 then @xmath168 @xmath169 . for existence , it suffices to establish existence of a local solution and use proposition [ boundednessipde ] to conclude the result .
it also suffices to restrict ourselves to the set @xmath96 $ ] .
let @xmath170 denote the hilbert space of real - valued sequences @xmath171 such that @xmath172 .
consider the operator @xmath173 where @xmath174 \times\ell^{2}\left ( \mathbb{r } \right ) $ ] defined by @xmath175 the set @xmath176 } d\left ( \mathcal{g}\left ( t,.\right ) \right ) $ ] is a nonempty dense subset of @xmath177 since it contains the set of stationary sequences converging to @xmath58 which is also dense in @xmath178 .
@xmath179 is the infinitesimal generator of the evolution system defined from @xmath96 \times\left [ t_{0}^{+},t_{1}\right ] \times\bigcap\nolimits_{t\in\left [ t_{0}^{+},t_{1}\right ] } d\left ( \mathcal{g}\left ( t,.\right ) \right ) $ ] to @xmath178 by @xmath180 , @xmath181 ) @xmath182 hence , if we set @xmath183 , @xmath184 then the solution of @xmath111 is given by @xmath185 where @xmath186 .
just as with the averaged model , the solution of @xmath1 is continuous with respect to the controls @xmath30 and @xmath187 .
in this section we consider the behavior of controlled solutions on a fixed time interval @xmath188 , $ ] where @xmath189 .
practically , this interval may be taken as representing one annual production period .
as in previous sections , we first consider the averaged model , and then use analogous methods to obtain results for the main model .
although we are optimizing only with respect to the impulsive strategy , we retain the continuous control function @xmath30 as a fixed function . in section [ sectioncouv ] , we will optimize also with respect to @xmath30 . the aim of this subsection is to characterize the impulsive strategy @xmath190 which minimizes the following cost functional for the averaged model:@xmath191 where @xmath192 is a @xmath193valued sequence of cost ratios related to the use of impulsive control and @xmath194 is also a cost related to the final inhibition rate .
the existence of such an optimal strategy is guaranteed by the following proposition .
[ prop : exist ] there is an optimal strategy @xmath195 which minimizes @xmath196 .
note that @xmath197 , so it is possible to define @xmath198 ^{k+1}}{\inf}j\left ( v\right ) $ ] , and there is a sequence @xmath199 such that the sequence @xmath200 converges to @xmath201 .
since @xmath0 ^{k+1}$ ] is compact and @xmath196 is continuous , there is a subsequence @xmath202 which converges to @xmath203 ^{k+1}$ ] such that @xmath204 . in the remainder of this subsection
we characterize the optimal control strategy in such a way that it may be computed .
we have mentioned above that the solution of @xmath68 is continuous with respect to control strategies @xmath30 and @xmath187
. now we add the assumption that the solution of @xmath68 is gteaux differentiable with respect to @xmath205 .
let @xmath206 be the solution of @xmath92 associated with a chosen control strategy @xmath207 , and let @xmath208 be its directional derivative , @xmath209 then @xmath208 may be computed as follows .
let @xmath210 be the semiflow corresponding to @xmath68 .
@xmath210 satisfies @xmath211 \tau _ { i},\tau_{i+1}\right ] , $ ] @xmath212 and @xmath213 on the other hand @xmath214 we have @xmath211 0,\tau_{1}\right ] $ ] , @xmath215 and @xmath211 \tau_{i},\tau_{i+1}\right ] , i\in\mathbb{n } ^{\ast}$ ] we get @xmath216
it follows that @xmath217 and @xmath218 , we have @xmath219 also , @xmath220 we have@xmath221 note that @xmath208 can be written in the form @xmath222 , where the coefficients @xmath223 may be inferred from the above expressions for @xmath208 .
let @xmath224 be the directional derivative of @xmath196 in the direction of @xmath205 for a chosen control strategy @xmath207 .
from ( [ eq : jdef ] ) we may compute @xmath225 then we have [ eq : optstrategyaverage]for any control strategy @xmath207 we have @xmath226 where @xmath227 , @xmath228 ^{k+1}\right\ } , \ ] ] and @xmath229 is solution of the following adjoint problem : @xmath230 0,t\right ] \setminus\left\ { \tau_{i}\right\ } , \text { } i\in\left [ 0,k\right ] \cap\mathbb{n } ^{\ast}\text { } ; \nonumber\\ \text { } p_{\underline{v}}\left ( t\right ) & = c_f,\text { } p_{\underline{v}}\left ( \tau_{i}\right ) = c_{i}\left ( 1-\underline{v}_{i}\right ) + p_{\underline{v}}\left ( \tau_{i}^{+}\right ) \underline{v}_{i}. \label{eqadjp2}\ ] ] in particular , in the case where @xmath231 is an optimal solution then @xmath232 an argument similar to that in the proof of proposition [ existenceglobalsolutionide ] shows that there is a unique absolutely continuous solution @xmath229 to the problem @xmath233 which satisfies @xmath211 \tau_{i},\tau_{i+1}\right ] , $ ] @xmath234 and @xmath211 \tau_{k},t\right ] $ ] @xmath235 using integration by parts we get @xmath236 on the other hand , from ( [ eq : deriv_z ] ) we get @xmath237 equating these two expressions yields ( after rearrangement ) @xmath238 plugging this into expression ( [ eq : jv ] ) for @xmath239 and using ( [ eqadjp2 ] ) completes the proof of ( [ eq : condoptstrategy0 ] ) . in the case where @xmath207 is an optimal strategy @xmath231 , then for an abitrary but fixed @xmath240 and @xmath241 sufficiently small , we have @xmath242 and consequently @xmath243 the result above is a version of the maximum principle which characterizes the optimal strategy , but does not provide an efficient way to compute it .
the following proposition provides a direct means for computing an optimal strategy in the case where @xmath244 .
[ eq : optstrategyaverage2 ] there exists an optimal strategy @xmath245 such that @xmath231 belongs to the set @xmath246 and @xmath247{l}0\text{~~if } p_{v^{\ast}}\left ( \tau_{i}^{+}\right ) > c_{i}\text { and } \theta_{v^{\ast}}\left ( \tau_{i}\right ) \geq\sigma^{\ast},\\ 1 \text{~~otherwise}.\end{array } \right .
\ ] ] by proposition [ prop : exist ] , we know that an optimal control strategy @xmath231 exists . in the case
where @xmath248 , then ( [ eq : condoptstrategy ] ) requires that @xmath249 , which necessarily leads to @xmath250 ( since @xmath251 by the definition of @xmath252 ) .
similarly , if @xmath253 then ( [ eq : condoptstrategy ] ) requires that @xmath254 , which corresponds necessarily to @xmath255 .
if @xmath256 , then from ( [ eq : condoptstrategy ] ) we have that @xmath257 .
supposing that @xmath258 , we may increase @xmath259 without changing the value of @xmath260 .
( here we should note that increasing @xmath259 will not affect the condition @xmath261 for @xmath262 because of monotonicity . ) in particular , we may choose @xmath263 and still obtain a solution @xmath231 that minimizes the cost function . in the case where @xmath264 , proposition [ eq : optstrategyaverage2 ] provides a constructive method for obtaining @xmath231 , by using ( [ eqadjp1 ] ) and ( [ eqadjp2 ] ) to solve for @xmath265 in a backwards time direction starting from @xmath266 .
in this subsection we generalize the results of the the previous subsection by establishing existence of and characterizing a strategy @xmath267 which minimizes the following cost functional:@xmath268 where @xmath269 ;
@xmath270 @xmath271 \right ) , $ ] @xmath272 ; and @xmath109 , @xmath273 is a @xmath193valued sequence of cost ratios related to the use of control .
@xmath194 is also a cost related to the final inhibition rate . before proving the existence of an optimal strategy , we recall the following lemma stated in @xcite .
[ lemma : mazur ] ( mazur ) let @xmath274 be a sequence taking its values in a real banach space @xmath275 that is weakly convergent to @xmath276 .
then there exists a @xmath277valued sequence @xmath278 which converges strongly to @xmath55 and such that @xmath149 @xmath279 is an element of the convex hull of @xmath274 .
[ existenceofanoptimalstrategyv]there is an optimal strategy @xmath280 which minimizes @xmath281 .
the problem can be reduced to finding @xmath282 ^{k+1}$ ] , since the terms @xmath63 with @xmath283 have no effect on @xmath196 .
note that @xmath284 let @xmath285 .
there is a @xmath286 \right ) \right ) ^{k+1}-$]valued sequence @xmath287 such that the sequence @xmath288 converges to @xmath201 . the sequence @xmath287 is bounded and there is a subsequence @xmath202 which converges weakly to a strategy @xmath289 \right ) \right ) ^{k+1}$ ] .
lemma [ lemma : mazur ] implies there is a sequence @xmath290 in @xmath291 \right ) \right ) ^{k+1}$ ] which converges strongly to @xmath231 .
since @xmath281 is continuous , it follows that @xmath204 . in the remainder of this subsection
we characterize the optimal control strategy in order to compute it .
the solution of @xmath1 is continuous with respect to control strategies @xmath187 .
now we add the assumption that the solution of @xmath1 is gteaux differentiable with respect to @xmath292 \right ) \right ) ^{k+1}$ ] .
let @xmath293 be the solution of @xmath1 associated to a chosen control strategy @xmath207 and let @xmath208 be its directional derivative , @xmath294 . in analogy to the derivation of ( [ eq : deriv_z ] ) and ( [ eq : zv_tau_averaged ] ) in the previous section , we may show that @xmath109 and @xmath295@xmath296 and @xmath297 @xmath298 as in the previous section , we may define @xmath239 as the directional derivative of the cost functional @xmath299 as defined in ( [ eq : j_maindef ] ) .
a straightforward computation yields @xmath300 we may then state the following analogy of proposition [ eq : optstrategyaverage ] : [ maintheorempulse ] if @xmath245 is an optimal strategy then @xmath301 @xmath302 where @xmath303 \right ) \right ) ^{k+1}\right\ } , \end{aligned}\ ] ] and @xmath304 is solution of the following adjoint problem @xmath305 : @xmath306 0,t\right ] \setminus\left\ { \tau_{i}\right\ } , \text { } i\in\left [ 0,k\right ] \cap\mathbb{n } ^{\ast}\text { } ; \label{eqadjp1pde}\\ p_{v^{\ast}}\left ( t , x\right ) & = c_{f}\left ( x\right ) , \label{eqadjp2pde}\\ \text { } p_{v^{\ast}}\left ( \tau_{i},x\right ) & = v_{i}^{\ast}\left ( x\right ) p_{v^{\ast}}\left ( \tau_{i}^{+},x\right ) + c_{i}\left ( x\right ) \left ( 1-v_{i}^{\ast}\left ( x\right ) \right ) ; \end{aligned}\ ] ] and @xmath307 the proof resembles that of proposition [ eq : optstrategyaverage ] .
we first argue as in proposition [ existenceglobalsolutionipde ] that there is a unique absolutely solution @xmath304 to the problem @xmath308 .
we may then evaluate @xmath309 by parts , and alternatively by using expression ( [ eqzv1pde ] ) for @xmath310 . after equating the two expressions , integrating over @xmath14 , and rearranging we
obtain @xmath311 and substituting this expression into expression ( [ j_vmain ] ) for @xmath312 , we find after simplification that @xmath313 for an arbitrary but fixed @xmath240 and @xmath241 sufficiently small we have @xmath314 and consequently ( as before ) @xmath315 .
consider in @xmath316 \right ) \right ) ^{k+1}$ ] the inner product@xmath317 with its associated norm @xmath318 note that @xmath319 @xmath320 this allows us to find @xmath231 using the topology given by @xmath321 . since @xmath312 is a continuous linear operator acting on @xmath205 , there is a unique @xmath322 such that @xmath323 from theorem [ maintheorempulse ] we find @xmath324 \cap\mathbb{n } , $ ] @xmath325 @xmath326 this identification is useful in the implementation of the gradient method .
we close this section with a proposition which , like proposition [ eq : optstrategyaverage2 ] , enables the direct computation of the optimal strategy when @xmath264 .
[ eq : optstrategypulsefull ] if @xmath245 is an optimal strategy then @xmath327 .
moreover , there exists an optimal strategy @xmath328 such that @xmath329{l}0,\text { if } p_{v^{\ast}}\left ( \tau_{i}^{+},x\right ) > c_{i}\text { } \\ \text{and } \left\vert \theta\left ( \tau_{i},.\right ) \right\vert _ { l^{2}\left ( \omega\right ) } \geq\sigma^{\ast}\left\vert \omega\right\vert , \\ \\
1,\text { otherwise}.\end{array } \right.\ ] ] the proof is similar to that of proposition [ eq : optstrategyaverage2 ] .
it follows from proposition [ eq : optstrategypulsefull ] that the optimal strategy may be found when @xmath244 by backwards - solving for @xmath330 .
in this section we consider the possibility of a continuous control strategy employed along with the pulse control . as in the previous section ,
we restrict the time of study to the set @xmath188 $ ] where @xmath331 corresponds practically to an annual production period . in this subsection
we demonstrate existence and provide a characterization of a strategy @xmath332 which minimizes the following cost functional:@xmath333 where @xmath334 is almost everywhere positive and @xmath335 .
@xmath336 and @xmath337 are time dependent cost ratios related to the use of the control strategy .
it is evident that only the @xmath338 first terms of @xmath205 and @xmath337 are significant .
@xmath194 is a cost related to the final inhibition rate .
we also require that @xmath339 ; \left [ 0,1\right ] \right ) $ ] .
existence of an optimal strategy is guaranteed by the following proposition , which uses virtually the same argument as theorem [ existenceofanoptimalstrategyv ] .
[ existenceofanoptimalstrategyuv]there is an optimal strategy @xmath340 which minimizes @xmath196
. we may take @xmath203 ^{k+1}$ ] , since @xmath63 with @xmath283 do not affect @xmath196 .
note that @xmath341 } { \sup}c\left ( s\right ) \right ) t+{\displaystyle\sum\nolimits_{i=0}^{k } } c_{i}.\ ] ] let @xmath342 ; \left [ 0,1\right ] \right ) \times\left [ 0,1\right ] ^{k+1}}{\inf}j\left ( u , v\right ) .\ ] ] then there is a @xmath343 ; \left [ 0,1\right ]
\right ) \times\left [ 0,1\right ] ^{k+1}-$]valued sequence @xmath344 such that the sequence @xmath345 converges to @xmath201 .
the sequence @xmath346 is bounded , so there is a subsequence @xmath347 which converges weakly to a strategy @xmath348 ; \left [ 0,1\right ] \right ) \times\left [ 0,1\right ] ^{k+1}$ ] . using the lemma of mazur ,
there is a sequence @xmath349 in @xmath350 ; \left [ 0,1\right ] \right ) \times\left [ 0,1\right ] ^{k+1}$ ] which converges strongly to @xmath351 . since @xmath196 is continuous , @xmath352 .
now let @xmath353 be the solution of @xmath74 associated to a chosen control strategy @xmath354 , and let @xmath355 and @xmath356 . then using ( [ eq : phi ] )
we may compute for @xmath108\tau_{i},\tau_{i+1}]$ ] , @xmath357 in analogy with ( [ eq : zv_computation ] ) , we may derive for @xmath108\tau _
{ i},\tau_{i+1}]$ ] @xmath358@xmath359 thus @xmath360 satisfies : @xmath361 and @xmath362 , @xmath363 we may also compute @xmath364 [ prop : jujvopt1 ] if @xmath365 is an optimal strategy then @xmath366@xmath367 where@xmath368 ; \mathbb{r } \right ) \times\mathbb{r } ^{k+1};\exists\varepsilon>0;\right .
\\ & \left .
\left ( u^{\ast}+\varepsilon u , v^{\ast}+\varepsilon v\right ) \in l^{\infty}\left ( \left [ 0,t\right ] ; \left [ 0,1\right ] \right ) \times\left [ 0,1\right ] ^{k+1}\right\ } , \end{aligned}\ ] ] and @xmath369 is the solution to the problem @xmath370 with @xmath245 and @xmath371 .
the proof parallels those of proposition [ eq : optstrategyaverage ] and theorem [ maintheorempulse ] .
we may evaluate @xmath372 in two ways : by parts , and using the derivative expression ( [ eqzu1 ] ) .
equating the two results gives an expression for @xmath373 , which we may then plug into ( [ eqju1 ] ) and simplify to obtain the above expression for @xmath374 .
the conclusion @xmath375 follows as before from the fact that @xmath376 for abitrary but fixed @xmath377 and @xmath241 sufficiently small . for @xmath378 ;
\mathbb{r } \right ) \supset l^{\infty}\left ( \left [ 0,t\right ] ;
\left [ 0,1\right ] \right ) , $ ] define the inner product@xmath379 with its associated norm @xmath380 . note that @xmath381 ; \mathbb{r } \right ) , $ ] @xmath382 } { ess\sup}\left\vert u\left ( t\right ) \right\vert _ { 3 } $ ] ; thus we may find @xmath383 using the topology given by @xmath384 .
since @xmath385 is a linear continuous operator acting on @xmath30 , it follows there is a unique @xmath386 ; \mathbb{r } \right ) $ ] such that @xmath387 indeed , from proposition [ prop : jujvopt1 ] we find @xmath388 [ prop : jujvopt12]if @xmath365 is an optimal strategy then @xmath389 .
moreover , @xmath247{l}0,\text { if } p_{u^{\ast},v^{\ast}}\left ( \tau_{i}^{+}\right ) > c_{i}\text { and } \theta_{u^{\ast},v^{\ast}}\left ( \tau_{i}\right ) \geq\sigma^{\ast}\\ 1,\text { otherwise},\end{array } \right.\ ] ] and@xmath390{l}0,\text { if } c>\sigma\alpha p_{u^{\ast},v^{\ast}}\theta_{u^{\ast},v^{\ast}}/\left ( 1-\sigma\right)\\ 1,\text{otherwise . }
\end{array } \right.\ ] ] the proof is similar to that of proposition [ eq : optstrategyaverage2 ] .
the extremality of the pulse strategy follows as before . to show extremality of the optimum continuous strategy
, we suppose that @xmath391 is an optimum strategy , and for real parameters @xmath392 define an alternate continuous control strategy @xmath393}(t),\ ] ] where @xmath394}$ ] denotes the characteristic function for the interval @xmath395 $ ] . we may then compute @xmath396 } } ( u^{\ast } + \xi\chi_{[s , s+\delta ] } , v^{\ast } ) d\xi \\ & = \int_0^{\zeta } \langle \overline{u}_{\xi , s,\delta}\chi_{[s , s+\delta ] } \rangle d\xi , \end{aligned}\ ] ] where @xmath397})\right)^2}\ ] ] substituting into ( [ eq : ju ] ) , we find ( for @xmath398 ) @xmath399 where @xmath400 since @xmath196 assumes its minimum at @xmath391 , it follows that @xmath401 is minimized when @xmath402 ( note that @xmath403 because of restrictions on the control @xmath404 ) .
it is easily shown that @xmath405 on that interval , which implies that the minimum occurs at one of the endpoints , that is @xmath406 or @xmath407 . to find which , we compute @xmath408 which leads directly to the condition ( [ eq : optucondition ] ) .
in this subsection we survey the existence and we characterize of a strategy @xmath409 for the main model which minimizes the following cost functional:@xmath410 where @xmath411 is almost everywhere positive , @xmath270 @xmath271 \right ) , $ ] @xmath412 @xmath109 , @xmath413 .
@xmath336 and @xmath337 are time dependent cost ratios related to the use of the control strategy .
@xmath194 is a cost related to the final inhibition rate .
as before , we need only consider the first @xmath338 terms of @xmath205 and @xmath337 .
we may also consider the restriction of @xmath30 to @xmath188 $ ] , so that @xmath414 \times\omega;\left [ 0,1\right ] \right ) $ ] .
the following theorem generalizes theorem [ existenceofanoptimalstrategyv ] and proposition [ existenceofanoptimalstrategyuv ] .
[ existenceofanoptimalstrategyuvpde]there is an optimal strategy @xmath340 which minimizes @xmath196 .
the proof uses the same argument as in theorem [ existenceofanoptimalstrategyv ] and proposition [ existenceofanoptimalstrategyuv ] , based on the lemma of mazur .
let @xmath415 be the solution of @xmath111 associated to a chosen control strategy @xmath416 , and let @xmath417 and @xmath418 .
using methods we have demonstrated above , it may be shown that for @xmath419 , @xmath420 where @xmath360 additionally satisfies @xmath421 and @xmath422 furthermore , we may show @xmath423 we obtain finally the following generalization of [ maintheorem ] if @xmath365 is an optimal strategy then @xmath366@xmath424 where@xmath425 \times\omega;\mathbb{r } \right ) \times\left ( l^{\infty}\left ( \omega;\mathbb{r } \right ) \right ) ^{k+1};\\ v&=\left\ { \left ( u , v\right ) \in \mathcal{l}\,|\,\exists\varepsilon>0 , \left ( u^{\ast}+\varepsilon u , v^{\ast}+\varepsilon v\right ) \in \mathcal{l}\right\},\end{aligned}\ ] ] and @xmath426is the solution to the problem @xmath427 the proof follows the same lines as proposition [ eq : optstrategyaverage ] , theorem [ maintheorempulse ] , and theorem [ prop : jujvopt1 ] , albeit the calculations are more complicated . for @xmath428 \times\omega;\mathbb{r } \right ) \supset l^{\infty}\left ( \left [ 0,t\right ] \times\omega;\left [ 0,1\right ] \right ) $ ] , define the inner product@xmath429 with associated norm @xmath430 .
note that @xmath431 \times\omega;\mathbb{r } \right ) , $ ] @xmath432 is less than @xmath433 } { ess\sup}\left\vert u\left ( t , x\right ) \right\vert $ ] ; this allows us to find @xmath383 using the topology given by @xmath434 .
since @xmath435 is a linear continuous operator acting on @xmath30 , it follows there is a unique @xmath386 \times\omega;\mathbb{r } \right ) $ ] such that@xmath436 .
indeed , from theorem [ maintheorem ] we have @xmath437 [ prop : jujvopt13]if @xmath245 is an optimal strategy then @xmath438 .
moreover , @xmath329{l}0 , \text { if } p_{v^{\ast}}\left ( \tau_{i}^{+},x\right ) > c_{i}\text { and } \\ \qquad\left\vert \theta_{u^{\ast},v^{\ast}}\left ( \tau_{i},.\right ) \right\vert _ { l^{2}\left ( \omega\right ) } \geq\sigma^{\ast}\left\vert \omega\right\vert ; \\ 1 , \text { otherwise},\end{array } \right.\ ] ] and @xmath439{l}0 , \text { if } c>\sigma\alpha p_{u^{\ast}}\theta_{u^{\ast},v^{\ast}}/\left ( 1-\sigma\right ) ; \\
1,\text { if } c<\sigma\alpha p_{u^{\ast}}\theta_{u^{\ast},v^{\ast}}/\left ( 1-\sigma\right),\\ \end{array } \right.\ ] ] where @xmath336 is a function of @xmath64 and @xmath440 , and @xmath441 are functions of @xmath442 .
the proof is similar to that of proposition [ eq : optstrategyaverage2 ] .
in this section we present simulations which verify that algorithms based on propositions [ eq : optstrategyaverage2 ] and [ eq : optstrategypulsefull ] are effective in determing optimal pulse - only control strategies for the spatially - averaged and space - dependent models , respectively .
we present first results from the spatially - averaged model , which are easier to display graphically .
following @xcite we use an inhibition pressure of the form@xmath466 with @xmath467 $ ] .
this function is shown in figure [ fig : inhib_press ] ; it reflects the seasonality of empirically - based severity index models found in the literature @xcite . ,
scaledwidth=50.0% ] parameters used in the simulation are summarized in table [ simulationparamtable ] .
we considered only the case where the cost of pulse intervention is independent of time ( that is , @xmath468 is constant ) : the general features of the solution are still manifest in this special case .
[ c]cccparameter & significance & value + & & + & & + & & + & & + & & + & & + & & + & & + & & + & & + & & + & & + & & simulation results for the averaged model are displayed in figure [ fig : pulse_strategy ] , [ fig : chem_control ] , and [ fig : pulse_cf ] . in figure
[ fig : pulse_strategy ] , the optimal pulse - only control strategy is shown for a range of pulse control costs .
( the optimal pulse control strategy was calculated based on theorem [ eq : optstrategyaverage2 ] , with @xmath469 calculated as the numerical solution to the adjoint problem ( [ eqadjp1])([eqadjp2 ] ) . as expected , as intervention cost increases the amount of intervention decreases for the optimal solution .
it is interesting to note in figure [ fig : pulse_strategy ] that the sets of intervention times corresponding to successively larger intervention costs form a sequence of nested subsets .
however , we have not proven that this is true in general . )
was set equal to 0.,scaledwidth=50.0% ] figure [ fig : chem_control ] shows optimal pulse control for the same unit pulse control costs , but with a constant chemical control @xmath470 and inhibition rate attractor @xmath471 .
the maximum inhibition rate is reduced by the chemical control , and the number of pulse interventions is also reduced compared to figure [ fig : pulse_strategy ] . .
the markings below the time axis indicate times at which pulse control is applied for the corresponding control cost .
the unit cost at harvest ( @xmath472 ) was set equal to 0.,scaledwidth=50.0% ] figure [ fig : pulse_cf ] illustrates the effect of unit cost at harvest ( @xmath472 ) on the optimal strategy . in this case , increases in final control cost lead to increases in the application of pulse control that effectively reduce the final cost . as in figure [ fig : pulse_strategy ] , the pulse control applications for different final control costs form a sequence of nested subsets . note that if @xmath473 then @xmath472 has no further influence on the optimal strategy , because then it is always preferable to use pulse control immediately before harvest rather than to incur the harvest cost @xmath472 .
we also simulated the main model , using the parameters specified in tables [ simulationparamtable ] and [ general simulationparamtable ] .
we first considered the case where the inhibition pressure is independent of location , while the initial inhibition rate varies with location according to the functional form @xmath474 this function indicates an initial infection that is concentrated towards the center of @xmath14 .
the constants @xmath475 and @xmath476 were chosen such that the average initial inhibition rate agrees with table [ simulationparamtable ] .
[ c]cccparameter & significance & value + & & + & & + & & + & & figures [ fig : maincontrol1 ] and [ fig : maincontrol2 ] show the evolution under optimal pulse strategy for all grid points in a @xmath477 grid .
the systems represented by the two figures have different diffusion matrices ( @xmath478 for figure [ fig : maincontrol1 ] , @xmath479 for figure [ fig : maincontrol2 ] ) .
the figures confirm that when the inhibition rate and costs are spatially indepent , then the optimal strategy is also spatially independent : pulse interventions are always applied to the entire region , and never to a proper subregion .
furthermore , the optimal strategy does not depend on the initial inhibition rate distribution , or on the diffusion matrix @xmath26 .
all of these characteristics may be rigorously proven using the fact that the optimal strategy is derived from the solution of the adjoint system ( [ eqadjp1pde])([eqadjp3pde ] ) , which is independent of @xmath480 and is also independent of @xmath55 as long as @xmath472 and @xmath337 are independent of @xmath55 . since the optimal strategy is space - independent
, we find that in this case the spatial average of the general model agrees exactly with the averaged model .
this illustrates the practical usefulness of the averaged model , in the case where inhibition pressure and costs are independent of spatial location .
grid , for a system with @xmath481 and other parameters as given in tables [ simulationparamtable ] and [ general simulationparamtable ] .
the space - averaged version of this system is associated with the @xmath482 curve in figure [ fig : pulse_strategy ] .
note the optimal pulse interventions ( indicated by the markings below the time axis ) are the same in both cases .
, scaledwidth=50.0% ] except with diffusion matrix @xmath483.,scaledwidth=50.0% ] next , we consider the case where the inhibition pressure depends on spatial location . we did this by choosing @xmath484 in ( [ eqinhibitionpressure ] ) for each grid point independently according to a uniform random distribution , then rescaling so that the average value of @xmath484 agrees with table [ simulationparamtable ] .
other parameters from table [ simulationparamtable ] remain unchanged , and are constant with respect to the space variable . in particular , the initial conditions @xmath20 were taken as constant with respect to the space variable .
figure [ fig : maincontrol3 ] shows the inhibition rate as a function of time for all grid points of a @xmath477 grid when optimal pulse control is applied . unlike the cases shown in figures [ fig : maincontrol1 ] and [ fig : maincontrol2 ] , in this case the optimal strategy depends on spatial position . except with constant initial inhibition pressure , and with space - dependent inhibition rate.,scaledwidth=50.0% ]
this paper significantly extends the results of @xcite in several respects .
first , the model has been generalized by imposing weaker smoothness conditions on the parameters .
parameters are only required to be measurable and essentially locally bounded on @xmath485 . under these conditions ,
we have showed existence and uniqueness of a solution which takes values in @xmath0 $ ] , thus establishing the model to be well formulated mathematically and epidemiologically .
second , we have included the possibility of a pulse control strategy , along with the continuous ( chemical ) control strategy studied in @xcite . the added pulse strategy @xmath486 represents the cultivational practices such as pruning old infected twigs , removing mummified fruits @xcite .
third , we have verified an explicit algorithm for finding the optimal pulse - only strategy in the case where @xmath264 .
numerical simulations for both the averaged version of the model and the full version confirm the practical applicability of this algorithm . as explained in subsection
[ subsectionidemodel ] , the averaged version of the model faithfully represents the average behavior on the bounded domain @xmath14 , when inhibition pressure and intervention costs are space - independent .
the author thank doctor martial ndeffo mbah from the yale school of medecine and the university of cambridge for all his useful remarks concerning the presentation of the paper .
they also thank the african mathematics millennium science initiative ( ammsi ) for the financial support they gave to the first author in the year 2014 .
bella - manga , bieyss d. , mouen b. , nyass s. , berry d. , elaboration dune stratgie de lutte durable et efficace contre lanthracnose des baies du cafier arabica dans les hautes terres de louest - cameroun : bilan des connaissances acquises et perspectives , colloque scientifique international sur le caf , * 19 * , trieste , mai 2001 .
bieysse d. , bella - manga d. , mouen b. , ndeumeni j. , roussel j. , fabre v. and berry d. , lanthracnose des baies une menace potentielle pour la culture mondiale de larabica .
plantations , recherche , dveloppement , pp 145 - 152 , 2002 .
dodd j. , estrada a. , matcham j. , jeffries p. , jeger j. , the effect of climatic factors on colletotrichum gloeosporioides , causal agent of mango anthracnose , in the philippines , plant pathology ( * 40 * ) , pp 568 - 575 , 1991 .
durand n. , bertrand b. , guyot b. , guiraud j. & fontana t. , study on the coffea arabica / colletotrichum kahawae pathosystem : impact of a biological plant protection product , j.plant dis.protect . , * 116 * ( * 2 * ) , pp 7885 , 2009 .
ganesh d. , petitot a. , silva m. , alary r. , lecouls a. , fernandez d. , monitoring of the early molecular resistance responses of coffee ( coffea arabica l. ) to the rust fungus ( hemileia vastatrix ) using real - time quantitative rt - pcr , plant science * 170 * , pp 10451051 , 2006 .
mouen b. , bieysse d. , nyasse s. , notteghem j. , and cilas c. , role of rainfall in the development of coffee berry disease in coffea arabica caused by colletotrichum kahawae , in cameroon , plant pathology , 2009 .
mouen b. , bieysse d. , njiayouom i. , deumeni j. , cilas c. , and notteghem j.,effect of cultural practices on the development of arabica coffee berry disease , caused by colletotrichum kahawae , eur j plant pathol .
* 119 * : 391400 , 2007 .
mouen b. , chillet m. , jullien a. , bellaire l. , le gainage prcoce des rgimes de bananes amliore la croissance des fruits et leur tat sanitaire vis - - vis de lanthracnose ( colletotrichum musae ) , fruits , vol . * 58 * , p. 7181
. mouen b. , njiayouom i. , bieysse d. , ndoumbe n. , cilas c. , and notteghem j. , effect of shade on arabica coffee berry disease development : toward an agroforestry system to reduce disease impact , phytopathology , vol .
* 98 * , no . * 12 * , 2008
. muller r. , gestin a. , contribution la mise au point des mthodes de lutte contrelanthracnose des baies du cafier darabie ( coffea arabica ) due une forme du colletotrichum coffeanum noack au cameroun , caf cacao th , vol . *
xi * ( * 2 * ) , 1967 .
muller r. , la lutte contre lanthracnose des baies du cafier arabica due une souche de colletotrichum coffeanum au cameroun , note technique , institut franais du caf , du cacao et autres plantes stimulantes , 1971 .
silva m. , vrzea v. , guerra - guimares l. , azineira g. , fernandez d. , petitot a. , bertrand b. , lashermes p. and nicole m. , coffee resistance to the main diseases : leaf rust and coffee berry disease , braz . j. plant physiol . , * 18*(*1 * ) : 119 - 147 , 2006 . | in this paper we propose and study a spatial diffusion model for the control of anthracnose disease in a bounded domain .
the model is a generalization of the one previously developed in @xcite .
we use the model to simulate two different types of control strategies against anthracnose disease .
strategies that employ chemical fungicides are modeled using a continuous control function ; while strategies that rely on cultivational practices ( such as pruning and removal of mummified fruits ) are modeled with a control function which is discrete in time ( though not in space ) . under weak smoothness conditions on parameters we demonstrate the well - posedness of the model by verifying existence and uniqueness of the solution for given initial conditions .
we also show that the set @xmath0 $ ] is positively invariant . we first study control by pulse strategy only , then analyze the simultaneous use of continuous and pulse strategies . in each case
we specify a cost functional to be minimized , and we demonstrate the existence of optimal control strategies that can be evaluated numerically using the gradient method presented in @xcite . we discuss the results of numerical simulations both for a spatially - averaged version of the model and for the full model .
* keywords * anthracnose modelling , nonlinear systems , impulsive pde , optimal control . * ams classification * 49j20 , 49j15 , 92d30 , 92d40 . |
accreting black holes are thought to power the relativistic jets that form in active galactic nuclei ( agns ) , microquasars , and gamma - ray bursts ( grbs ) .
a plausible production mechanism for those jets is magnetic extraction of the spin energy of a kerr black hole@xcite .
a key feature of this process is a double trans - magnetosonic plasma flow which is launched from a stagnation radius located between the inner and outer light surfaces , and sustained by a plasma source in the magnetosphere . in general , there is a range of plasma injection rates within which the blandford - znajek ( bz ) process can be activated ; it has to be sufficiently high to provide the charge density required to establish an mhd flow , but low enough to avoid overloading of magnetic field lines that leads to a shutdown of the bz process .
the nature of the plasma source in most of the relativistic systems mentioned above is not well understood yet . in agns
the problem seems to be how to inject enough charged particles on open magnetic field lines .
direct feeding by the surrounding accretion flow seems unlikely , as charged particles would have to cross magnetic field lines on timescale shorter than the accretion time in order to reach the polar outflow .
free neutrons that may be produced in a radiative inefficient accretion flow ( riaf ) can cross field lines , however , they will decay over a distance typically much shorter than the horizon scale of a supermassive black hole .
thus , even if existent at sufficient quantity they will not reach the inner regions .
pair production on open filed lines via annihilation of mev photons is a plausible plasma source , but requires sufficiently hot accretion flow , and may be relevant only to faint sources like m87 and sgr @xmath0 .
it has been shown that the density of the charges thereby injected depends sensitively on the accretion rate and the conditions in the riaf @xcite .
naive estimates@xcite , although highly uncertain , suggest that in case of m87 this process can not provide complete screening at accretion rates that correspond to the inferred jet power , and to a fit of the observed sed by an advection - dominated accretion flow ( adaf ) model . as a consequence ,
an intermittent gap may form at the base of the flow in which the injected density is amplified to the required level by copious pair cascades induced by the potential drop in the gap .
this gap may be the source of the variable tev emission observed in m87 and , conceivably , some other non - blazar agns@xcite . in powerful blazars
the accretion flow is much cooler , hence direct charge injection should not ensue .
however , the disk luminosity is much higher and it could be that once pair creation is initiated in the gap by a stray charge , it will be sustained forever in a cyclic process , owing to inverse compton scattering and pair creation on the dense target photon field , as demonstrated in the case of pulsars@xcite . in grbs
the plasma injection rate is anticipated to be always well above that required to establish a mhd flow . however , under certain conditions , it may lead to overloading and a consequent shutdown of the bz process . in what follows
, we describe some recent studies@xcite wherein the role of plasma injection in the magnetosphere of a kerr black hole has been carefully examined .
we consider relativistic jets that are created near the polar region of a kerr black hole .
as mentioned above , the magnetosphere consists of a double - flow structure , whereby two plasma streams are launched , along every magnetic flux tube , in opposite directions from a stagnation radius @xmath1 located between the two light surfaces , where @xmath2 is the inclination angle of the flux tube . to analyze this structure ,
we have constructed@xcite a semi - analytic model that incorporates plasma injection in a self - consistent manner .
the injection process is modeled by prescribed source terms in the mhd equations , that determine the rate of change in mass , energy , angular momentum and entropy along streamlines .
this generalize the equations of ref . to the kerr geometry .
we identified two distinct types of solutions , that correspond to regimes where the bz process is switch - on or switch - off ( see figure [ f1 ] ) .
these two types of flows are characterized by the sign of the specific energy on the horizon , henceforth denoted by @xmath3 . for negative energy solutions ( @xmath4 ) the energy flux
is always directed outwards , implying energy extraction of the black hole spin energy . at sufficiently low injection rates
the emitted power is shown to converge to the bz result derived in the force free limit ( see below ) .
the dynamics of the flow in this regime is governed entirely by the frame dragging potential induced by the black hole . in case of positive energy solutions ( @xmath5 )
the dynamics of the flow is dictated by the external plasma source .
if the plasma injected in the magnetosphere is relativistically hot , then a pressure driven , double - flow is launched from the stagnation radius , whereby a fraction of the injected energy is absorbed by the black hole and the rest emerges at infinity in the form of a relativistic jet .
if the injected plasma is cold , an outflow may not form at all .
which type of flow will form under given conditions depends merely on the plasma injection rate , as discussed in the next section .
takahashi et al.@xcite have shown that two conditions must be satisfied in order for energy to be extracted from a kerr black hole : ( i ) the frame dragging potential must exceed the angular velocity of the magnetic field lines near the horizon , and ( ii ) the alfvn point of the inflow must be located inside the ergosphere . when these conditions are satisfied , the specific energy of a fluid element near the horizon , as measured at infinity , is negative , @xmath6 , implying an outward energy flux on the horizon .
a question of interest is how the efficiency of the extraction process depends on the load .
this question was not addressed in ref . .
in order to analyze the effect of the load , we computed the structure of the ideal mhd inflow emanating from the stagnation radius , for different plasma injection rates . to simplify the analysis
, we invoked an infinitely thin injection zone , whereby the mass injection profile is given by @xmath7 , and likewise for the energy - momentum source terms .
then , the specific energy @xmath8 is conserved at @xmath9 , and its value is uniquely determined by the regularity condition at the fast magnetosonic point .
the value of the enthalpy flux injected at the stagnation radius fixes the location of the slow point .
for further details see ref . . to elucidate key features , we consider first the zero temperature limit
an example is shown in figure [ f2 ] for an equatorial flow , where the extracted power @xmath10 is plotted against the injected mass flow rate @xmath11 .
the numerical values were computed assuming magnetic flux of @xmath12 g @xmath13 .
as seen , the extracted power converges to the force - free result , @xmath14 , derived in ref .
( marked by the horizontal dashed line ) at sufficiently small loads , but is strongly suppressed as the load approaches the critical value @xmath15 . from this , we concluded that a rotating black hole can transfer its rotational energy to the outflow only along field lines on which the accretion rate satisfies @xmath16 where @xmath17 $ ] . in ref .
it is shown that similar results are obtained in the general case of a hot flow .
the critical condition generalizes to @xmath18 , where @xmath19 is total power injected in the magnetosphere which , in the zero temperature limit , reduces to @xmath20 .
the immediate consequence of the above results for long grbs is that following the stellar collapse , the polar region must be devoid of matter in order for a jet to form .
but even then , there is another plasma source in the magnetosphere , namely annihilation of mev neutrinos that emanate from the hyper - accretion disk surrounding the black hole .
the plasma thereby deposited is relativistically hot , and so a polar outflow will be driven either by the black hole or by the pressure of the injected plasma , provided that the central region is baryon poor , as explained above .
mev neutrinos are emitted from the inner disk region , with a sensitive dependence of the neutrino luminosity on accretion rate .
recent calculations of the annihilation rate around a kerr black hole@xcite yield a net energy deposition rate of @xmath21 erg s@xmath22 , for accretion rates ( henceforth measured in units of @xmath23 s@xmath22 ) in the range @xmath24 , where @xmath25 is the radius of the marginally stable orbit in units of @xmath26 . from our analysis
we estimate that for accretion rates @xmath27 the jet is powered by the black hole ( the first flow type ) , whereas for higher rates it is powered by the neutrino source .
the function @xmath28 satisfies @xmath29 , but otherwise depends weakly on @xmath30 . for @xmath31 it varies between 1 and @xmath32 in the range @xmath33 . in a future work , we will intend to generalize the analysis of ref . to realistic injection profiles , as those computed in ref . .
in a preliminary work@xcite , we already analyzed the structure of a neutrino driven flow in a schwarzschild geometry .
we restricted our analysis to unmagnetized and non rotating flows , since the hole itself is non rotating ( @xmath34 ) and the particles follow radial geodesics near the horizon .
we adopted an energy deposition profile of the form @xmath35 with @xmath36 for @xmath37 and @xmath38 for @xmath34 , where @xmath30 is the normalized black hole spin@xcite .
we derived the double transonic structure and computed the power of the outflow for different energy deposition profiles@xcite .
we requiered that both the inflow and the outflow solutions pass smootly through their sonic points .
we started the integration at the inner sonic point and adjusted the stagnation pressure to cross the outer one .
we found that for a given choice of the energy deposition profile , there exists a unique solution that passes through the inner and the outer sonic points .
near the horizon , the inflow moves along radial geodesics , while the position of the outer sonic point is determined by the pressure at the stagnation point .
we concluded that the outflow production efficiency @xmath39 , defined as the fraction of @xmath40 that emerges at infinity , is typically large , with @xmath41 , @xmath42 , @xmath43 for @xmath44 , @xmath45 and @xmath46 , respectively .
we also found that the specific entropy in the outflow is larger than usually thought ( see figure 2 of ref . ) , owing to the delayed acceleration of the flow , and pointed out the implications for prompt emission .
0 r. d. blandford and r. l. znajek , _ mon . not .
r. astron .
soc . _ * 179 * , 433 ( 1977 ) .
a. levinson and f. rieger , _ astrophys .
j. _ * 730 * , 123 ( 2011 ) .
m. mocibrodzka , c. f. gammie , j. c. dolence and h. shiokawa , _ astrophys . j. _ * 735 * , 9 ( 2011 ) .
a. levinson , _ phys .
_ , * 85 * , 912 ( 2000 ) .
a. neronov and f. a. , aharonian , _ astrophys
. j. _ * 671 * , 85 ( 2007 ) .
a. timokhin , _ mon . not .
r. astron .
_ * 408 * , 209 ( 2010 ) .
a. timokhin and j. arons , _ mon . not .
r. astron .
soc . _ * 429 * , 429 ( 2013 ) .
a. levinson and n. globus , _ astrophys .
j. _ * 770 * , 159 ( 2013 ) .
n. globus and a. levinson , loaded mhd flows in kerr spacetime , to appear in _ phys .
d15_. a. levinson , _ astrophys .
j. _ * 648 * , 510 ( 2006 ) .
m. takahashi , s. nitta , y. tatematsu and a. tomimatsu , _ astrophys .
j. _ * 363 * , 206 ( 1990 ) .
chen and a. beloborodov , _ astrophys .
j. _ * 657 * , 383 ( 2007 ) .
i. zalamea and a. beloborodov , _ mon . not .
r. astron .
soc . _ * 410 * , 2302 ( 2011 ) . | we discuss the role plasma injection plays in the formation of outflows in kerr spacetime . using a model for the double flow established in the polar region of a rotating black hole , we study the interplay between the different processes that can power the outflow . in particular , we find two types of flows with distinct properties that depend on the rate at which energy is deposited in the magnetosphere .
we discuss the implications of this result for gamma ray bursts outflows . |
we investigate the connections between radix representations for @xmath0 , self - affine tilings of @xmath1 , and haar - like scaling functions for multiresolution analyses and associated wavelet sets . in a separate paper
, we investigate the idea , also introduced by jeong @xcite , of radix representations for vectors in @xmath0 , or general point lattices @xmath2 ( @xmath3 a nondegenerate @xmath4 matrix ) @xcite .
we wish to consider expanding matrices which preserve @xmath5 . without loss of generality
, we may assume @xmath6 .
a matrix which preserves @xmath0 must have integer entries .
a _ dilation matrix _ for @xmath0 is an @xmath4 matrix @xmath7 with integer entries , all of whose eigenvalues @xmath8
satisfy @xmath9 .
note that for a dilation matrix @xmath7 , @xmath10 is an integer , with @xmath11 .
then @xmath12 has nontrivial cokernel .
let @xmath13 be a complete set of coset representatives of @xmath12 .
we call the elemen ts of @xmath13 _ digits_. we may associate a sequence of digits with each @xmath14 by the euclidean algorithm , as follows .
each @xmath14 is in a unique coset of @xmath12 , thus there exist unique @xmath15 and @xmath16 such that @xmath17 similarly , for each @xmath18 , @xmath19 , there exist unique @xmath20 and @xmath21 such that @xmath22 formally , we write @xmath23 if there exists a nonnegative integer @xmath24 such that @xmath25 for all @xmath26 , then the euclidean algorithm terminates and we say that @xmath27 has a radix representation with radix @xmath7 and digit set @xmath13 . let @xmath7 be a dilation matrix .
we say that the matrix @xmath7 _ yields a radix representation with digit set @xmath13 _
if for every @xmath14 there exists a nonnegative integer @xmath28 and a sequence of digits @xmath29 in @xmath13 such that @xmath30 that is , a dilation matrix @xmath7 yields a radix representation with digit set @xmath13 if _ every _ @xmath14 has a radix representation with radix @xmath7 and digit set @xmath13 .
let @xmath7 be a dilation matrix , and define @xmath31 let @xmath32 be a fundamental domain for @xmath0 , centered at the origin , @xmath33 in @xcite , we give the following two results about radix representations . [ rad_rep ]
let @xmath7 be an @xmath4 dilation matrix . if @xmath34 then @xmath7 yields a radix representation of @xmath0 with digit set @xmath35 .
[ big_enough ] for every dilation matrix @xmath7 , there exists a positive integer @xmath36 such that @xmath37 yields a radix representation with digit set @xmath38 .
radix representations are closely related to self - affine tilings of @xmath1 .
a measurable set @xmath39 gives a self - affine tiling of @xmath1 under translation by @xmath0 if 1 .
@xmath40 , and the intersection @xmath41 has measure zero for any two distinct @xmath42 ( tiling ) ; and 2 .
there is a collection of @xmath43 vectors @xmath44 that are distinct coset representatives of @xmath12 such that @xmath45 set @xmath46 with the digits @xmath47 for some digit set @xmath13 .
one can easily check that @xmath48 is a self - affine set .
we would like to be able to think of the elements of @xmath48 as the fractional parts of vectors in @xmath1 in the same way that the fractional par ts of real numbers lie in @xmath49 $ ] .
this is an accurate interpretation if @xmath48 is congruent to @xmath50 .
thought of another way , we would like @xmath48 to tile @xmath1 under translation by @xmath0 . [ radix<=>tiling&0 ]
let @xmath7 be a dilation matrix , and let @xmath13 be a digit set for @xmath7 .
then @xmath7 yields a radix representation with digit set @xmath13 if and only if the set @xmath51 tiles @xmath1 under translation by @xmath0 and the origin @xmath52 is in the interior of @xmath48 .
we split the proof of this theorem into a few lemmas . [ radix=>tiling ]
let @xmath7 be a dilation matrix , and let @xmath13 be a digit set for @xmath7 . if @xmath7 yields a radix representation with a digit set @xmath13 , then the set @xmath51 tiles @xmath1 under translation by @xmath0 . applying proposition 5.19 from @xcite the dilation matrix @xmath7 , digit set @xmath13 , and corresponding set @xmath53 satisfy 1 .
@xmath48 is a compact subset of @xmath1 ; 2 .
@xmath54 ; 3 .
@xmath55 ; and 4 .
@xmath48 contains an open set . to show that @xmath48 tiles @xmath1 under translation by @xmath0
, we must show that @xmath56 ( where @xmath57 denotes lebesgue measure ) .
we extend an idea of lagarias and wang ( @xcite , p.31 ) to show that @xmath58 for any distinct @xmath59 for which there is a radix representation with radix @xmath7 and digit set @xmath60 .
note that for any subset @xmath61 of @xmath1 , @xmath62 ( where @xmath43 ) .
in particular , @xmath63 additionally , property @xmath64 implies that @xmath65 for all @xmath66 .
then @xmath67 thus @xmath68 for all distinct @xmath69 .
since @xmath70 , @xmath71 for all @xmath66 .
so @xmath72 for all @xmath73 .
now consider @xmath74 for any @xmath14 . by hypothesis ,
@xmath75 for some @xmath76 , with @xmath77 and the @xmath47 .
then @xmath78 and @xmath79 then @xmath77 implies that @xmath80 if the set @xmath53 tiles @xmath1 , it is not necessarily true that @xmath7 yields a radix representation .
for example , in the case where @xmath81 , we can find a radix representation for all nonnegative integers with the digit set @xmath82 , o r for all non - positive integers with the digit set @xmath83 , but we can not represent all integers with a radix representation using any digit set @xcite . yet
@xmath84 $ ] ( with digit set @xmath85 ) , does tile @xmath86 under translation by @xmath87 .
similarly , if @xmath7 is the twin dragon matrix @xcite , @xmath88}\ ] ] then @xmath7 is a dilation matrix .
a digit set for @xmath7 is @xmath89 } , d_1 = { \left [ \begin{array}{r}{1}\\{0}\end{array } \right]}\},\ ] ] and the set @xmath48 generated by the twin dragon matrix with this digit set tiles @xmath90 under translation by @xmath0 @xcite , yet @xmath7 does not yield a radix representation of @xmath0 ( for example , the vector @xmath91}\ ] ] does not have a radix representation ) @xcite .
for the twin dragon matrix . ] in both of the examples , the origin @xmath52 is on the boundary of the tile @xmath48 , so that @xmath92 .
we claim in theorem [ radix<=>tiling&0 ] that this must be the case for all dilation matrices @xmath7 which give tiles @xmath48 but do not yield radix representations .
we first introduce a technical lemma .
[ technical ] let @xmath7 be a dilation matrix , and let @xmath13 be a digit set for @xmath7 . if @xmath93 then there exists an increasing subsequence @xmath94 of the positive integers and a sequence of vectors @xmath95 converging to @xmath52 such that @xmath96 for all @xmath19 . in the proof of lemma [ radix=>tiling ]
, we noted that the set @xmath48 is compact .
thus for any @xmath97 such that @xmath98 , @xmath99 , and there exists an open ball centered at @xmath100 , @xmath101 , such that @xmath102 .
let @xmath103 be a sequence of vectors in @xmath1 converging to @xmath52 with @xmath104 for all @xmath19 .
set @xmath105 , and notice that @xmath106 , with @xmath107 .
set @xmath108 and @xmath109 for @xmath110 .
then @xmath111 is a decreasing sequence of positive numbers , @xmath112 , and the open balls @xmath113 of radius @xmath114 centered at the vectors @xmath115 satisfy @xmath116 for all @xmath117 . by construction , if @xmath118 is any point in the ball @xmath113 for each @xmath19 , then @xmath119 and @xmath120 for all @xmath19 .
in particular , there exists an increasing subsequence @xmath94 of the positive integers such that @xmath121 is a fine enough lattice to ensure that @xmath122 for each @xmath19 .
we may choose some @xmath123 for each @xmath19 . by construction , @xmath96 for each @xmath19 , and
the sequence @xmath124 converges to @xmath52 .
[ radix=>0 ] let @xmath7 be a dilation matrix , and let @xmath13 be a digit set for @xmath7 .
if @xmath7 yields a radix representation with digit set @xmath13 , then @xmath52 is in the interior of the set @xmath48 generated by @xmath7 and @xmath13 .
we prove the desired result by contradiction .
assume that @xmath93 .
by lemma [ technical ] , let @xmath125 be a sequence of vectors in @xmath1 converging to @xmath52 such that @xmath126 ( for some increasing subsequence @xmath94 of the positive integers ) but @xmath96 for all @xmath19 .
we may write @xmath127 for some @xmath128 for each @xmath129 . since @xmath7 yields a radix representation
, there exists an integer @xmath130 for each @xmath18 and digits @xmath131 such that @xmath132 thus @xmath133 with @xmath134 and @xmath135 .
since @xmath48 is compact , @xmath136 is bounded above by @xmath137 for some @xmath138 .
then , since @xmath139 converges to @xmath52 , @xmath140 is also bounded above for sufficiently large @xmath129 .
we use the rough estimate that there exists an integ er @xmath141 such that for all @xmath142 , @xmath143 .
thus for @xmath142 , the integer vectors @xmath144 all belong to a finite subset of @xmath0 .
this implies that there exists an integer @xmath145 such that @xmath146 for all @xmath147 . for @xmath148 , @xmath149 , and thus @xmath150 for all sufficiently large @xmath129 .
this contradicts the choice of @xmath139 , thus our assumption that @xmath93 must be false .
we have shown that if a dilation matrix @xmath7 yields a radix representation with digit set @xmath13 , then the set @xmath53 tiles @xmath1 under translation by @xmath0 , and @xmath151 .
next we prove the converse .
[ 0&tiling=>radix ] let @xmath7 be a dilation matrix , and let @xmath13 be a digit set for @xmath7 . if the set @xmath53 tiles @xmath1 under translation by @xmath0 and if @xmath151 , then @xmath7 yields a radix representat ion with digit set @xmath13 . following the notation of @xcite , set @xmath152 the set of vectors that can be expressed with a radix representation of length less than or equal to @xmath153 .
note that @xmath154 . by construction , @xmath54 .
thus @xmath155 by the tiling hypothesis , @xmath156 for all distinct @xmath157 .
thus @xmath158 for all @xmath159 with @xmath160 , and @xmath161 let @xmath162 be an open ball centered at the origin such that @xmath163
. then @xmath164 the sets @xmath165 are expanding , with @xmath166 .
thus @xmath167 the opposite containment is true as well , since @xmath168 for each @xmath153 .
thus @xmath169 .
we have now completed the proof of theorem [ radix<=>tiling&0 ] .
recall that @xmath170 is the smallest singular value of the dilation matrix @xmath7 , and that @xmath32 is our canonical fundamental domain of @xmath0 , @xmath171 .
combining theorems [ rad_rep ] and [ radix=>tiling ] , we also have t he following corollary .
let @xmath7 be a dilation matrix , and let @xmath35 .
if @xmath34 then the set @xmath172 tiles @xmath1 under translation by @xmath0 .
self - affine tiles allow us to construct multivariable wavelet sets associated with multiresolution analyses .
we review some basic definitions from wavelet theory here .
1 . @xmath175 ; 2 .
@xmath176 ; 3 .
@xmath177 if and only if @xmath178 for all @xmath179 ; 4 . @xmath180 if and only if @xmath181 for all @xmath182 ; and 5
. there exists a function @xmath183 , called a _ scaling function _
, such that @xmath184 is a complete orthonormal basis for @xmath185 .
the existence of multiresolution analyses in dimension @xmath186 has been studied by a number of authors . in @xcite , grchenig and
madych showed that if @xmath187 is a scaling function for a multiresolution analysis , then @xmath61 mus t be an affine image of a self - affine tiling of @xmath1 under translation by @xmath0 .
they showed also that if @xmath48 is a set of the form @xmath172 with @xmath7 a dilation matrix and @xmath13 a digit set for @xmath7 , then @xmath188 is the scaling function for a multiresolution analysis ( note that @xmath189 ) .
lagarias and wang noted that all self - affine tiles @xmath48 which tile @xmath1 under translatio n by @xmath0 must be of this form @xcite . a scaling function that is the characteristic function of some measurable set
is called a haar - like scaling function ( after the haar scaling function , which is @xmath190}$ ] ) .
lagarias and wang studied necessary conditions for sets of the form @xmath51 to tile @xmath1 under translation by @xmath0 in a series of papers ( @xcite , @xcite , @xcite , @xcite ) .
much of their work in these papers concerned t he question of when a set @xmath51 tiles @xmath1 under translation by a sublattice of @xmath0 . in @xcite and @xcite
, they showed that if @xmath13 is a complete set of coset representatives of @xmath12 , then @xmath191 is a self - affine tile of @xmath1 under translation by some sublattice of @xmath0 .
they also studied some properties of the tiling set @xmath51 .
he and lau @xcite studied sets @xmath51 which tile @xmath1 under translation by a sublattice of @xmath0 as well ; in particular , they looked at possible digit sets @xmath13 . in order for @xmath192 to be a scaling function for a multiresolution analysis , however , we need @xmath48 to tile @xmath1 under translation by all of @xmath0 . lagarias and wang gave some necessary conditions for a dilation matrix @xmath7 to yield a haar - like scaling function in @xcite and @xcite .
they showed that all dilation matrices in dimensions @xmath193 and @xmath194 yield haar - like scaling functions .
our results below give a sufficient condition for a dilation matrix @xmath7 to yield a haar - like scaling function , in any dimension .
note that for dilation matrices that yield a haar - like scaling function , strichartz has shown that multiresolution analyses and associated wavelet bases with arbitrary regularity can be constructed @xcite .
let @xmath7 be a dilation matrix , and let @xmath13 be a digit set for @xmath7 such that @xmath7 yields a radix representation for @xmath0 with digit set @xmath13 .
let @xmath48 be the set depending on @xmath7 and @xmath13 defined above .
then @xmath188 is the scaling function for a multiresolution analysis . in particular , if @xmath7 satisfies @xmath34 and if @xmath13 is the set @xmath35 with @xmath195 , then @xmath188 is the scaling function for a multiresolution analysis .
let @xmath7 be a dilation matrix
. then there exists a positive integer @xmath36 such that for all integers @xmath196 there exists a multiresolution analysis associated with the dilation matrix @xmath197 . | we investigate the connection between radix representations for @xmath0 and self - affine tilings of @xmath1 .
we apply our results to show that haar - like multivariable wavelets exist for all dilation matrices that are sufficie ntly large . |
diffusion of a molecule to a spherical trap is a classical problem important in chemical kinetics . in an unbounded domain , the problem reduces to the smoluchowski theory of reaction kinetics . in the context of biological processes , intracellular transport of biomolecules and chemical reactions
occur within closed domains with complex geometries @xcite . as a first passage time problem , this is closely related to the narrow escape problem , where a diffusing molecule escapes a closed domain through a small opening on the boundary , and the long time behavior has been studied using matched asymptotics @xcite .
there are many examples of this type of first passage time problem in biological modeling , including transport of receptors on the plasma membrane of a dendrite @xcite , intracellular virus trafficking @xcite , molecular motor transport @xcite , binding of a transcription factor to a segment of dna within a nucleus @xcite , and export of newly transcribed mrna through nuclear pores @xcite .
consider a bounded domain @xmath0 , containing a small , absorbing spherical trap , @xmath1 , with radius @xmath2 centered at @xmath3 .
we denote by @xmath4 the exterior boundary surface to @xmath5 , and by @xmath6 the exterior boundary to @xmath7 .
the non - trap portion of @xmath5 is denoted by @xmath8 .
consider a molecule undergoing brownian motion within @xmath9 .
we denote by @xmath10 the probability density that the molecule is at position @xmath11 at time @xmath12 and has not yet encountered the trap . for @xmath13 the diffusion constant of the molecule
, @xmath10 satisfies the diffusion equation [ eq:50 ] @xmath14 where @xmath15 denotes the partial derivative in the outward normal direction , @xmath16 , to the boundary .
let @xmath17 label the random variable for the time at which the molecule first reaches @xmath18 . the first passage time cumulative distribution is defined as @xmath19 = 1 - \int_{\omega}p({{\boldsymbol{r } } } , t )
d{{\boldsymbol{r}}}.\ ] ] the solution to can be written in terms of an eigenfunction expansion with @xmath20 where the eigenfunctions and eigenvalues satisfy [ eq:8 ] @xmath21 and the eigenfunctions are orthonormal in @xmath22 .
we order the eigenvalues so that @xmath23 . in the limit that the radius of the trap vanishes , the smallest eigenvalue , subsequently called the principal eigenvalue , also vanishes ( i.e. , @xmath24 ) .
similarly , the corresponding eigenfunction , subsequently called the principal eigenfunction , approaches @xmath25 as @xmath26 .
corresponding to these limits , the first passage time @xmath27 and @xmath28 as @xmath29 . in what follows we let @xmath30 and @xmath31 denote the diameter and volume of the set @xmath32
. for @xmath33 the asymptotics of the principal eigenvalue are known , and given by @xmath34 @xcite ( see also @xcite ) .
note that to first order in @xmath2 , @xmath35 depends only on the volume of @xmath5 and not the domain geometry .
higher order terms which depend on other properties of the domain are discussed in the next section .
the small @xmath2 asymptotics of @xmath36 motivate a large - time approximation of @xmath10 , based on a separation of time scales .
truncating the eigenfunction expansion after the first term gives the long time approximation , @xmath37 note , however , that the initial condition is not satisfied by this expansion . instead , the initial condition is modified so that the molecule starts from a uniformly distributed initial position with @xmath38 in other words , the long time behavior depends very little on the initial position of the molecule because it is likely to explore a large portion of the domain before locating the trap .
the first passage time density is @xmath39 , where @xmath40 is given by . the long - time ,
@xmath41 , approximation of the first passage time density is then @xmath42 the first passage time is therefore approximately an exponential random variable , with mean @xmath43 } \sim \frac{1}{\lambda_{0 } } & , \quad \text{for } \lambda_{1}t \gg 1 ,
\\ \sim \frac{{\left\vert \omega \right\vert}}{4 \pi d \epsilon } & , \quad\text{for } \epsilon \ll \operatorname{diam}\omega .
\end{aligned}\ ] ] an exponentially - distributed first passage time is an important assumption in course - grained models , such as the reaction - diffusion master equation ( rdme ) @xcite .
( the rdme is a lattice stochastic reaction diffusion model which assumes that reacting chemicals are well mixed within a computational voxel . ) more broadly , exponential waiting times are essential for jump processes to be markovian .
the above long - time approximation motivates several questions .
first , when is the non - exponential , short - time behavior of the first passage time important ?
second , how does changing the initial position of the molecule effect the approximation ?
it follows from that the mean binding time is approximately independent of the initial position . on the other hand ,
as we show here the most likely binding time , called the mode , depends strongly on the initial position . recently
, the importance of the initial position in first passage times in confined domains has been studied in the context of chemical reactions @xcite , and shown to play a role in quantifying the difference between two or more identically distributed first passage times @xcite . more generally , to estimate spatial statistics for the position of the diffusing molecule it is necessary to obtain expansions of not just the first passage time density , @xmath44 , but also the solution to the diffusion equation , @xmath10 .
the first passage time problem in a confined domain has also been studied from the perspective of a continuous time random walk ( ctrw ) on a finite graph of size @xmath45 @xcite .
meyer and coworkers obtain exact results for the laplace transform , which is the moment generating function for the first passage time distribution , and expand the moments for large @xmath45 .
they then reconstruct the large @xmath45 expansion of the first passage time distribution from the moments .
these results can also be interpreted as an approximation of the first passage time distribution in the large volume limit .
this perspective is closely related to the one considered here ; instead of an expansion in large volume , we assume the domain volume is @xmath46 and expand in terms of the small radius of the target . motivated by these and other examples
, we develop a uniform in time asymptotic approximation as @xmath26 of the probability density , @xmath10 ( see ) , and the first passage time density , @xmath44 ( see ) , that accounts for non - exponential , small time behavior and the initial position of the molecule .
the paper is organized as follows . in section
[ sec : unif - asympt - appr ] we further develop the long time approximation and present the complimentary short time _ correction _ based on a pseudopotential approximation . adding these two estimates we derive a uniform in time asymptotic expansion of @xmath10 for small @xmath2 .
it must be emphasized that what we call the `` short time '' correction is _ not _ an asymptotic expansion of @xmath10 as @xmath47 , but instead is a correction that when added to the long time expansion for any fixed @xmath12 and @xmath48 gives an asymptotic expansion of @xmath10 in @xmath2 . in section [ sec : first - passage - time ] we use the results of section [ sec : unif - asympt - appr ] to derive a small @xmath2 expansion of the first passage time density ( through terms of order @xmath49 ) . finally , in section [ sec : spher - trap - conc ] these approximations
are compared to the exact solution , exact first passage time density , and several other statistics for a spherical trap concentric to a spherical domain .
our basic approach is to first split @xmath10 into two components : a large time approximation that will accurately describe the behavior of @xmath10 for @xmath41 , and a short time correction to this approximation when @xmath50 .
note , both are defined for all times , but the latter approaches zero as @xmath51 , and so only provides a significant contribution for @xmath52 .
it should be stressed that the short time correction is not an asymptotic approximation of @xmath10 as @xmath47 , but instead serves as a correction to the long time expansion for @xmath52 .
we write @xmath10 as @xmath53 where @xmath54 is the large time approximation and @xmath55 is the short time correction . we will take @xmath56}$ ] to be the long time approximation of the eigenfunction expansion of @xmath10 . with this choice , @xmath54 and @xmath55
satisfy the projected initial conditions @xmath57 here we have dropped the subscript and subsequently identify @xmath58 and @xmath59 as the principal eigenfunction and eigenvalue respectively . using and as initial conditions , and setting @xmath60 in , then gives @xmath61 as required . in the next two sections we derive asymptotic expansions of @xmath54 and @xmath55 for @xmath62 .
the expansion of @xmath54 is based off the principal eigenvalue and eigenfunction expansions developed in @xcite .
the expansion of @xmath55 adapts the pseudopotential method we first used in @xcite , where uniform in time expansions of @xmath10 and the first passage time cumulative distribution , @xmath63}$ ] , were obtained for @xmath64 and @xmath65 the origin .
we have found that a direct pseudopotential approximation of in bounded domains with neumann boundary conditions breaks down for large , but finite times .
for example , the direct pseudopotential based expansion of @xmath63}$ ] can become negative for large times .
this inaccuracy in the pseudopotential approximation arises from the non - zero steady state solution to the limiting @xmath66 equation . as we see in section [ sec : unif - asympt - appr - short - time ] , by projecting out the principal eigenfunction this problem is removed when expanding @xmath55 .
this motivated our use of the splitting @xmath67 . since the initial condition is an eigenfunction of the laplacian , the long time density is given by @xmath68 as discussed in the introduction ,
there are well known asymptotic approximations for small @xmath2 of @xmath69 and @xmath59 . these then determine the small @xmath2 behavior of @xmath70 .
the expansions of @xmath69 and @xmath59 are typically given in terms of the corresponding no - trap problem where @xmath71 .
let @xmath72 denote the fundamental solution to the diffusion equation in @xmath5 ( _ i.e. _ the @xmath71 problem ) , then @xmath73 with the no - flux neumann boundary condition @xmath74 we will also need the corresponding solution to the time - independent problem , the pseudo - green s function @xmath75 , satisfying @xmath76 with the no - flux neumann boundary condition @xmath77 and the normalization condition @xmath78 within the derivative terms in we can replace @xmath72 by @xmath79 . integrating the resulting equation in @xmath12 on @xmath80 , and using the uniqueness of the solution to with the boundary condition and the normalization , we find @xmath81 here the term @xmath82 is necessary to guarantee convergence of the integral .
finally , we denote by @xmath83 the value of the regular part of @xmath84 at @xmath85 , @xmath86.\ ] ] let @xmath87 .
as derived in @xcite , the asymptotic expansions of the principal eigenvalue and eigenfunction for small @xmath2 are @xmath88 and @xmath89 + \epsilon^2 { \bar{\psi}}. \end{aligned}\end{aligned}\ ] ] here @xmath90 denotes the spatial average of the second order term and is given by @xcite @xmath91 note , the second order term in is not explicitly derived in @xcite , but can be found by solving equation ( 2.20 ) ( of @xcite ) with the normalization .
the corresponding expansion of the initial condition , @xmath92 , in @xmath2 is @xmath93 where the functions @xmath94 are obtained from substituting the asymptotic approximations of the principal eigenfunction and eigenvalue into and collecting terms in @xmath2 .
we find that @xmath95 @xmath96 \\ & - \frac{{\hat{k}}^2}{{\left\vert \omega \right\vert}^2 } \int_{\omega } \left[u({{\boldsymbol{r}}},{{\boldsymbol{r } } } ' ) + u({{\boldsymbol{r}}_{0}},{{\boldsymbol{r}}}')\right ] u({{\boldsymbol{r}}}',{{\boldsymbol{r}}_{\rm{b } } } ) d { { \boldsymbol{r } } } ' , \end{split}\ ] ] so that the small @xmath2 expansion of @xmath70 is then @xmath97 } e^{-{\lambda_{\text{\tiny{lt}}}}t}. \ ] ] to construct an asymptotic approximation to @xmath98 for small @xmath2 , we replace the trap boundary condition by a sink term in the pde involving a fermi pseudopotential operator @xcite , subsequently denoted by @xmath99 .
the boundary condition @xmath100 for @xmath101 is replaced by the sink term @xmath102_{{{\boldsymbol{r}}}={{\boldsymbol{r}}_{\rm{b}}}}\\ \times\delta({{\boldsymbol{r}}}-{{\boldsymbol{r}}_{\rm{b } } } ) \end{gathered}\ ] ] for @xmath103 , in the special case that @xmath104 is the origin , this reduces to @xmath105}_{{{\boldsymbol{r}}}= { \boldsymbol{0 } } } \delta({{\boldsymbol{r}}}).\ ] ] before we proceed with the pseudopotential approximation , it is instructive to consider why for an equivalent problem in 1d , replacing the absorbing boundary condition with a sink term makes the problem easier to solve . to see how this idea breaks down in higher dimensions , and to motivate the pseudopotential operator ,
we apply the laplace transform to ( with @xmath106 replaced by @xmath55 and the initial condition modified to ) .
we replace the dirichlet boundary condition by a delta function absorption term on the right hand side .
if @xmath107 denotes the laplace transform of @xmath98 , then we find @xmath108 so that absorption by the target occurs when the center of the target is reached ( at some rate @xmath109 that is to be determined ) . using the green s function of the diffusion equation ,
we can write the solution as @xmath110 to solve the above equation , we need only take the limit @xmath111 and solve for @xmath112 . however , we observe that for dimensions greater than one @xmath113 , which is why the naive approach breaks down .
there are a few different methods for adapting this idea to work in higher dimensions , namely matched asymptotics @xcite and pseudo potential operators , which is the approach that we use here .
the pseudopotential operator , @xmath99 , was developed so that the operator @xmath114 on @xmath5 provides an asymptotic approximation in @xmath2 of @xmath115 on @xmath9 with a zero dirichlet boundary condition on @xmath116 @xcite .
it was originally constructed for approximating hard core potentials in quantum mechanical scattering problems @xcite , but has also been used in the estimation of diffusion - limited reaction rates in two - dimensional periodic systems @xcite .
the operator was derived in @xcite by expanding the eigenfunctions , @xmath117 , in a basis of spherical harmonics and then analytically continuing the domain of definition of each eigenfunction into the interior of the sphere , @xmath7 .
on @xmath5 , it was found that formally @xmath118 when @xmath64 , it has been shown that the asymptotic expansion for small @xmath2 of the solution to the diffusion equation with pseudopotential interaction agrees with the direct asymptotic expansion in @xmath2 of the exact solution to the diffusion equation with a zero dirichlet boundary condition , @xmath119 for @xmath101 , up through terms of order @xmath49 @xcite .
the pseudopotential approximation for @xmath98 is that @xmath120 for @xmath121 , with the initial condition and a no - flux neumann boundary condition on @xmath4 . in @xcite several approaches
are developed for rigorously defining pseudopotential interactions ( usually called point interactions or singular perturbations of the laplacian in those works ) .
following these works , in particular @xcite , we split @xmath122 into a regular part , @xmath123 , and a singular part , @xmath124 , so that @xmath125 here it is assumed that @xmath123 is `` nice '' as @xmath126 . in appendix [ ap : ppsolutrep ] we give a more detailed motivation for this representation . to find the asymptotic expansion of @xmath98 for small @xmath2 , we begin by formulating a closed integral equation for @xmath123 .
as described in @xcite , we can separate @xmath84 into a component that is regular at @xmath85 , denoted by @xmath127 , and a singular part , @xmath128 , so that @xmath129 note that the pseudopotential applied to the singular part of @xmath84 is zero .
the action of the pseudopotential on the representation is therefore @xmath130 } & = \epsilon{\hat{k}}{\left [ { \phi}({{\boldsymbol{r}}_{\rm{b}}},t ) + { \gamma}q(t)\right ] } \delta({{\boldsymbol{r}}}-{{\boldsymbol{r}}_{\rm{b}}}),\end{aligned}\ ] ] as @xmath131 by . substituting the representation of @xmath98 into , we find @xmath132 we enforce the point boundary condition that the delta function terms should cancel @xcite so that @xmath133 after substituting into and rearranging terms we find that @xmath134 if the starting position is close to the target , the last term on the right hand side is expected to be small .
it follows that space and time are approximately decoupled , which is consistent with the results of the ctrw approach found in @xcite . by , equation simplifies to @xmath135 using duhamel s principle we find that @xmath136 integrating by parts we find @xmath137 while the no - flux boundary condition implies @xmath138 using the two preceding identities , it follows that simplifies to @xmath139 } d{{\boldsymbol{r}}}'.\end{aligned}\ ] ] eliminating @xmath140 with the point boundary condition gives @xmath141.\end{gathered}\ ] ] we now use the integral equation to find an asymptotic expansion of @xmath98 in @xmath2 .
let @xmath142 similarly , we define the expansion of the regular part of @xmath98 by @xmath143 using we identify the expansion terms of @xmath98 as @xmath144 the principal eigenvalue and eigenfunction expansions of the previous section imply that @xmath145 substituting this expansion into yields @xmath146 @xmath147 and @xmath148 evaluating requires the calculation of @xmath149 .
let @xmath150 .
using we have that @xmath151 } g_0 ( { { \boldsymbol{r}}},{{\boldsymbol{r}}_{\rm{b}}},s ) ds \\ & \quad\qquad+ { { { \phi}^{(0)}}}({{\boldsymbol{r}}_{\rm{b}}},t ) \int_t^{\infty } g_0({{\boldsymbol{r}}},{{\boldsymbol{r}}_{\rm{b}}},s ) ds\\ & \quad\qquad - \frac{1}{{\left\vert \omega \right\vert } } \int_0^t { { { \phi}^{(0)}}}({{\boldsymbol{r}}_{\rm{b}}},s ) ds .
\end{split}\ ] ] combining this expression with the identity @xmath152 reusing , and taking the limit @xmath153 , we find @xmath154 g_0({{\boldsymbol{r}}_{\rm{b } } } , { { \boldsymbol{r}}_{\rm{b } } } , s)ds . \end{split}\ ] ] note , in the first integral @xmath155 will scale like @xmath156 as @xmath157 .
this singularity is weakened by the @xmath158 term , which formally scales like @xmath159 as @xmath157 ( for fixed @xmath160 ) . as such
, the overall singularity in @xmath159 is integrable .
similarly , @xmath161 will cancel the effective singularity in @xmath12 of the last integral .
we therefore find the recursive expansion formula that [ thm : shorttimedensityexpansion ] the asymptotic expansion of @xmath98 for @xmath62 is given by [ eq : shorttimedensexpan ] @xmath162 as a short time correction to @xmath70 , we expect as @xmath163 , @xmath164 ( away from the singularity at @xmath165 ) . using that @xmath166 , , and
it is immediate that @xmath167 for @xmath168 . in appendix
[ ap : pst2longtime ] we show that @xmath169 for @xmath168 .
let @xmath170 combining theorem [ thm : shorttimedensityexpansion ] with the long time expansion we find , [ thm : diffdensexpansion ] for @xmath62 , @xmath171}\\ & \quad + \epsilon \frac{{\hat{k}}}{{\left\vert \omega \right\vert } } \int_\omega g({{\boldsymbol{r}}},{{\boldsymbol{r}}}',t ) u({{\boldsymbol{r}}}',{{\boldsymbol{r}}_{\rm{b } } } ) d { { \boldsymbol{r } } } ' \\ & \quad - \epsilon { \hat{k}}\int_0^t g({{\boldsymbol{r}}},{{\boldsymbol{r}}_{\rm{b}}},t - s ) { f_{\text{\tiny{st}}}}(s ) ds .
\end{split}\ ] ] note , based on theorem [ thm : shorttimedensityexpansion ] one can derive an expansion of @xmath10 valid through terms of @xmath49 .
that said , this expression is of sufficient complexity that we do not summarize it here .
denote by @xmath17 the first passage time ( fpt ) for the diffusing molecule to exit through @xmath6 .
the fpt cumulative distribution is defined as @xmath172 = 1-\int_{\omega } p({{\boldsymbol{r}}},t)d{{\boldsymbol{r}}}.\ ] ] substituting into , we find that @xmath173 , where @xmath58 and @xmath59 are the principal eigenfunction and eigenvalue satisfying for @xmath174 . from it
follows that @xmath175 , so that @xmath176 define the cumulative distribution of a standard exponential random variable as @xmath177 then , the fpt cumulative distribution corresponding to the leading order asymptotic expansion of the long time approximation can be written as @xmath178 ( see introduction and ) . since @xmath179 , we write the uniform approximation to the fpt cumulative distribution in terms of the two time scales @xmath12 and @xmath180 . here
@xmath181 denotes a shrunken time - scale .
notice from that at leading order , @xmath182 . substituting , , and into and collecting terms in powers of @xmath2 yields @xmath183 , where @xmath184 y(\tau ) \\
+ \left(\epsilon { \hat{k}}- \epsilon^{2}{\hat{k}}^{2 } \gamma\right)\int_{0}^{t}{f_{\text{\tiny{st}}}}(s ) ds + \epsilon^2 { \hat{k}}\int_0^t { { { \phi}^{(1)}}}({{\boldsymbol{r}}_{\rm{b}}},s ) ds.\end{gathered}\ ] ] here @xmath185 and @xmath186 are defined in and @xmath187 is given by . in evaluating the various spatial integrals we have made use of the identities @xmath188 and @xmath189 .
an explicit asymptotic expansion of @xmath40 can then be obtained by using that @xmath190 .
the uniform approximation of the fpt cumulative distribution is therefore @xmath191 . by definition ,
the fpt density function is @xmath192 .
we denote the expansion of the long time scale approximation , @xmath193 , by @xmath194 ( see ) . formally differentiating the asymptotic expansion @xmath195
, we find the asymptotic expansion of @xmath44 for @xmath62 is given by @xmath196{f_{\text{\tiny{lt}}}}(t ) \\ + { \left(\epsilon { \hat{k}}- \epsilon^{2}{\hat{k}}^{2 } \gamma\right ) } { f_{\text{\tiny{st}}}}(t ) + \epsilon^2 { \hat{k}}{{{\phi}^{(1)}}}({{\boldsymbol{r}}_{\rm{b}}},t).\end{gathered}\ ] ] since we have derived the expansion of @xmath44 by formal differentiation of the expansion of @xmath40 , we obtain terms that are of higher order than @xmath49 in ( as @xmath197 is @xmath198 ) .
however , for brevity we ignore the @xmath2 dependence of @xmath199 when referring to the order of the approximation . in other words , when referring to the `` leading order '' , `` first order '' , or `` second order '' expansion of @xmath44 , we mean those terms arising from the derivative of the corresponding order expansion of @xmath200 , treating @xmath201 as @xmath46 . as such ,
the `` leading order '' expansion of @xmath44 will be @xmath197 , the `` first order '' expansion will be @xmath202 and the `` second order '' expansion will be .
to illustrate our asymptotic results we consider the problem of a diffusing molecule searching for a small spherical trap of radius @xmath2 centered at the origin .
we assume the trap is contained within a larger , concentric , spherical domain with unit radius .
as this problem is exactly solvable , we will use the exact solution formulae summarized in this section to study the accuracy of our asymptotic expansions from the preceding sections as both @xmath2 and the number of expansion terms are varied .
denote by @xmath203 the spherically symmetric probability density for a diffusing molecule to be a distance @xmath204 from the origin at time @xmath12 .
we assume the trap is centered at the origin , so that @xmath104 , and let @xmath205 , @xmath206 . for @xmath207
, we have that @xmath208 where @xmath209 denotes the boundary of the unit sphere .
the advantage of this geometry is that an exact solution to the diffusion equation is known @xcite .
we find @xmath210 where @xmath211},\ ] ] @xmath212 , and the eigenvalue @xmath213 is given implicitly by @xmath214 the corresponding first passage time density is @xmath215 where @xmath216\\ \times\left(\frac{(1 + \frac{\lambda_{n}}{d})\sin((r_{0}-\epsilon)\sqrt{\frac{\lambda_{n}}{d}})}{(1-\epsilon)(1+\frac{\lambda_{n}}{d } ) - 1}\right).\end{gathered}\ ] ] in the remainder of this section , we list the quantities necessary to compute the asymptotic expansions of @xmath217 and @xmath44 for small @xmath2 .
recalling that @xmath104 , @xmath218 is then given by @xcite @xmath219 it follows from that @xmath220 and from that @xmath221 the fundamental solution @xmath222 , where @xmath223 denotes the spherically - symmetric green s function for the @xmath71 neumann problem ( see appendix [ s : neugfappendix ] ) , is given by @xmath224 where @xmath225 with @xmath226 .
the eigenvalues , @xmath227 , satisfy @xmath228 note that by comparing to it follows that @xmath229 . integrating over the unit sphere we find
@xmath230}\\ & \quad + \frac{\epsilon { \hat{k}}}{{\left\vert \omega \right\vert } } \int_{0}^{1 } g(r , r',t ) u({{\boldsymbol{r}}}',{\boldsymbol{0 } } ) ( r')^{2 } dr ' \\ & \quad - \epsilon { \hat{k}}\int_0^t g(r,0,t - s ) { f_{\text{\tiny{st}}}}(s ) ds .
\end{split}\ ] ] the asymptotic expansion of the first passage time density , @xmath44 , can be evaluated directly from .
here we use and to express @xmath231 as an eigenfunction expansion .
we find that @xmath232 the short time correction to the first passage time density is given by @xmath233 while @xmath234 to evaluate the time convolution , @xmath235 g_0({{\boldsymbol{r}}_{\rm{b } } } , { { \boldsymbol{r}}_{\rm{b } } } , s)ds,\ ] ] in we use the python ` quad ` routine .
the integral is split into a short time portion , @xmath236 , and a long time portion , @xmath237 .
@xmath238 is chosen sufficiently small that @xmath239 can be approximated by a gaussian evaluated at the origin , @xmath240 , with the same absolute error tolerance we use in evaluating the preceding series ( see appendix [ s : numericspoints ] ) .
we now study the error between the exact spatial and first passage time densities from the preceding section , @xmath217 and @xmath44 , and their asymptotic approximations for small @xmath2 . in what follows
we keep @xmath241 , @xmath242 , and vary @xmath2 between @xmath243 and @xmath244 .
the tolerances we used in evaluating the various series of the previous section are given in appendix [ s : numericspoints ] . while we are interpreting our spatial and time units as non - dimensionalized , these choices are also consistent with using spatial units of @xmath245 and time units of seconds . with these units
the overall domain has roughly the radius of a yeast cell nucleus .
we may therefore interpret the trap as a dna binding site that a diffusing protein is searching for .
while trap radii for dna binding sites are not generally experimentally measured , the width of some dna binding potentials have been measured .
for example , the lexa protein binding potential was found to have a width of approximately @xmath246 @xcite .
the long time approximation of the first passage time density is the single exponential @xmath247 , with the time - scale @xmath248 .
the principal eigenvalue @xmath59 is given implicitly by ( with @xmath174 ) and has the asymptotic approximation @xmath190 ( see also ) .
hence , for small @xmath2 , the long time approximation of the first passage time density is asymptotic to @xmath249 . as described at the end of section [ sec : first - passage - time ] , we refer to @xmath197 as the leading order approximation of @xmath44 as @xmath250 .
( we will also interchangeably refer to @xmath197 as either the large time or long time approximation . ) , by @xmath199 .
observe that the error decreases like @xmath251 as expected from .,width=283 ] the implicit equation can be solved numerically to calculate @xmath59 to arbitrary precision by a root finding algorithm ( e.g. , newton s method ) . in fig .
[ fig : lambda_error ] we compute the relative error , @xmath252 , of the asymptotic approximation , @xmath199 , as compared to the numerically estimated value of @xmath59 ( computed to machine precision ) .
we see that as @xmath26 , the relative error between the two decreases like @xmath253 , as expected from .
( black curve ) and its asymptotic expansions for small @xmath2 at several time points .
the blue ( light gray ) curve gives the leading order expansion , @xmath254 , while the green dashed curve gives the first order expansion .
we use a logarithmic @xmath255-axis to emphasize the solution behavior near the target . @xmath256 and @xmath257 ( similar to the width of measured dna binding potentials @xcite ) . ,
width=302 ] in fig .
[ fig : space ] , we show the leading order spatial density approximation ( blue or light gray curve ) , the first order expansion ( green dashed curve ) , and the exact spatial density ( black curve ) .
these curves plot @xmath258 the expansion , and @xmath217 respectively .
the spatial density is shown as a function of @xmath204 at four different time points .
for this figure we set @xmath259 and @xmath257 .
the density is initially concentrated around the initial position at @xmath260 and slowly fills the region @xmath261 until the density is approximately uniform at @xmath262 .
the only visible difference between the leading order approximation and the exact result is near the absorbing boundary , @xmath263 , where the exact solution displays a boundary layer that is lost in the leading order approximation .
the first order expansion reintroduces this boundary layer and is indistinguishable from the exact solution at the scale of the graph . in the remainder of this section we focus on the approximation of the first passage time .
the only free parameters in the model are the radius of the trap , @xmath2 , and the initial distance from the trap , @xmath264 .
the long time approximation , @xmath197 , is independent of @xmath264 .
it follows that the accuracy of @xmath197 in approximating @xmath44 improves when the initial distance from the trap is large ( i.e. , @xmath265 ) . in other words ,
the long time approximation is best when the particle is likely to explore a large portion of the domain before locating the trap .
when the initial distance from the trap is small ( i.e. , @xmath266 ) , we might expect the short time contribution to be significant since there is a higher probability that the particle will quickly locate the trap before exploring the rest of the domain . in fig .
[ fig : density1 ] , we show the asymptotic expansion of the first passage time density for @xmath267 and @xmath268 .
with this choice the initial distance of the particle from the trap is small .
moreover , since the accuracy of the expansion should decrease as @xmath2 increases , taking @xmath267 demonstrates the worst case behavior of the expansion for biologically relevant values of @xmath2 . in fig .
[ fig : density1](left ) the density function is shown with @xmath12 on a log scale to accentuate the small time behavior .
there is a significant difference between the long time approximation ( near - flat , bottom , light blue curve ) and the exact solution ( uppermost , green curve ) .
the first and second - order uniform approximations correct for this difference . the large - time behavior is shown in fig . [ fig : density1](right ) with @xmath269 on a log scale . for all except the shortest times the curve is linear , reflecting the exponential long - time behavior . we see that on this time - scale there is very little visible difference between each curve . fig .
[ fig : density2 ] is the same as fig .
[ fig : density1 ] , except that @xmath270 so that the initial distance from the trap is larger . in this case
the peak in the density occurs at a larger time . in both cases ,
the qualitative difference between the exact solution and the long - time approximation is a time lag before the exponential long time behavior dominates .
the time - scale for this time lag is roughly the diffusive transit time to cover the initial distance from the trap ( i.e. , @xmath271 ) . the absolute error of these approximations is shown in fig .
[ fig : density_error1 ] for @xmath268 and @xmath270 . in both cases ,
the maximum error is noticeably decreased as the order of the asymptotic expansion is increased . comparing the first and second order expansions , we see the main increase in accuracy results for times less than @xmath262 . points in time where one of the approximations crosses the exact solution result in locally increased accuracy ( the cusp - like drops in the expansion errors ) .
interestingly , when @xmath272 the long time approximation is more accurate for large times than the first- or second - order uniform approximations .
note , however , the error in each expansion at these times is substantially smaller than for short to moderate times . .
solid curves show the error of the long - time approximation @xmath197 .
the dashed curves show the second order uniform approximation .
note that the @xmath273 and @xmath274 curves for the large - time approximation are indistinguishable.,width=302 ] finally , we examine the max norm error , @xmath275 , as a function of @xmath2 for different values of @xmath264 .
the time points this error was numerically evaluated over are the same as those used for the graphs in fig .
[ fig : density_error1 ] , and are given in appendix [ s : numericspoints ] .
the result shown in fig .
[ fig : nerror ] confirms the asymptotic convergence of the approximation as @xmath26 . the large - time approximation error ( solid lines )
shows linear convergence , while the second order uniform approximation error ( dashed line ) , which includes short time behavior , shows cubic convergence .
as stated in the introduction , the mean binding time is well approximated by the @xmath264-independent large - time approximation .
that is , @xmath276 } \sim 1/\lambda$ ] , where @xmath59 is given by . however , other statistics may be of interest that depend strongly on @xmath264 .
one example is the mode , defined as the most likely binding time , call it @xmath277 , where @xmath278 . since the large - time approximation is an exponential distribution ,
the corresponding approximation of the mode is @xmath279 . .
solid curves show the exact solution , dash dotted curves the first order approximation , and the dashed curves the 2nd order approximation.,width=302 ] in figure fig .
[ fig : mode ] , we compute the mode by numerically maximizing the first passage time density . the exact mode is compared to first ( dash dotted curves ) and second order ( dashed curves ) approximations of the mode as a function of @xmath264 . each of the indicated curves are drawn for three different values of @xmath2 .
for @xmath280 , the difference between each curve is indistinguishable .
notice that as @xmath2 decreases the mode increases , particularly for larger values of @xmath264 , indicating that the large time approximation of the mode becomes less accurate as @xmath29 .
although the first passage time of a brownian particle in a confined geometry is a well - studied problem , an analytical characterization that includes short - time behavior of the survival probability density has been unresolved .
the asymptotic approximation of the long - time behavior establishes a link between the spatial characteristics of the problem ( i.e. , the starting position of the particle and the space dependent survival probability density ) and the short time behavior .
that is , the long time approximation loses information about the initial position and treats the survival probability density as uniform in space . consequently , the long time approximation is insufficient if one is interested in statistics that depend on these spatial characteristics . using a multiple time - scale perturbation approach
, we develop a long time expansion and a corresponding short time correction to this expansion of the solution to the diffusion equation in a bounded domain containing a small , absorbing spherical trap .
the long time approximation is derived from the matched asymptotic expansions of @xcite , while the short time correction is derived by modification of the pseudopotential method used in @xcite . combining these expansions
we develop a uniformly accurate ( in time ) approximation of the survival time cumulative distribution and the first passage time density . to study the accuracy of our method , we consider a example problem where the domain and trap are concentric spheres . by assuming radial symmetry
, we have available for comparison the exact solution to the example problem .
our results show excellent quantitative agreement for all times over a range of physiologically realistic values of @xmath2 .
moreover , they demonstrate the applicability of our expansions to estimating statistics that depend critically on the initial position of the diffusing particle . our approach should also be applicable to two - dimensional systems and multiple targets .
pseudopotentials have already been used to approximate rates of diffusion limited reactions in two - dimensional periodic systems @xcite .
likewise , pseudopotentials were originally developed to study many - particle scattering problems @xcite . while we are unaware of their use for approximating first passage processes in many - body / target systems , it should be feasible to adapt the techniques previously used in the quantum mechanical scattering context , allowing the extension of our work to multi - target systems .
sai was supported by nsf grant dms-0920886 .
jmn was supported in part by the mathematical biosciences institute and the national science foundation under grant dms-0931642 .
we thank the referees for their helpful comments and suggestions .
in @xcite several approaches for rigorously defining pseudopotential - like interactions are presented ( usually called point interactions or singular perturbations of the laplacian in those works ) . in the approach of @xcite , the laplacian plus point interaction operator , @xmath281 , is rigorously constructed so as to be equivalent to the laplacian with pseudopotential , @xmath282 ( see for the definition of the pseudopotential , @xmath99 , and @xcite for details on the construction of @xmath283 ) .
the splitting is rigorously justified by these works , and in the context of the diffusion equation goes back at least as far as @xcite .
we now give a _
motivation for the splitting by studying the laplace transform of .
again , we refer to the references @xcite for the rigorous justification .
our analysis is similar to that given in section ii.a of @xcite ( where @xmath284 ) .
denote by @xmath285 the laplace transform of a function , @xmath286 . taking the laplace transform of
we find @xmath287 for @xmath121 and @xmath288 , with the neumann boundary condition that @xmath289 for @xmath290 .
we assume the coefficient of @xmath291 within the pseudopotential term is finite , and subsequently denote it by @xmath292 ( as in @xcite ) , @xmath293_{{{\boldsymbol{r}}}= { { \boldsymbol{r}}_{\rm{b}}}}\\ & \phantom{= } \quad \times \delta({{\boldsymbol{r}}}-{{\boldsymbol{r}}_{\rm{b}}}).\end{aligned}\ ] ] recalling that @xmath294 is the green s function for the @xmath66 problem , we may then write @xmath295 where @xmath296 subsequently denotes the first two terms . substituting the preceding equation for @xmath107 into the definition of @xmath292 we find @xmath297}.\ ] ] here we have split @xmath298 into a part that is regular at @xmath299 , @xmath300 , and an explicit singular part so that @xmath301 solving for @xmath292 we find @xmath302 here we see how the pseudopotential corrects the naive point sink approximation , as given by .
the addition of the radial derivative in the definition of @xmath99 allows the pseudopotential to remove @xmath303 type singularities in three - dimensions .
this allows the unknown coefficient , @xmath292 , to be determined . using the last equation for @xmath292
, we find that @xmath304 we may write @xmath305 where @xmath306 and @xmath307 as the singular part of @xmath84 is @xmath308 @xcite , @xmath309 is regular at @xmath85 for @xmath288 .
_ formally _ , taking an inverse laplace transform of gives the representation of @xmath98 .
in this appendix we show that as @xmath311 , @xmath312 for @xmath168 . as in the last appendix , @xmath285 will denote the laplace transform of a function , @xmath286 .
we first collect some basic identities that will aid in evaluating the limit : @xmath313 the first identity follows immediately from the definition of @xmath314 and . in the right hand side of we replace the @xmath315 terms with time integrals of @xmath316 by , switch the order of integration , and evaluate the spatial integral using the semigroup property of @xmath316 to find that @xmath317 } ds dt .
\end{gathered}\ ] ] as @xmath318 } ds = \\
u({{\boldsymbol{r}}_{0}},{{\boldsymbol{r}}_{\rm{b } } } ) - \int_{0}^{t } { \left[g({{\boldsymbol{r}}_{0}},{{\boldsymbol{r}}_{\rm{b}}},s ) - \frac{1}{{\left\vert \omega \right\vert}}\right ] } ds , \end{gathered}\ ] ] recalling the definition of @xmath319 we see that @xmath320 then follows by definition of the laplace transform .
finally , by we have that @xmath321 using , we have that @xmath322 . a dominated convergence argument then implies .
we are now ready to evaluate the limit of @xmath323 as @xmath311 . by dominated convergence and ,
it is immediate from that @xmath324 where the last line follows by . by definition of the laplace
transform , @xmath325 from the definition of @xmath326 we find that @xmath327 } \\ & \phantom{=}\quad + \frac{{\hat{k}}}{{\left\vert \omega \right\vert } s } { \left [ u({{\boldsymbol{r}}_{0}},{{\boldsymbol{r}}_{\rm{b } } } ) - { { \tilde{{\phi}}^{(0)}}}({{\boldsymbol{r}}_{\rm{b}}},s)\right ] } \\ & \phantom{=}\quad + \frac{{\hat{k}}}{{\left\vert \omega \right\vert } } \int_{\omega } { \tilde{g}}({{\boldsymbol{r}}},{{\boldsymbol{r}}}',s ) u({{\boldsymbol{r}}}',{{\boldsymbol{r}}_{\rm{b } } } ) d { { \boldsymbol{r}}}'.\end{aligned}\ ] ] substituting into , and using , , and the definition of @xmath90 we find @xmath328}.\end{gathered}\ ] ] the limit of the bracketed term can be evaluated by splitting @xmath315 and @xmath329 into regular and singular parts ( at @xmath85 ) .
we write that @xmath330 where @xmath331 is finite as @xmath47 . using we see that @xmath332 where @xmath333 denotes the laplace transform of @xmath334 . as such , @xmath335 } &
= { \tilde{r}}_0({{\boldsymbol{r}}_{\rm{b}}},{{\boldsymbol{r}}_{\rm{b}}},0 ) \\ & \phantom{= } - { \tilde{r}}_0({{\boldsymbol{r}}_{\rm{b}}},{{\boldsymbol{r}}_{\rm{b}}},s ) + \frac{1}{{\hat{k } } } \sqrt{\frac{s}{d}}.\end{aligned}\ ] ] evaluating the @xmath159 limit in , it follows that as @xmath47 , @xmath312 .
let @xmath223 denote the spherically symmetric solution to the diffusion equation , satisfying @xmath336 } , \quad r \in \left[0,1 \right),\\ { \frac{\partial g}{\partial r } } & = 0 , \quad r = 1,\end{aligned}\ ] ] with the initial condition that @xmath337 . with this choice ,
@xmath338 for @xmath339 the boundary of the unit sphere . here
@xmath340 denotes the solution to the corresponding three dimensional diffusion equation .
note also the normalization that @xmath341 by eigenfunction expansion
we find @xmath342 where the eigenvalues @xmath343 satisfy and we use the convention that @xmath344
when evaluating the series for the exact solution and asymptotic approximation , we sum until the magnitude of the last added term drops below a given error threshold .
we used an error threshold of @xmath345 for the exact solution and @xmath346 for the uniform approximation .
the figures are generated with 1000 equally - spaced points for @xmath347 and 500 equally - spaced points for @xmath348 .
34ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1103/revmodphys.85.135 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ , ed . , , vol .
( , , ) @noop _ _ ( , , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ , ed .
( , , ) @noop * * , ( ) @noop _ _ , no . ( , , ) @noop * * , ( ) @noop _ _ , ed .
( , , ) @noop * * , ( ) @noop * * , ( ) | the problem of the time required for a diffusing molecule , within a large bounded domain , to first locate a small target is prevalent in biological modeling . here
we study this problem for a small spherical target .
we develop uniform in time asymptotic expansions in the target radius of the solution to the corresponding diffusion equation .
our approach is based on combining expansions of a long - time approximation of the solution , involving the first eigenvalue and eigenfunction of the laplacian , with expansions of a short - time correction calculated by pseudopotential approximation .
these expansions allow the calculation of corresponding expansions of the first passage time density for the diffusing molecule to find the target .
we demonstrate the accuracy of our method in approximating the first passage time density and related statistics for the spherically symmetric problem where the domain is a large concentric sphere about a small target centered at the origin . |
most star - forming galaxies follow a relation between their stellar masses and star formation rates that has been well characterized up to @xmath16 ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the very existence and tightness of this relation suggests that these galaxies live in a state of equilibrium where their ability to form stars is regulated by the availability of gas and the amount of material they return to the circum - galactic medium through outflows ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
simultaneously , it downplays the importance of galaxy mergers in the global star formation budget of the universe @xcite and highlights the influence of secular processes with longer duty cycles such as gas accretion , bar formation and bulge growth @xcite .
we refer to this general framework as the equilibrium model " for galaxy evolution , the formalism of which is detailed in recent work ( e.g. * ? ? ?
* ; * ? ? ?
this paradigm was ushered in by a combination of large imaging surveys , detailed kinematics studies , molecular gas measurements , and theoretical efforts .
for example , large optical and infrared surveys have contributed by allowing for the accurate measurement of stellar masses and star formation rates in very large galaxy samples , often supported by significant spectroscopic observing campaigns .
recently , _
observations in the wavelength range of 70 - 500@xmath17 m with the pacs and spire instruments have further contributed , by providing direct calorimetric sfrs that have been used to improve and recalibrate other indicators ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
at the same time , near - infrared integral field spectroscopy measurements at @xmath14 have started revealing that a significant fraction of high redshift galaxies are rotation - dominated discs ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
these observations convincingly demonstrate that the high sfrs measured in these galaxies are generally not caused by major mergers , as was previously assumed by analogy with the local ulirgs which have comparable sfrs and are all major mergers @xcite . instead
, the high sfrs of @xmath14 galaxies are caused by high molecular gas fractions ,
well above the @xmath18% typically observed at @xmath10 @xcite .
some of the strongest direct evidence in favor of the equilibrium model above indeed comes from molecular gas observations . in the local universe , with the exception of ulirgss
, it has now been directly observed that the location of a galaxy in the sfr-@xmath4 plane is mostly determined by its supply of molecular gas , with variations in star formation efficiency playing a second order role @xcite .
similar conclusions have also been reached based on high - redshift galaxy samples , whether directly using co data @xcite , or indirectly using far - infrared photometry to estimate gas masses @xcite . also , outflows of molecular material , which are an important element in setting the balance between gas and star formation in the models ,
have now been directly observed in a range of objects ( e.g. * ? ? ?
but mostly , it is now possible to detect co line emission in normal star - forming galaxies at @xmath19 ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
these observations convincingly show that the rapid decline in the specific sfr of galaxies since @xmath14 can be explained by the measured gas fractions and a slowly varying depletion timescale ( @xmath20 , * ? ? ?
* ) . out
to @xmath21 it is possible to detect far - infrared emission and co lines in individual objects with current instrumentation .
however , even in the deepest _ herschel _ fields and with long integrations at the iram pdbi , it is still not possible to measure directly the dust and gas contents of individual normal star - forming galaxies with masses @xmath1@xmath2 at @xmath22 . a proven way to study molecular gas in high redshift galaxies with lower masses and lower
sfrs is instead to target objects that are gravitationally lensed @xcite .
samples of lensed star - forming galaxies have for example also been used to extend to higher redshifts and lower stellar masses studies of the kinematics and stellar populations of disc galaxies , of the mass - metallicity relation ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and of the origin of metallicity gradients in galaxy discs @xcite . in this paper
, we explore the relation between dust , gas and star formation at @xmath23 using a sample of 17 uv - bright lensed galaxies targeted for deep _ herschel _ pacs and spire observations and having sfrs and stellar masses characteristic of main - sequence objects at their redshifts .
we analyze the results of these observations in the context of the equilibrium model .
in addition , we present new iram pdbi observations that more than double the number of published lensed galaxies with co line measurements . after we describe the sample in [ sample ] , the multi - wavelength observations are presented in [ data ] , including the new herschel and iram pdbi observations . in
[ quantities ] , we describe how key quantities such as stellar masses , star formation rates , dust masses , dust temperatures , gas masses and metallicities were calculated homogeneously . the key results of this study are presented in
[ results ] and summarized in [ summary ] ; in short , we find that the @xmath5 lensed galaxies have low dust and gas masses , but high dust temperatures as a consequence of an efficient conversion of their gas into stars .
gas mass fractions and depletion times follow a redshift evolution out to @xmath24 that is consistent with the expected scaling relations under the equilibrium model .
all rest - frame and derived quantities in this work assume a @xcite imf , and a cosmology with @xmath25 km s@xmath26 mpc@xmath26 , @xmath27 and @xmath28 .
all molecular gas masses ( @xmath29 ) and derived quantities such as gas fractions and depletion times , presented and plotted include a factor of 1.36 to account for the presence of helium .
the main sample consists of 17 lensed galaxies that were selected for deep _ herschel _ pacs / spire imaging ( details of these observations are presented in [ herscheldata ] ) .
these galaxies , selected from the literature , have been discovered as bright blue arcs of conspicuous morphology and then spectroscopically confirmed to be high redshift lensed objects . as shown in
[ samplebias ] , their intrinsic properties are similar to uv - selected lyman break galaxies ( lbgs ) , or their @xmath14 bx / bm analogs , but the observed fluxes are strongly amplified .
the sources come both from the traditional method of searching for galaxies lensed by massive clusters , and from the recent searches for bright blue arcs in the sloan digital sky survey ( sdss ) , with lenses that typically are individual luminous red galaxies ( e.g. * ? ? ?
in addition , we also consider the submm - identified eyelash galaxy @xcite , as it is located within the field of view of our _ herschel| _ observations of the cosmic eye .
the sample is by no means a complete census of uv - bright galaxies , but instead samples some of the best - known objects with rich multi - wavelength observations , including near - infrared imaging and spectroscopy and millimeter continuum and line measurements . in table
[ sampletab ] the basic properties of the sample are given ( coordinates , redshifts and amplification factors ) .
the galaxies have redshifts in the range @xmath30 , with the exception of four galaxies with @xmath21 ( median redshift is 2.3 ) , and most have stellar masses in the range @xmath31 ( median , @xmath32@xmath2 ) , as will be presented in [ mstarsection ] .
lccclcl 8:00arc & 00h22m40.9s & + 14d31m14.0s & 2.730 & @xcite & @xmath33 & @xcite + j0712 & 07h12m17.5s & + 59d32m16.3s & 2.646 & @xcite & [email protected] & @xcite + j0744 & 07h44m47.8s & + 39d27m25.7s & 2.209 & @xcite & [email protected] & @xcite + j0900 & 09h00m03.3s & + 22d34m07.6s & 2.032 & @xcite & [email protected] & @xcite + j0901 & 09h01m22.4s & + 18d14m32.3s & 2.256 & @xcite & [email protected] & @xcite + j1133 & 11h33m14.2s & + 50d08m39.5s & 1.544 & @xcite & [email protected] & @xcite + j1137 & 11h37m40.4s & + 49d36m35.8s & 1.411 & @xcite & [email protected] & this work ( appendix a ) + horseshoe & 11h48m33.1s & + 19d30m03.2s & 2.379 & @xcite & [email protected] & @xcite + j1149 & 11h49m35.2s & + 22d23m45.9s & 1.491 & @xcite & [email protected] & @xcite + clone & 12h06m02.1s & + 51d42m29.5s & 2.001 & @xcite & [email protected] & @xcite + j1226 & 12h26m51.3s & + 21d52m20.3s & 2.925 & @xcite & [email protected] & @xcite + j1343 & 13h43m34.0s & + 41d55m09.0s & 2.092 & @xcite & [email protected] & @xcite + j1441 & 14h41m49.4s & + 14d41m21.3s & 1.433 & @xcite & [email protected] & @xcite + cb58 & 15h14m22.2s & + 36d36m25.2s & 2.729 & @xcite & [email protected] & @xcite + j1527 & 15h27m45.2s & + 06d52m19.2s & 2.762 & @xcite & [email protected] & @xcite + eye & 21h35m12.7s & -01d01m42.9s & 3.074 & @xcite & [email protected] & @xcite + eyelash & 21h35m11.6s & -01d02m52.0s & 2.326 & @xcite & [email protected] & @xcite as outlined in the introduction , the power of gravitational lensing is used here to extend the galaxy population where direct measurements of both molecular gas and dust masses can be performed . in particular , most galaxies in the sample have a stellar mass of @xmath1@xmath2 and modest sfrs , as determined from ir and uv photometry , of @xmath35 @xmath2 yr@xmath26 , given their median redshift of @xmath36 .
we stress that the only way to directly detect the far - infrared emission of such galaxies across multiple _ herschel _ bands is to target systems that are gravitationally lensed . even in the deepest pacs and spire blank fields , this mass / redshift / sfr regime is barely accessible through stacking of tens of galaxies @xcite . therefore , it is impossible to directly compare the fir properties of our lensed galaxies to those of a well - matched , un - lensed reference sample .
such a comparison is however desirable .
since the lensed galaxies were selected visually based on their rest - frame uv light , there is a possibility that they represent a biased sub - sample of the high - redshift star - forming galaxy population . for example
, we could expect the selection technique to preferentially isolate objects that are particularly uv - bright and therefore dust- and metal - poor . to answer this question , we create a control sample of un - lensed galaxies , matching on mass , redshift , and ir luminosity , where the latter is obtained either from pacs or mips 24@xmath17 m photometry when available , or else derived indirectly from the optical / uv photometry .
we then compare the uv - to - ir ratio between the lensed and control samples .
more exactly , for each lensed galaxy , we extract a control galaxy from the goods fields catalog @xcite , within 0.2 dex in @xmath4 , 0.2 dex in @xmath37 and 0.2 in @xmath38 . to check whether the lensed galaxies are less extincted than the control sample , the distributions of @xmath39 are compared . as described in
[ sfrsection ] , for the lensed galaxies , we calculate sfr@xmath40 from the 160@xmath17 m fluxes and sfr@xmath41 from the b- or v - band photometry .
for the goods sample , @xcite used a ladder of star formation indicators to infer the respective contributions of obscured and un - obscured star - formation , even in the absence of flux measurements at fir wavelengths , therefore also allowing a measurement of @xmath39 .
a ks test gives a probability @xmath42 that the distributions of @xmath39 in the lensed and control samples are representative of the same parent population . as a second test
, we study the uv slope @xmath43 of the lensed galaxies . in figure [ irxbeta ]
the lensed galaxies are shown in the @xmath44 plane , where @xmath45 is the effective uv attenuation , @xmath46 the uv slope is measured from hst optical photometry using bands corresponding to rest - frame wavelengths redwards of lyman @xmath47 .
the observed values are interpolated to derive the magnitudes at 1600 and 2800 ( rest - frame ) which are then used to compute @xmath43 . as a comparison ,
we use again the goods fields sample for which @xcite have derived values of the uv slope @xmath43 . this reference sample
is shown in fig .
[ irxbeta ] as gray circles .
as these are pacs - detected galaxies only in the goods fields , the sample is biased towards large values of @xmath45 . to circumvent the necessity of detecting individual objects in several herschel bands
, @xcite used stacks of uv - selected galaxies to probe galaxies with typically lower masses and sfrs .
their results are shown in fig .
[ irxbeta ] for two different bins of @xmath43 .
the lensed galaxies are located in the same region of the @xmath44 as the bx / bm galaxies of @xcite , and are also seen to follow the @xcite extinction law ( solid line ) .
the only exception is the eyelash , which is an outlier with high uv attenuation as expected since it is the only submillimeter - selected galaxy in the sample . using the sample of @xcite (
both herschel - detected and undetected galaxies ) we can also define a control sample for each of our lensed galaxies by matching on @xmath4 , sfr and redshift . in 10 of the 12 galaxies where we have sufficient data to conduct this experiment
the median of the @xmath43 distribution in the control sample agrees with the measured value of @xmath43 for the lensed galaxy within its uncertainty .
the two exceptions are j0900 and j1527 where measured values of @xmath43 are -1.8 while the control samples suggest -0.6 and -1.2 , respectively .
we also note that while previous studies of the cosmic eye and cb58 suggested that these objects were better represented by an smc extinction law than by the @xcite relation ( see e.g. * ? ? ?
* ) , the new herschel data revise the ir luminosities upward and find these galaxies in agreement with the meurer / calzetti relation . the lack of significant differences in either the effective uv attenuation ( @xmath45 ) or the uv slope ( @xmath43 ) between the lensed galaxies of this study and their respective matched control samples suggests that they are not a biased subsample of the underlying population .
we therefore proceed under the assumption that they are representative of the bulk of the star - forming galaxy population at their specific redshifts and modest masses . in table
[ sampletab ] , we give the magnification factors used to correct measured quantities for the lensing effect .
the magnification factors are derived from optical imaging ( typically from hst ) , and represent the average factor across the discs of these extended galaxies .
however , in reality , different regions of the galaxies may be lensed differentially , with regions falling on or near caustics being more strongly magnified .
therefore , components of the galaxy having different spatial extents and compactness may not be magnified uniformly .
this effect is known as differential lensing ( e.g. * ? ? ?
* ) . in practice
, this could impact our analysis since both warm dust and high@xmath48 co transitions are typically emitted in denser , more compact regions than cold dust or co ( 1 - 0 ) .
both the far - infrared sed and the co spectral line energy distribution ( sled ) might be biased , and standard calibrations to measure dust masses and temperatures , as well as excitation corrections to extrapolate total gas masses from a ( 3 - 2 ) line flux , may not apply @xcite . for the specific case of galaxy lenses selected in large area , flux - limited submm / fir surveys , there is a bias towards galaxies with very compact dusty star - forming regions , as these are most likely to benefit from the strongest magnifications @xcite . with the exception of the eyelash ,
our lenses are not submm / fir - identified , but rather selected from cluster fields and sdss imaging . in the discovery papers , sdss galaxy lenses were typically identified visually as blue arcs .
this includes the arc length , driven by the basic geometric configuration of lens and background object , as well as flux
. they will therefore not suffer the specific biases discussed by @xcite , but the rest - uv emission used to derive the lensing model may still differ in extent or centroid from that at longer wavelengths .
it is difficult to directly assess the impact of differential lensing , as the exercise requires detailed magnification maps , and high resolution imaging of the different components ( optical / uv continuum , fir continuum , co line emission , ... ) .
the effect has therefore been so far quantified only based on simulations and very few , special galaxies . in particular ,
most of the literature on this concentrates on the special case of submm- or fir - selected galaxies such as the eyelash ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
as can be seen in figure [ irxbeta ] , the eyelash is a very special object , the only one in our sample to be submm - selected , and any conclusions drawn from such extreme dusty systems may not apply to all normal star - forming galaxies .
we can however use these studies to get a sense of the amplitude of the issue in the most extreme cases . using a submillimeter galaxy as a model , @xcite estimate the impact of differential lensing on fir - selected galaxy samples .
they find that even when comparing quantities measured from rest - frame optical / nir wavelengths ( e.g. stellar masses ) and fir observations ( e.g. sfrs ) , the median differential magnification ratio is @xmath49 with small dispersion . therefore even in these cases , differential lensing does not affect the position of a galaxies in the sfr - mass plane significantly .
@xcite find a stronger effect of differential lensing on the co sled , but the effect manifests itself mostly for transitions with @xmath50 .
indeed , for the few galaxies in our sample where co(1 - 0 ) measurements are also available , a typical ( 3 - 2)/(1 - 0 ) line ratio of @xmath51 is retrieved @xcite , similar to what is measured in unlensed galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
a common explanation for this ubiquitous value of the ( 3 - 2)/(1 - 0 ) line ratio is that the lines are emitted from the same moderately excited component of the ism , but @xcite argue that it may instead reflect a generic feature of multi - component star - forming isms , in which the different lines are emitted by different but well mixed , optically thick and thermalized components with different but characteristic filling factors .
since our _ herschel _ and iram observations are mostly unresolved , it is not possible to directly assess the impact of differential lensing on our measurements . generally , this effect has not yet been extensively studied for normal star - forming high - redshift discs ( as opposed to submm- or fir - selected galaxies ) . however , based on the arguments above and the hypothesis that the fir continuum and the co line emission originate from similar physical regions ( see however * ? ? ?
* ) , we can conclude that if occurring , differential lensing is unlikely to significantly affect the results of this study .
it may be biasing our gas masses high if the co sleds are affected , but by no more than 40% , comparing our adopted excitation correction @xmath52 and the value of 1.4 indicated by the observations of the eye , eyelash and cb58 ( * ? ? ?
* ; * ? ? ?
* see also [ mh2section ] ) .
similarly , differential lensing could be biasing the inferred dust properties .
if the more compact , hotter regions of the galaxies are more strongly lensed than the diffuse component , the inferred dust temperatures could be too high ( although * ? ? ?
* show that the effect can be reversed in systems with lower magnification factors of @xmath53 ) .
this would result in under - estimated dust masses as well .
however , the fact that we do not observe a strong effect on the co sled argues against a strong effect , but we can not rule out the possibility of a small bias and include this caveat in the discussion in [ evolution ] .
we add to the @xmath14 lensed galaxy sample additional objects from the literature to serve as a comparison point .
first , we use a compilation of 16 galaxies in the goods - north field for which co line fluxes as well as deep herschel photometry are available @xcite .
these galaxies are located mostly at @xmath54 and sample well the sfr - stellar mass plane .
we also use the compilation of submillimeter galaxies ( smgs ) from @xcite , which expand the sample toward higher sfrs at fixed stellar mass and redshift .
all the smgs from the @xcite compilation are used in the analysis when only dust measurements are involved , and the subset of these also found in @xcite whenever co measurements are also needed .
all the galaxies in the comparison sample have _ herschel _ pacs and spire photometry , which was processed identically as for the lensed galaxy sample in order to derive infrared luminosities , sfrs , dust temperatures and dust masses ( see details in [ firproducts ] ) .
the lensed and comparison galaxy samples are shown in figure [ sfrmass ] in the sfr-@xmath4 plane for two redshift intervals , @xmath55 and @xmath56 , using the goods and egs catalogs of @xcite to provide a reference and define the main sequence ( ms ) .
these are the samples onto which our measurements of stellar mass and star formation rate are calibrated ( sections [ mstarsection ] and [ sfrsection ] ) , making them the best matched reference catalogs .
the ms we derive from these is @xmath57 where @xmath58 $ ] and @xmath59 $ ] for the three redshift intervals of @xmath60 $ ] , @xmath61 $ ] and @xmath62 $ ] , respectively . when discussing the molecular gas properties of the lensed galaxies , we also use additional references .
in particular , we make use of the data from the cold gass survey @xcite , which includes co(1 - 0 ) measurements for a representative sample of 365 sdss - selected galaxies at @xmath63 with @xmath64@xmath2 . for high redshift galaxies ,
the phibss sample @xcite is by far the largest compilation of co measurements for normal star - forming galaxies , and is the perfect reference to study e.g. the redshift evolution of the gas contents of galaxies ( see section [ evolution ] ) .
lcccccccc 8:00arc & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + j0712 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & + j0744 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & + j0900 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & + j0901 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + j1133 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & + j1137 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & + horseshoe & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & + j1149 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & + clone & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & + j1226 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & + j1343 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & + j1441 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & + cb58 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + j1527 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & + eye & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 19.1@xmath34 10.0 & [email protected] + eyelash & [email protected] & [email protected] & [email protected] & 335.4@xmath34 19.5 & [email protected] & 353.1@xmath34 10.3 & [email protected] we have obtained 70 , 100 and 160@xmath17 m ` mini - scanmaps ' of our targets using the pacs instrument @xcite on board the herschel space observatory @xcite . total observing time per source was 1 hour at 70 and 100@xmath17 m each , and 2 hours at 160@xmath17 m wavelength , which is observed in parallel to both of the shorter wavelengths .
the resulting pacs maps cover an area of @xmath65 at more than half of the peak coverage , and have useful information over an area @xmath66 .
we processed the pacs data to maps using standard procedures similar to those described for the pep project in @xcite , in build 7.0.1786 of the herschel hipe software @xcite .
the _ herschel _
blind pointing accuracy is @xmath67 rms @xcite . to secure the astrometry , we therefore inspected the pacs maps to identify far - infrared sources clearly associated with counterparts having accurate astrometry ( typically from sdss ) .
these reference positions were used to correct the astrometry of the pacs data to sub - arcsecond accuracy .
we used the spire instrument @xcite to simultaneously obtain 250 , 350 , and 500@xmath17 m ` small maps ' of our sources , using 14 repetitions and a total observing time of 35 minutes per source .
maps were produced with the standard reduction pipeline in hipe ( version 4.0.1349 ) .
following the recommendation in the spire photometer instrument description , the maps are scaled with the appropriate flux correction factors of 1.02 , 1.05 , and 0.94 at 250@xmath17 m , 350@xmath17 m , and 500@xmath17 m , respectively .
the typical calibration accuracy of spire maps is 15% .
a preliminary source extraction at 250@xmath17 m was performed , and the sources with counterparts in the pacs images ( shifted as described above to the appropriate coordinate zero point ) were used to correct the astrometry in all spire bands .
we developed a measurement technique that combines aspects of blind source extraction , and guided extraction using prior information on the position of the sources .
the procedure is aimed at measuring reliable fluxes for the lensed sources across the six _ herschel _ bands , rather than produce complete catalogs in any given band . the first step is to perform a blind extraction with starfinder @xcite on the 100@xmath17 m , 160@xmath17 m and 250@xmath17 m maps .
the resulting catalogs are used as prior information on the position of sources brighter than 3@xmath68 , which is used to extract accurate fluxes .
the main advantages of this psf - fitting technique are that it accounts for most of the blending which could be an issue in these typically crowded fields at the longer wavelengths , and allows for photometry on any specific object that is reliable across the different bands @xcite . for the pacs images , we use the merged blind 100@xmath17 m and 160@xmath17 m catalogs as a prior , while the 250@xmath17 m blind catalog is used as the prior for the spire bands . in cases where the lensed galaxy of interest is detected at less than 3@xmath68
, the position of this galaxy is added in the prior catalogs based on its coordinates at shorter wavelengths .
in most cases , the closest object with a 3@xmath68 detection in the priors catalog , be it the lens or another object , is located 20 - 40 from the lensed galaxy and therefore easily separable at both 160 and 250@xmath17 m . only in the cases of cb58 ,
the cosmic eye and j1133 is the closest neighboring object closer , at a distance of 10 - 15 . even in these cases the objects
are well separated in the pacs imaging , which we use as a guide for the spire priors .
the _ herschel _ fluxes obtained from the psf - fitting are finally aperture - corrected using factors of ( 0.883 , 0.866 , 0.811 , 0.835 , 0.848 , 0.898 ) , derived specifically for the psfs used at ( 70 , 100 , 160 , 250 , 350 , 500)@xmath17 m .
unless the lens is detected in the pacs / spire images with @xmath69 , it is assumed that its fir emission is negligible and does not contaminate the measurement for the lensed galaxy of interest .
the final _ herschel _ fluxes for the 17 lensed galaxies are given in table [ firtab ] .
values @xmath70 should be interpreted as upper limits , which occurs for 5 galaxies at both pacs 160@xmath17 m and spire 250@xmath17 m .
photometric observations in the 1.2 mm continuum were obtained for part of our sample during the pool observing sessions at the iram 30 m telescope in the winters 2006/2007 and 2007/2008 .
we used the 117 element version of the max planck millimeter bolometer ( mambo ) array @xcite .
on - off observations were typically obtained in blocks of 6 scans of 20 subscans each , and repeated in later observing nights unless a detection was already reached .
the data were reduced with standard procedures in the mopsic package developed by r. zylka , using the default calibration files for the applicable pool periods .
table [ firtab ] lists the measured 1.2 mm fluxes and their statistical uncertainties .
we add the mambo detection which we already obtained for cb58 @xcite and the sma flux obtained by @xcite for the eyelash . for j0901
, we obtained a clear mambo detection at the target position centered on the southern bright lensed component but both the pacs maps and iram - pdb co maps clearly indicate that the 11 mambo beam is missing flux . based on the co and pacs 100@xmath17 m maps , the ratio between the total flux of j0901 and the flux in the southern component is independently measured to be 2.4 and 2.5 , respectively .
we hence scale the mambo flux from the southern component by a factor of 2.4 to infer the total flux before further use .
molecular gas mass measurements through observations of the co(3 - 2 ) line have been performed for 10 of the 17 galaxies in the _ herschel _ sample , and we report here on both the previously published measurements and those coming from new pdbi observations . in sections [ results ] and [ summary ] ,
the full sample of 17 galaxies is used for the elements of the analysis that do not invoke molecular gas , and this subset of 10 galaxies for the rest of the analysis .
the co line has been previously observed in seven galaxies from our _ herschel _ sample .
we briefly review the specifics of the observations of each object below , while a summary of the co measurements is given in table [ cotable ] .
[ [ eye ] ] eye + + + the cosmic eye was observed in the co(3 - 2 ) line with the iram pdbi @xcite .
the line width is @xmath71 km s@xmath26 and the total line flux integrated over the line is @xmath72 jy km s@xmath26 .
the co emission appears to be spatially associated with component b@xmath73 of the system , and @xcite therefore suggest that the appropriate magnification correction factor for the co line flux is 8 rather than the value of 30 found for the entire system @xcite . however , @xcite report a detection of the co(1 - 0 ) with the vla that is spatially consistent with the bulk of the rest - frame uv emission , leading them to conclude that the total magnification value of the system should be used .
this is the approach we adopt here .
[ [ eyelash ] ] eyelash + + + + + + + multiple co lines of both @xmath74co and @xmath75co were observed for this object @xcite .
although a @xmath74co(1 - 0 ) flux has been measured , for uniformity with the rest of the sample we adopt for the eyelash the flux of @xmath76 jy km s@xmath26 measured in the ( 3 - 2 ) line .
[ [ j0901 ] ] j0901 + + + + + @xcite reports on both evla co(1 - 0 ) and pdbi co(3 - 2 ) observations .
for consistency with the rest of the dataset , we adopt the co(3 - 2 ) flux of @xmath77 jy km s@xmath26 , which is obtained after primary beam correction given the large angular size of the source .
[ [ cb58 ] ] cb58 + + + + we adopt for this object the co(3 - 2 ) flux of [email protected] jy km s@xmath26 measured by @xcite , which itself is consistent with an upper limit previously set by @xcite and with the vla co(10 ) detection by @xcite for the value of @xmath78 we adopt for the bulk of our sample ( see [ mh2section ] ) .
[ [ arc ] ] 8:00arc + + + + + + + the
eight oclock arc " ( thereafter the 8:00arc ) was observed in 2007 may with the iram pdbi in compact configuration , using the 3 mm sis receivers to target the redshifted co(3 - 2 ) line ( baker et al . in prep . ) .
following a standard reduction process , a 0.45 mjy continuum source associated with the lens was subtracted ( see also * ? ? ?
* ) , and a final line flux of [email protected] jy km s@xmath26 is measured .
[ [ clone - horseshoe ] ] clone , horseshoe + + + + + + + + + + + + + + + + baker et al .
( in prep .
) report on iram pdbi compact configuration observations of these two sources , combining datasets taken in july to october 2007 and april 2009 . using standard reductions and integrating over the spatially extended co emission , total line fluxes for the redshifted co(3 - 2 ) line of @xmath79 ( clone ) and @xmath80 ( horseshoe ) are derived . in july - october 2011
we obtained co(3 - 2 ) maps ( rest frequency of 345.998 ghz ) for three additional uv - bright lensed objects with the iram pdbi @xcite .
the three targets , j0900 , j1137 and j1226 , were chosen to extend the parameter space of objects with molecular gas measurements .
specifically , j0900 has the lowest metallicity of the sources with reliable _ herschel _ flux measurements , and j1137 and j1226 are at the very low and very high ends of the redshift range of the sample , respectively .
the lower redshift source , j1137 , was observed in the 2 mm band , while for the other two objects the co(3 - 2 ) line was visible in the 3 mm band .
observations were carried out under average to mediocre summer conditions with five of the six 15-m antennae in operation and in compact configuration .
a total of 6 - 10 tracks of various duration per object were necessary to reach the required line sensitivities .
data were recorded with the dual polarization , large bandwidth widex correlators , providing spectral resolution of 1.95 mhz over a total bandwidth of 3.6 ghz .
the data were calibrated using the clic package and maps produced with mapping , within the iram gildas software environment .
a standard passband calibration scheme was first applied , followed by phase and amplitude calibration . due to the poor observing conditions during some of the runs ( e.g. high precipitable water vapour , strong winds , low elevation )
, particular care was taken to flag data with high phase noise .
the absolute flux calibration was done using observation of reference bright quasars , and is typically accurate to better than 20% @xcite .
the data cubes were examined for sources at the expected spatial and spectral positions . the co(3 - 2 ) line
is clearly detected in j1137 , but not in j0900 and j1226 . assuming 200 km s@xmath26 line widths , the @xmath81 upper limit on the line flux is 0.16 jy km s@xmath26 for j1226 and 0.44 jy km s@xmath26 for j0900 .
the measured co(3 - 2 ) integrated line flux of j1137 is @xmath82 jy km s@xmath26 , and the line fwhm is 137 km s@xmath26 as determined by a gaussian fit .
the integrated co(3 - 2 ) line map and spectrum of j1137 are shown in figure [ j1137spectrum ] .
stellar masses found in the literature can vary significantly for the same object depending on the measurement technique , and specific assumptions made regarding star formation histories , metallicities , and stellar population ages .
for example there is an order of magnitude difference for cb58 between @xcite and @xcite , for the eye between @xcite and @xcite , and for the 8:00arc between @xcite and @xcite or @xcite .
since homogeneity is paramount for the analysis we conduct here , we derive new stellar masses consistently for all the lensed galaxies in our sample .
this is done with _
spitzer_/irac imaging , which is available for all the lensed galaxies with the exception of j1441 .
the calibrated images at 3.6 and 4.5 @xmath17 m were retrieved from the _ spitzer _ archive , and fluxes measured using a custom - made pipeline . since most of the lensed galaxies appear as resolved arcs in the irac images , and since they are also often situated in the wings of bright sources ( generally , the lensing galaxy ) , standard photometric tools are not adequate .
the details of the extraction technique are given in appendix [ iracphoto ] , and the measured 3.6 and 4.5 @xmath17 m fluxes ( un - corrected for lensing ) are given in table [ masstab ] .
lccccc 8:00arc & 165.4@xmath34 6.4 & 200.3@xmath34 8.3 & 10.88@xmath34 0.21 & 11.20@xmath34 0.40 &
@xcite + j0712 & & & & 10.23@xmath34 0.48 & @xcite + j0744 & 21.7@xmath34 1.3 & 23.0@xmath34 1.4 & 9.70@xmath34 0.11 & 9.76@xmath34 0.08 & @xcite + j0900 & 44.2@xmath34 5.1 & 41.1@xmath34 7.1 & 10.45@xmath34 0.16 & 10.28@xmath34 0.20 & @xcite + j0901 & 477.6@xmath34 20.2 & 556.3@xmath34 26.4 & 11.49@xmath34 0.14 & & + j1133 & 112.2@xmath34 26.0 & 130.5@xmath34 16.7 & 10.17@xmath34 0.15 & & + j1137 & 197.0@xmath34 22.5 & 192.6@xmath34 14.0 & 10.21@xmath34 0.13 & & + horseshoe & 158.2@xmath34 11.0 & 123.1@xmath34 16.0 & 10.30@xmath34 0.13 & 10.56@xmath34 0.19 & @xcite + j1149 & 27.0@xmath34 2.0 & 25.2@xmath34 1.6 & 9.26@xmath34 0.13 & & + clone & 230.8@xmath34 8.2 & 218.7@xmath34 9.9 & 10.39@xmath34 0.04 & & + j1226 & 52.8@xmath34 6.0 & 47.2@xmath34 4.2 & 9.72@xmath34 0.15 & @xmath83 & @xcite + j1343 & 69.6@xmath34 7.7 & & 9.76@xmath34 0.10 & 9.40@xmath34 0.20 & @xcite + j1441 & & & & & + cb58 & 79.0@xmath34 4.3 & 85.2@xmath34 5.2 & 10.03@xmath34 0.18 & 9.69@xmath34 0.14 & @xcite + j1527 & 36.3@xmath34 2.6 & 38.2@xmath34 4.1 & 10.02@xmath34 0.15 & @xmath84 & @xcite + eye & 204.7@xmath34 6.7 & 185.2@xmath34 9.1 & 10.51@xmath34 0.09 & 10.53@xmath34 0.08 & @xcite + eyelash & 132.8@xmath34 8.3 & 206.8@xmath34 12.3 & 10.34@xmath34 0.10 & 10.28@xmath34 0.29 & @xcite to compute stellar masses from these irac fluxes that will be consistent with the masses of the comparison sample , we use the catalog of @xcite for the goods fields as a calibration set .
for each of our lensed galaxies , we extract from the calibration set all galaxies within @xmath85 that are star forming based on their location in the sfr-@xmath4 plane .
this subset of galaxies is used to determine the relation between observed 3.6 and 4.5 @xmath17 m fluxes and stellar mass .
the scatter in these empirical relations varies from 0.13 dex for the lensed galaxies at @xmath21 to 0.19 dex at the highest redshifts .
these uncertainties are comparable to the typical variations in stellar masses derived though sed modeling under varying assumptions ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
stellar masses for the lensed galaxies are then obtained by taking the observed irac fluxes ( table [ masstab ] ) , correcting them for the lensing magnification , and then applying the empirical calibration determined for each object .
the masses derived from the 3.6 and 4.5 @xmath17 m images are consistent within the errors , and we adopt as our stellar masses the mean between the two values . these values are summarized in table [ masstab ] with the errors quoted obtained by propagating the uncertainties on the parameters of the fit to the calibration data set , the measurement errors on the irac fluxes , and the uncertainty on the magnification factor .
the measurement and calibration uncertainties account for @xmath86 of the error budget , the rest coming from the uncertainty on the magnification .
when available , we compare in figure [ masses ] previously published stellar masses ( typically from sed - fitting ) to our irac - derived masses . for cb58 , the cosmic eye and the 8:00arc
, two previously published values differing by @xmath87one order of magnitude are shown .
the irac - derived masses are always consistent with the higher of the two values .
adopting these higher estimates for these three galaxies , the scatter between the irac- and sed - derived stellar masses is 0.23 dex .
this scatter and the outlier points are caused by the uncertainty on the calibration of our irac - based stellar masses , and by different assumptions about star formation histories in the sed modeling .
the far - infrared seds from the pacs , spire and mambo photometry are shown in figure [ sed ] .
they represent some of the highest fidelity _ individual _ herschel seds of star - forming galaxies at @xmath88 .
these seds are used to derive dust temperatures , dust masses , and total infrared luminosities .
the procedure is identical to the one applied on the comparison samples .
we summarize the key elements here , with the full details given in @xcite .
dust masses are calculated using the models of @xcite ( dl07 ) .
a grid of models is created to sample the expected values of pah abundances , radiation field intensities and dust fractions in the diffuse ism . at each grid point , the model sed is compared with the herschel photometry , with the dust mass given by the normalization of the sed minimizing the @xmath89 . for each galaxy ,
the final dust mass assigned is the mean value over the grid points where @xmath90 . to derive dust temperatures , a single modified black - body model with dust emissivity @xmath91
is then fit to all the sed points with @xmath92 m .
dust temperatures and masses , along with their measured uncertainties , are given in table [ irproducts ] .
while infrared luminosities can be derived from the dl07 modeling at the same time as the dust masses , we adopt here the values that are derived from the 160@xmath17 m fluxes ( although for completeness we also give in tab .
[ irproducts ] the values of @xmath37 obtained by integrating in the wavelength range 8 - 1000@xmath17 m the best fitting dl07 model sed ) .
@xcite have demonstrated that for our @xmath93 galaxies this method is robust , as the uncertainties on the 160@xmath17m - to-@xmath37 conversion factors provided by the @xcite template library are @xmath94 dex . even in the cases where the lensed galaxies are not detected in the spire bands ,
making the fitting of dl07 model templates difficult , there is always a @xmath95 detection at 160@xmath17 m .
for the galaxies with good detections in all pacs and spire bands , there is an excellent agreement between the infrared luminosities derived by the dl07 modeling and by the extrapolation from 160@xmath17 m flux ( see also * ? ? ? * ) .
lcccccc 8:00arc & 50@xmath96 & 31.00@xmath97 & 33.44@xmath98 & 342.6@xmath34 13.9 & 354.1@xmath34 14.4 & [email protected] + j0712 & 49@xmath99 & 1.91@xmath100 & 1.29@xmath101 & 9.8@xmath34 23.7 & 12.7@xmath34 23.8 & [email protected] + j0744 & & & 1.20@xmath102 & 40.5@xmath34 159.2 & 41.8@xmath34 159.2 & [email protected] + j0900 & 54@xmath103 & 4.89@xmath104 & 6.22@xmath105 & 68.2@xmath34 22.1 & 129.9@xmath34 26.2 & [email protected] + j0901 & 36@xmath96&281.67@xmath106 & 67.56@xmath106 & 584.9@xmath34 26.5 & 608.3@xmath34 27.0 & [email protected] + j1133 & 32@xmath107 & 3.82@xmath108 & 0.65@xmath109 & [email protected] & [email protected] & [email protected] + j1137 & 45@xmath110 & 2.00@xmath111 & 1.65@xmath112 & [email protected] & [email protected] & [email protected] + horseshoe & 46@xmath113 & 2.99@xmath114 & 1.89@xmath115 & [email protected] & [email protected] & [email protected] + j1149 & 34@xmath116 & 2.86@xmath117 & 0.34@xmath118 & [email protected] & [email protected] & [email protected] + clone & 47@xmath119 & 3.58@xmath120 & 3.25@xmath121 & [email protected] & [email protected] & [email protected] + j1226 & 63@xmath122 & 1.42@xmath123 & 1.61@xmath124 & [email protected] & [email protected] & [email protected] + j1343 & & & 0.38@xmath125 & [email protected] & [email protected] & [email protected] + j1441 & & & 1.20@xmath126 & 19.1@xmath34 15.3 & 27.0@xmath34 15.5 & [email protected] + cb58 & 55@xmath96 & 2.74@xmath127 & 3.34@xmath128 & [email protected] & [email protected] & [email protected] + j1527 & & & 1.78@xmath129 & 26.0@xmath34 13.9 & 49.5@xmath34 14.8 & [email protected] + eye & 50@xmath130 & 4.30@xmath131 & 4.88@xmath132 & [email protected] & [email protected] & [email protected] + eyelash & 37@xmath96 & 77.99@xmath106 & 19.31@xmath106 & [email protected] & [email protected] & [email protected] dust - obscured sfrs can be obtained simply from the infrared luminosities as @xmath133 , with @xmath37 in units of solar luminosity and sfr@xmath40 in @xmath2yr@xmath26 , assuming a chabrier imf . at high redshifts and at high total sfr ( sfr@xmath134 ) , sfr@xmath40 is the dominant contribution ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and it is commonly assumed that @xmath135 , neglecting the un - obscured component ( sfr@xmath41 ) .
since the lensed galaxies in the sample were mostly selected for being bright blue arcs in optical images ( i.e. bright in rest frame uv ) , and since due to their large magnification factors their total intrinsic sfrs are modest , it is important in this case to estimate the contribution of sfr@xmath41 to sfr@xmath134 .
following @xcite , we measure this for a chabrier imf as @xmath136 .
the rest - frame 1600 luminosity ( in erg / s / hz ) is taken to be : @xmath137 where @xmath138 is the luminosity distance in cm , @xmath17 is the magnification factor from table [ sampletab ] , and @xmath139 the apparent ab - magnitude at a rest wavelength 1600 .
the most accurate approach to obtain @xmath139 would be to interpolate between all available bands as was done for example by @xcite , but its main drawback for the lensed galaxy sample is in the lack of homogenous , multi - wavelength photometry . while some galaxies have published photometry in several hst bands , some have been observed only with a single hst filter , while some others only have ground based photometry available .
given the nature of the available data and the fact that the seds of galaxies are typically relatively flat at these wavelengths , we directly adopt for @xmath139 the observed magnitude in the available optical band the closest to rest - frame 1600 . given the typical uv slopes @xmath43 in the range from -2.0 to -1.0 ( see fig .
[ irxbeta ] ) , the conversion factor from ab magnitude to sfr@xmath41 has only a very weak wavelength dependence , and therefore this approximation of adopting the closest band leads to systematic uncertainties of no more than 10% .
we therefore adopt for the lensed galaxies a total sfr given by @xmath140 . for the goods - n comparison sample ,
we retrieve the acs photometry from the multi - wavelength catalog of @xcite and compute sfr@xmath41 from eq .
[ eq_sfruv ] using either b- or v - band magnitudes , depending on the redshift of each galaxy . for the smg sample ,
we assume that @xmath141 , based on their high ir luminosities and location in the sfr-@xmath4 plane , indicating a very high attenuation of the uv light and therefore a negligible contribution of sfr@xmath41 @xcite . in order to obtain a homogenous set of metallicities for all galaxies in our herschel / iram sample
, we have compiled from the literature [ nii ] and h@xmath47 line fluxes ( table [ metaltab ] ) .
such fluxes were available for 11 of the 17 lensed galaxies . using the observed [ nii]@xmath142 ratio , we compute the nebular abundance using the calibration from @xcite .
this is the calibration that produces the least amount of scatter in the mass - metallicity relation @xcite and that was used by @xcite to derive a prescription for the metallicity - dependence of the co - to - h@xmath143 conversion factor . in the absence of a measured [ nii]@xmath142 ratio ,
metallicities measured from the strong line indicator @xmath144 are adopted for the eye , j1226 ( see appendix [ j1226metal ] ) and j1441 . these individual measurements are converted to the @xcite scale using the appropriate relation given in @xcite . for the remaining three galaxies ( eyelash , j1133 and j1137 ) , we use the mass - metallicity
( mz ) relation as given in @xcite to infer a metallicity . for the special case of j0901
, we adopt the metallicity value derived from the mz relation even in the presence of a measured [ nii]/h@xmath47 ratio as this value is likely affected by the presence of an agn ( details in [ j0901agn ] ) . for galaxies in the comparison sample , the latter method is used as explained in @xcite .
all the metallicities , as well as the h@xmath47 luminosities and [ nii]/h@xmath47 line ratios when available , are summarized in table [ metaltab ] .
lccccc 8:00arc & 74.5@xmath34 0.9 & 0.153@xmath34 0.005 & 8.52@xmath34 0.10 & n2 & @xcite + j0712 & 10.7@xmath34 0.3 & 0.210@xmath34 0.019 & 8.63@xmath34 0.11 & n2 & @xcite + j0744 & 15.4@xmath34 0.4 & 0.299@xmath34 0.017 & 8.74@xmath34 0.10 & n2 & @xcite + j0900 & 84.7@xmath34 3.2 & 0.054@xmath34 0.011 & 8.20@xmath34 0.13 & n2 & @xcite + j0901 & 72.0@xmath34 1.0 & 0.889@xmath34 0.019 & 8.91@xmath34 0.15 & mz & @xcite + j1133 & & & 8.55@xmath34 0.20 & mz & this work + j1137 & & & 8.61@xmath34 0.20 & mz & this work + horseshoe & 43.4@xmath34 0.9 & 0.092@xmath34 0.016 & 8.36@xmath34 0.13 & n2 & @xcite + j1149 & & 0.112@xmath34 0.050 & 8.43@xmath34 0.22 & n2 & @xcite + clone & 202.0@xmath34 5.0 & 0.193@xmath34 0.011 & 8.60@xmath34 0.10 & n2 & @xcite + j1226 & & & 8.27@xmath34 0.19 & r23 & @xcite + j1343 & 155.0@xmath34 10.0 & 0.148@xmath34 0.028 & 8.52@xmath34 0.13 & n2 & @xcite + j1441 & & & 8.14@xmath34 0.40 & r23 & @xcite + cb58 & 125.6@xmath34 3.7 & 0.091@xmath34 0.021 & 8.36@xmath34 0.14 & n2 & @xcite + j1527 & & @xmath145
0.224@xmath34 0.075 & @xmath145 8.65@xmath34 0.18 & n2 & @xcite + eye & & & 8.64@xmath34 0.14 & r23 & @xcite + eyelash & & & 8.61@xmath34 0.20 & mz & this work starting from @xmath146 , the integrated co(3 - 2 ) line fluxes in jy km s@xmath26 presented in table [ cotable ] , we calculate the co luminosity of the lensed galaxies , @xmath147 , in units of ( k km s@xmath26 pc@xmath148 ) following @xcite : @xmath149 where @xmath17 is the magnification factor as given in table [ sampletab ] , @xmath138 is the luminosity distance in mpc and @xmath150 the observed frequency in ghz .
the factor @xmath151 is the excitation correction to extrapolate the co(1 - 0 ) line flux from our co(3 - 2 ) observations . as in @xcite
, we adopt a value of @xmath152 based on recent studies of the co spectral line energy distribution ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
these studies have targeted both normal high-@xmath38 star - forming galaxies and smgs , with similar results pointing to a characteristic value of @xmath153 that may be due to low excitation or to the typical filling factors of the two lines .
we therefore apply this value of @xmath78 uniformly across our sample , and assume a conservative 20% uncertainty on this value compared to the typical errors presented in these individual studies .
total molecular gas masses are inferred from @xmath147 using the co - to - h@xmath143 conversion factor , @xmath154 ( @xmath155 ) .
the specific value of @xmath154 to be applied for each galaxy must be determined with care , as most have sub - solar metallicities ( @xmath156 ; * ? ? ?
* ) . under such low metallicity conditions ,
both observations and models suggest that the value of @xmath154 increases , as the co molecule becomes a poor proxy for h@xmath143 ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . here ,
different methods to estimate @xmath154 are investigated .
the first is the
inverse kennicutt - schmidt relation " method . under the assumption that a tight relation exists between the surface density of molecular gas and the sfr surface density , the value of @xmath154 can be estimated knowing @xmath157 and the co luminosity . using a compilation of local and high redshift galaxies , @xcite calibrated a relation between @xmath154 and metallicity using this approach . since most recent studies suggest that the ks relation is near - linear ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , the problem is further simplified and @xmath154 can be estimated as : @xmath158 where @xmath147 is the co(1 - 0 ) line luminosity in ( k km s@xmath26 pc@xmath148 ) , and @xmath154 the co - to - h@xmath143 conversion factor in @xmath2 ( k km s@xmath26 pc@xmath159 . in applying eq .
[ xcoks ] , we adopt a redshift - dependent depletion time , @xmath160@xmath161 , as suggested by @xcite and further supported here to @xmath23 in [ evolution ] . in figure [ xco ] , left panel , the value of @xmath162 is plotted against metallicity , for the lensed galaxies .
the values of @xmath154 scatter around the relation derived by @xcite using this method and a sample of both local and high redshift galaxies , @xmath163 + 11.8 .
\label{xcog12}\ ] ] the second approach , the gas - to - dust ratio method " , relies on a measurement of the dust mass and a motivated choice of a gas - to - dust ratio .
this method has been used successfully in the local universe ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and shown to be applicable also at high redshifts @xcite . in the case of high redshift galaxies , the conversion factor can be estimated simply as : @xmath164 where @xmath165 is in solar masses , and we adopt a metallicity - dependent gas - to - dust ratio , @xmath166 , from @xcite .
all the assumptions required to apply equation [ xcodust ] to high redshift galaxies are extensively described in 5.1 of @xcite . in particular
, it needs to be assumed that the co lines and the fir continuum are emitted from the same physical regions of the galaxies given that _
herschel _ does not resolve them , and that at high redshift @xmath167 and therefore that the atomic component of the ism can be neglected in eq .
[ xcodust ] ( see [ missinggas ] for a justification of this assumption ) .
a significant additional uncertainty lies in the assumption that the @xmath166 relation of @xcite , calibrated on a handful of very nearby galaxies , applies directly at @xmath168 .
furthermore , other studies of nearby galaxies indicate that the scatter in the @xmath169 relation may be larger than suggested by this specific @xmath10 sample , especially at low metallicities @xcite . in figure [ xco ] , right panel , @xmath170 measured from eq .
[ xcodust ] is shown as a function of metallicity .
the @xmath5 galaxies follow the inverse relation between @xmath154 and metallicity seen in the @xmath10 data and in the empirical relation of @xcite , although with a small systematic offset to lower values of @xmath154 at fixed metallicity , the implications of which are discussed in
[ gdrsection ] .
although affected by different sets of uncertainties and assumptions , the two methods of estimating @xmath154 produce consistent results , which in the mean and within their errors reproduce the metallicity - dependence of @xmath154 recently calibrated at @xmath19 by @xcite , and previously also observed locally ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . throughout the rest of this paper
, we therefore adopt as a consensus between @xmath162 and @xmath170 the value of @xmath154 obtained from the prescription of @xcite , as given in eq .
[ xcog12 ] , using our best estimates of metallicities given in table [ metaltab ] .
these values of @xmath154 , as well as the co luminosities and derived molecular gas masses are presented in table [ cotable ] .
lccccccccc 8:00arc & [email protected] & baker in prep . &
5.18@xmath34 2.35 & 27.46 & 8.52 & 9.41 & 10.69 & 0.39 & 137 + j0900 & & this work & @xmath145 4.13 & @xmath171 15.58 & @xmath171 3.21 & 24.50 & @xmath145 11.00 & & @xmath145 777 + j0901 & [email protected] & @xcite & 133.91@xmath34 40.22 & 2.09 & 1.41 & 3.05 & 11.61 & 0.57 & 671 + j1137 & [email protected] & this work & 1.57@xmath34 0.45 & 6.90 & 1.53 & 7.33 & 10.06 & 0.42 & 661 + horseshoe & [email protected] & baker in prep .
& 0.87@xmath34 0.42 & 19.02 & 6.71 & 15.05 & 10.12 & 0.40 & 351 + clone & [email protected] & baker in prep . &
0.75@xmath34 0.28 & 29.70 & 5.91 & 7.58 & 9.75 & 0.19 & 127 + j1226 & & this work & @xmath145 0.34 & @xmath171 33.51 & @xmath171 9.89 & 19.82 & @xmath145 9.82 & & @xmath145 226 + cb58 & [email protected] & @xcite & 0.87@xmath34 0.34 & 28.38 & 6.20 & 15.27 & 10.12 & 0.55 & 216 + eye & [email protected] & @xcite & 1.53@xmath34 0.41 & 19.32 & 3.20 & 6.72 & 10.01 & 0.24 & 128 + eyelash & [email protected] & @xcite & 23.17@xmath34 5.64 & 3.03 & 4.06 & 7.33 & 11.23 & 0.89 & 1089
having derived accurate and homogeneous measures of @xmath4 , sfr , @xmath172 , @xmath165 , @xmath29 and @xmath173 across the lensed galaxies and comparison samples , we now investigate the gas and dust properties of galaxies as a function of their position in the sfr-@xmath4-@xmath38 parameter space . in figure
[ tdustfig ] , we plot the dust temperature against the offset from the star formation main sequence and the total infrared luminosity , both for the lensed and comparison galaxy samples . while we recover with the comparison sample the known trend between @xmath172 and these two quantities @xcite , the @xmath5 lensed galaxies occupy a significantly different region of the plot .
dust temperatures in these galaxies are very high ( @xmath172@xmath35k ) , even though they are located on the main sequence and have modest infrared luminosities ( @xmath175 ) given their redshifts and stellar masses .
plotting dust temperatures against specific star formation rate rather than main sequence offset produces qualitatively equivalent results .
there is evidence that the tightness of the @xmath176 relation observed in classical samples of smgs ( e.g. * ? ? ?
* see also fig .
[ tdustfig]b ) is due to selection biases @xcite . in particular
, the sub - millimeter selection technique favors colder objects , especially at low @xmath37 .
when also considering samples of dusty high - redshift star - forming galaxies selected through other techniques ( e.g. the optically - faint radio galaxies of * ? ? ?
* ) , dust temperatures can be significantly higher at fixed @xmath37 .
the lensed galaxies shown in fig .
[ tdustfig]b extend these results to galaxies with even lower ir luminosities and higher dust temperatures . to confirm the high @xmath172 among the lensed galaxies ,
we also show in figure [ tdustfig]c - d similar relations using the infrared color , defined as the ratio between rest - frame 60@xmath17 m and 100@xmath17 m fluxes ( @xmath177 , where @xmath178 and @xmath179 are determined by linear interpolation of the herschel photometry ) .
this quantity , commonly used in iras - based studies , is a proxy for @xmath172 but is independent of any model assumptions .
the same behavior is observed for the @xmath5 lensed galaxies , with these objects having unusually high @xmath177 ratios given their modest infrared luminosities . among the @xmath5
lensed galaxies , there are however two exceptions , j0901 and the eyelash . with @xmath172@xmath180k and @xmath181 ,
these two galaxies follow the general trend between @xmath172 and ms offset ( or @xmath37 ) traced by the various comparison samples @xcite .
these two galaxies also differ from the rest of the @xmath5 lensed galaxy sample in other regards .
for example , they have the largest dust masses ( @xmath165@xmath182@xmath2 as compared to @xmath183 @xmath2 ) , and although no direct metallicity measurement is available for the eyelash , j0901 has the highest metallicity of all the galaxies in the sample ( but see appendix [ j0901agn ] ) . the high dust temperatures in the rest of the @xmath5 lensed galaxies therefore seem to be linked to their low dust contents compared to their sfrs .
as we could not find evidence in [ samplebias ] that the @xmath5 lensed galaxies form a biased subsample of the underlying galaxy population , this observations suggests that @xmath1@xmath2 galaxies at @xmath23 have higher dust temperatures and lower dust masses than similar galaxies at lower redshifts .
using deep herschel pacs and spire blind fields and a stacking technique , the mean @xmath172 of normal star - forming galaxies with @xmath184@xmath2 at @xmath5 is just barely measurable .
such studies indeed suggest a rise in temperature and intensity of the radiation field on the main sequence in the mass / redshift regime that we have now probed directly with the lensed galaxies @xcite . in the local universe ,
the typical gas - to - dust ratio , @xmath15 , for star - forming galaxies with solar metallicities is of order 100 @xcite .
this ratio has been shown to increase in low metallicity environments ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , as predicted by dust formation models ( e.g. * ? ? ?
the value of @xmath15 is also expected to vary in high density environments , such as the nuclear regions of starbursts , but observations suggest only a mild decrease of no more than a factor of 2 @xcite . in the previous section
, we inferred that the high dust temperatures of the @xmath5 lensed galaxies were due to their low dust masses and metallicities . to understand whether the dust content of these galaxies is abnormally low for their other properties we plot in figure [ gdr ] the measured value of @xmath15 as a function of metallicity for the galaxies within 0.5 dex of the star formation main sequence .
the gas - to - dust ratio is measured as @xmath185 , where @xmath186 is the molecular gas mass from table [ cotable ] calculated from the co luminosity using the metallicity - dependent conversion factor of @xcite , and @xmath165 is derived from the herschel photometry using the dl07 models ( see [ firproducts ] ) .
figure [ gdr ] reveals that the high redshift galaxies have a gas - to - dust ratio that scales inversely with metallicity like in the local universe .
however , fitting the @xmath169 relation of the @xmath5 galaxies while keeping the slope fixed to the @xmath10 reference relation reveals an increase in the mean @xmath15 at fixed metallicity by a factor of 1.7 .
since measuring co in normal star - forming galaxies at high redshifts is challenging , there has been significant interest recently in using dust masses as a proxy for the total molecular gas contents ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the method consists in measuring @xmath165 using far - infrared and/or sub - mm photometry , and then applying the estimated gas - to - dust ratio ( @xmath166 , see
[ mh2section ] ) to arrive at the gas mass .
the systematic offset between the @xmath5 normal star - forming galaxies and the reference relation of @xmath166 therefore needs to be taken into consideration .
there are at least three possible explanations : [ [ are - dust - properties - different - at - z2 ] ] are dust properties different at @xmath5 ? + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + as detailed in [ firproducts ] , dust masses for the lensed galaxies
are computed using the model of dl07 . in this
model , dust is considered to be a combination of carbonaceous and amorphous silicate grains with a specific size distribution set to reproduce the milky way extinction curve .
the model also assumes that a large fraction of the dust is found in the diffuse ism , the rest in photodissociation regions , with the two components exposed to different radiation fields , the parameters of which are set to reproduce milky way conditions or left as free parameters . as argued by @xcite , not only are dust grain properties poorly understood even in the local universe , the lack of a clear understanding of how dust is exactly produced and destroyed leads to significant uncertainties in applying the dl07 model to the denser , hotter interstellar medium of high redshift galaxies .
[ [ is - alpha_cofor - the - lensed - galaxies - smaller - than - expected ] ] is @xmath154 for the lensed galaxies smaller than expected ? + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + if we instead assume that the dust masses are accurately measured in the lensed galaxies , the offset in @xmath166 compared to @xmath10 could be explained by a conversion factor @xmath154 smaller on average by a factor of @xmath187 .
the lensed galaxies are located on the main - sequence , which is interpreted as being the locus of galaxies where star formation takes place in virialized gmcs rather than in a centrally - concentrated , dense starburst mode . under such conditions ,
@xmath154 should scale as @xmath188 @xcite .
assuming that gas and dust are thermalized , the large measured values of @xmath172 may indicate a reduction of @xmath154 , unless the gas density @xmath189 also increases in proportion .
the two @xmath5 lensed galaxies with normal temperatures of @xmath190k ( j0901 and the eyelash , see fig .
[ tdustfig ] ) also have gas - to - dust ratios that are elevated compared to the @xmath10 relation , arguing that the ism in the rest of the sample is not only warm but also denser such that @xmath154 is not affected by the temperature variations .
[ [ is - the - gas - to - dust - ratio - really - higher - at - z2 ] ] is the gas - to - dust ratio really higher at @xmath5 ? + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + as a last step
, we can assume that none of these concerns apply and that both @xmath186 and @xmath165 are accurately measured in the lensed galaxies , leaving us with the conclusion that the gas - to - dust ratio is higher at fixed metallicity than in the local universe .
this could happen for example if a smaller fraction of the metals are locked up in dust grains under the specific conditions prevailing in the ism of the high redshift galaxies , or if the far - infrared emission used to compute @xmath165 and the rest - frame optical line emission used to compute the metallicity are not emitted from the same physical regions .
however , irrespectively of which one of these possible explanations is valid , we can conclude that when measuring dust and gas masses though standard techniques , an offset of 0.23 dex is obtained between the measured gas - to - dust ratio and the standard @xmath10 prescription of @xcite .
care must therefore be taken in applying the gas - to - dust ratio method to estimate molecular gas masses of high redshift galaxies , as small but systematic differences with co - based measurements may otherwise occur . in [ hightdust ]
above , we suggested that the high @xmath172 of the lensed galaxies is caused by their low dust contents . to further illustrate this
, we repeat the analysis done in figure [ tdustfig ] , but this time normalizing @xmath37 by the dust or the gas mass .
we first show in figure [ tdustsfe ] how both @xmath172 and the @xmath177 ratio depend on the star formation efficiency ( sfe@xmath191 ) . a correlation between these quantities
is expected as @xmath172 increases as a function of @xmath192 and @xmath37 ( fig .
[ tdustfig ] ) , and there is also a positive correlation between sfe and @xmath192 @xcite .
figure [ tdustsfe ] reveals that when the gas mass in each galaxy is taken into account , most of the offset between the @xmath5 lensed galaxies and the comparison samples disappears , although the @xmath5 normal star - forming galaxies still appear to have dust temperatures higher by @xmath193k at fixed sfe compared to the reference sample ( gray symbols ) .
the offset is strongest for those galaxies with the lowest metallicities .
while metallicity may account for part of the remaining offset in @xmath172 at fixed sfe , there could also be a contribution from differential lensing ( see [ difflens ] ) , which if present at all , could manifest itself in an overestimation of the dust temperatures . in figure [ tdustmdust ]
we then show a similar relation , but using the dust rather than the gas mass as the normalization factor , allowing us to include a larger number of galaxies for which dust masses are measured , but gas masses are not available . in this case again , some of the lensed galaxies appear to be systematically offset to higher @xmath172 at fixed value of sfr@xmath194 .
the @xmath5 lensed galaxies having typically low metallicities , each dust grain will be exposed to more uv photons for a fixed sfr per unit gas mass , and therefore will have a hotter temperature .
there is evidence for this in the fact that the galaxies with @xmath87solar metallicities follow more closely the relation traced by the reference sample , in particular when using the @xmath177 ratio ( see fig .
[ tdustmdust ] ) .
the combination of figures [ gdr]-[tdustmdust ] thus argues that in these galaxies , the balance between @xmath165 , @xmath186 and @xmath195 follows expected relations ( modulo the possible change of @xmath15 with redshift ) , and that the high dust temperatures are the result of the high sfrs per unit gas and dust mass , i.e. the high star formation efficiencies .
this assertion that the @xmath5 lensed galaxies have high star formation efficiencies ( or put differently , short molecular gas depletion timescales , @xmath160 ) needs to be investigated in more detail . in the local universe ,
normal star - forming galaxies have @xmath160@xmath7 gyr @xcite . in the phibss survey ,
@xcite show that at @xmath196 the depletion time for main - sequence galaxies is reduced to @xmath87700 myr ( with 0.24 dex scatter in the @xmath197@xmath160 distribution , and assuming a galactic conversion factor @xmath154 ) .
based on these observations , @xcite infer a redshift - dependence of the form @xmath160@xmath161 .
this redshift evolution is only slightly slower than the dependence of @xmath198 expected if @xmath160 is proportional to the dynamical timescale @xcite .
the lensed and reference galaxies at @xmath5 can be used to track the behavior of the depletion time at even higher redshifts . in figure
[ fgas ] , left panel , the redshift evolution of @xmath160 is shown .
the gray band shows the expected trend based on the empirical @xmath11 dependence ( upper envelope ) , and the analytical expectation of a @xmath198 dependence ( lower envelope ) .
the mean values in four different redshift intervals are shown . for the highest redshift interval ( @xmath30 ) , we combine galaxies from the lensed and comparison samples as well as from phibss . while the mean depletion time is @xmath87700 myr at @xmath199 , it decreases further to @xmath6 myr at @xmath200 , consistently with the predicted redshift evolution .
we caution however that the sample at @xmath5 is by no means complete or representative of all normal star forming galaxies , but in terms of stellar masses and main sequence offset it is at least comparable to the @xmath201 phibss sample and the @xmath10 reference sample from cold gass .
this measurement at @xmath5 suggests that the redshift - dependence of the molecular gas depletion time for main sequence galaxies , as predicted analytically ( e.g. * ? ? ?
* ; * ? ? ?
* ) and recently observed up to @xmath21 @xcite , extends to @xmath202 . using the lensed galaxies and the various comparison samples , we can also trace the redshift evolution of the molecular gas mass fraction , @xmath203 in what follows , we restrict the samples only to galaxies found within 0.5 dex of the star formation main sequence at the appropriate redshift , i.e. we study the gas contents of normal , star - forming discs as a function of redshift .
several studies have now reported a rapid increase of @xmath204 with redshift @xcite .
the most robust analysis so far was performed by @xcite . in that work , the phibss data at @xmath205 and @xmath14 were corrected for incompleteness and compared to a matched local control sample extracted from the cold gass catalog , revealing an increase of @xmath204 from 8% at @xmath10 to 33% at @xmath205 and 47% at @xmath14 .
these three secure measurements are reproduced in figure [ fgas ] ( right panel ) .
there are very few galaxies at @xmath206 with published co measurements , but in fig .
[ fgas ] the few systems found in @xcite and @xcite are compiled ( co measurements for several galaxies with @xmath207 are published also in @xcite , but we do not include them here as they are above - main sequence objects ) . as the phibss sample extends to @xmath36 ,
we combine all the main - sequence galaxies in our lensed and comparison samples above that redshift to derive a mean gas fraction of @xmath208% at @xmath209 .
we then apply the methodology of @xcite to correct for sample incompleteness . as the sample of lensed galaxies with @xmath210 is richer in on- and below - main sequence galaxies , accounting for this
bias rises the mean gas fraction to 44% .
our observations therefore suggest that the trend for increasing gas fraction with redshift does not extend beyond @xmath14 , and may even be reversing .
can this flattening of the relation between gas fractions and redshift at @xmath5 be expected under the equilibrium model ?
the definition of the gas fraction ( eq . [ fgaseq ] ) can be re - expressed as : @xmath211 and the best predictions available for the redshift evolution of @xmath212 and ssfr used to compute the expected behavior of @xmath213 . as explained in
[ evolution ] , it is estimated that @xmath214 , \label{tdepeq}\ ] ] with @xmath47 measured to be -1.0 by @xcite and predicted to be -1.5 in the analytic model of @xcite .
the relation is normalized to the typical depletion time of 1.5 gyr observed in local galaxies @xcite .
based on studies of the slope and redshift evolution of the star formation main sequence , the typical ssfr ( in gyr@xmath26 ) of a star - forming galaxy of mass @xmath4 at redshift @xmath38 is @xmath215 the above equation is presentd by @xcite based on results from a number of recent high - redshift imaging surveys @xcite .
the expected redshift evolution of the gas fraction for galaxies of a given stellar mass can then be obtained by combining equations [ fgaseq]-[ssfreq ] . for galaxies in the mass range @xmath216 and for @xmath217 in eq .
[ tdepeq ] , the expected trend is shown in figure [ fgas ] as the light gray band . at @xmath5 ,
@xmath204 flattens out because of the shallower evolution of ssfr with redshift mostly canceling out the @xmath11 term from the @xmath160 relation ( eq .
[ tdepeq ] ) .
this model predicts a very modest evolution of the mean gas fraction from 47% at @xmath218 to 49% at @xmath12 , consistently with our measurement . a different behavior for @xmath213 can be obtained by modifying the redshift - dependence of either @xmath212 or ssfr ( eq . [ fgaseq ] ) .
for example , if a value of @xmath219 is used to set the evolution of @xmath212 , the mean gas fraction expected at any redshift is lower than in the first model .
however , since there is no observational evidence for a redshift evolution of @xmath212 different from what has been assumed so far , we turn instead our attention to ssfr .
equation [ ssfreq ] assumes that at fixed stellar mass , the specific star formation rate increases as @xmath220 out to @xmath221 , and that this evolution then slows down to @xmath222 . if instead we took ssfr to keep increasing steadily with redshift , for example as ssfr@xmath223@xmath4@xmath224 @xcite , the predicted mean gas fraction at @xmath12 would become 57% , in tension with our measured value of @xmath13 .
alternatively , it has been reported that the characteristic ssfr of main - sequence galaxies reaches a plateau at @xmath14 @xcite .
if we assume that ssfr increases up to @xmath221 according to eq .
[ ssfreq ] and then remains constant at @xmath5 , the mean gas fraction is predicted to drop to 41% at @xmath12 . while this behavior is consistent with the observations presented here ,
the existence of such a plateau in ssfr at @xmath225 has recently been challenged . after accounting for contamination by strong line emission ,
different authors suggest an increase of the mean ssfr by a factor of 2 - 5 between @xmath221 and @xmath226 ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this decrease in @xmath204 at @xmath5 should therefore be considered as a lower limit , with the relation based on eq .
[ ssfreq ] providing the most realistic prediction of the redshift evolution of the gas fraction , as supported by our observations .
another factor that could contribute to the redshift evolution of the gas fraction is the relative contribution of atomic hydrogen to the cold gas budget of the galaxies ( m@xmath227m@xmath228 ) at the different epochs . for star forming galaxies with @xmath229 at @xmath230 such as those in the phibss sample
, it is generally assumed that @xmath231 , at least within the parts of the discs where star formation is actively taking place .
this assumption is based on the observed high surface densities , above the characteristic threshold for the atomic - to - molecular conversion . on the other hand , for nearby galaxies with @xmath232
, it is observed that @xmath233 , albeit with large galaxy - galaxy variations @xcite .
an important fraction of this atomic gas is located in the outer regions of galaxies , outside of the actively star - forming disks , but even within the central regions a significant fraction of the cold gas is in atomic form as the hi - to - h@xmath143 transition typically occurs at a radius of @xmath234 @xcite .
we can therefore estimate that on average within @xmath235 , @xmath236 , raising the cold gas mass fraction from 8% when only the molecular phase was considered to @xmath237 .
we can not directly quantify the fraction of atomic gas in our high redshift lensed galaxies , but it could be significant given the low stellar masses .
not only do we know locally that the hi fractions increase as stellar masses decrease @xcite , it can also be expected that the gas surface densities are lower in the @xmath5 , @xmath238 lensed galaxies than in others at the high - mass end of the star - formation main sequence .
therefore , an additional interpretation for the low mean value of @xmath239 at @xmath209 is that a higher fraction of the cold gas mass is in neutral form than in the sample at @xmath240 and 2.2 .
however , were we missing a significant cold gas component , we would have measured lower than expected gas - to - dust ratios , as any dust should be mixed with both the molecular and atomic phases of the ism .
figure [ gdr ] shows the opposite behavior , suggesting that the lower gas fraction measured at @xmath12 is not the result of neglecting the atomic component .
the main observational results presented in this paper can be summarized as : 1 .
we measure a mean molecular gas depletion time of 450 myr at @xmath241 .
this decrease of @xmath212 by a factor of 5 since @xmath10 and 1.5 since @xmath9 is consistent with the expected scaling of @xmath242 ( @xmath243 ) if the depletion time is linked to the dynamical timescale .
our results validate up to @xmath202 and down to @xmath4@xmath1@xmath2 the calibration of this relation established by @xcite out to @xmath244 using data from the phibss and cold gass surveys and a metallicity - dependent conversion factor .
the mean gas fraction measured at @xmath209 is @xmath13 ( 44% after accounting for sample incompleteness ) , suggesting that the trend of increasing gas fraction with redshift ( @xmath245 $ ] at @xmath246 $ ] ) does not extend beyond @xmath14 .
this observation is consistent with recent studies suggesting that the evolution of the mean ssfr of main - sequence galaxies slows down or even reaches a plateau beyond @xmath221 .
3 . the lensed galaxies at @xmath5 exhibit high dust temperatures , with values of @xmath35k such as found typically only in galaxies with extreme ir luminosities .
the high values of @xmath172 are a consequence of the fact that the lensed galaxies have low metallicities and short gas depletion times , as expected for their high redshifts and low stellar masses .
4 . using co line luminosities , dust masses , and direct metallicity measurements , the conversion factor @xmath154 for the lensed galaxies is estimated using the inverse kennicutt - schmidt " and the gas - to - dust ratio methods , and in both cases agrees with a scaling of @xmath154@xmath247 as parametrized by @xcite .
the gas - to - dust ratios in the @xmath5 lensed galaxies exhibit the same metallicity dependence as observed in the local universe , but with a systematic offset ( fig .
[ gdr ] ) . at fixed metallicity
, we observe @xmath15 to be larger by a factor of 1.7 at @xmath5 , suggesting that applying the local calibration of the @xmath15-metallicity relation to infer the molecular gas mass of high redshift galaxies may lead to systematic differences with co - based estimates .
most of these results could be explained by assuming that our lensed galaxies sample , mostly selected in sdss imaging based on high luminosities at rest - uv wavelengths , is heavily biased toward dust - poor objects .
this could serve to explain for example the high dust temperatures , the low dust and gas masses , and the short depletion times .
however , we have conducted extensive tests ( see [ samplebias ] ) , and could not find evidence of any such bias .
instead , the sample appears to be representative of the overall population of main sequence galaxies at @xmath241 with @xmath248@xmath2 . while it should be kept in mind that sample biases may be present , in the absence of evidence to the contrary we have proceeded with the analysis under the assumption that our results apply to the bulk of the high redshift galaxy population .
the combined herschel and iram observations presented in this paper then suggest that main sequence galaxies with modest stellar masses ( @xmath249 ) at @xmath241 have high star formation efficiencies and a molecular gas mass fraction no larger than measured at @xmath230 , consistently with a simple model where the redshift evolution of the characteristic ssfr of main - sequence galaxies can be explained by a slowly varying gas depletion time and the measured gas fractions .
the short depletion times and the possible redshift evolution of the gas - to - dust ratio imply that these high redshift galaxies have less dust at fixed sfr , producing the high @xmath172 values we measure as each dust grain is exposed to more radiation ( fig .
[ tdustfig ] ) . before concluding , we wish to point out that the equilibrium model on which this analysis relies requires that the gas depletion timescale is shorter than the accretion timescale .
this balance is reached at a redshift @xmath250 . at @xmath251 ,
the star formation process can not keep up with the accretion of new gas , and this is thus the epoch where the gas reservoirs of the galaxies are filling up with gas fractions expected to be high .
the exact value of @xmath250 is however still debated .
for example , @xcite found that @xmath252 in the absence of outflows , but that even modest outflows raised the threshold to @xmath253 , the latter value being in agreement with the analytical work of @xcite . on the other hand ,
@xcite suggest that @xmath252 , and @xcite estimate that @xmath254 , using a sample of lyman break galaxies and an indirect measurement of gas fractions . in this study , we have used direct measurements of co line fluxes and the power of gravitational lensing to push to higher redshifts the study of the redshift evolution of gas fractions . after correcting for sample incompleteness , @xcite measure a mean gas fraction of 47% at @xmath200 . in this study
, we directly measure a mean gas fraction of 40% at @xmath209 , which is then corrected upward to 44% after accounting for sample incompleteness .
this observation suggests that @xmath204 does not increase significantly between @xmath221 and 3 .
a similar conclusion was reached by @xcite based on co observations of two @xmath202 lbgs . since gas fractions
are expected to be high during the gas accretion phase , these results may indicate that we have not yet reached observationally the regime where galaxies are out of equilibrium , and therefore that @xmath255
. the improved determination of stellar mass and star formation rates in large samples of galaxies beyond @xmath221 , as well as the direct measurement of gas masses in normal star - forming galaxies at @xmath256 with alma and noema will be essential to refine this picture .
we thank c. feruglio for assistance with the reduction of the iram data and j. richard for sharing the magnification factor for j1133 ahead of publication .
we also thank t. jones and c. sharon for useful input , as well as the anonymous referee for constructive comments .
pacs has been developed by a consortium of institutes led by mpe ( germany ) and including uvie ( austria ) ; ku leuven , csl , imec ( belgium ) ; cea , lam ( france ) ; mpia ( germany ) ; inaf - ifsi / oaa / oap / oat , lens , sissa ( italy ) ; iac ( spain ) .
this development has been supported by the funding agencies bmvit ( austria ) , esa - prodex ( belgium ) , cea / cnes ( france ) , dlr ( germany ) , asi / inaf ( italy ) , and cicyt / mcyt ( spain ) .
spire has been developed by a consortium of institutes led by cardiff university ( uk ) and including univ .
lethbridge ( canada ) ; naoc ( china ) ; cea , lam ( france ) ; ifsi , univ .
padua ( italy ) ; iac ( spain ) ; stockholm observatory ( sweden ) ; imperial college london , ral , ucl - mssl , ukatc , univ .
sussex ( uk ) ; and caltech , jpl , nhsc , univ .
colorado ( usa ) .
this development has been supported by national funding agencies : csa ( canada ) ; naoc ( china ) ; cea , cnes , cnrs ( france ) ; asi ( italy ) ; mcinn ( spain ) ; snsb ( sweden ) ; stfc ( uk ) ; and nasa ( usa ) .
196 natexlab#1#1 , s. s. , tucker , d. l. , lin , h. , diehl , h. t. , annis , j. , buckley - geer , e. j. , & frieman , j. a. 2007 , , 662 , l51 , m. , grevesse , n. , sauval , a. j. , & scott , p. 2009
, , 47 , 481 , a. j. , lutz , d. , genzel , r. , tacconi , l. j. , & lehnert , m. d. 2001 , , 372 , l37 , a. j. , tacconi , l. j. , genzel , r. , lehnert , m. d. , & lutz , d. 2004 , , 604 , 125 , a. , blitz , l. , bolatto , a. , bureau , m. , teuben , p. , wong , t. , & wright , m. 2013 , , 763 , 64 , v. , evans , n. w. , moiseev , a. , king , l. j. , hewett , p. c. , pettini , m. , wyrzykowski , l. , mcmahon , r. g. , smith , m. c. , gilmore , g. , sanchez , s. f. , udalski , a. , koposov , s. , zucker , d. b. , & walcher , c. j. 2007 , , 671 , l9 , s. , magnelli , b. , nordon , r. , lutz , d. , wuyts , s. , altieri , b. , andreani , p. , aussel , h. , &
2011 , , 532 , a49 , e. , & arnouts , s. 1996 , , 117 , 393 , f. , fan , x. , bechtold , j. , mcgreer , i. d. , just , d. w. , sand , d. j. , green , r. f. , thompson , d. , peng , c. y. , seifert , w. , ageorges , n. , juette , m. , knierim , v. , & buschkamp , p. 2010
, , 725 , 1877 , f. , leroy , a. , walter , f. , brinks , e. , de blok , w. j. g. , madore , b. , & thornley , m. d. 2008 , , 136 , 2846 , f. , leroy , a. k. , walter , f. , brinks , e. , de blok , w. j. g. , kramer , c. , rix , h. w. , schruba , a. , schuster , k .- f . , usero , a. , & wiesemeyer , h. w. 2011 , , 730 , l13 + , g. a. , heiderman , a. , gebhardt , k. , evans , n. j. , & adams , j. 2009 , , 704 , 842 , r. d. , & narayan , r. 1992 , , 30 , 311 , l. , fukui , y. , kawamura , a. , leroy , a. , mizuno , n. , & rosolowsky , e. 2007 , protostars and planets v , 81 , a. d. , leroy , a. k. , jameson , k. , ostriker , e. , x gordon , k. , lawton , b. , stanimirovi , s. , israel , f. p. , madden , s. c. , hony , s. , sandstrom , k. m. , bot , c. , rubio , m. , winkler , p. f. , roman - duval , j. , van loon , j. t. , oliveira , j. m. , & indebetouw , r. 2011 , , 741 , 12 , a. d. , wolfire , m. , & leroy , a. k. 2013 , arxiv e - prints , a. , lequeux , j. , & gavazzi , g. 2002 , , 384 , 33 , m. s. , smail , i. , chapman , s. c. , genzel , r. , ivison , r. j. , tacconi , l. j. , alaghband - zadeh , s. , bertoldi , f. , blain , a. w. , casey , c. m. , cox , p. , greve , t. r. , lutz , d. , neri , r. , omont , a. , & swinbank , a. m. 2013 , , 563 , n. , dekel , a. , genzel , r. , genel , s. , cresci , g. , frster schreiber , n. m. , shapiro , k. l. , davies , r. i. , & tacconi , l. 2010 , , 718 , 1001 , r. j. , illingworth , g. d. , oesch , p. a. , franx , m. , labb , i. , trenti , m. , van dokkum , p. , carollo , c. m. , gonzlez , v. , smit , r. , & magee , d. 2012 , , 754 , 83 , v. , iglesias - pramo , j. , seibert , m. , burgarella , d. , charlot , s. , martin , d. c. , xu , c. k. , heckman , t. m. , & et al . 2005 , , 619 , l51 , d. , armus , l. , bohlin , r. c. , kinney , a. l. , koornneef , j. , & storchi - bergmann , t. 2000 , , 533 , 682 , c. m. , chapman , s. c. , beswick , r. j. , biggs , a. d. , blain , a. w. , hainline , l. j. , ivison , r. j. , muxlow , t. w. b. , & smail , i. 2009 , , 399 , 121 , c. m. , chen , c .- c . , cowie , l. , barger , a. , capak , p. , ilbert , o. , koss , m. , lee , n. , le floch , e. , sanders , d. b. , & williams , j. p. 2013 , arxiv:1302.2619 , b. , schiminovich , d. , kauffmann , g. , fabello , s. , hummels , c. , lemonias , j. , moran , s. m. , wu , r. , cooper , a. , & wang , j. 2012 , , 544 , a65 , b. , schiminovich , d. , kauffmann , g. , fabello , s. , wang , j. , & et al . 2010 , , 403 , 683 , g. 2003 , , 115 , 763 , s. c. , blain , a. w. , smail , i. , & ivison , r. j. 2005 , , 622 , 772 , s. c. , helou , g. , lewis , g. f. , & dale , d. a. 2003 , , 588 , 186 , s. c. , smail , i. , blain , a. w. , & ivison , r. j. 2004 , , 614 , 671 , r. , & elbaz , d. 2001 , , 556 , 562 , d. l. , dunne , l. , & eales , s. 2010 , , 403 , 274 , f. , garcia - burillo , s. , braine , j. , schinnerer , e. , walter , f. , & colina , l. 2012 , arxiv e - prints , k. e. k. , swinbank , a. m. , neri , r. , cox , p. , smail , i. , ellis , r. s. , geach , j. e. , siana , b. , teplitz , h. , dye , s. , kneib , j .- p . , edge , a. c. , & richard , j. 2007 , , 665 , 936 , l. , catinella , b. , boissier , s. , boselli , a. , & heinis , s. 2011 , , 415 , 1797 , e. , bournaud , f. , walter , f. , dannerbauer , h. , carilli , c. l. , dickinson , m. , elbaz , d. , morrison , g. e. , riechers , d. , onodera , m. , salmi , f. , krips , m. , & stern , d. 2010 , , 713 , 686 , e. , dickinson , m. , morrison , g. , chary , r. , cimatti , a. , elbaz , d. , frayer , d. , renzini , a. , pope , a. , alexander , d. m. , bauer , f. e. , giavalisco , m. , huynh , m. , kurk , j. , & mignoli , m. 2007 , , 670 , 156 , e. , elbaz , d. , walter , f. , bournaud , f. , salmi , f. , carilli , c. , dannerbauer , h. , dickinson , m. , monaco , p. , & riechers , d. 2010 , , 714 , l118 , d. a. , helou , g. , contursi , a. , silbermann , n. a. , & kolhatkar , s. 2001 , , 549 , 215 , t. m. , hartmann , d. , & thaddeus , p. 2001
, , 547 , 792 , a. l. r. , swinbank , a. m. , smail , i. , cox , p. , edge , a. c. , weiss , a. , harris , a. i. , baker , a. j. , de breuck , c. , geach , j. e. , ivison , r. j. , krips , m. , lundgren , a. , longmore , s. , neri , r. , & flaquer , b. o. 2011 , , 410 , 1687 , h. , daddi , e. , riechers , d. a. , walter , f. , carilli , c. l. , dickinson , m. , elbaz , d. , & morrison , g. e. 2009 , , 698 , l178 , r. , finlator , k. , & oppenheimer , b. d. 2012 , , 421 , 98 , r. , oppenheimer , b. d. , & finlator , k. 2011 , , 415 , 11 , g. , terlevich , r. , & terlevich , e. 2002 , , 330 , 69 , m. , christensen , l. , dodorico , s. , schaerer , d. , & richard , j. 2011 , , 533 , a15 , h. t. , allam , s. s. , annis , j. , buckley - geer , e. j. , frieman , j. a. , kubik , d. , kubo , j. m. , lin , h. , tucker , d. , & west , a. 2009 , , 707 , 686 , e. , bendinelli , o. , bonaccini , d. , close , l. , currie , d. , & parmeggiani , g. 2000 , , 147 , 335 , d. , solomon , p. m. , & radford , s. j. e. 1995 , , 453 , l65 , b. t. , dale , d. a. , bendo , g. , gordon , k. d. , smith , j. d. t. , & et al .
2007 , , 663 , 866 , b. t. , & li , a. 2007 , , 657 , 810 , s. , smail , i. , swinbank , a. m. , ebeling , h. , & edge , a. c. 2007 , , 379 , 308 , m. g. 2001 , , 328 , 223 , d. , daddi , e. , le borgne , d. , dickinson , m. , alexander , d. m. , chary , r. , starck , j. , brandt , w. n. , kitzbichler , m. , macdonald , e. , nonino , m. , popesso , p. , stern , d. , & vanzella , e. 2007 , , 468 , 33 , d. , dickinson , m. , hwang , h. s. , daz - santos , t. , magdis , g. , magnelli , b. , le borgne , d. , galliano , f. , pannella , m. , chanial , p. , & et al .
2011 , , 533 , a119 + , d. , hwang , h. s. , magnelli , b. , daddi , e. , aussel , h. , altieri , b. , amblard , a. , andreani , p. , & et al .
2010 , , 518 , l29 + , c. w. , rieke , g. h. , gordon , k. d. , smith , j .- d .
t. , werner , m. w. , moustakas , j. , willmer , c. n. a. , & vanzi , l. 2008 , , 678 , 804 , b. , tasca , l. , amram , p. , contini , t. , le fvre , o. , queyrel , j. , vergani , d. , garilli , b. , kissler - patig , m. , moultaka , j. , paioro , l. , tresse , l. , bournaud , f. , lpez - sanjuan , c. , & perret , v. 2012 , , 539 , a92 , r. , allam , s. s. , baker , a. j. , lin , h. , lutz , d. , shapley , a. e. , shin , m .- s . , allyn smith , j. , strauss , m. a. , & tucker , d. l. 2010 , , 723 , 729 , r. 2013 , , 433 , 1910 , r. , gnedin , n. y. , & kravtsov , a. v. 2012 , , 747 , 124 , s. l. , papovich , c. , rudnick , g. , egami , e. , le floch , e. , rieke , m. j. , rigby , j. r. , & willmer , c. n. a. 2009 , , 700 , 376 , n. m. , genzel , r. , bouch , n. , cresci , g. , davies , r. , buschkamp , p. , shapiro , k. , tacconi , l. j. , hicks , e. k. s. , genel , s. , shapley , a. e. , erb , d. k. , steidel , c. c. , lutz , d. , eisenhauer , f. , gillessen , s. , sternberg , a. , renzini , a. , cimatti , a. , daddi , e. , kurk , j. , lilly , s. , kong , x. , lehnert , m. d. , nesvadba , n. , verma , a. , mccracken , h. , arimoto , n. , mignoli , m. , & onodera , m. 2009 , , 706 , 1364 , n. m. , genzel , r. , lehnert , m. d. , bouch , n. , verma , a. , erb , d. k. , shapley , a. e. , steidel , c. c. , davies , r. , lutz , d. , nesvadba , n. , tacconi , l. j. , eisenhauer , f. , abuter , r. , gilbert , a. , gillessen , s. , & sternberg , a. 2006 , , 645 , 1062 , n. m. , van dokkum , p. g. , franx , m. , labb , i. , rudnick , g. , daddi , e. , illingworth , g. d. , kriek , m. , moorwood , a. f. m. , rix , h .- w . , rttgering , h. , trujillo , i. , van der werf , p. , van starkenburg , l. , & wuyts , s. 2004 , , 616 , 40 , d. t. , papadopoulos , p. p. , bechtold , j. , seaquist , e. r. , yee , h. k. c. , & scoville , n. z. 1997 , , 113 , 562 , h. , jullo , e. , cooray , a. , bussmann , r. s. , ivison , r. j. , prez - fournon , i. , djorgovski , s. g. , scoville , n. , & et al .
2012 , , 753 , 134 , m. , madden , s. c. , galliano , f. , hony , s. , bendo , g. j. , & sauvage , m. 2011 , , 532 , a56 , j. e. , smail , i. , moran , s. m. , macarthur , l. a. , lagos , c. d. p. , & edge , a. c. 2011 , , 730 , l19 , s. , genzel , r. , bouch , n. , sternberg , a. , naab , t. , schreiber , n. m. f. , shapiro , k. l. , tacconi , l. j. , lutz , d. , cresci , g. , buschkamp , p. , davies , r. i. , & hicks , e. k. s. 2008 , , 688 , 789 , r. , burkert , a. , bouch , n. , cresci , g. , frster schreiber , n. m. , shapley , a. , shapiro , k. , tacconi , l. j. , buschkamp , p. , cimatti , a. , daddi , e. , davies , r. , eisenhauer , f. , erb , d. k. , genel , s. , gerhard , o. , hicks , e. , lutz , d. , naab , t. , ott , t. , rabien , s. , renzini , a. , steidel , c. c. , sternberg , a. , & lilly , s. j. 2008 , , 687 , 59 , r. , tacconi , l. j. , combes , f. , bolatto , a. , neri , r. , sternberg , a. , cooper , m. c. , bouch , n. , & et al .
2012 , , 746 , 69 , r. , tacconi , l. j. , eisenhauer , f. , frster schreiber , n. m. , cimatti , a. , daddi , e. , bouch , n. , davies , r. , lehnert , m. d. , lutz , d. , nesvadba , n. , verma , a. , abuter , r. , shapiro , k. , sternberg , a. , renzini , a. , kong , x. , arimoto , n. , & mignoli , m. 2006 , , 442 , 786 , r. , tacconi , l. j. , gracia - carpio , j. , sternberg , a. , cooper , m. c. , & et al .
2010 , , 407 , 2091 , s. c. o. , & mac low , m .- m .
2011 , , 412 , 337 , a. , marconi , a. , cresci , g. , maiolino , r. , mannucci , f. , calura , f. , cimatti , a. , cocchia , f. , grazian , a. , matteucci , f. , nagao , t. , pozzetti , l. , & troncoso , p. 2011
, , 528 , a88 , v. , bouwens , r. , llingworth , g. , labbe , i. , oesch , p. , franx , m. , & magee , d. 2012 , arxiv:1208.4362 , v. , labb , i. , bouwens , r. j. , illingworth , g. , franx , m. , kriek , m. , & brammer , g. b. 2010 , , 713 , 115 , p. , braine , j. , rodriguez - fernandez , n. j. , israel , f. p. , schuster , k. f. , brouillet , n. , & gardan , e. 2010 , , 512 , a68 , m. j. , abergel , a. , abreu , a. , ade , p. a. r. , andr , p. , augueres ,
j .- l . , babbedge , t. , bae , y. , & et al .
2010 , , 518 , l3 , s. , delannoy , j. , downes , d. , greve , a. , guelin , m. , lucas , r. , morris , d. , radford , s. j. e. , wink , j. , cernicharo , j. , forveille , t. , garcia - burillo , s. , neri , r. , blondel , j. , perrigourad , a. , plathner , d. , & torres , m. 1992 , , 262 , 624 , k. n. , shapley , a. e. , kornei , k. a. , pettini , m. , buckley - geer , e. , allam , s. s. , & tucker , d. l. 2009 , , 701 , 52 , a. i. , baker , a. j. , zonak , s. g. , sharon , c. e. , genzel , r. , rauch , k. , watts , g. , & creager , r. 2010 , , 723 , 1139 , s. , buat , v. , bthermin , m. , aussel , h. , bock , j. , boselli , a. , burgarella , d. , conley , a. , & et al .
2013 , , 429 , 1113 , y. d. , marrone , d. p. , & holder , g. p. 2012
, , 761 , 20 , s. , haynes , m. p. , giovanelli , r. , & brinchmann , j. 2012 , , 756 , 113 , l. , bianchi , s. , & maiolino , r. 2005 , , 434 , 849 , h. s. , elbaz , d. , magdis , g. , daddi , e. , symeonidis , m. , altieri , b. , amblard , a. , andreani , p. , & et al .
2010 , , 409 , 75 , f. p. 1997
, , 328 , 471 , r. j. , papadopoulos , p. p. , smail , i. , greve , t. r. , thomson , a. p. , xilouris , e. m. , & chapman , s. c. 2011 , , 412 , 1913 , t. , ellis , r. , jullo , e. , & richard , j. 2010 , , 725 , l176 , t. , ellis , r. s. , richard , j. , & jullo , e. 2013 , , 765 , 48 , t. a. , swinbank , a. m. , ellis , r. s. , richard , j. , & stark , d. p. 2010
, , 404 , 1247 , s. , cohen , s. , windhorst , r. a. , silk , j. , oconnell , r. w. , dopita , m. a. , dekel , a. , hathi , n. p. , straughn , a. , & rutkowski , m. 2013 , , 429 , l40 , jr .
, r. c. 1998 , , 36 , 189 , l. j. , dopita , m. a. , sutherland , r. s. , heisler , c. a. , & trevena , j. 2001 , , 556 , 121 , l. j. , & ellison , s. l. 2008 , , 681 , 1183 , c. s. , keeton , c. r. , & mcleod , b. a. 2001 , , 547 , 50 , b. p. , gladders , m. d. , hennawi , j. f. , sharon , k. , wuyts , e. , rigby , j. r. , bayliss , m. b. , & dahle , h. 2010 , , 723 , l73 , r. , schneider , p. , & bartelmann , m. 1994 , , 284 , 285 , e. , gemuend , h .- p . , gromke , j. , haslam , c. g. , reichertz , l. , haller , e. e. , beeman , j. w. , hansen , v. , sievers , a. , & zylka , r. 1998 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol .
3357 , society of photo - optical instrumentation engineers ( spie ) conference series , ed .
t. g. phillips , 319325 , m. r. , & dekel , a. 2012 , , 753 , 16 , j. m. , allam , s. s. , annis , j. , buckley - geer , e. j. , diehl , h. t. , kubik , d. , lin , h. , & tucker , d. 2009 , , 696 , l61 , r. , mcgaugh , s. s. , & de blok , w. j. g. 2004 , , 355 , 887 , a. k. , bolatto , a. , gordon , k. , sandstrom , k. , gratier , p. , rosolowsky , e. , engelbracht , c. w. , mizuno , n. , corbelli , e. , fukui , y. , & kawamura , a. 2011 , , 737 , 12 , a. k. , walter , f. , brinks , e. , bigiel , f. , de blok , w. j. g. , madore , b. , & thornley , m. d. 2008 , , 136 , 2782 , e. m. , kewley , l. j. , & larson , k. l. 2010 , , 139 , 712 , s. j. , carollo , c. m. , pipino , a. , renzini , a. , & peng , y. 2013 , arxiv e - prints , h. , buckley - geer , e. , allam , s. s. , tucker , d. l. , diehl , h. t. , kubik , d. , kubo , j. m. , annis , j. , frieman , j. a. , oguri , m. , & inada , n. 2009 , , 699 , 1242 , r. c. , jones , t. , richard , j. , bower , r. g. , ellis , r. s. , swinbank , a. m. , rigby , j. r. , smail , i. , arribas , s. , rodriguez zaurin , j. , colina , l. , ebeling , h. , & crain , r. a. 2012 , , 427 , 688 , d. , poglitsch , a. , altieri , b. , andreani , p. , aussel , h. , berta , s. , bongiovanni , a. , brisbin , d. , & et al .
2011 , , 532 , a90 + , g. e. , daddi , e. , bthermin , m. , sargent , m. , elbaz , d. , pannella , m. , dickinson , m. , dannerbauer , h. , da cunha , e. , walter , f. , rigopoulou , d. , charmandaris , v. , hwang , h. s. , & kartaltepe , j. 2012 , , 760 , 6 , g. e. , daddi , e. , elbaz , d. , sargent , m. , dickinson , m. , dannerbauer , h. , aussel , h. , walter , f. , hwang , h. s. , charmandaris , v. , hodge , j. , riechers , d. , rigopoulou , d. , carilli , c. , pannella , m. , mullaney , j. , leiton , r. , & scott , d. 2011 , , 740 , l15 , g. e. , daddi , e. , sargent , m. , elbaz , d. , gobat , r. , dannerbauer , h. , feruglio , c. , tan , q. , rigopoulou , d. , charmandaris , v. , dickinson , m. , reddy , n. , & aussel , h. 2012 , , 758 , l9 , b. , elbaz , d. , chary , r. r. , dickinson , m. , le borgne , d. , frayer , d. t. , & willmer , c. n. a. 2009 , , 496 , 57 . 2011 , , 528 , a35 , b. , lutz , d. , berta , s. , altieri , b. , andreani , p. , aussel , h. , castaeda , h. , cava , a. , & et al .
2010 , , 518 , l28 + , b. , lutz , d. , santini , p. , saintonge , a. , berta , s. , albrecht , m. , altieri , b. , & andreani , p. e. a. 2012 , , 539 , a155 . 2013 , submitted , b. , saintonge , a. , lutz , d. , tacconi , l. j. , berta , s. , bournaud , f. , charmandaris , v. , dannerbauer , h. , elbaz , d. , frster - schreiber , n. m. , graci - carpio , j. , ivison , r. , maiolino , r. , nordon , r. , popesso , p. , rodighiero , g. , santini , p. , & wuyts , s. 2012 , , 548 , a22 , c. , pforr , j. , renzini , a. , daddi , e. , dickinson , m. , cimatti , a. , & tonini , c. 2010 , , 407 , 830 marshall , p. j. , treu , t. , melbourne , j. , et al .
2007 , , 671 , 1196 , s. s. 1991 , , 380 , 140 , g. r. , heckman , t. m. , & calzetti , d. 1999 , , 521 , 64 , s. f. , genzel , r. , frster schreiber , n. m. , shapiro griffin , k. , mancini , c. , lilly , s. j. , renzini , a. , bouch , n. , burkert , a. , buschkamp , p. , carollo , c. m. , cresci , g. , davies , r. , eisenhauer , f. , genel , s. , hicks , e. k. s. , kurk , j. , lutz , d. , naab , t. , peng , y. , sternberg , a. , tacconi , l. j. , wuyts , s. , zamorani , g. , & vergani , d. 2013 , , 767 , 104 , k. g. , weiner , b. j. , faber , s. m. , papovich , c. , koo , d. c. , somerville , r. s. , bundy , k. , conselice , c. j. , & et al .
2007 , , 660 , l43 , r. , lutz , d. , genzel , r. , berta , s. , wuyts , s. , & et al .
2012 , , 745 , 182 , r. , lutz , d. , saintonge , a. , berta , s. , wuyts , s. , frster schreiber , n. m. , genzel , r. , magnelli , b. , poglitsch , a. , popesso , p. , rosario , d. , sturm , e. , & tacconi , l. j. 2013 , , 762 , 125 , r. , lutz , d. , shao , l. , magnelli , b. , berta , s. , altieri , b. , andreani , p. , aussel , h. , & et al .
2010 , , 518 , l24 + , s. 2010 , in astronomical society of the pacific conference series , vol .
434 , astronomical data analysis software and systems xix , ed .
y. mizumoto , k .-
morita , & m. ohishi , 139 , m. , carilli , c. l. , daddi , e. , mccracken , h. j. , owen , f. n. , renzini , a. , strazzullo , v. , civano , f. , koekemoer , a. m. , schinnerer , e. , scoville , n. , smoli , v. , taniguchi , y. , aussel , h. , kneib , j. p. , ilbert , o. , mellier , y. , salvato , m. , thompson , d. , & willott , c. j. 2009 , , 698 , l116 , c. , finkelstein , s. l. , ferguson , h. c. , lotz , j. m. , & giavalisco , m. 2011 , , 412 , 1123 , c. y. , ho , l. c. , impey , c. d. , & rix , h .- w .
2002 , , 124 , 266 . 2010 , , 139 , 2097 , c. y. , impey , c. d. , rix , h .- w .
, kochanek , c. s. , keeton , c. r. , falco , e. e. , lehr , j. , & mcleod , b. a. 2006 , , 649 , 616 , m. , christensen , l. , dodorico , s. , belokurov , v. , evans , n. w. , hewett , p. c. , koposov , s. , mason , e. , & vernet , j. 2010 , , 402 , 2335 , g. l. , riedinger , j. r. , passvogel , t. , crone , g. , doyle , d. , gageur , u. , heras , a. m. , jewell , c. , metcalfe , l. , ott , s. , & schmidt , m. 2010 , , 518 , l1 , a. , waelkens , c. , geis , n. , feuchtgruber , h. , vandenbussche , b. , rodriguez , l. , krause , o. , renotte , e. , & et al .
2010 , , 518 , l2 , m. , tajer , m. , maraschi , l. , trinchieri , g. , lonsdale , c. j. , chiappetti , l. , andreon , s. , pierre , m. , & et al .
2007 , , 663 , 81 , n. , bolatto , a. d. , xue , r. , wong , t. , leroy , a. k. , walter , f. , bigiel , f. , rosolowsky , e. , fisher , d. b. , vogel , s. n. , blitz , l. , west , a. a. , & ott , j. 2012 , , 745 , 183 , n. a. , erb , d. k. , pettini , m. , steidel , c. c. , & shapley , a. e. 2010 , , 712 , 1070 , n. a. , pettini , m. , steidel , c. c. , shapley , a. e. , erb , d. k. , & law , d. r. 2012 , , 754 , 25 , j. , jones , t. , ellis , r. , stark , d. p. , livermore , r. , & swinbank , m. 2011 , , 413 , 643 , d. a. , carilli , c. l. , walter , f. , & momjian , e. 2010 , , 724 , l153 , a. r. , bell , e. f. , skelton , r. e. , mcintosh , d. h. , somerville , r. s. , zheng , x. , rix , h .- w . , bacon , d. , & et al .
2009 , , 704 , 324 , g. , cimatti , a. , gruppioni , c. , popesso , p. , andreani , p. , & et al .
2010 , , 518 , l25 + , g. , daddi , e. , baronchelli , i. , cimatti , a. , renzini , a. , aussel , h. , popesso , p. , lutz , d. , & et al .
2011 , , 739 , l40 + , d. j. , santini , p. , lutz , d. , shao , l. , maiolino , r. , & et al .
2012 , , 545 , a45 , i. g. , ivison , r. j. , greve , t. r. , amblard , a. , arumugam , v. , & et al .
2012 , , 419 , 2758 , e. , engargiola , g. , plambeck , r. , & blitz , l. 2003 , , 599 , 258 , a. , kauffmann , g. , kramer , c. , tacconi , l. j. , buchbender , c. , catinella , b. , fabello , s. , graci - carpio , j. , & et al .
2011 , , 415 , 32 , a. , kauffmann , g. , wang , j. , kramer , c. , tacconi , l. j. , buchbender , c. , catinella , b. , graci - carpio , j. , & et al .
2011 , , 415 , 61 , a. , tacconi , l. j. , fabello , s. , wang , j. , catinella , b. , & et al .
2012 , , 758 , 73 , s. , rich , r. m. , charlot , s. , brinchmann , j. , johnson , b. d. , & et al .
2007 , , 173 , 267 , d. j. , treu , t. , smith , g. p. , & ellis , r. s. 2004 , , 604 , 88 , d. b. , & mirabel , i. f. 1996 , , 34 , 749 , k. m. , leroy , a. k. , walter , f. , bolatto , a. d. , croxall , k. v. , & et al .
2012 , arxiv:1212.120 , p. , maiolino , r. , magnelli , b. , silva , l. , grazian , a. , & et al .
2010 , , 518 , l154 , m. t. , daddi , e. , bthermin , m. , aussel , h. , magdis , g. , hwang , h. s. , juneau , s. , elbaz , d. , & da cunha , e. 2013 , arxiv:1303.439 , n. z. 2012 , arxiv:1210.6990 , s. , saglia , r. p. , bender , r. , hopp , u. , belloni , p. , & ziegler , b. 1998 , , 298 , 945 , s. 2012 , , 424 , 2429 , j. l. 1968 , atlas de galaxias australes , k. l. , genzel , r. , frster schreiber , n. m. , tacconi , l. j. , bouch , n. , cresci , g. , davies , r. , eisenhauer , f. , johansson , p. h. , krajnovi , d. , lutz , d. , naab , t. , arimoto , n. , arribas , s. , cimatti , a. , colina , l. , daddi , e. , daigle , o. , erb , d. , hernandez , o. , kong , x. , mignoli , m. , onodera , m. , renzini , a. , shapley , a. , & steidel , c. 2008 , , 682 , 231 , a. e. , steidel , c. c. , erb , d. k. , reddy , n. a. , adelberger , k. l. , pettini , m. , barmby , p. , & huang , j. 2005 , , 626 , 698 , c. e. 2013 , phd thesis , rutgers , the state university of new jersey , r. , glover , s. c. , dullemond , c. p. , & klessen , r. s. 2011 , , 412 , 1686 , r. , kelly , b. c. , rahman , n. , bigiel , f. , bolatto , a. d. , clark , p. c. , klessen , r. s. , & konstandin , l. k. 2013 , arxiv:1306.2951 , b. , teplitz , h. i. , chary , r .-
r . , colbert , j. , & frayer , d. t. 2008 , , 689 , 59 , i. , swinbank , a. m. , richard , j. , ebeling , h. , kneib , j .-
p . , edge , a. c. , stark , d. , ellis , r. s. , dye , s. , smith , g. p. , & mullis , c. 2007 , , 654 , l33 , g. p. , ebeling , h. , limousin , m. , kneib , j .- p . ,
swinbank , a. m. , ma , c .- j . , jauzac , m. , richard , j. , jullo , e. , sand , d. j. , edge , a. c. , & smail , i. 2009 , , 707 , l163 , p. m. , downes , d. , radford , s. j. e. , & barrett , j. w. 1997 , , 478 , 144 , v. , mannucci , f. , cresci , g. , maiolino , r. , marconi , a. , nagao , t. , baroni , a. , & grazian , a. 2012 , , 539 , a136 , d. p. , ellis , r. s. , bunker , a. , bundy , k. , targett , t. , benson , a. , & lacy , m. 2009 , , 697 , 1493 , d. p. , schenker , m. a. , ellis , r. s. , robertson , b. , mclure , r. , & dunlop , j. 2012 , arxiv:1208.3529 , e. , gonzlez - alfonso , e. , veilleux , s. , fischer , j. , graci - carpio , j. , hailey - dunsheath , s. , contursi , a. , & poglitsch , a. e. a. 2011 , , 733 , l16 , a. m. , smail , i. , longmore , s. , harris , a. i. , baker , a. j. , de breuck , c. , richard , j. , edge , a. c. , & et al .
2010 , , 464 , 733 , m. , vaccari , m. , berta , s. , page , m. j. , lutz , d. , & et al .
2013 , , 431 , 2317 , l. j. , genzel , r. , neri , r. , cox , p. , cooper , m. c. , shapiro , k. , bolatto , a. , bouch , n. , & et al .
2010 , , 463 , 781 , l. j. , genzel , r. , smail , i. , neri , r. , chapman , s. c. , & et al .
2008 , , 680 , 246 , l. j. , neri , r. , genzel , r. , combes , f. , bolatto , a. , & et al .
2013 , , 768 , 74 , h. i. , mclean , i. s. , becklin , e. e. , figer , d. f. , gilbert , a. m. , graham , j. r. , larkin , j. e. , levenson , n. a. , & wilcox , m. k. 2000 , , 533 , l65 , s. , kim , d .- c . , & sanders , d. b. 2002 , , 143 , 315 , f. , wucknitz , o. , mckean , j. p. , & garrett , m. a. 2010 , , 524 , a79 , j. l. , cooray , a. , de bernardis , f. , amblard , a. , arumugam , v. , aussel , h. , baker , a. j. , bthermin , m. , & et al .
2013 , , 762 , 59 , s. m. , neistein , e. , & dekel , a. 2011 , , 417 , 2737 , a. , downes , d. , walter , f. , & henkel , c. 2007 , in astronomical society of the pacific conference series , vol .
375 , from z - machines to alma : ( sub)millimeter spectroscopy of galaxies , ed .
a. j. baker , j. glenn , a. i. harris , j. g. mangum , & m. s. yun , 25+ , k. e. , van dokkum , p. g. , brammer , g. , & franx , m. 2012 , , 754 , l29 , c. d. 1995 , , 448 , l97 + , c. d. , petitpas , g. r. , iono , d. , baker , a. j. , peck , a. b. , krips , m. , warren , b. , golding , j. , atkinson , a. , armus , l. , & et al .
2008 , , 178 , 189 , e. , glazebrook , k. , blake , c. , wyder , t. , martin , c. , poole , g. b. , sharp , r. , couch , w. , kacprzak , g. g. , brough , s. , colless , m. , contreras , c. , croom , s. , croton , d. , davis , t. , drinkwater , m. j. , forster , k. , gilbank , d. g. , gladders , m. , jelliffe , b. , jurek , r. j. , li , i .-
h . , madore , b. , pimbblet , k. , pracy , m. , woods , d. , & yee , h. k. c. 2011 , , 417 , 2601 , s. a. , larkin , j. e. , barczys , m. , erb , d. k. , iserlohe , c. , krabbe , a. , law , d. r. , mcelwain , m. w. , quirrenbach , a. , steidel , c. c. , & weiss , j. 2007 , , 658 , 78 , e. , rigby , j. r. , gladders , m. d. , gilbank , d. g. , sharon , k. , gralla , m. b. , & bayliss , m. b. 2012 , , 745 , 86 , e. , rigby , j. r. , sharon , k. , & gladders , m. d. 2012 , , 755 , 73 , s. , frster schreiber , n. m. , lutz , d. , nordon , r. , berta , s. , altieri , b. , andreani , p. , & et al .
2011 , , 738 , 106 , s. , frster schreiber , n. m. , van der wel , a. , magnelli , b. , guo , y. , genzel , r. , lutz , d. , aussel , h. , & et al .
2011 , , 742 , 96 , t .-
t . , kewley , l. j. , swinbank , a. m. , riterchard , j. , & livermore , r. c. 2011 , , 732 , l14
j1137 is a galaxy - scale gravitational lens system , where a luminous red galaxy ( i.e. the lens galaxy ) at @xmath257 is forming a bright blue arc of the background source ( i.e. the source galaxy ) .
follow - up spectroscopy of the bright blue arc by @xcite confirms a primary source galaxy redshift of @xmath258 and shows evidence of a secondary source galaxy at @xmath259 .
as shown in figure 2 of @xcite , sdss imaging of the blue arc resembles two split knots and does not distinguish between the two background sources that are very close in redshift , forming nearly overlapping multiple images .
however , multiwavelength _
hst_-wfpc2 imaging of sdss j1137 indicates that the bright blue arc is comprised of several images , as shown in figure [ fig : sdssj1137 + 4936](_left _ ) .
figure [ fig : sdssj1137 + 4936](_right _ ) demonstrates the subtraction of the lens galaxy light profile , showing the lensed features more clearly : * features * a * through * c * , which resemble three small knots merging together to form a faint secondary arc , result from the lensing of the source galaxy at @xmath259 * feature * d * is an extraneous , non - lensing object that is masked during the initial lens modeling process * feature * e * corresponds to the primary lensed arc , resulting from the lensing of the source galaxy at @xmath258 .
therefore , an accurate lens model must account for all images except * d*. we use lensfit @xcite , an extension of the galaxy decomposition software galfit @xcite for strong gravitational lens analysis , to derive the lens model .
we refer the reader to @xcite and bandara et al .
( 2013 , apj submitted ) , for a detailed overview of lensfit and its application for the analysis of galaxy - scale gravitational lenses discovered in the sloan lens acs ( slacs ) survey , and summarize our methodology below . to describe the mass distribution of the lens galaxy ,
we assume a singular isothermal ellipsoid mass model ( sie , * ? ? ?
* ) with an external shear field to model the tidal effects by nearby objects .
the projected mass density of the sie model is , @xmath260^{-1/2 } \label{eq : sie}\end{aligned}\ ] ] where @xmath261 is the mass scale and @xmath262 is the axis ratio of the mass model .
the mass scale parameter ( @xmath261 ) is approximately the einstein radius of the lens ( denoted as @xmath263 ) ; however , this relation is only exact at the @xmath264 limit @xcite . at this limit
, @xmath263 is related to the physical quantities of the mass model by , @xmath265 where @xmath266 is the velocity dispersion of the mass model and @xmath267 and @xmath268 are angular diameter distances from the lens to the source galaxy and from the observer to the source galaxy respectively @xcite . the sie mass model of lensfit
is characterized by the following parameters : mass model centroid ( @xmath269 ) , einstein radius ( @xmath263 ) , axis ratio of the mass model ( @xmath270 ) , position angle of the major axis measured e from n ( @xmath271 ) , external shear ( @xmath272 ) and the position angle of the external shear component measured e from n ( @xmath273 ) .
furthermore , lensfit is a parametric lensing code that describes the _ unlensed _ source galaxy light profile through a set of parametric functions . to model the source galaxies of j1137
, we use s@xmath274rsic profiles @xcite , @xmath275 } ) \label{eq : sersic}\end{aligned}\ ] ] where @xmath276 is the surface brightness at a given radius @xmath277 , @xmath278 is the half - light radius ( i.e. also referred to as the effective radius , @xmath279 ) , @xmath280 is the pixel surface brightness at the effective radius and @xmath281 is the concentration parameter .
the elliptically symmetric @xmath282 profile is characterized by the following parameters : position of the @xmath282 component ( @xmath283 ) , half - light radius ( @xmath278 ) , apparent magnitude ( @xmath284 ) , the @xmath282 index ( _ n _ ) , axis ratio of the elliptical profile ( @xmath262 ) and the position angle of the major axis measured e from n ( @xmath271 ) .
we perform the lens modeling of j1137 in the _
hst_-wfpc2 filters f814w , f606w and f450w .
since the gravitational lensing phenomenon is achromatic , the mass - model should be identical across multiple filters within the systematic uncertainties .
therefore , when modeling a gravitational lens in multiple filters , the mass - model parameters are typically fixed to those determined by the highest signal - to - noise filter ( in this case , the f814w filter ) . however , we allow the sie mass model parameters to vary freely for the f606w filter ( which also has high signal - to - noise ) , such that we can test whether the lens modeling process is sufficiently robust to converge to the same mass model parameters for different filters .
we find that the sie mass model parameters from f814w and f606w imaging are virtually indistinguishable , thus confirming that the lens model solution is robust .
since f450w imaging has a lower signal - to - noise , we initially fix the sie mass model parameters to those constrained by f814w and f606w imaging . during the final iteration of the f450w lens model , we allow the mass model to vary freely and find that the fractional difference between f450w parameters and those obtained from f814w / f606w imaging is less than @xmath285 .
the morphology of the @xmath258 _ unlensed _ source galaxy ( which forms the lensed arc * e * on the image plane ) is best described by three @xmath282 components , in the f814w and f606w filters , and two @xmath282 components in the f450w filter .
in addition , the morphology of the @xmath259 unlensed source galaxy ( which forms features a through c on the image plane ) is best described by a single @xmath282 component in all three filters .
figure [ fig : lensfit_breakdown1 ] shows the results of lens modeling of sdss j1137 in the f814w filter . as indicated by the `` double residual '' image of sdss j1137 , shown in the fourth panel of figure [ fig : lensfit_breakdown1 ]
, the complete lens model predicts images * a * through * c * and * e * but not * d * ( which was unmasked during the final steps of the lensing analysis ) .
lensfit output parameters include the unlensed flux of each @xmath282 component of the source galaxy light profile . due to the circular feedback mechanism between the mass model and the source galaxy light profile , the use of multiple @xmath282 components to describe the unlensed source galaxy yields the best - fit mass model .
in other words , it is important to minimize the residuals of the source galaxy light profile to fully constrain the mass model , using multiple @xmath282 components if necessary .
however , from the standpoint of computing the overall intrinsic properties of the source galaxy ( e.g. for comparison with non - lensed galaxy samples ) , we require a simplified representation that mimics the analysis techniques of high - redshift studies which typically use a single @xmath282 component in galaxy fitting . therefore , we compute the _ global _ properties of the unlensed @xmath258 source galaxy ( i.e. the flux of a _ single _ @xmath282 component that best minimizes the overall residuals ) by performing an additional lensfit iteration using single @xmath282 components . the sie mass model , which was fully constrained by the use of multiple @xmath282 components to define the @xmath258 source galaxy , and the @xmath259 source galaxy light profile parameters are fixed to those implied by the best - fit lens model . for this study ,
the quantity of interest is the magnification factor of the source galaxy at @xmath258 resulting in lensed image * e * , defined as @xmath286 we use the sextractor photometry package @xcite to deblend and measure the flux within the lensed image * e * in all three filters .
we then combine the total flux of the lensed image , obtained from the sextractor photometry , and the unlensed flux of the _ global _ source galaxy profile to derive the magnification factor , according to equation [ eq : magnification ] .
the total magnification factor of sdss j1137 is the mean value of f814w , f606w and f450w filters and corresponds to @xmath287 . to estimate the uncertainty on this measurement , we propagate the errors coming from the determination of both the unlensed and lensed galaxy fluxes , of which the magnification factor is the ratio .
the former includes the error on the slope of the mass model @xcite , the error caused by the lens galaxy subtraction , and the error associated with the psf model , while the lensed galaxy flux uncertainty comes only from the sextractor photometry . including all these sources of uncertainty , we arrive at a total error of 18@xmath288 for the magnification factor of j1137 .
this also justifies our choice of a blanket 20% uncertainty on the magnification factors taken from the literature when published without an uncertainty value ( see table [ sampletab ] ) .
@xmath289{fig12a.eps } & \includegraphics[width=0.48\textwidth]{fig12b.eps } \end{array}$ ] @xmath290{fig13a.eps } & \includegraphics[width=0.23\textwidth]{fig13b.eps } & \includegraphics[width=0.23\textwidth]{fig13c.eps } & \includegraphics[width=0.23\textwidth]{fig13d.eps } \end{array}$ ]
the _ spitzer _ archive was queried for observations of the lensed galaxies in the 3.6 and 4.5@xmath17 m bands , and the calibrated images retrieved .
the only lensed galaxy in our sample without irac coverage is j1441 .
the 3.6@xmath17 m images of the other 16 galaxies are shown in figure [ stamps ] .
most of the lensed galaxies appear as extended arcs in the irac images , and they are generally located within the wings of the foreground lenses , which are typically bright at these wavelengths .
for these two main reasons , standard photometry packages are not optimal and instead a custom idl script was written to measure accurate fluxes .
we define an annulus , typically centered on the foreground lens , that encompasses fully the arc .
the section of this annulus containing the arc is used to integrate the flux of the object , and the remaining section is used to estimate the background , as indicated n figure [ stamps ] .
the advantages of this approach are twofold : ( 1 ) the shape of the aperture naturally matches well that of the lensed arcs , and ( 2 ) the background is estimated over a region with equivalent noise properties and contamination from the bright lens .
the exact parameters defining the aperture ( inner and outer radii , position angle and opening angle ) are fixed using curve of growth arguments . in figure
[ stamps ] , the curves of growth obtained by changing the outer radius of the apertures are shown . the background level within the aperture
is estimated by measuring the mode of the sky pixels in the background aperture " and multiplying it by the area of the aperture .
the mode is calculated with the mmm task in idl ( based of the daophot task of the same name ) , allowing an accurate measure of the sky properties even in the presence of bright point sources within the background aperture " .
the error on the total flux of the lensed galaxy is obtained by adding in quadrature the uncertainty on the flux in the aperture from the irac error map , the scatter in the sky values , and the uncertainty in the mean sky brightness . some of the lensed galaxies are however best modeled by a simple circular aperture than with an arc aperture .
this is the case of the eye , the eyelash , cb58 , j0744 , j1149 and j1226 . in those cases ,
a simple circular aperture is used , with a radius determined from the curve of growth ( see fig . [ stamps ] ) .
the background is determined from an annulus around this aperture , using the same technique described above .
for the lensed galaxy j0712 , neither technique produces reliable results , since the faint blue arc is heavily blended with the brighter lens , as shown in fig .
[ stamps ] . for that galaxy ,
we adopt the stellar mass of @xmath291 derived by @xcite from sed fitting to the hst photometry .
the final , background subtracted fluxes at 3.6 and 4.5@xmath17 m are summarized in table [ masstab ] , and our methodology to derive stellar masses from these fluxes is described in [ mstarsection ] .
the lensed galaxy j0901 has been reported by @xcite to harbor an agn , based on a high [ nii]/h@xmath47 ratio , above the range expected from star forming galaxies , suggesting excitation of the lines through agn and/or shocks ( e.g. * ? ? ?
* ; * ? ? ?
care must therefore be taken that the agn emission does not affect our estimates of stellar mass and star formation rate . since
our stellar masses are derived from the irac photometry at 3.6 and 4.5 @xmath17 m , we compare in figure [ j0901sed ] the shape of the observed near - infrared spectral energy distribution to templates from the swire library @xcite .
the emission observed in the irac bands is fully consistent with the starburst templates , suggesting that the 3.6 and 4.5 @xmath17 m fluxes ( and therefore the stellar mass derived from them ) are not affected by the presence of an agn .
@xcite also concluded from _
spitzer_/irs spectroscopy that the contribution of the agn to the mid - infrared emission is insignificant and that the system is instead starburst - driven . in the absence of an agn signature at near- and mid - infrared wavelengths , it can be safely concluded that there is no contamination at even longer wavelengths and therefore that the sfr and dust mass / temperature derived from the herschel photometry are secure ( see * ? ? ?
* for a discussion of agn contamination in the far - infrared ) .
however , given the un - physically high [ nii]/h@xmath47 ratio for star - forming regions , we adopt for this galaxy the metallicity given by the mz relation , @xmath292 .
this value is consistent with the lower limit of 1.3@xmath293 reported by @xcite based on argon line fluxes from _ spitzer _ mid - infrared spectroscopy . in the absence of a [ nii]/h@xmath47 ratio measurement
, we computed the metallicity for j1226 using the @xmath144 and @xmath294 indicators , using h@xmath43 , [ oiii ] and [ oii ] line fluxes from the spectrum of @xcite , further corrected for extinction ( j.r . rigby , private communication ) .
the values derived are @xmath295 and @xmath296 . to compute a metallicity from these line ratios
, we use the calibration of @xcite as parametrized by @xcite .
the measured values of @xmath144 and @xmath297 put this galaxy at the transition between the two branches of the relation , with 12@xmath298o / [email protected] and 8.45 for the lower and upper branches , respectively .
we can therefore obtain a reliable metallicity estimate , even without [ nii]/h@xmath47 to discriminate between the two branches .
finally , we use the relation of @xcite to convert this metallicity on the @xcite scale , for consistency with the rest of the sample ( section [ metalsection ] ) .
the final value we adopt for j1226 is therefore 12@xmath298o / h@xmath300 , where the uncertainty is the sum in quadrature of the propagated measurement errors , and the average scatter of 0.15 in the @xcite calibration . | we combine iram plateau de bure interferometer and _ herschel _ pacs and spire measurements to study the dust and gas contents of high - redshift star forming galaxies .
we present new observations for a sample of 17 lensed galaxies at @xmath0 , which allow us to directly probe the cold ism of normal star - forming galaxies with stellar masses of @xmath1@xmath2 , a regime otherwise not ( yet ) accessible by individual detections in _ herschel _ and molecular gas studies .
the lensed galaxies are combined with reference samples of sub - millimeter and normal @xmath3 star - forming galaxies with similar far - infrared photometry to study the gas and dust properties of galaxies in the sfr-@xmath4-redshift parameter space .
the mean gas depletion timescale of main sequence galaxies at @xmath5 is measured to be only @xmath6myr , a factor of @xmath7 ( @xmath8 ) shorter than at @xmath9 ( @xmath10 ) , in agreement with a @xmath11 scaling .
the mean gas mass fraction at @xmath12 is @xmath13 ( 44% after incompleteness correction ) , suggesting a flattening or even a reversal of the trend of increasing gas fractions with redshift recently observed up to @xmath14 .
the depletion timescale and gas fractions of the @xmath5 normal star - forming galaxies can be explained under the equilibrium model " for galaxy evolution , in which the gas reservoir of galaxies is the primary driver of the redshift evolution of specific star formation rates . due to their high star formation efficiencies and low metallicities , the @xmath5 lensed galaxies have warm dust despite being located on the star formation main sequence . at fixed metallicity
, they also have a gas - to - dust ratio 1.7 times larger than observed locally when using the same standard techniques , suggesting that applying the local calibration of the @xmath15-metallicity relation to infer the molecular gas mass of high redshift galaxies may lead to systematic differences with co - based estimates . |
deep imaging surveys for high redshift studies have , in recent years , been carried out over an increasing range of wavelengths from the x - ray to the radio regimes .
an important motivation is the recognition that surveys at any given wavelength range may provide an incomplete and biased census of high redshift galaxy populations .
moreover , a multi - wavelength approach is highly desirable to assess the nature of and establish links between high redshift samples selected at various wavelengths , and to constrain the properties of their stars and gas contents as indicators of their evolutionary state .
deep surveys at near - infrared wavelengths ( nir ; @xmath25 ) have emerged as an important way to investigate the high redshift universe ( see , e.g. , @xcite for a review , and references therein ) .
nir observations allow one to access the rest - frame optical light of galaxies at @xmath26 .
nir surveys provide a vital complement , for instance , to optical surveys , which probe the rest - frame uv light at these redshifts ( e.g. * ? ? ?
* ; * ? ? ?
* ) : the rest - frame optical is less affected by the contribution from young massive stars and by dust obscuration , and is expected to be dominated by longer - lived lower - mass stars .
selection in the nir , compared to the optical , thus provides a more reliable means of tracing the bulk of the stellar mass at high redshift , of probing earlier star formation epochs , and of identifying galaxies with spectral energy distributions ( seds ) similar to those of galaxies locally .
a significant advantage of nir surveys is that they enable , in the same rest - frame optical , direct and consistent comparisons between @xmath26 samples and the well - studied populations at @xmath27 . with the sensitive instrumentation and large format nir detectors mounted on ground - based @xmath28 telescopes or space - based facilities available , it is possible to carry out nir surveys of comparable depth and quality as in the optical .
this greatly helps in the identification of counterparts in optical observations and in the construction of accurate seds .
yet , nir surveys with sufficient depth and angular resolution remain challenging .
for instance , a typical normal nearby @xmath29 galaxy would have an apparent @xmath30-band magnitude @xmath31 if placed at @xmath32 @xcite . among current nir surveys ,
those reaching such faint limits are still scarce .
they include the _ hubble space telescope _
( hst ) nicmos fields of the hubble deep field north ( hdf - n ; * ? ? ?
* ; * ? ? ?
* ) and hubble ultra deep field ( udf ; * ? ? ?
* ) , the hubble deep field south ( hdf - s ) parallel nicmos field @xcite , and the two udf parallel nicmos fields @xcite .
ground - based surveys include the subaru deep field @xcite , and several blank fields imaged at the keck telescopes @xcite . in this context , we initiated the faint infrared extragalactic survey ( fires ; * ? ? ?
* ) carried out at the eso _ very large telescope _
the survey consists of very deep imaging in the @xmath0 , @xmath1 , and @xmath2 bands with the isaac instrument @xcite of two fields with existing deep optical hst wfpc2 data : the hdf - s and @xmath3 , a field around a massive cluster at @xmath4 . in total
, 103 h of isaac observations were spent in a single @xmath33 pointing for hdf - s and 77 h in a @xmath5 mosaic of four pointings for @xmath3 .
the @xmath2-band @xmath6 limiting total ab magnitudes for point sources are @xmath34 for hdf - s and @xmath35 for the @xmath3 field , and the effective resolution of the coadded maps in @xmath0 , @xmath1 , and @xmath2 is @xmath36 fwhm for both fields . together with the hst wfpc2 data and additional vlt fors1 optical imaging for the @xmath3 field
, this provides a unique data set with seven or eight band photometry covering @xmath37 over a total area of @xmath38 . with the hdf - s data , fires reaches fainter limits in all three nir bands than any other existing ground - based survey and provides the deepest @xmath30-band map to date .
the @xmath3 field is an essential complement as it considerably increases the survey area to an @xmath39 brighter limit .
the data and source catalogues of the hdf - s are described by @xcite .
several key results of both fields have been published , including sed modeling , spectroscopy , and clustering analysis of the new population of `` distant red galaxies '' at @xmath40 @xcite , determination of the stellar rest - frame optical luminosity density , colours , and mass density at @xmath41 @xcite , determination of the rest - frame optical sizes of galaxies out to @xmath42 @xcite , and the discovery of six large disk - like galaxies at @xmath43 @xcite . in this paper , we present the data and source catalogues of the @xmath3 field . as for hdf - s ,
the @xmath3 raw and reduced data sets are available from the fires web site .
sections [ sect - obs ] and [ sect - red ] describe the observations and data reduction .
sect [ sect - images ] discusses the final reduced images .
section [ sect - det ] describes the @xmath2-band limited source detection and photometric measurements methods and [ sect - redshifts ] the photometric redshift determinations .
section [ sect - catalogue ] lists the catalogue parameters .
section [ sect - analysis ] presents the completeness analysis , number counts , magnitude and colour distributions of the @xmath2-band selected sources .
section [ sect - conclu ] summarizes the paper .
all magnitudes are expressed in the ab photometric system @xcite unless explicitely stated otherwise .
our strategy for the fires survey was to complement our very deep nir imaging of hdf - s with a wider field , allowing for better statistics on source populations down to flux levels twice brighter than the faint limits of hdf - s .
the @xmath3 field was ideally suited for this purpose because of available hst wfpc2 imaging offering a unique combination of field size and depth .
the wfpc2 data consist of a mosaic of six pointings covering @xmath45 observed through the f606w and f814w filters ( effective wavelengths @xmath46 and 8040 , and effective widths @xmath47 and 1540 , respectively ) .
we will denote these bandpasses as @xmath48 and @xmath49 .
the total integration time was 6500 s per band and pointing . to extend the optical wavelength coverage , we obtained @xmath50 imaging with the vlt fors1 instrument . the @xmath3 field benefits from an extensive amount of existing data over the electromagnetic spectrum , including a deep mosaic at 850 taken with the submillimeter common - user bolometric array ( scuba ; k. k. knudsen et al .
, in preparation ) , deep radio 5 ghz imaging from the very large array ( vla ; * ? ? ?
* ) , and x - ray observations with the acis camera onboard the chandra observatory @xcite . optical imaging with acs onboard hst
has also now become publicly available .
spectroscopic redshifts have been measured for over 400 galaxies in the field from data collected at the keck and gemini telescopes and at the vlt ( @xcite ; s. wuyts et al . , m. kriek et al . , in preparation
these spectroscopic determinations enabled us to calibrate our photometric redshifts ( [ sect - redshifts ] ) .
the @xmath4 @xmath3 cluster itself has but a small influence over the area surveyed .
the majority of galaxies detected in the mosaics lie in front of or behind the cluster .
the cluster mass distribution has been modeled accurately by @xcite from weak lensing analysis , so that the gravitational lensing can be well accounted for .
the average background magnification effects over the field of view covered by the fires observations range from 5% to 25% between @xmath51 and 4 , and are largest in the cluster s central regions .
we emphasize that colours and seds are not affected since gravitational lensing is achromatic .
moreover , the surface brightness is preserved and , to first order , the surface density to a given apparent magnitude is insensitive to lensing ( the exact effect depends on the slope of the luminosity function of the sources under consideration ) .
table [ tab - obs ] summarizes the properties of the @xmath3 data set and figure [ fig - fovs ] illustrates the fields of view of the different instruments .
the observations and reduction of the hst wfpc2 mosaics are presented by @xcite . in this and the following section , we focus on the data taken with isaac and fors1 , both mounted at the vlt antu telescope .
all vlt observations were acquired in service mode as part of the eso large program 164.o-0612 ( p.i .
m. franx ) .
the @xmath3 field was observed with isaac @xcite in imaging mode through the @xmath0 , @xmath1 , and @xmath2 filters ( @xmath52 , 1.65 , and 2.16 , and @xmath53 , 0.30 , and 0.27 , respectively ) .
the observations were carried out on 36 nights , in 1999 december , 2000 may , and 2001 february through june .
the short - wavelength camera of isaac is equipped with a rockwell hawaii @xmath54 hgcdte detector array and provides a scale of @xmath55 for a total field of view of @xmath33 .
the wfpc2 mosaic was covered with a square of @xmath56 pointings ( referred to as f1 to f4 ; see figure [ fig - fovs ] ) .
nominal pointings ensured that adjacent fields overlap by @xmath57 .
the data were obtained in `` observing blocks '' ( obs ) consisting of sequences of 30 exposures totaling 1 h integration time in @xmath0 and @xmath1 , and 40 exposures totaling 40 m in @xmath2 .
successive exposures within each ob were dithered randomly in a 20 box to allow the construction of sky frames with minimal object contamination . typical exposure times in @xmath0 , @xmath1 , and @xmath2 were of 120 , 120 , and 60 s , split into 4 , 6 , and 6 sub - integrations , respectively .
the entire data set includes 3207 frames grouped in 93 obs , with total integration times of @xmath58 per field and per band .
we rejected about 3% of the frames , representing 3% of the total integration time , because they did not meet our atmospheric seeing or transmission requirements , or because they were affected by strong instrumental artefacts . throughout the paper , we consider only the 3104 frames from 92 obs with sufficient quality .
most of the data were obtained under excellent seeing , with median full - width at half maximum @xmath59 in @xmath0 and @xmath60 in both @xmath1 and @xmath2 .
ninety percent of the data have a seeing better than @xmath61 .
figure [ fig - seeing ] gives the seeing distribution of the individual frames .
the fwhm was determined by fitting moffat profiles to bright , isolated , and unsaturated stars in each image ( mostly three to five depending on the field , on the signal - to - noise ratio of the stars , and on whether they lie within the field of view of individual dithered frames ) , and averaging the fwhms per image .
the conditions were photometric for about half of the observing nights , or 53% of the obs .
in the course of the 19-month interval during which the nir observations were taken , the antu primary mirror was twice re - aluminized , in 2000 february and 2001 march .
we will refer to the three periods separated by these events as periods p1 , p2 , and p3 .
the effects of re - aluminization are noticeable in our data , with an increase in sensitivity between successive periods as deduced from bright stars counts in the @xmath3 data and reflected in the photometric zero points .
based on the standard stars data used for flux calibration ( [ sub - red_isaac ] ) , the increase in @xmath0 , @xmath1 , and @xmath2 amounts to 37% , 47% , and 50% between p1 and p2 , and 29% , 26% , and 19% between p2 and p3 , for a net gain of @xmath62 .
seven obs were taken in p1 , 41 in p2 , and 45 in p3 .
the isaac pixel scale also varied slightly over time , by a maximum of 0.27% .
figure [ fig - obs_param ] shows the variations with time and airmass of the relative count rates of stars and of the background levels in the individual frames .
the stars are the same as used for the seeing measurements .
their counts were integrated in a 6-diameter circular aperture in the sky - subtracted frames and normalized to the median values in period p3 .
the background was computed as the median count rates over the raw frames .
the figure also plots the seeing fwhm versus airmass .
the stars count rates increase sharply after each realuminization and , over period p2 , are consistent with the long - term progressive loss of sensitivity between two realuminizations . in contrast
, the background levels have overall remained fairly constant , with no systematic trend with period and the variations reflecting mostly the nightly observing conditions .
this constancy in average background levels was also noticed in the hdf - s data set and presumably indicates that light is scattered in as much as out of the beam .
it implies a substantial increase in telescope efficiency and in source signal - to - noise ratio ( s / n ) achievable in background - limited data after realuminization .
the nir background can vary on timescales of minutes ; it is dominated by telluric oh airglow lines in the @xmath0 and @xmath1 bands , and by thermal emission from the telescope , instrument , and atmosphere in the @xmath2 band .
the nightly variations in our data are generally largest and most rapid in the @xmath1 band and show little correlation with the hour of observation .
the @xmath30-band background has comparable levels but smaller short - term variations and is systematically higher at the beginning and end of the night .
the @xmath0 background is the lowest and most stable , and peaks at the beginning of the night .
variations in stars count rates within each period reflect changes in sky transparency and there is no apparent airmass dependence .
this indicates that the atmospheric extinction is negligible for airmasses of @xmath63 as covered by the observations .
the seeing is expected to correlate with airmass ( roughly as @xmath64^{0.6}$ ] ) but no obvious trend is seen in our data , partly due to the small range of airmass and because the seeing is largely determined by the actual observing conditions .
the data suggest an overall increase in background levels with airmass , most clearly in @xmath1 and @xmath2 , which could be explained by the increasing path length of the thermal and oh - line emitting layers .
however , for successive observations such as within obs , variations in airmass are directly coupled with time so that part of possible trends with airmass may be attributed to nightly patterns in the background . the @xmath3 field was observed with fors1 @xcite in direct imaging mode on six nights between 2000 january and march using the bessel @xmath13 , @xmath14 , and @xmath15 filters ( @xmath65 , 4330 , and 5550 , and @xmath66 , 870 , and 1070 , respectively ) .
fors1 employs a @xmath67 tektronix ccd detector .
the standard resolution collimator was selected for the observations , giving a @xmath68 scale and a @xmath69 field of view covering entirely the isaac and wfpc2 mosaics .
the data were collected in obs of six successive single exposures of 550 , 400 , and 300 s in @xmath13 , @xmath14 , and @xmath15 .
the total integration times are 4.6 , 2.0 , and 0.5 h , respectively , distributed over 5 , 3 , and 1 obs .
this excludes the few aborted or complete obs with insufficient quality , lying well outside our observing conditions specifications . within each
ob , the exposures were dithered in a box of @xmath70 .
the average seeing fwhm of the data is @xmath71 in @xmath13 , @xmath72 in @xmath14 , and @xmath73 in @xmath15 .
the seeing was determined as for the isaac data by fitting moffat profiles to five bright , isolated , unsaturated stars in the individual frames .
the @xmath14- and @xmath15-band data were all taken prior to the 2000 february realuminization while the @xmath13-band observations were carried out both before and after .
our @xmath3 data indicate an increase in @xmath13-band sensitivity of @xmath74 based on stars count rates in a 6-diameter aperture and small rotation angle differences in the camera orientation ( @xmath75 between various obs ) .
the data were obtained during photometric nights except for the three @xmath13-band obs taken in 2000 march .
we reduced the isaac data using the dimsum package within iraf , the eclipse software , and additional routines we have developed specifically for our fires observations .
we followed the general procedure for the hdf - s described by @xcite with adjustments necessary because of the different characteristics of the @xmath3 data set .
in particular , the observations were taken over a longer period where the instrument behaviour changed noticeably . also , the shallower data and more crowded field around @xmath3 required optimization of various steps to minimize residual noise and systematic effects .
we generated dark@xmath76bias images for use in making flat fields by averaging together individual dark frames taken with each detector integration time ( dit ) employed for the flat fields .
the typical level of the dark frames for a given dit varied significantly over time with five distinct epochs identified in our data .
we thus created dark@xmath76bias images appropriate for each epoch in addition to those made from all available dark frames .
we constructed flat fields from sequences of successive exposures of the sky taken during twilight , on nights where the @xmath3 field was observed or as close as possible in time , using the eclipse task `` flat . ''
this task also produced the initial bad ( dead ) pixel masks .
we made three flat fields per night and band as follows : ( 1 ) by forcing eclipse / flat to compute the dark@xmath76bias by itself , ( 2 ) by providing the average dark@xmath76bias image from all available data for the corresponding dit , and ( 3 ) by providing the dark@xmath76bias image for the corresponding dit and observing epoch .
the flat fields often exhibited abrupt jumps at mid - array ( row 512 ) with amplitude up to 20% and significant differences in large - scale gradients between the three methods , typically at the 5% level .
these features carry the imprint of the detector bias , which dominates the signal in the dark frames .
the bias level is very uniform along rows but varies importantly along columns , with a steep initial decline flattening progressively from bottom to top of each of the lower and upper halves of the array frame .
the bias depends not only on the dit but also on the detector illumination during and prior to an exposure , so that for a given dit the dark@xmath76bias image may not represent well the dark@xmath76bias present in the twilight frames . for each night and band , we chose the flat field for which the bias signatures were minimized .
this was generally the one obtained with method 1 , which better accounted for the intrinsic dark@xmath76bias of the twilight frames , but in a few cases method 3 was the most successful .
the nightly flat fields also varied noticeably between epochs where the dark@xmath76 bias changed most but were otherwise very similar .
we created `` master flat fields '' by averaging the adopted nightly flat fields for each epoch .
all flat fields were normalized to unit mean over the image . for each science exposure in a given ob
, we constructed a `` sky '' image to remove the sky and telescope background , and the dark@xmath76bias signal .
we used a minimum of three and a maximum of eight temporally adjacent frames depending on when the science exposure was taken during the ob sequence . in a `` first pass , ''
the individual sky frames were scaled to a common median level equal to that of the science frame , averaged together with minimum - maximum rejection algorithm at each pixel , and subtracted from the science frame .
cosmic rays and hot pixels were identified using the `` l. a. cosmic '' routine @xcite and negative outliers at @xmath77 below the mean residual background were added to the bad pixel lists .
the sky - subtracted frames within each ob were aligned using integer - pixel shifts and coaveraged . from this intermediate combined ob image ,
an object mask was created by applying an appropriate threshold to identify astronomical sources . in a second `` mask pass , ''
the sky subtraction procedure was repeated using the object masks to minimize the contribution of sources in the background computation .
the mask - pass sky - subtracted frames were divided by the master flat field for the corresponding epoch of observation and band , and combined as before .
associated weight maps were generated , proportional to the total integration time of all frames contributing at each pixel .
we normalized the final ob images to a 1s integration time and the weight maps to a maximum value of 1 .
non - linearity ( raw counts exceeding 10000 adu , for which eso reports 99% linearity ) is not a concern except for the very few brightest sources ( @xmath78 per isaac field ) in the frames obtained under the best seeing and transparency conditions .
we did not apply any correction for non - linearity but excluded those stars with raw counts well above 10000 adu in the flux calibration and psf determination described below . in coaveraging the reduced frames , bad pixels and cosmic rays
were masked out and no clipping or rejection algorithm was applied .
we verified that no strongly deviant pixel , either persistent or as single event , was missed by comparing the ob images to versions obtained by median - combining the registered sky - subtracted frames without masking and from the behaviour of individual pixels throughout the stack of unaligned unmasked sky - subtracted frames .
in several obs , we noticed satellite tracks , moving targets ( presumably kuiper belt objects ) , and spurious features produced by reflection effects of bright sources falling on the edges of the detector array .
we removed them with customized masking .
we inspected the individual reduced frames to identify those with significant residual artefacts due to poor sky subtraction or instrumental features .
we implemented additional mask - pass routines in dimsum to improve the quality of the affected data , described in what follows .
_ bias residuals
particular processing was required to account for the time and illumination - dependent bias level of the detector .
the median - scaling applied in constructing the sky images accounts for the global intensity of the dark@xmath76bias but not for its spatial structure .
because the bias variations are strong along columns but negligible along rows , we removed the residuals by subtracting the median row by row in all sky - subtracted frames . in many obs , the bias varied so strongly and rapidly that it led to spurious `` numerical ghosts '' with shapes reminiscent of the masked sources .
these features were most conspicuous in regions of steepest spatial gradients in the bias ( lower parts of the bottom and top half of the frames ) and during periods of steepest gradients in time ( related to the illumination history ) .
the average of the scaled sky frames did not represent accurately the local background in the science frame and the effect was worsened by the object masking which reduced locally the number of frames used to estimate the background .
the computed background in the regions with objects masked in one or more of the sky frames thus differed significantly from that of adjacent regions masked in none of the sky frames , imprinting positive ghosts for a locally underestimated background or vice - versa .
we eliminated the numerical ghosts while constructing the sky frames : in addition to the median - scaling , we accounted for an additive component representing the bias offsets between a science frame and each of its corresponding scaled sky frames by fitting a median to each row in the difference maps .
we applied this procedure to the 28 obs where ghosts were clearly recognized by eye inspection , mostly in the @xmath0 and @xmath1 bands .
_ residual large - scale gradients
_ for about half of the data , significant large - scale gradients remained in the sky - subtracted frames .
they were caused primarily by highly variable sky and thermal background emission , and stray light .
the fractions of obs affected in each band reflect the background behaviour seen in figure [ fig - obs_param ] : 92% in @xmath1 , 46% in @xmath2 , and 27% in @xmath0 .
we first attempted to remove these residuals by fitting two - piece cubic splines directly to the object - masked frames , along rows or columns depending on the gradient pattern . owing to crowding in the @xmath3 field ,
the fits were however generally poorly constrained .
we thus created template `` flat images '' by registering and coaveraging all sky - subtracted frames unaffected by residual gradients , taken from all obs of the appropriate field and band to increase the s / n ratio .
we subtracted the flat templates from the affected frames .
this largely eliminated sources , leaving residual cores due to small mismatches in alignment and seeing .
we used these difference maps to create new masks and fit the two - piece cubic splines . because the masked area was substantially reduced , the fits were much better constrained , allowing successful removal of the residual gradients .
we performed this second - order sky subtraction for the 29 obs where the gradients were strongest and apparent in the combined ob images . _ instrumental features and other artifacts _ a high spatial frequency pattern known as the `` odd - even column effect '' appeared in part of the data taken after the 2001 march realuminization ( 19 obs ) .
it is characterized by an offset in level between adjacent columns , occasionally further modulated over a width of about 20 pixels and generally different between detector quadrants .
when present , we corrected for this artifact by subtracting the median of the background along columns and for each quadrant separately in the object - masked sky - subtracted frames .
finally , high to moderate spatial frequency ripples presumably from moonlight reflections were present in the background of 41 obs , with important frame - to - frame variations in intensity , structure , and orientation .
a semi - circular high - frequency time - dependent interference pattern was also obvious in 3 obs .
the complexity and variability of these artifacts make them very difficult to correct for . since they generally averaged out in the combined ob images , we did not treat them in any particular way except by excluding the most affect frames which left noticeable patterns in the ob images .
we corrected the ob images for the documented isaac geometrical distortion while simultaneously registering and interpolating the isaac data to the wfpc2 maps at @xmath79 .
this `` rectification '' was done using the same algorithm as for the hdf - s data described by @xcite .
briefly , the transformation corrects for distortion , aligns the isaac ob images ( allowing for sub - pixel shifts ) , and applies the best - fit linear transformation determined from the positions of stars and point - like sources across the @xmath0-band images and the wfpc2 @xmath49-band mosaic .
the images are resampled using third - order polynomial interpolation , and only once to minimize the effects on the noise properties .
we rescaled the rectified ob images to ensure flux conservation .
since the isaac field distortions have small amplitude and thus negligible effects on the photometry , we used a constant flux scaling factor for the entire images .
the rectification of the ob images instead of the individual frames affects the point - spread function ( psf ) because of the dithering and integer - pixel registration within each ob .
figure [ fig - seeing_obs ] plots the psf fwhm measured in the rectified ob images versus the median of the fwhms in the single frames of the corresponding ob .
the effect is negligible at @xmath80 and increases towards lower fwhm up to a maximum of 25% at @xmath81 , the limit of undersampling .
we determined the instrumental photometric zero points ( zps ) using observations of standard stars .
all stars were selected from the lco / palomar nicmos list @xcite except for one taken from the ukirt faint standards list ( * ? ? ?
* observed on 1999 december 18 ) .
we measured all fluxes and magnitudes for photometric calibration in a 6-diameter circular aperture , which proved to be adequate by monitoring the growth curve of the standard stars and of reference stars in the @xmath3 field .
the standard stars were observed in sequences of five dithered exposures .
we integrated the count rates of the stars within the 6-diameter aperture in each frame of a sequence after sky subtraction and flat - fielding ( using the appropriate flat field for the epoch of observation and band ) .
we computed single - star fluxes by averaging , with @xmath82-clipping , the count rates for each sequence .
typical uncertainties , taken as the median of the standard deviations of single - star fluxes , were of 1.7% ( average of 2.1% ) . using the calibrated magnitudes , we derived single - star zps .
we identified as non - photometric the nights where single - star zps varied by more than twice the typical uncertainties ( i.e. 3.4% ) , and excluded them from the subsequent analysis . for each of the remaining nights , we derived nightly zps by averaging the single - star zps . because the nightly zps exhibit a large systematic increase after realuminization events , we computed period zps by averaging with @xmath82-clipping the nightly zps for each of p1 to p3 .
nights rejected by the @xmath82-clipping were added to the initial list of non - photometric nights . in total , 17 out of 36 nights were photometric according to our criteria .
the median night - to - night dispersion of the zps is 0.024 mag and varies from 0.007 and 0.060 mag depending on the band and period .
the average zps per period in @xmath0 , @xmath1 , and @xmath2 increased by 0.340 , 0.415 , and 0.441 mag from p1 to p2 , and by 0.276 , 0.255 , and 0.188 mag from p2 to p3 .
we adopted as reference zps for our @xmath3 data those of p3 , which are reported in table [ tab - zps ] .
we did not account for atmospheric extinction or for colour terms .
the calibration stars observations cover airmasses @xmath83 , at which 95% of the @xmath3 data were obtained . as for the @xmath3 stars count rates ( figure [ fig - obs_param ] ) , the standard stars data do not show any measurable correlation with airmass .
the isaac @xmath1 and @xmath2 filters match well those used to establish the `` lco '' standard system of @xcite while the @xmath0 filter has a slightly redder effective wavelength than the lco @xmath84 filter . for a colour term @xmath85 , the colour coefficient @xmath86 is theoretically expected to be @xmath87 in @xmath1 and @xmath2 , and @xmath88 in @xmath0 @xcite but these values have not been verified experimentally yet .
the zps derived for the individual stars do not reveal any dependence on @xmath89 colour over the range @xmath90 covered by the standard stars set , with a scatter much larger than the expected colour effects .
we calibrated the @xmath3 rectified ob images using a set of @xmath91 bright , isolated , and unsaturated stars for each field .
we first considered only the obs taken during photometric nights .
we integrated the stars count rates in a 6-diameter aperture and applied the nightly zps to obtain individual magnitudes .
we adopted as calibrated magnitudes the @xmath82-clipped averages of all individual measurements .
we then included all obs and computed for each star and ob the appropriate scaling factor based on the star s magnitude and the adopted zps of table [ tab - zps ] .
we calibrated the ob images individually using the @xmath82-clipped average of the scaling factors of all stars per ob . for each reference star in the @xmath3 field
, we computed the @xmath92 dispersion of their calibrated magnitudes in the photometric obs .
the median of the dispersions of all reference stars is 0.030 , 0.032 , and 0.025 mag in @xmath0 , @xmath1 , and @xmath2 , respectively .
we adopt 0.03 mag as the internal accuracy of the photometric calibration of the nir @xmath3 data set .
we produced the final @xmath0- , @xmath1- , and @xmath2-band images of @xmath3 in two steps : we first combined all ob images per field and then combined the four field maps into the large mosaic .
we proceeded in this way because of systematic differences in psf of the ob images between the data sets of each field , which would be problematic for the overlap areas . to create the separate field maps , we weight - averaged the calibrated rectified ob images per band , without clipping or rejection algorithm .
the weight applied at each pixel @xmath93 was proportional to the integration time @xmath94 , scaled by the inverse of the pixel - to - pixel variance of the background noise in each image @xmath95 and of the squared @xmath96 of the ob psf : @xmath97 the @xmath94 term corresponds to the weight maps of the obs multiplied by their respective total integration time , and thus accounts for pixels that were excluded because bad / hot or affected by other artefacts .
the background noise was estimated on source - free areas of the calibrated but pre - rectified ob images to avoid the spatially - correlated noise introduced by rectification .
the @xmath98 term represents the noise within an aperture of the size of the seeing disk ( assuming purely gaussian noise ) so that the weighting scheme optimizes the s / n ratio for point sources .
we created weight maps for each field by averaging those of the individual obs applying the same weighting terms ( i.e. , given by eq .
[ eq - weight ] ) .
these maps , hereafter denoted @xmath99 , give the weighted exposure time per pixel .
they reproduce accurately the spatial variations of the background noise within each field , as indicated by the constant pixel - to - pixel rms over the `` noise - equalized '' maps obtained by multiplying the images with the squared - root of the associated weight maps .
the sky subtraction algorithm in dimsum and our additional optimization routines have introduced small negative biases in the background visible as orthogonal stripes at @xmath100 , and as dark areas around large sources and in crowded regions .
they are due to an overestimate of the background levels because of insufficient source masking .
the object masks were defined from the shallower ob images , such that emission in the extended psf wings around sources was not fully accounted for and very faint sources were missed . since this effect is systematic in all obs , it is enhanced and becomes apparent in the deeper field maps . to reduce these residuals , we rotated a copy of the field maps , made new object masks using sextractor ( as described in [ sect - det ] ) , fitted the background in the unmasked regions with bi - cubic splines along rows and columns , rotated the fits back , and subtracted them from the field maps .
our aim was to produce final mosaics as uniform as possible in effective angular resolution and optimize the s / n ratio in the overlap areas .
we determined the effective psf of the field maps by averaging sub - images centered on bright , isolated , unsaturated stars ( normalized to the stars peak flux ) .
table [ tab - obs ] lists the fwhm of the best - fit moffat profile to the psfs . for each band separately , we convolved the three highest resolution fields to bring all fields to a common resolution .
we computed the kernels with a lucy - richardson deconvolution algorithm .
we quantified the accuracy of the psf matching from the curve - of - growth analysis of the convolved psfs , measured from the same stars in the smoothed field maps . for each band
, the fraction of enclosed flux agrees to better than 1% in circular apertures with diameters in the range @xmath101 , and within 4% for diameters between @xmath102 ( @xmath103 of the psf of the smoothed field maps ) and 6 ( the reference aperture for photometric calibration ) . to optimize the s / n ratio in the overlap areas , we applied a similar weighting scheme as in eq .
[ eq - weight ] , with however some differences .
we omitted the fwhm - dependent term since it is identical among the psf - matched field maps .
we needed to account for the effective background noise levels properly . using the pixel - to - pixel background rms determined over the whole field images resulted in significant global differences between the four fields in the noise - equalized mosaics .
this indicates that the relative rms - weighted total integration times between the fields does not reflect the differences in effective noise .
we attribute this to the systematics discussed in [ sub - red_isaac ] , which affect in different proportions and with varying severity the data of each field .
this is evident in the results of the noise analysis described in
[ sub - depths ] , where the deviations from the behaviour for ( uncorrelated ) pure gaussian noise of the real background rms as a function of aperture size is noticeably different between the fields . for the inverse variance term of the weights
, we used the effective noise derived from this analysis for an aperture diameter of @xmath104 .
this led to the most uniform noise - equalized mosaics in terms of background pixel - to - pixel rms , and again optimizes for point sources .
we created three sets of weight maps for the mosaics , all based on the fields weight maps @xmath99 .
we made one set by averaging the @xmath99 maps , applying the same weights as for the images .
this set accounts for the variations of effective background noise over the entire mosaic relevant for photometry and source detection based on a s / n threshold ( `` effective weight maps , '' with values denoted by @xmath105 ) .
we made a second set weighting only by the cumulative number of frames contributing to the flux ( i.e. , with only the @xmath94 and no variance term ) , as record of the relative total integration time per pixel ( `` exposure time weight maps , '' @xmath106 ) .
for the third set , we simply averaged the @xmath99 maps assigning equal weights so that in the mosaic , the maximum weight in the deepest part of all four fields is identical .
this set is appropriate for detecting sources using a surface brightness threshold as we do in [ sub - det ] ( `` adjusted weight maps , '' @xmath107 ) .
we renormalized all the weight maps for the mosaics to a maximum of unity .
the bias subtraction and flat - fielding of the optical fors1 imaging were carried out using pipeline reduction tasks developed for the fors instruments2 can be found at + http://www.eso.org/observing/dfo/quality/fors/pipeline/pipe_reduc.html ] within the eso - midas system @xcite .
bias subtraction was performed using the nightly master bias frame scaled to the average level of the science frame in the pre- and over - scan regions .
flat - fielding was performed with the nightly master sky flat image obtained from series of exposures taken during twilight .
we further reduced the science frames within iraf as follows .
we subtracted a constant sky level corresponding to the mode of each frame . in one @xmath13-band
ob , there was a residual offset in background level between the lower and upper half of the array , which we eliminated by adding to the top half the difference in median over 20 rows below and above the mid - array row 1024 .
we identified cosmic rays with l.a.cosmic @xcite and negative bad pixels by applying a @xmath108 threshold below zero .
we updated the resulting masks to flag out additional artefacts identified by visual inspection ( e.g. , occasional bad rows and columns , transient positive source - like features and broad faint tracks , edges of the low sensitivity `` holes '' intrinsic to the fors1 array not adequately flat - fielded out ) .
we removed residual large - scale patterns ( at a few percents of the subtracted constant sky level ) using a background map generated with the sextractor program @xcite run on the initial sky - subtracted frames .
we specified a grid mesh size of 64 pixels and median filter size of 10 meshes for the background estimation , which produced the best results .
it is unclear whether these residuals were due to flat field inaccuracies or spatial structure in the background .
either division by or subtraction of the sextractor background map reduced the residuals to @xmath109 .
the two methods led to essentially indistinguishable results , with differences in remaining gradients well below 1% and in photometry of several stars in the field smaller than @xmath110 .
as reported on the eso fors1@xmath762 web pages , flat fielding with the master sky flat can leave large - scale gradients of a few percents , so we adopted the sky - subtracted frames divided by the sextractor background maps .
we determined the photometric zps from observations of standard stars taken during the same nights as the @xmath3 imaging .
these were available for the single ob in the @xmath15 band , and two obs in each the @xmath14 and @xmath13 bands .
the stars were selected from the catalogue of @xcite .
several exposures were obtained through each night , and each exposure includes multiple standard stars .
the stars data were bias - subtracted and flat - fielded in the same way as the @xmath3 data .
we determined single - star zps from individual star counts , and accounted for the non - negligible effects of airmass and colour terms using the atmospheric extinction and colour coefficients for the corresponding period of observations reported on the eso / fors1 web pages .
we computed the nightly zps by averaging with 2@xmath111-clipping the single - star zps .
the @xmath92 dispersions range from 0.013 to 0.049 mag , depending on the night and band .
we also computed single - frame zps as the 2@xmath111-clipped average of the multiple stars zps per frame .
frame - to - frame variations in zps are smaller than the single - star zps dispersion and there is no trend with time , indicating that the nights were photometric . as for the isaac data , we used a circular aperture of diameter 6 for the photometric calibration .
we verified that no measurable systematic variations with airmass or colours were left .
the adopted zps as well as the extinction and colour coefficients used are given in table [ tab - zps ] .
for the @xmath3 data collected on nights without appropriate standard stars observations ( one ob in @xmath14 and three obs in @xmath13 ) , we proceeded as for the isaac data taken during non - photometric nights .
we first determined the magnitudes of five bright , isolated , unsaturated stars from the subset of calibrated @xmath3 frames , averaging with @xmath82 clipping the individual measurements ( the mean of the @xmath92 dispersions are 0.013 mag in @xmath14 and 0.036 mag in @xmath13 ) .
we then computed for each star and frame the appropriate scaling factor based on the stars magnitudes and adopted zps .
we calibrated the frames individually using the @xmath82-clipped average of the scaling factors of all stars per frame .
we estimate the internal accuracy of the photometric calibration of the fors1 optical @xmath3 data set to be 0.015 mag in @xmath15 , 0.02 mag in @xmath14 , and 0.04 mag in @xmath13 .
we aligned all calibrated frames allowing for integer - pixel shifts .
we then co - averaged all registered calibrated frames for each band , applying the weighting scheme of eq .
[ eq - weight ] with the various terms determined from the individual frames .
this was done before rectification to the wfpc2 maps . because of the smaller number of frames and the poorer seeing conditions for the fors1 observations ( compared to the isaac observations ) , the smearing due to the combination of dithering , uncorrected field distortion , and sub - pixel misalignment are not critical . the co - averaging , rectification , and creation of the weight maps were done in the exact same way as for the isaac field maps .
the final reduced images have a pixel size of @xmath112 and are aligned relative to the wfpc2 @xmath49 mosaic , with north oriented nearly up at @xmath113 anticlockwise from the vertical axis ( see [ sub - astrometry ] for the exact astrometry ) .
the images are all normalized to instrumental counts per second .
the fully reduced images as well as the raw data are available electronically on the fires web site .
the isaac nir mosaics and fors1 optical maps are shallower towards the outer edges , reflecting lower total exposure times because of the dithering process in the observations . in the nir mosaics ,
the largest weights are reached in the regions where the adjacent fields overlap ( 12% of the total area mapped with isaac ) .
in addition , the combination of different integration times and noise levels results in significant global offsets in weight between the four fields .
the largest effective weights in the central parts of the fields are @xmath114 for all nir bands ( in f1 for @xmath0 and @xmath2 , and f4 for @xmath1 ) compared to the normalized maximum of 1 reached in the overlap regions .
field f2 has the lowest maximum effective weight , reaching @xmath115 , 0.5 , and 0.39 in @xmath0 , @xmath1 , and @xmath2 , respectively
. this can bias area selection when applying weight criteria . to take advantage of the field size
, it may be more appropriate to use the adjusted weight maps where the maximum weights are equal in all fields ( see [ sub - red_isaac ] ) or to treat the fields individually using their respective weight maps @xmath99 .
the area in the @xmath2-band mosaic with @xmath116 , 0.25 , and 0.01 ( corresponding approximately to @xmath117 , 0.30 , and 0.01 in at least one of the four isaac fields ) is 16.3 , 26.2 , and @xmath118 . as quality cut for photometry
, we adopt the @xmath119 area defined by @xmath120 ( approximately @xmath121 in at least one isaac field ) in each nir band , and @xmath122 in each optical band . the noise - equalized @xmath2-band mosaic ( i.e. the @xmath2-band map multiplied by the squared root of its @xmath105 weight map ) is shown in figure [ fig - fovs ] .
it illustrates the richness in faint sources and the spatial uniformity in normalized background rms achieved with the effective weight map .
figure [ fig - ijk ] shows the rgb colour composite image of the central @xmath123 of the field , constructed from the @xmath49 , @xmath0 , and @xmath2 maps .
the three maps have been convolved to an angular resolution of @xmath124 to match the broadest psf of all images ( in the @xmath13 band ; see [ sub - imquality ] below ) .
the colour image is displayed with a linear stretch adjusted to enhance the fainter objects .
the variety of colours is clearly visible over a wide range of brightnesses , even at the faintest levels .
this indicates that the maps reach comparable depths , and reflects the combination of different rest - frame colours and different redshifts of the sources .
we characterized the psf in the final combined maps using bright , isolated , and unsaturated stars ( the sets were partly different in the various bands and included eight to thirteen stars ) .
the psf in the isaac and fors1 images is stable and symmetric over the entire field of view , with a gaussian core profile and wings consistent with a lorentzian profile .
we adopted the average of the peak - normalized star profiles as the final psf in each band . from moffat profile
fits , the psf fwhm in the @xmath0- , @xmath1- , and @xmath2-band mosaics is @xmath125 , @xmath11 , and @xmath12 , with ellipticities of 0.02 , 0.04 , and 0.03 , respectively . for the optical bands , the psf fwhms and ellipticities are @xmath126 and 0.06 in @xmath13 , @xmath127 and 0.04 in @xmath14 , @xmath61 and 0.02 in @xmath15 , @xmath128 and 0.02 in @xmath48 , and @xmath129 and 0.02 in @xmath49 .
the fwhm of the individual stars are within @xmath130 of the final psf fwhm in each of the isaac and fors1 images , and within @xmath131 for the wfpc2 images .
the psf fwhms are reported in table [ tab - obs ] . for consistent photometry across all bands , we convolved all images to a common angular resolution corresponding to that of the @xmath13-band map , which has the poorest resolution ( @xmath124 ) .
the convolution kernels between the @xmath13-band psf and that of the other bands were computed using a lucy - richardson deconvolution algorithm . from the curve - of - growth analysis of the convolved psfs , the fraction of enclosed flux agrees to @xmath132 in circular apertures with diameters between 1 and 2 , relevant for the colour measurements ( see [ sub - phot ] ) .
the accuracy of the psf - matching is @xmath133 for diameters in the range @xmath134 .
accurate registration among the various maps is important for correct cross - identification of sources , precise colour measurements , and proper comparisons of the morphology in different wavebands .
we verified the relative alignment using 20 bright stars and compact sources throughout the field .
the median rms variation in position of individual objects across the bands is @xmath135 ( or 0.7 pixel in the final images ) , with a range from @xmath136 to @xmath137 .
relative to the positions in the wfpc2 @xmath49 or @xmath48 maps , there is no systematic trend in the position offsets in any of the bands .
the amplitude of the offsets in the @xmath0 , @xmath1 , and @xmath2 mosaics has a @xmath92 dispersion of @xmath138 , corresponding to a fraction 0.47 of a pixel at the original pre - rectified scale of the isaac data ( @xmath55 ) . for the @xmath13 , @xmath14 , and @xmath15 maps ,
the offsets dispersion is @xmath139 , or 0.6 pixel at the original fors1 scale ( @xmath137 ) .
we verified the absolute astrometry using stars from the usno catalogue ( v. a2.0 , * ? ? ? * ) and ucac2 catalogue @xcite .
we solved for the positions determined from the wfpc2 @xmath49 mosaic , our reference image for the relative registration .
the astrometric solution is @xmath140 where @xmath141 and @xmath142 are the pixel coordinates in the final images , for which the precise spatial scale is @xmath143 .
the orientation is such that north and east coincide nearly with the vertical and horizontal axes , respectively , with a small rotation angle of @xmath144 counterclockwise ( following the usual astronomical convention , east is @xmath145 counterclockwise from north ) .
the offsets @xmath146 and @xmath147 are expressed in arcseconds and relative to the reference coordinates @xmath148 : @xmath149 , @xmath150 : @xmath151 .
we estimate that the absolute astrometry is accurate to @xmath152 . in the raw data ,
the background noise properties are expected to be well described by the dispersion of the signal measured in each pixel since both the poisson and read - out noise should be uncorrelated . for such uncorrelated gaussian noise ,
the effective background rms for an aperture of area @xmath153 is simply the pixel - to - pixel rms @xmath154 scaled by the linear size @xmath155 of the aperture , @xmath156 .
instrumental features , the data reduction , and the rectification , combination , and psf - matching procedures have added significant systematics and correlated noise in our final data . the determination of the limiting depths and photometric uncertainties depends critically on the accurate description of the noise properties of the images . for this purpose , we followed the same empirical approach as for the hdf - s data @xcite .
we derived the function @xmath157 directly from the data by measuring the fluxes of 2800 non - overlapping circular apertures .
we placed the apertures at random in the maps but so as to avoid all pixels associated with sources detected in the @xmath2 band and within their @xmath158 isophote ( the surface brightness threshold applied for source detection ; see [ sect - det ] ) .
we used the same positions for all maps , and repeated the measurements with different aperture diameters ranging from @xmath102 to 3 .
because the noise properties vary importantly between the four fields of the nir mosaics , we analyzed them separately ( each field has about 700 aperture positions ) .
we applied the procedure on the psf - matched images used for the photometry , allowing us to compute realistic photometric uncertainties .
we also ran the procedure on the set of non - concolved wfpc2 and isaac images to estimate more representative intrinsic depths , which we can also better compare with those of the hdf - s data ( see below ) . for each aperture size , the distribution of fluxes is symmetric around the background value ( about 0 ) , and is well reproduced with a gaussian profile .
we thus determined the background rms variations from the fwhm of the best - fit gaussian to the observed distribution for each aperture size . while at any fixed spatial scale
, the noise properties over the images are consistent with a pure gaussian , the variations with @xmath159 deviate appreciably from gaussian scaling , being systematically larger and increasing non - linearly . as for the hdf - s data @xcite , the real noise behaviour in the @xmath3 data set can be well approximated by the polynomial model : @xmath160 where @xmath161 designates the band ( and the field for the isaac data ) and the weight term accounts for the spatial variations in noise level related to the exposure time .
the coefficient @xmath162 represents the effects of correlated noise , which are dominated by the resampling of the images to the wfpc2 pixel scale and , for the psf - matched data , the convolution applied .
the curvature in the @xmath163 relations quantified by the coefficient @xmath164 indicates a noise contribution that becomes increasingly important on larger scales .
it presumably originates from objects below the detection level , instrumental features , and residuals from the sky subtraction and flat - fielding .
the noise behaviour is qualitatively the same in all of the images .
figure [ fig - noise ] shows the results for the four fields of the psf - matched @xmath2-band mosaic , with the background rms measurements for the various apertures and the best - fit polynomial models ( eq . [ eq - noise ] ) compared to the expected linear relationship for uncorrelated gaussian noise .
the flux distributions from which the background rms is derived are also plotted for selected aperture sizes . with the empty areas defined from the @xmath2-band map , sources that are undetected in @xmath2 but detectable in the other bands
will contribute to the aperture flux distributions .
this is reflected in the distributions for bands other than @xmath2 , more obviously for the optical bands , with a small excess of positive flux measurements . however , the bias introduced in the results is very small because the positive tail affects very little the gaussian fits .
table [ tab - noise ] summarizes the results for the psf - matched images , along with the corresponding @xmath92 background - noise limiting magnitudes .
table [ tab - depth ] gives the limiting magnitudes for the rectified but unconvolved images .
the limiting magnitudes refer to a `` point - source aperture '' with a diameter @xmath165 that maximizes the s / n of photometric measurements in unweighted circular apertures of point - like sources .
compared to the hdf - s data set , the @xmath3 field limiting magnitudes are @xmath166 brighter , depending on the band .
this comparison is rather crude because it does not account for the partly different bandpasses in the optical ( @xmath13 and @xmath14 in particular ) or for the different angular resolution between the data sets .
for the isaac nir data , the @xmath0- , @xmath1- , and @xmath2-band psf - matched maps are @xmath167 shallower than the hdf - s images .
the latter were matched to the resolution of the isaac @xmath1-band map of @xmath10 and the point - source aperture has @xmath168 @xcite .
the intrinsic resolution of the @xmath3 nir rectified but unconvolved mosaics is similar , enabling a better comparison .
the limiting magnitudes derived from this set are @xmath169 brighter than for the hdf - s .
the differences expected based only on the integration time are of @xmath170 for similar observing conditions ( seeing , atmospheric transmission , and night - sky oh and continuum emission ) .
the depths achieved for the @xmath3 field exceed slightly these expectations .
various factors can explain this performance .
the most important one is probably that about half of the @xmath3 data were taken after the 2001 march realuminization ( period p3 ) , when the effective sensitivity had increased by @xmath171 compared to earlier epochs since the end of 1999 ( p1 and p2 ) during which the hdf - s observations were carried out ( [ sub - obs_isaac ] ; see also * ? ? ?
we also analyzed the distributions of flux ratios between bands .
the results obtained for ratios measured in random pairs of apertures ( i.e. , not necessarily spatially coincident ) correspond closely to the predictions based on eq .
[ eq - noise ] assuming the noise is uncorrelated between bands .
in contrast , the scaling relations derived for ratios within registered apertures for several band pairs lie systematically below such predictions on scales of @xmath172 ( although still well above expectations for pure uncorrelated gaussian noise ) .
this suggests that the background noise pattern is correlated between the bands , as was also seen in the hdf - s data set @xcite .
the effects are most important and up to @xmath74 among the maps obtained with the same instrument .
this is likely due to the large - scale residuals remaining after the data processing ( e.g. , from the bias , sky , and flat - field ) , which are similar for a given instrument .
interestingly , this effect is also present at the 10% level for ratios involving the fors1 @xmath14 or @xmath15 bands and the wfpc2 @xmath48 or @xmath49 bands , as well as between the wfpc2 @xmath49 and isaac @xmath0 bands .
this is probably the signature of sources undetected in @xmath2 but clearly present in the other bands contaminating the `` empty '' background flux measurements .
confusion noise was argued in the case of the hdf - s data but is unlikely for the shallower @xmath3 maps .
the implications of this analysis is that our uncertainties on colours are probably slightly overestimated because we propagated those of each bandpass neglecting the cross - correlation terms .
we did not attempt to account for these effects as they are much smaller than the difference between uncorrelated gaussian scaling and our empirical relationships .
our aim was to construct a @xmath2-band selected source catalogue as uniform as possible over the entire @xmath3 field , optimized for point - like sources , and with reliable detections and optical - nir photometry . because of the different noise properties between the four isaac fields , detection with a constant s / n threshold would result in different limiting magnitudes and limiting surface brightnesses from one field to the other .
we therefore applied a constant surface brightness threshold @xmath173 instead .
we compromised between completeness and reliability by choosing a @xmath173 criterion so as to keep the contamination by spurious ( faint ) sources below @xmath174 in all fields .
the general procedure for source detection and photometry is analoguous to that followed by @xcite in making the hdf - s catalogue .
we carried out the source detection using the sextractor software version 2.2.2 @xcite .
the objects were detected in the rectified but non psf - matched @xmath2-band map .
to optimize for point - like sources , we filtered the detection map with a gaussian kernel of @xmath175 approximating well the core of the @xmath2-band psf .
the detection criterion was that at least one pixel be brighter than @xmath176 .
this corresponds to @xmath17 of the background noise in the filtered @xmath2-band mosaic over the two fields with intermediate depth ( f3 and f4 ) .
the threshold adopted is rather conservative but ensures that the spurious source contamination is small .
the resulting catalogue contains 1858 objects .
we estimated the fraction of false detections by running sextractor with the exact same parameters on the inverse detection map ; it is @xmath177 over the entire mosaic and @xmath178 in the shallowest field , f2 .
sextractor identifies and separates blended sources based on the distribution of the @xmath2-band light in the detection map ; 18% of the sources were flagged as blended .
we specified deblending parameters so as to avoid oversplitting of objects with relatively complex morphologies .
the spatial filtering affects the detection process and the isophotal parameters .
our filtering with a kernel reproducing the @xmath2-band psf introduces a small bias against faint extended sources . as for the hdf - s catalogue
( see * ? ? ?
* ) , we preferred not to combine multiple catalogues obtained with different kernel sizes because their subsequent merging is complicated and subjective , and does not lead to a substantial gain in sensitivity for larger objects . the psf - like kernel we adopted is fully adequate for our purposes since most of the faint sources detected are compact or unresolved in the nir and it reduces the problems due to blending when convolving with larger kernels .
we used sextractor in dual - image mode , which allows for spatially accurate and consistent photometry across all bands . while the sources were extracted and the isophotal parameters determined from the filtered @xmath2-band map , the photometry was performed on the psf - matched maps and within the same set of apertures in all bands for each object .
of the 1858 @xmath2-band detected sources , 1663 have full eight - band photometry .
we used circular apertures with 30 different diameters ranging from @xmath168 to 3 .
we also used isophotal apertures defined by the detection threshold of @xmath176 and autoscaling apertures , which , following @xcite , are elliptical apertures scaled based on the first moments of the filtered @xmath2-band light profile ( the `` mag_iso '' and `` mag_auto '' computed by sextractor ) . adopting the notation of @xcite
, we will refer to these apertures as aper(@xmath179 ) , aper(iso ) , and aper(auto ) , respectively .
we also defined customized `` colour '' and `` total '' apertures aper(colour ) and aper(total ) .
these were introduced for the hdf - s catalogue to allow detailed control of the photometry relying on well - defined criteria and simple but robust treatment of blended sources .
we considered a source as blended if the sextractor flags `` blended '' or `` bias '' were set ( see * ? ? ?
the colour aperture is optimized for accurate and consistent colours , and relies on the @xmath2-band isophotal aperture of each object .
the criterion to select the appropriate colour aperture uses the equivalent circularized isophotal diameter @xmath180 . for isolated sources , @xmath181 for blended sources , @xmath182 where @xmath183 is the factor by which we reduce the size of the circular apertures centered on blended objects to minimize flux contamination from the close neighbour(s ) .
the optimal value of @xmath183 needs be determined from experimentation , and we found that the same @xmath184 as for hdf - s was appropriate for the @xmath3 field data .
the smallest colour aperture allowed , aper(10 ) , has a diameter @xmath104 of the psf - matched images .
this maximizes the s / n of flux measurements for point sources in unweighted circular apertures and avoids smaller apertures for which the photometry becomes less reliable .
the largest colour aperture considered , aper(20 ) , was selected to avoid the uncertainties of very large isophotal areas and to reduce possible contamination of the fluxes by adjacent sources undetected in the @xmath2 band but present in other bands .
the total apertures were used to derive the integrated fluxes of the objects in the @xmath2-band .
isolated and blended sources were again treated differently : @xmath185 equivalent circularized apertures were also defined for the autoscaling kron - like apertures with @xmath186 and for the total apertures with @xmath187 . to compute the total fluxes and magnitudes
, we applied an aperture correction to the measurements in aper(total ) .
the correction factor is the ratio of the fluxes enclosed within circular apertures of diameters 6 ( used for the photometric calibration ) and @xmath188 , inferred from the growth curve of the @xmath2-band psf constructed from bright stars ( [ sub - imquality ] ) .
this correction is significant for most objects , reaching 0.58 mag for faint point - like sources with flux measured in a @xmath189 aperture .
figure [ fig - magtot_comp ] compares various methods of estimating integrated @xmath2-band magnitudes . for point - like sources
, the total magnitude is expected to correspond to the flux in the point - source aper(10 ) scaled by the aperture correction .
the isophotal and autoscaling apertures tend to decrease in size at fainter magnitudes , leading to more severe underestimate of the total magnitudes due to extended emission intrinsic to the sources or in the psf wings outside of the apertures . for isophotal magnitudes , the turnover sets in at @xmath190 , with the difference @xmath191 $ ] growing rapidly and reflecting mainly the shrinking aper(iso ) defined by the constant surface brightness threshold .
magnitudes within aper(auto ) come closer but still miss significant flux in the psf wings at @xmath192 . with the method based on eq .
[ eq - total ] , we best recover the integrated magnitudes expected for faint point - like sources . our treatment of isolated and blended sources also reduces the dispersion of @xmath193 around @xmath194 .
we derived the @xmath92 photometric uncertainties from eq .
[ eq - noise ] , using the @xmath195 and @xmath196 coefficients given in table [ tab - noise ] , the linear size @xmath197 of the aperture considered , and the average weight within the aperture .
these uncertainties properly account for the actual background noise in the maps influencing directly the flux within the various apertures .
they may overestimate the uncertainties of colours by a small amount ( [ sub - depths ] ) . from the variations in photometric zps determined from standard stars or inferred from stars within the @xmath3 field of view ( [ sub - red_isaac ] and [ sub - red_fors ] ) , we estimate that uncertainties of the absolute flux calibration are @xmath198 .
the photometry may further be affected by possible biases from the sky and background subtraction procedure , or surface brightness biases ; these additional sources of errors are not included in our uncertainties .
for all objects in the catalogue , we derived photometric redshifts ( @xmath18 ) using a technique described in detail by @xcite and @xcite .
briefly , the algorithm involves linear combinations of redshifted spectral templates of galaxies of various types ( from normal ellipticals and spirals to magellanic irregulars and young starbursts ) .
the mean absorption by neutral hydrogen from the intergalactic medium along the line of sight is accounted for ( based on * ? ? ?
the @xmath18 uncertainties are estimated from monte - carlo simulations , accounting for uncertainties in the fluxes as well as template mismatch .
a key feature of the technique is that the derived uncertainties reflect not only the @xmath92 confidence interval around the best solution but also the presence of secondary solutions with comparable likelihood .
spectroscopic redshifts ( @xmath21 ) have been measured for over 400 galaxies in the @xmath3 field ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* s. wuyts et al . , in preparation ) , of which about 330 are included in our @xmath2-band selected catalogue .
the majority ( @xmath199 ) lie at @xmath200 , mostly members of the @xmath3 cluster at @xmath4 . of the remaining , 32 galaxies have @xmath201 and 11 have @xmath202 .
the much larger number of @xmath21 determinations in the @xmath3 field compared to hdf - s allows a more robust assessment of the accuracy of our photometric redshift technique .
figure [ fig - zph_zsp ] compares the @xmath18 and @xmath21 for galaxies with spectroscopic determinations , and shows the @xmath18 and @xmath21 distributions of sources at @xmath203 , for which @xmath204 .
overall , the accuracy of the photometric redshifts is @xmath205 . in the redshift ranges @xmath206 , @xmath207 , and @xmath40 , @xmath208 , 0.127 , and 0.040 , respectively . at @xmath207 , the @xmath18 s
are less well constrained , as reflected by the larger @xmath92 confidence intervals .
this is mainly because the fitting procedure there relies heavily on the presence of the balmer/4000 break in the seds ( characteristic of @xmath209 old stellar populations ) , and the break falls in the large gap between the @xmath49 and @xmath0 bandpasses . at @xmath40 ,
the agreement between @xmath18 and @xmath21 is best , although the number of galaxies with available @xmath21 in this redshift interval is still small .
our @xmath18 s slightly underestimate the true redshifts by @xmath210 on average . in the rest of the paper
, we will use the spectroscopic redshift whenever available , otherwise the photometric redshift .
we identified stars using a combination of three criteria .
the main criterion was that the raw @xmath211 for fits using stellar seds was lower than that of the best - fit combination of galaxy seds from the @xmath18 determination .
the stellar templates are synthetic seds of main - sequence stars with effective temperatures from 3000 to 10000 k computed with the nextgen model atmospheres of @xcite
. we did not consider cooler dwarfs or evolved stars ( giants , supergiants ) because non local thermodynamical equilibrium ( non - lte ) effects and extensive molecular absorption become important and model atmospheres still suffer from large uncertainties in these regimes .
moreover , adequate empirical spectral atlases with complete and consistent optical - nir coverage are scarce for these stellar classes .
as argued for the hdf - s , very late m - type , methane , t- , and l- dwarfs as well as normal post - main - sequence or variable o- to c - rich stars on the asymptotic giant branch are not expected to be as red in the nir as most redshifted galaxy types and/or are very unlikely to be present in the small survey areas at high galactic latitude covered by the hdf - s and @xmath3fields . as noted by @xcite ,
stars are well separated from galaxies in the @xmath212 versus @xmath213 diagram ( see also [ sect - analysis ] ) ; our second criterion was that the candidates lie on the locus of stars in this colour space .
we then verified the size of each candidate in the wfpc2 @xmath214 mosaic ( non psf - matched ) and excluded those that are spatially resolved .
the final list includes 84 stars down to @xmath215 .
the source catalogue is available electronically on the fires web site .
the catalogue format and entries are similar to those of the hdf - s catalogue @xcite , and described here .
_ i d _ running identification number in the catalogue order as reported by sextractor .
_ x , y _ pixel positions of the objects in the images , based on the @xmath2-band detection map .
the reference is the @xmath216 frame of the rectified , registered maps at @xmath79 .
_ ra , dec _ right ascension and declination coordinates for equinox @xmath217 ( see eq . [ eq - astrom ] ) . _
@xmath218 , @xmath219 _ flux measured in the colour aperture ( [ sub - phot ] ) and the associated uncertainty derived from the noise analysis ( [ sub - depths ] ) in each of the bandpasses @xmath220 . the units are @xmath221 . _
@xmath222 , @xmath223 _ same as above for measurements in the isophotal aperture defined by the surface brightness threshold of @xmath176 in the @xmath2-band detection map ( the mag_iso computed by sextractor ) . _
@xmath224 , @xmath225 _ @xmath2-band flux in the total aperture scaled by the aperture correction ( [ sub - phot ] ) and the associated uncertainty from the noise analysis ( [ sub - depths ] ) .
the units are @xmath221 .
the aperture correction is based on the @xmath2-band psf profile outside of the total aperture . _
@xmath226 , @xmath227 _ same as above for measurements in the kron - like autoscaling aperture based on the light distribution in the @xmath2-band detection map ( the mag_auto computed by sextractor ) . _
@xmath228 , @xmath229 _ flux in the bandpass @xmath161 and the associated uncertainty derived from the noise analysis ( [ sub - depths ] ) measured in each of the 30 circular apertures with diameter @xmath230 pixel of the rectified maps ( @xmath231 ) .
the units are @xmath221 .
_ apcol _ type of aperture used for measuring the colour flux ( [ sub - phot ] ) , encoded as follows . 1 : minimum circular aperture considered with diameter @xmath189 used whenever the isophotal aperture is smaller .
2 : maximum circular aperture considered with diameter
@xmath232 used whenever the isophotal aperture is larger .
3 : isophotal aperture defined by the surface brightness threshold of @xmath176 in the @xmath2-band detection map .
4 : circular aperture with diameter @xmath233 where @xmath184 is the aperture reduction factor applied for blended sources . _
aptot _ type of aperture used for measuring the total @xmath2-band flux ( [ sub - phot ] ) , encoded as follows . 1 : kron - like autoscaling aperture based on the light distribution in the @xmath2-band detection map
2 : colour aperture used in the case of blended sources . _
apcorr _ scaling factor corresponding to the aperture correction applied to the @xmath2-band flux measured in the total aperture in order to estimate the total integrated flux . _
@xmath234 , @xmath235 _ circularized radii @xmath236 from the area of the apertures used to measure the colour and total fluxes .
units are pixels in the rectified images ( @xmath231 ) . _
@xmath237 , @xmath238 _ area of the aperture enclosed within the @xmath176 isophote , and of the elliptical kron - like autoscaling aperture ( @xmath239 , where @xmath162 and @xmath164 are the semi - major and semi - minor axis lengths ) .
units are squared pixels in the rectified images ( @xmath231 ) . _
@xmath240 , @xmath241 _ fwhm of sources from the @xmath2-band detection map assuming a gaussian profile , and from the non psf - matched wfpc2 @xmath49 mosaic .
the latter was obtained by running sextractor on the @xmath49 map , and cross - correlating the @xmath49- and @xmath2-band selected catalogues .
units are pixels in the rectified images ( @xmath231 ) . _
_ average of the effective weight in the bandpass @xmath161 within the point - source aperture for the psf - matched set of maps ( diameter of @xmath243 ) . for the optical fors1 and wfpc2 data ,
the effective weight is simply proportional to the total exposure time at each pixel while for the nir isaac mosaics , it corresponds to the inverse variance of the effective background noise resulting from the integration time and the systematics from the raw data and the reduction procedure ( [ sub - red_isaac ] and [ sub - red_fors ] ) .
the weights are normalized to a maximum of unity over the maps . _
@xmath244 _ same as above for the weights in each of the fields @xmath245 of the nir isaac data separately .
these weights are simply proportional to the total exposure time per pixel , and normalized to a maximum of unity within each field map . _ bias , blended , star
_ flags indicating the following if set to 1 ( and otherwise set to 0 ) .
_ bias _ is set if the kron - like autoscaling aperture aper(auto ) contains more than 10% of bad pixels or if the flux measurement within this aperture is affected by neighbouring sources as determined from the light distribution in the @xmath2-band detection map .
_ blended _ is set if the source overlaps with adjacent objects based on the @xmath2-band detection map .
_ star _ is set if the star identification criteria are satisfied ( sed better fit by a stellar template than a combination of galaxy templates , colours lying on the stellar locus in the @xmath212 versus @xmath213 diagram , spatially unresolved light profile in the non psf - matched wfpc2 mosaics ; [ sect - redshifts ] ) . _
@xmath18 , @xmath246 _ photometric redshift and associated 68% confidence intervals ( @xmath247 ; [ sect - redshifts ] ) .
we derived the completeness limits in the @xmath2 band from simulations in which we added point sources to the mosaic and analyzed the recovery fraction as a function of input magnitude .
we performed the simulations on the rectified ( @xmath79 ) and non psf - matched ( @xmath175 ) image used for the source detection .
we treated the four isaac fields separately because of their different effective depths .
we considered only the deepest part of each field with nearly uniform image quality , where the effective weights exceed 95% of the maximum within the given field ( i.e. , @xmath248 ) .
we used the psf profile ( created by averaging the images of bright , isolated , unsaturated stars , see [ sub - imquality ] ) to generate artificial point sources with total @xmath249-band magnitudes between @xmath250 and 28 .
for realistic simulations , we distributed the sources according to the power - law @xmath251 .
we adopted @xmath252 as determined from the raw number counts at @xmath253 , where @xmath254 and where incompleteness does not yet play a significant role . for each field
, we added 30000 simulated point sources to the image , drawn at random from the power - law distribution and placed at random locations . to avoid overlap among the simulated sources , we ran the simulations by adding only 25 of them at a time to the image . for each realization
, we extracted the sources as described in
[ sub - det ] except for the application of a fainter surface brightness detection threshold of @xmath255 ( @xmath256 of the background noise in the two intermediate - depth fields ) to ensure reliable derivation of the 50% completeness limits .
we considered a source as recovered if its measured position lay within 3.5 pixels of its input position , although experimentation showed that the results are little sensitive to the exact choice of this criterion .
we accounted for blending with real sources following eq .
[ eq - total ] when determining the total @xmath2-band magnitudes of the simulated sources .
we did not apply any particular treatment for simulated sources that may have been lost on or falsely recovered as real sources ( `` unmasked case '' ) . to quantify this effect
, we repeated the analysis excluding all regions within the isophotal areas of real sources ( `` masked case '' ) .
figure [ fig - complete ] shows the completeness curves for each field and the output versus input total @xmath2-band magnitudes for the unmasked case .
table [ tab - complete ] gives the 90% and 50% completeness limits for both the unmasked and masked cases . for the unmasked case , the 90% completeness levels vary by up to 0.8 mag between the fields , with a mean of @xmath257 .
the 50% completeness levels are essentially the same for the four fields at @xmath258 ( and close to the @xmath6 detection limits ) .
the shape and slope of the completeness curve for the shallowest field f2 at faint magnitudes is distinctly different from that of the other three fields , which reach fainter and more similar depths .
this is caused by a more important contribution from spurious detections in f2 because of higher amplitude noise fluctuations .
not only the depth of the images but also the crowding of real sources influences the derived limits , through loss and confusion of simulated sources overlapping with real objects .
this effect is significant in the @xmath3 field compared to the hdf - s because of the presence of the cluster .
we found this alters mainly the absolute level and slope of the completeness curves above 90% , and therefore the inferred 90% limits .
for comparison , in the masked case the bright end of the curves is flatter , and the 90% completeness levels increase to @xmath259 and reflect more closely the relative depth between the fields as derived from the background noise analysis in [ sub - depths ] . in contrast , the steepest part of the curves is more robust against loss and confusion of simulated sources with real sources , with derived 50% limits increasing marginally to @xmath260 .
our assumption of point - like simulated sources optimizes the detection efficiency , which is further enhanced by our filtering of the detection map with a psf - like kernel ( see [ sub - det ] ) .
consequently , the completeness levels derived above constitute upper limits . in reality
, extended sources would have brighter completeness levels that would also depend on the actual sizes and light profiles .
an analysis as a function of source size and morphology is presented by @xcite . in correcting the raw counts below
, we followed the simple conservative approach based on the point - source simulations presented above , and used the results for the unmasked case .
we computed the raw @xmath2-band source counts over the same area of the mosaic as considered for the simulations presented above ( where @xmath248 for each individual field ) .
we excluded stars , used 0.5 mag wide bins , and extended the counts to the 50% completeness level ( @xmath261 ) . in correcting the raw counts for incompleteness and spurious detections
, we accounted for the non - uniform depth and noise properties over the mosaic as follows .
we first treated each field separately , computing the raw counts , applying the corresponding corrections , and weighting by the effective area where the depth equals the faint edge of each magnitude bin .
we then combined the corrected counts from the four fields and normalized by the total effective area over the mosaic as a function of depth to obtain the surface densities .
we derived the effective area to a given depth based on the mosaic s effective weight map ( @xmath105 ;
[ sub - red_isaac ] ) , which accounts for the variations in absolute noise levels over the entire field of view .
this procedure yields the most reliable results for our @xmath3 data , especially because of the significant offset in central depth of f2 compared to the other fields .
accounting for the effective area as a function of depth has little impact within each field because we only considered the deepest regions within each of them .
however , it becomes important when combining the fields counts .
in particular , at the faintest magnitude bin considered ( centered at @xmath262 ) , the raw counts as well as the point - source simulations in f2 suffer from a larger contamination by spurious detections so that the derived corrections are less reliable .
our treating the fields separately and accounting for the effective area per magnitude interval avoids a strong bias from f2 at the faint end .
figure [ fig - numcounts1 ] shows the differential @xmath2-band galaxy counts for the @xmath3 field .
the raw counts are plotted along with the counts corrected for incompleteness only and for both incompleteness and spurious detections .
the incompleteness correction is simply the inverse of the detection efficiency of simulated sources given by the completeness curves .
the correction for incompleteness and spurious detections is taken as the ratio of the number of simulated sources per input magnitude bin to the number of detections per output magnitude bin .
since spurious sources contribute to the number of detections ( as for the raw counts ) , the full correction is effectively smaller than that for incompleteness only .
as the figure shows , little correction is needed out to the faintest magnitude , where the incompleteness and full corrections amount to 30% and 10% , respectively .
the presence of the @xmath4 cluster has minimal effect on the counts at the faint end .
the average magnification factor from gravitational lensing over the entire mosaic is small and ranges from 1.05 to 1.25 for background source redshifts between @xmath51 and 4 ( h. hoekstra , priv .
comm . ; based on @xcite ) . moreover , the flux and area magnification cancel out to first order by shifting the counts in the surface densities@xmath263magnitude plane along a path that is roughly parallel to the observed distribution ( as indicated in figure [ fig - numcounts1 ] for @xmath264 with average magnification factor of 1.20 ) . at brighter magnitudes ,
the cluster galaxies produce the bump around @xmath265 ( the approximate cluster contribution is illustrated in figure [ fig - numcounts1 ] , taken as the counts of all sources with @xmath266 , which are dominated by the cluster galaxies ; see figure [ fig - zph_zsp ] ) .
accurate corrections for the lensing effects and direct contribution to the counts of the cluster are beyond the scope of this paper ( and would depend , e.g. , on the luminosity function of the background sources , which is unknown , and on a reliable and complete census of the cluster members ) .
figure [ fig - numcounts2 ] compares the fires counts with those of selected deep surveys from the literature . in the range @xmath267 ( @xmath268 ) ,
the @xmath3 field counts follow a @xmath269 relation with logarithmic slope @xmath252 ( it varies from 0.19 to 0.21 between the raw and incompleteness - corrected cases ) .
this is comparable to the faint - end slope derived from our @xmath39 deeper @xmath2-band map of hdf - s @xcite .
overall , the counts in the two fires fields agree quite well out to the 50% completeness limit of the shallower @xmath3 field , with the exception of the bump due to the @xmath3 cluster itself . at the faint end
, the good agreement suggests that faint cluster galaxies probably do not significantly affect the count slope in the @xmath3 field .
the @xmath270 times wider area of the @xmath3 field also leads to more robust statistics compared to the deeper but smaller hdf - s . compared to other @xmath30-band surveys ,
the @xmath3 field is not the deepest nor the widest , but is unique in its combination of depth and area surveyed .
there are appreciably large differences among the various counts shown in figure [ fig - numcounts2 ] .
in particular , the faint - end slope of the fires fields generally lies at the flatter end of the range from the other surveys ( @xmath271 at @xmath272 , or @xmath273 ; @xcite ) .
it seems possible that the @xcite counts may have been overestimated .
the significance of the differences between studies is however difficult to ascertain in the faint regime , where model - dependent corrections become important .
a complex combination of effects is probably at play , including intrinsic field - to - field variations ( which can be large since all the fields are relatively small , covering a few arcminutes squared to a few tens of arcminutes squared ) as well as differences in the photometric measurements ( e.g. , estimates of total magnitudes ) , the filter bandpasses ( @xmath30 , @xmath2 , @xmath274 ) , the derivations and application of the corrections to the counts ( spurious sources contaminations , non - uniform depth , size and morphology dependencies ) .
this makes any comparison between surveys far from straightforward .
figures [ fig - colmag ] and [ fig - colcol ] show the distribution of @xmath2-band selected sources from the @xmath3 field in various @xmath275 colour - magnitude and colour - colour diagrams .
the sources are selected to have a @xmath276 on their @xmath2-band colour flux , and to lie within the well - exposed areas with @xmath121 in each nir band and @xmath277 in each optical map ( implying @xmath278 or @xmath279 for the resulting sample ) .
all colours refer to measurements in the colour apertures , as described in
[ sub - phot ] .
the colour distributions are unaffected by the ( achromatic ) gravitational lensing of background sources from the @xmath4 cluster .
the cluster sequence is seen in all plots , in the tight relationships at brighter magnitudes .
the stars identified in [ sect - redshifts ] follow a well - defined blue locus in @xmath212 and @xmath280 versus @xmath193 , and in @xmath212 versus @xmath213 and @xmath280 colours , well separated from the distributions for galaxies .
as for the hdf - s @xcite , we find large numbers of sources with very red @xmath212 colours in the @xmath3 field .
in particular , the criterion @xmath281 effectively selects evolved or dust - obscured galaxies at @xmath40 , the `` distant red galaxies '' ( drgs , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
their average rest - frame optical colours fall within the range covered by normal galaxies locally , unlike the lyman - break galaxies ( lbgs ) selected in the optical as @xmath13-dropouts , which are typically much bluer ( @xcite ; see also @xcite ) .
this indicates that drgs have on average higher mass - to - light ratios than lbgs at similar redshifts .
the surface density of drgs is @xmath282 and @xmath283 in the @xmath3 and hdf - s fields , respectively , at @xmath284 , and their contribution to the integrated rest - frame @xmath15-band luminosity is @xmath285 .
this suggests that drgs may be a significant constituent of the @xmath286 universe in terms of stellar mass ( @xcite ; rudnick et al .
, in preparation ) .
field - to - field variations are likely large for drgs , as suggested by the high clustering amplitude derived for the hdf - s sample @xcite and by the fact that drgs lie at the high - mass end of galaxies at @xmath40 @xcite .
these variations may explain why such red @xmath287 selected galaxies have not been recognized as an important high redshift population in previous nir surveys ; those deep enough to detect very red and faint high redshift sources were limited to very small areas .
another class of red objects , the extremely red objects ( eros ) at @xmath288 selected by their very red @xmath289 or @xmath290 colours , has been well - studied in the past decade ( see * ? ? ? * for a review ) .
figure [ fig - colcol ] shows that at @xmath284 in the @xmath3 field , there is some overlap between samples selected with the drg criterion @xmath291 and with the ero criterion @xmath292 .
about 60% of @xmath291 objects also have @xmath292 colours .
however , @xmath131 only of @xmath292 objects meet the drg criterion .
this reflects primarily a redshift effect : the drg criterion is efficient at isolating evolved or very dusty systems at @xmath40 , as it is intended to , while the ero criterion is more sensitive to galaxies at lower redshifts , down to @xmath293 .
figure [ fig - zhist ] compares the redshift distributions of the samples in the @xmath3 field .
we have presented the observations , reduction , source detection and photometry of the field around the @xmath3 cluster observed as part of fires .
we have described the main properties of the final images and of the @xmath2-selected source catalogue .
the @xmath3 field complements our hdf - s data , covering an area @xmath270 times wider with very similar image quality in the nir and reaching limiting total magnitudes for point sources about 0.7 mag brighter .
we have compared the photometric redshifts with the extensive set of spectroscopic redshifts available for the @xmath3field and found very good agreement with @xmath23 .
the @xmath3 field confirms our initial findings from the hdf - s , with large numbers of red nir - selected high redshift objects , and substantially increases the statistical significance of the samples , in particular for the distant red galaxies at @xmath40 selected by their @xmath291 colours .
this paper completes the initial phase of the fires survey .
figure [ fig - surveys ] places fires in a broader context by comparing it in the area - depth plane to other published deep nir surveys .
the hdf - s data constitute the deepest ground - based imaging at nir wavelengths to date . in @xmath84 and @xmath1 band ,
it is only surpassed in depth by space - based imaging with hst nicmos while in @xmath30-band it represents the currently deepest observations even from space .
the @xmath3 field occupies a unique position in terms of area - depth combination , being the widest for its depth or conversely the deepest for its size .
we have exploited our deep optical - nir imaging of the hdf - s and @xmath3 fields in several papers to investigate the properties of the @xmath2-selected galaxies and various sub - samples , focusing on their seds , their rest - frame optical luminosities , colours , sizes , their stellar masses , their clustering , and the derived redshift evolution of both their integrated light and stellar mass @xcite .
the results of this initial phase have guided the directions of our current and planned follow - up programs .
the scientific inferences from the hdf - s and @xmath3 fields remain yet limited by the effects of field - to - field variations , and the size and low spectroscopic completeness of the high - redshift samples in particular .
we have thus actively undertaken the addition of more disjoint fields and extensive spectroscopy .
current and planned follow - ups also include near - infrared integral field spectroscopy , extension of the imaging to other wavebands ( especially in the mid - ir with spitzer ) , and hst acs and nicmos high angular resolution imaging to constrain the nature and detailed spectral , dynamical , and morphological properties of the galaxy samples ( e.g. , @xcite ; k. k. knudsen et al . , in preparation ) . with its unique data set ,
fires has been influential to our understanding of galaxy evolution , and will continue to be so as it is extended by its various follow - up programs .
it sets an important basis in the framework of major current and planned surveys such as , for instance , goods and cosmos . we would like to thank the eso staff for obtaining the high quality isaac and fors1 data in service mode and for making them readily available to us .
we also thank henk hoekstra for kindly computing for us the lensing magnifications in the @xmath3 field .
n.m.f.s . and g.r .
acknowledge generous travel support from the leids kerkhoven - bosscha fonds .
g.r . acknowledges the financial support of sonderforschungsbereich 375 from the deutsche forschungsgemeinschaft .
g.d.i . acknowledges financial support by the nasa grant nag 5 - 7697 .
amico , p. , cuby , j .-
, devillard , n. , yung , y. , & lidman , c. 2002 , isaac data reduction guide , version 1.5 , ( garching : eso ) appenzeller , i. , et al . , 1998 , the messenger , 94 , 1 arnouts , s. , et al . , 2001 , , 379 , 740 bershady , m. a. , lowenthal , j. d. , & koo , d. c. 1998 , , 505 , 50 bertin , e. , & arnouts , s. 1996 , , 117 , 393 best , p. , van dokkum , p. g. , franx , m. , & rttgering , h. j. a. 2002 , , 330 , 17 bouwens , r. j. , illingworth , g. d. , thompson , r. i. , & franx , m. 2005 , , 624 , l5 casali , m. , & hawarden , t. 1992 , jcmt - ukirt newsl . , no . 4 , 33 chen , h .- w .
, et al . , 2002 , , 570 , 54 cimatti , a. , et al . , 2002 , , 392
, 395 corbin , m. r. , oneil , e. , thompson , r. i. , rieke , m. j. , & schneider , g. 2000 , , 120 , 1209 daddi , e. , et al . , 2003 , , 588 , 50 dickinson , m. , et al . , 2003 , in preparation djorgovski , s. , et al . , 1995 , , 438 , l13 drory , n. , feulner , g. , bender , r. , botzler , c. s. , hopp , u. , maraston , c. , mendes de oliveira , c. , & snigula , j. 2001 , , 325 , 550 frster schreiber , n. m. , et al . , 2004 , , 616 , 40 franx , m. , et al
. , 2000 , the messenger , 99 , 20 franx , m. , et al . , 2003 , , 587 , l79 giavalisco , m. , et al . , 2004 , , 600 , l93 hauschildt , p. h. , allard , f. , ferguson , j. , baron , e. , & alexander , d. r. 1999 , , 525 , 871 heidt , j. , et al . , 2003 , , 398 , 49 hoekstra , h. , franx , m. , & kuijken , k. 2000 , , 532 , 88 jeltema , t. e. , canizares , c. r. , bautz , m. w. , malm , m. r. , donahue , m. , & garmire , g. p. 2001
, , 562 , 124 kron , r. g. 1980 , , 43 , 305 kuchinski , l. e. , madore , b. f. , freedman , w. l. , & trewhella , m. 2001 , , 122 , 729 labb , i. , et al . , 2003a , , 125 , 1107 labb , i. , et al . ,
2003b , , 591 , l95 labb , i. , et al . , 2005 , , 624 , l81 landolt , a. u. 1992 , , 104 , 340 madau , p. 1995
, , 441 , 18 maihara , t. , et al . , 2001 , , 53 , 25 mccarthy , p. j. , 2004 , , 42 , 477 meisenheimer , k. 1998 , astronomische gesellschaft meeting abstracts , 14 , 3 monet , d. g. , et al . , 1998 , vizier online data catalog , i/252 moorwood , a. f. m. , et al . , 1998 , the messenger , 94 , 7 moustakas , l. a. , davis , m. , graham , j. r. , silk , j. , peterson , b. a. , & yoshii , y. 1997 , , 475 , 445 oke , j. b. 1971 , , 170 , 193 papovich , c. , dickinson , m. , & ferguson , h. c. 2001 , , 559 , 620 persson , s. e. , murphy , d. c. , krzeminski , w. , roth , m. , & rieke , m. j. 1998 , , 116 , 2475 rubin , k. h. r. , van dokkum , p. g. , coppi , p. , johnson , o. , frster schreiber , n. m. , franx , m. , & van der werf , p. 2004
, , 613 , l5 rudnick , g. , et al . , 2001 , , 122 , 2205 rudnick , g. , et al . , 2003 , , 599 , 847 saracco , p. , dodorico , s. , moorwood , a. f. m. , buzzoni , a. , cuby , j .-
g . , & lidman , c. 1999 , , 349 , 751 shapley , a. e. , steidel , c. c. , adelberger , k. l. , dickinson , m. , giavalisco , m. , & pettini , m. 2001 , , 562 , 95 thompson , r. i. , storrie - lombardi , l. j. , weymann , r. j. , rieke , m. j. , schneider , g. , stobie , e. , & lytle , d. 1999 , , 117 , 17 thompson , r. i. , et al . , 2005 , , in press ( astro - ph/0503504 ) tran , k .- v .
h. , et al . , 2003 , in preparation tran ,
h. , kelson , d. d. , van dokkum , p. g. , franx , m. , illingworth , g. d. , & magee , d. 1999 , , 522 , 39 trujillo , i. et al .
, 2003 , , 604 , 521 trujillo , i. et al .
, 2005 , , submitted ( astro - ph/0504225 ) toft , s. , van dokkum , p. g. , franx , m. , thompson , r. i. , illingworth , g. d. , bouwens , r. j. , & kriek , m. 2005 , , 624 , l9 van dokkum , p.g .
2001 , , 113 , 1420 van dokkum , p.g .
, franx , m. , fabricant , d. , illingworth , g.d .
, & kelson , d.d .
2000 , , 541 , 95 van dokkum , p. g. , kriek , m. , rodgers , b. , franx . m. , & puxley , p. 2005
, , 622 , l13 van dokkum , p. g. , et al .
2003 , , 587 , l83 van dokkum , p. g. , et al .
2004 , , 611 , 703 warmels , r. h. , 1991 , `` the eso - midas system , '' in astronomical data analysis software and systems i ( pasp conf .
series ) , 25 , 115 williams , r. e. , et al . , 2000 , , 120 , 2735 zacharias , n. , urban , s. e. , zacharias , m. i. , wycoff , g. l. , hall , d. m. , germain , m. e. , holdenried , e. r. , & winter , l. 2004 , , 127 , 3043 cccccc + + + fors1 & @xmath13 & 1 & 30 & 16500 & 0.69 + fors1 & @xmath14 & 1 & 18 & 7200 & 0.57 + fors1 & @xmath15 & 1 & 6 & 1800 & 0.65 + wfpc2 & f606w & 6 & 12 & 6500 & 0.21 + wfpc2 & f814w & 6 & 12 & 6500 & 0.22 + isaac & @xmath0 & 4 & 776 & 93120 & 0.48 + isaac & @xmath1 & 4 & 739 & 87780 & 0.46 + isaac & @xmath2 & 4 & 1588 & 95280 & 0.52 + + + + isaac & @xmath0 & f1 & 205 & 24600 & 0.44 + & & f2 & 211 & 25320 & 0.46 + & & f3 & 180 & 21600 & 0.45 + & & f4 & 180 & 21600 & 0.46 + isaac & @xmath1 & f1 & 196 & 23520 & 0.46 + & & f2 & 180 & 20700 & 0.42 + & & f3 & 184 & 22080 & 0.43 + & & f4 & 179 & 21480 & 0.42 + isaac & @xmath2 & f1 & 473 & 28380 & 0.46 + & & f2 & 398 & 23880 & 0.50 + & & f3 & 358 & 21480 & 0.41 + & & f4 & 359 & 21540 & 0.39 + lcc fors1 @xmath13 & 24.127 & 24.828 + fors1 @xmath14 & 26.746 & 26.634 + fors1 @xmath15 & 27.131 & 27.145 + wfpc2 @xmath48 & 22.900 & 23.020 + wfpc2 @xmath49 & 21.660 & 22.090 + isaac @xmath0 & 24.970 & 25.872 + isaac @xmath1 & 24.846 & 26.226 + isaac @xmath2 & 24.309 & 26.169 cccccc fors1 @xmath13 & & @xmath294 & 1.82 & 0.025 & 29.01 + fors1 @xmath14 & & @xmath295 & 2.31 & 0.093 & 29.60 + fors1 @xmath15 & & @xmath296 & 2.30 & 0.034 & 28.78 + wfpc2 @xmath48 & & @xmath297 & 2.01 & 0.234 & 28.84 + wfpc2 @xmath49 & & @xmath298 & 2.19 & 0.224 & 28.10 + isaac @xmath0 & f1 & 0.00504 & 2.57 & 0.121 & 27.80 + & f2 & 0.00591 & 2.13 & 0.140 & 27.70 + & f3 & 0.00531 & 1.96 & 0.143 & 27.87 + & f4 & 0.00489 & 2.65 & 0.109 & 27.84 + isaac @xmath1 & f1 & 0.01054 & 3.29 & 0.101 & 27.20 + & f2 & 0.01192 & 2.79 & 0.128 & 27.14 + & f3 & 0.00957 & 3.06 & 0.111 & 27.34 + & f4 & 0.00939 & 2.71 & 0.136 & 27.40 + isaac @xmath2 & f1 & 0.00926 & 2.45 & 0.128 & 27.45 + & f2 & 0.01323 & 2.54 & 0.115 & 27.07 + & f3 & 0.00985 & 2.76 & 0.125 & 27.30 + & f4 & 0.00975 & 2.73 & 0.143 & 27.28 + ccccc fors1 @xmath13 & & 1.035 & 29.01 & 27.24 + fors1 @xmath14 & & 1.035 & 29.60 & 27.83 + fors1 @xmath15 & & 1.035 & 28.78 & 27.01 + wfpc2 @xmath48 & & 0.31 & 30.00 & 28.28 + wfpc2 @xmath49 & & 0.33 & 29.24 & 27.46 + isaac @xmath0 & f1 & 0.72 & 28.00 & 26.20 + & f2 & 0.72 & 27.77 & 25.97 + & f3 & 0.72 & 27.95 & 26.15 + & f4 & 0.72 & 27.98 & 26.18 + isaac @xmath1 & f1 & 0.69 & 27.39 & 25.57 + & f2 & 0.69 & 27.31 & 25.49 + & f3 & 0.69 & 27.61 & 25.79 + & f4 & 0.69 & 27.63 & 25.81 + isaac @xmath2 & f1 & 0.78 & 27.61 & 25.73 + & f2 & 0.78 & 27.16 & 25.28 + & f3 & 0.78 & 27.50 & 25.63 + & f4 & 0.78 & 27.54 & 25.67 + | we present deep near - infrared @xmath0 , @xmath1 , and @xmath2 band imaging of a field around @xmath3 , a massive cluster at @xmath4 .
the observations were carried out with the isaac instrument at the eso very large telescope ( vlt ) as part of the faint infrared extragalactic survey ( fires ) .
the total integration time amounts to 25.9 h in @xmath0 , 24.4 h in @xmath1 , and 26.5 h in @xmath2 band , divided nearly equally between four pointings covering @xmath5 .
the @xmath6 total limiting ab magnitudes for point sources from the shallowest to the deepest pointing are @xmath7 , @xmath8 , and @xmath9 .
the effective spatial resolution of the coadded images has @xmath10 , @xmath11 , and @xmath12 in @xmath0 , @xmath1 , and @xmath2 , respectively .
we complemented the isaac data with deep optical imaging using existing hubble space telescope wfpc2 mosaics through the f606w and f814w filters and additional @xmath13 , @xmath14 , and @xmath15 band observations we obtained with the vlt fors1 instrument .
we constructed a @xmath2-band limited multicolour source catalogue down to @xmath16 ( @xmath17 for point sources ) .
the catalogue contains 1858 objects , of which 1663 have eight - band photometry .
we describe the observations , data reduction , source detection , and photometric measurements method .
we also present the number counts , colour distributions , and photometric redshifts @xmath18 of the catalogue sources .
we find that our @xmath2-band counts at the faint end @xmath19 , with slope @xmath20 , lie at the flatter end of published counts in other deep fields and are consistent with those we derived previously in the hubble deep field south ( hdf - s ) , the other fires field .
spectroscopic redshifts @xmath21 are available for @xmath22 sources in the @xmath3 field ; comparison between the @xmath18 and @xmath21 show very good agreement , with @xmath23 .
the @xmath3 field observations complement our hdf - s data set with nearly five times larger area at @xmath24 brighter magnitudes , providing more robust statistics for the slightly brighter source populations . |
in studying the thermodynamic state of a physical system , the free energy @xmath0 is a quantity of fundamental importance .
it describes the equilibrium properties of systems that may exchange energy with their environments .
formally , it is related to the internal energy , @xmath1 , by a legendre transform @xmath2 , where @xmath3 is the temperature and @xmath4 the entropy .
the free energy is a state function and hence , for any process connecting two equilibrium states , the respective change of the free energy @xmath5 , solely depends on the final and initial states without regard to the particular process connecting them .
in contrast , the work @xmath6 done on the system and the heat @xmath7 exchanged with the environment are process - dependent . yet , their sum yielding the change in internal energy , @xmath8 , does not depend on the details of the path connecting the final with the initial state .
recently , jarzynski found a relation between the path dependent work and the path independent free energy change in terms of the following sum rule @xcite , @xmath9 where @xmath10 denotes the probability density function ( pdf ) of the work that is performed on the system .
the process from which this work results , starts out in a state of thermal equilibrium at temperature @xmath11 , and is induced by the action of forces , or more generally by changes of parameters characterizing the hamiltonian of the considered system .
these parameter changes are supposed to follow a specific protocol , on the details of which the work pdf will depend in general . in eq .
( [ jarzynski ] ) , @xmath12 , denotes the difference between the free energies @xmath13 and @xmath14 of the initial thermal equilibrium state and the thermal equilibrium state of the system with the final parameter values , respectively .
both equilibrium states are at the same temperature @xmath3 . in general
, the second equilibrium state differs from the actual state that is reached at the end of the protocol . in principle
, this `` associated thermal equilibrium '' will be approached when the system stays in contact with a heat bath at temperature @xmath3 upon completion of the protocol . as a difference of a state function
, @xmath15 depends on the initial and final parameter values but is independent of the details of the protocol .
the random nature of the work manifested in the pdf @xmath10 is a consequence of the inherent randomness of the thermal initial state and additionally also due to a possible randomness of the dynamics , be it of quantum or classical , stochastic nature . in the latter case ,
randomness and dissipation must be properly balanced by fluctuation dissipation relations eventually imposing thermal equilibrium at the initial inverse temperature @xmath16 at constant parameter values .
any application of the jarzynski equality ( [ jarzynski ] ) to experiments hence requires repeating experiments with the same protocol many times in order to generate a sufficient statistics . the same requirement to generate sufficiently many data
must also be met in numerical implementations of the jarzynski equality aiming at the determination of free energy changes . the feasibility of this scheme with the goal to determine the free energy change @xmath15 was demonstrated in various experimental systems such as for single molecules @xcite and classical oscillators @xcite .
yet the practical applicability of the jarzynski equality is severely limited because the estimate of the exponential work average strongly relies on how well the tail of the work distribution with @xmath17 is sampled @xcite .
in general the exponential average @xmath18 gives rise to a bias of the jarzynski free energy estimate @xcite .
it is now well recognized that the best convergence can be obtained from slow protocols associated with small dissipation such that @xmath19 , i.e. , for almost reversible processes in which the system passes through a succession of quasi - equilibrium states @xcite .
for fast switching , the finite sampling error becomes substantial both concerning the bias and the variance .
the respective behavior in the large sampling limit was investigated in several studies @xcite . an alternative strategy to determine free energy differences
is based on the crooks relation @xcite , @xmath20 which allows to infer @xmath15 without the average process required in eq.([jarzynski ] ) . here
the two work pdfs @xmath21 and @xmath22 refer to the original , or forward ( @xmath23 ) protocol and to the backward protocol ( @xmath24 ) which is started in the associated equilibrium state , i.e. , at the initial temperature of the forward process and at those parameter values that have finally been reached in the forward process and retraces its parameter values .
the free energy change can be read off from a plot displaying the functions @xmath21 and @xmath25 as the work value at which the two distributions cross each other .
a reliable estimate of the crossing point requires sufficient sampling of work with @xmath17 for the forward and respectively with @xmath26 for the backward process .
it is worthwhile mentioning here that the _ average _ work is never smaller than the free energy change , i.e. @xmath27 , as a consequence of jensen s inequality , stating that @xmath28 .
hence , the work applied to the system in an isothermal process is at least as large as the change of free energy , in accordance with the second law of thermodynamics .
the amount of work that is equal to the free energy change can be applied to the system in an isothermal reversible process , while any surplus @xmath29 is `` dissipated work '' .
the dissipated work can also be interpreted in terms of the entropy change of the total system including the heat - bath , @xmath30 @xcite , which , again is positive in accordance with the second law . for rapid protocols ,
the dissipated work becomes large and therefore those realizations of the process with @xmath31 may become extremely unlikely , such that even large samples obtained from experiments or numerical investigations may result in an unpopulated no man s land between the forward and the backward work pdfs and , hence , in an unreliable estimate of the free energy change based on crooks crossing criterion .
dividing both sides of the crooks relation ( [ crooks ] ) by the factor of @xmath32 , one obtains @xmath33 this equation was originally suggested by bennett @xcite as an efficient basis to estimate partition function ratios by means of monte carlo sampling . in bennett s
derivation , the fermi - function - like weights in eq .
( [ bennett ] ) resulted as acceptance ratios from the requirement of minimal variance of the partition function estimator in the large sample size limit . from the gaussian assumption which this argument underlies one would be let to believe that the bennett estimator yields minimal variance only if the overlap region of the forward and backward pdfs @xmath21 and @xmath25 is substantially populated .
however , shirts et al .
@xcite demonstrated that bennett s acceptance ratio method always yields a maximum likelihood estimate of the free energy change by the use of forward and backward data .
as such , it allows to extract @xmath15 with the smallest variance compared to any other free energy estimator , even if the two pdfs , @xmath34 and @xmath35 do not overlap .
therefore , the estimation of the free energy change from eq .
( 3 ) is less restricted than the jarzynski method and crooks crossing criterion as has been confirmed in various recent studies @xcite . in this work ,
we investigate the statistical behavior of the free energy change estimation by the bennett method , discuss its limitations imposed by the difference of the amounts of dissipated work in the forward and backward protocol , and compare it with the methods proposed by jarzynski and crooks .
in our discussion we lay particular emphasis on the practical limitations resulting from the nonequilibrium nature that is imposed on the system by time dependent perturbations .
we begin in section ii with a brief review on how the dissipated work and the time asymmetry constrain the jarzynski and crooks method . in the following section iii ,
we shortly review the maximum likelihood argument @xcite leading to eq .
( [ bennett ] ) and examine its solutions in some limiting cases .
in particular when the jarzynski and crooks methods are both hampered by large dissipated work , we show that the bennett method simplifies to the `` 1/2-formula '' expressing the free energy change as the arithmetic mean of the jarzynski estimation obtained from the forward and the backward process . in sec .
iv , we take as an example a gaussian chain , and consider processes in which work is done by extending or compressing the chain at a constant speed .
reliable estimations of @xmath15 based on crooks crossing criterion or the jarzynski estimator do not exist if the dissipated work becomes large .
we present regions in a pulling - speed versus sample - size plane , in which the errors of these estimators are smaller than @xmath36 .
for the gaussian chain , the dissipated works in the forward and backward processes are the same , and , as a consequence the bennett method gives precise estimations of @xmath15 . in sec .
vi , we consider a chain of monomers interacting via pairing potentials to form a hairpin - like structure .
keeping one end fixed and pulling at the other end with constant speed gives rise to a non - gaussian work distribution with different dissipated works for the forward and the backward protocols .
the difference increases with growing pulling speed and leads to a finite bias of the bennett estimator .
the results are summarized in sec .
we consider a time dependent perturbation of a system in terms of a control parameter @xmath37 that varies in time @xmath38 according to a prescribed protocol .
the protocol can be performed in bidirectional way . for the forward protocol during a time interval @xmath39 , the system departs from its initial thermal equilibrium state at temperature @xmath3 and control parameter @xmath40 and reaches a terminal state with @xmath41 , which , in general , is a nonequilibrium state . the work done during this process will be referred to as the forward work . for the backward protocol ,
the system starts out in thermal equilibrium at the same inverse temperature @xmath16 with the control parameter at @xmath41 . in this case
, the parameter retraces its values in time - reversed order and finally reaches @xmath42 as terminal parameter value . after repeating infinitely many times both protocols , two probability distribution functions , @xmath34 for the forward and @xmath43 for the backward protocol , can be assembled . to obtain a work smaller than the free energy change in a realization of the forward process
is displayed as a function of the dissipated work @xmath44 for a gaussian process as given by eq .
almost reversible protocols with small dissipated work give a probability @xmath45 close to the maximum @xmath46 . in the opposite limit of large dissipated work @xmath45 becomes exponentially small . ] as mentioned in the introduction , the probability for observing work less than the free energy change is the crucial factor for the applicability of the crooks crossing criterion . for the forward process
this probability is given by @xmath47 and , accordingly , for the backward process , by @xmath48 if these probabilities are too low to get a sufficient number of realizations of work below the free energy change within a reasonable total number of experiments , the crooks crossing criterion fails to give @xmath15 .
moreover , a poor sampling of work data below @xmath15 will lead to a substantial underestimation of the exponential average @xmath18 resulting in a too large jarzynski estimate of @xmath15 . as an example , we consider a gaussian work pdf @xmath21 for the forward process . as a consequence of the crooks relation ( [ crooks ] )
the backward work pdf @xmath22 is also gaussian with the same variance @xmath49 .
hence the two pdfs are : @xmath50 , \quad \alpha = f , b\:.\ ] ] in this case the free energy change becomes @xmath51 .
the probability of finding work less than the free energy change in the forward protocol then becomes @xmath52 where @xmath53 denotes the complementary error function , see fig . 1 .
if the process is weakly dissipative , i.e for @xmath54 , the forward work distribution becomes narrow and the argument of the error function is small . because of @xmath55 for @xmath56 , the probability @xmath57 approaches the value @xmath46 in the limit of vanishing dissipated work @xcite . in other words , on average , every other measurement probes a work that is less than the free energy change ( fig . 1 ) . in the opposite limit of a strongly dissipative process being characterized by @xmath58 ,
the work pdfs become broad because then @xmath59 . using the asymptotic behavior of @xmath53 for @xmath60 , @xmath61
, we obtain an exponentially small probability @xmath62 to find a work in the forward protocol that is less than the free energy change .
defined by eq . ( 6 ) for a gaussian pdf monotonicly varies as a function of the hysteresis @xmath63 from vanishingly small values at small @xmath64 to a maximal value at large @xmath64 .
note that the hysteresis @xmath64 gives the average dissipated work for the forward and the backward protocol , and , for a gaussian work distribution , characterizes the separation of the peaks of the forward and backward work pdfs @xmath21 and @xmath25 , respectively .
the pdfs at the values indicated by the arrows in panel ( a ) are displayed in the panels ( b ) and ( c ) .
( b ) for weak dissipation ( @xmath65 ) , the time asymmetry is small ( @xmath66 ) and the work pdfs display a large overlapping area near @xmath15 .
( c ) for strong dissipation ( @xmath67 ) , the time asymmetry approaches its maximum ( @xmath68 ) and the work pdfs are well - separated . ]
a measure quantifying the difference between the forward and the backward work pdfs @xmath21 and @xmath25 based on the jensen - shannon divergence of the two distributions is given by the so - called time asymmetry @xcite reading @xmath69 where @xmath70 and @xmath71 denote averages with respect to the forward and the backward distributions , respectively .
this quantity vanishes when @xmath21 and @xmath25 agree with each other and otherwise falls in the range @xmath72 .
therefore , if the measured work values are mostly populated in the close vicinity of the free energy change , @xmath15 for the forward protocol and @xmath73 for the backward protocol , the time asymmetry approaches its minimum value @xmath74 .
in this case , @xmath34 and @xmath35 have a large overlap .
on the other hand , the time asymmetry reaches its maximum @xmath75 when @xmath34 and @xmath35 are perfectly separated by a large gap between the respective regions with @xmath76 for the forward protocol and @xmath77 for the backward protocol .
we exemplify these behaviors for gaussian work pdfs , given by eq .
( [ gauss ] ) : figure 2 ( a ) displays the time asymmetry as a function of the peak separation between @xmath34 and @xmath35 , that is as a function of the hysteresis @xmath64 defined as @xcite @xmath78 for small values of @xmath64 , the time asymmetry is close to zero , and the corresponding pdfs substantially overlap in an area near @xmath15 as displayed in fig .
2(b ) for @xmath79 .
in contrast , when @xmath64 is large the time asymmetry approaches its maximum , and seemingly , the forward and backward pdfs @xmath34 and @xmath35 no longer overlap with each other as exemplified in fig .
2(c ) for @xmath80 .
strictly speaking , due to the crooks relation ( [ crooks ] ) @xmath34 and @xmath35 must share the same support and therefore always have a finite overlap , which , however , may become virtually invisible .
little overlap of the forward and the backward work pdfs indicates that the process involved in the work generation is highly irreversible , and hence , the direction of time flow is unambiguous giving the information @xmath75 of a single bit .
also , for this case , large dissipated work is associated with both processes , and the probabilities @xmath81 and @xmath82 to find work values being less than the free energy change become extremely small ( fig .
2(c ) ) . in summary
, the more pronounced is the direction of the time - arrow in a nonequilibrium process , the rarer it becomes to find events with work smaller than the free energy change and the more erroneous both estimators by jarzynski and crooks turn out to be .
finally , we note a relation between the time asymmetry and the hysteresis @xmath64 specifying the average dissipated work of the forward and the backward protocol . because the free energy does not change upon completion of a cyclic process , the total dissipated work agrees with twice the above introduced work hysteresis , @xmath83 . as pointed out in ref .
@xcite , the hysteresis gives a rough estimate for the number of measurements @xmath84 required for a relatively reliable estimation of @xmath15 based on the jarzynski equality , as @xmath85 . on the other hand the hysteresis
is related to the time asymmetry by the following inequality @xcite , @xmath86 as a consequence the required number of measurements becomes infinitely large if the time asymmetry approaches the limiting value , @xmath75 .
therefore not only the jarzynski estimator ceases to work but also crooks crossing criterion fails for processes with unambiguous arrows of time .
as mentioned in the introduction , shirts et al .
@xcite obtained the bennett relation ( [ bennett ] ) by means of the statistical concept of maximum likelihood . here , this approach is based on a transformation of the crooks relation ( [ crooks ] ) into an expression for the probability @xmath87 with which the forward protocol is drawn from an ensemble of equally many realizations of both protocols under the condition that the work performed on the system is @xmath88 , for the forward , and @xmath89 for the backward protocol .
this conditional probability takes the form of a fermi function , reading @xcite @xmath90}\ : .
\label{pfw}\ ] ] the complementary probability @xmath91 for finding @xmath92 in a realization of the backward process then becomes @xmath93}\ : .
\label{pbw}\ ] ] according to the maximum likelihood method , the most likely value of the free energy change compatible with the work values of m realizations of the forward and equally many realizations of the backward protocol maximizes the likelihood defined as the joint probability @xmath94 evaluated at the actual outcomes @xmath95 , @xmath96 , of the forward , @xmath97 , and the backward , @xmath98 protocols .
this leads to @xmath99 presenting a non - linear equation in @xmath15 which , for given data @xmath100 and @xmath101 , has a uniquely defined solution .
it can numerically be solved by means of the newton algorithm . in the continuum limit of @xmath102
this equation can be written as @xmath103 yielding eq .
( [ bennett ] ) upon a change of the variable @xmath104 on the right hand side . in the limiting cases of slow and rapid protocols ,
( [ bennettfinite ] ) and ( [ contb ] ) have simple solutions .
when the work protocol is performed quasi - statically so that the associated dissipation is small , the majority of work values are localized near the free energy difference such that @xmath105 for the forward , and @xmath106 for the backward protocol .
the expansions of the exponential factors in eq .
( [ contb ] ) in terms of their small arguments lead to @xmath107 yielding for the free energy change @xmath108 for a slow work protocol with small dissipations , the bennett estimate of the free energy difference is given by the difference of averaged works of forward and backward protocols . on the other hand , for a fast protocol generating large dissipation ,
the overwhelming majority of forward data satisfies @xmath109 and accordingly @xmath110 for the backward data .
then , we can expand the fermi - functions in eq .
( [ bennettfinite ] ) giving in leading order @xmath111 this yields a central result of this work @xmath112 where the jazynski estimates @xmath113 are determined from @xmath114 realizations of the forward ( @xmath115 ) and the backward ( @xmath116 ) protocols as , @xmath117\:.\ ] ] the bennett estimate @xmath15 in the large dissipation regime is given by half of the difference of the forward and the backward jarzynski estimates , @xmath118 , which we denote as @xmath119 . due to the finite sampling of work values , the estimates @xmath113 are still random and
hence can differ from experiment to experiment . in order to characterize the statistics of the jarzynski estimator for a finite number @xmath114 of work data
, one considers @xmath120 repetitions of @xmath114 work measurements .
the totality of work data then consists of @xmath121,[w^{(2)}_{j}],\cdots , [ w^{(m)}_{j}]\}$ ] where @xmath122 $ ] is the data set from the @xmath123-th experiment : @xmath122\equiv \{w^{(\ell)}_{1},w^{(\ell)}_{2},\cdots w^{(\ell)}_{m}\ } $ ] .
based on the resulting set of the jarzynski estimates for @xmath114 data according to eq .
( [ sb ] ) the statistics of these estimates can be analyzed .
in particular , one obtains a block averaged estimate of the free energy difference from @xmath120 experiments reading @xmath124.\ ] ] it was proven that the block averaged jarzynski estimator , @xmath125 , with a finite @xmath114 is bounded from below by the true value of @xmath15 and from above by the average work @xmath126 , i.e. , it lies in the range @xcite @xmath127 unfortunately , neither a lower nor an upper bound of the bennett estimator is known .
the underlying difficulty results from the structure of the eq .
( [ half ] ) , which is based on the _ difference _ between the forward and backward jarzynski estimates .
therefore the inequalities ( [ mono ] ) valid for the individual terms do not translate to the bennett estimation in the large dissipation limit .
we illustrate the aforementioned features by considering a one - dimensional gaussian chain of @xmath128 beads connected by harmonic springs .
additionally , the beads experience strong friction and fluctuating forces stemming from a heat bath at temperature @xmath3 .
the potential energy of the system is given by @xmath129 where @xmath130 denotes the position of the @xmath131th bead ; the first bead with @xmath132 is fixed at the origin ( @xmath133 ) . a simple way to perform work on the system
is to pull the last bead ( @xmath134 ) with a constant speed . the forward protocol acts during a time interval @xmath135 , starting from the thermal equilibrium state of the chain with its end at the origin , @xmath136 , and proceeds by increasing @xmath137 linearly in time as @xmath138 until the chain end reaches the designated position @xmath139 . for the backward protocol , the chain is initially in the thermal equilibrium while its end is fixed at @xmath140 and the work is done by changing the chain end position as @xmath141 .
the jarzynski work for the forward protocol is given by @xmath142,\ ] ] and the same expression with @xmath143 gives the work for the backward protocol .
the overdamped motion of the beads ( @xmath144 ) with the friction constant @xmath145 can be described by a position langevin equation : @xmath146 where the gaussian white noises @xmath147 model thermal random forces exerted on the chain .
accordingly , they satisfy @xmath148 and @xmath149 .
( [ legc ] ) describes the rouse model of a polymer in a viscous fluid where the hydrodynamic interactions are neglected . as defined in eq .
( [ hysteresis ] ) is displayed as a function of the time scale ratio @xmath150 representing the pulling speed , for @xmath151 , and for different lengths of the chain , @xmath152 ( @xmath153 ) and @xmath154 ( @xmath155 ) in the regime of slow pulling speeds .
note that @xmath156 agrees with the work variance of the gaussian chain . for vanishing pulling speed ( @xmath157 )
the variance vanishes . at higher pulling speeds ,
the chain - length dependence is more pronounced .
the panel ( b ) shows the deviations of the block averaged jarzynski estimator as defined in eq .
( [ jeba ] ) from the exact value of the free energy change as a function of pulling speed .
the data were sampled from the exactly known gaussian work distributions given by the eqs .
( [ gauss ] ) and ( [ avesig ] ) , rather than by simulations of the langevin equations ( [ legc ] ) . the sampling size was chosen as @xmath158 ( @xmath159 for @xmath152 and @xmath160 for @xmath154 ) and @xmath161 ( @xmath153 for @xmath152 and @xmath155 for @xmath154 ) . in both cases
the number of @xmath162 blocks led to well converged block averages .
as the pulling speed increases , the error due to the finite sampling size becomes significant .
only the protocol taking much longer than the relaxation time of the system ( @xmath163 ) yields an error less than @xmath36 . in both panels , the lines connecting the symbols are guides to the eye . ] for pulling protocols with @xmath164 .
the duration of time in which the final end to end distance was reached was chosen as @xmath165 with @xmath166 varying from @xmath167 up to @xmath168 for different block sizes , @xmath158 ( @xmath169 ) and @xmath161 ( @xmath153 ) .
averages are performed over @xmath170 blocks .
irrespective of the block size @xmath114 , the bias remains within @xmath36 when the time asymmetry is less than @xmath75 ( represented by the vertical dotted line ) , and abruptly rises in the limit of maximal time asymmetry , @xmath171 .
the broken and solid lines connecting equal symbols serve as guides to the eye .
the vertical bars at the symbols indicate the according variances of the jarzynski estimator . ]
this model is exactly solvable as shown by dhar @xcite .
the free energy difference associated with stretching in the forward process is given by @xmath172 and accordingly @xmath73 for the backward process .
the probability distribution of the work is of gaussian form as given by the eq .
( [ gauss ] ) , with the forward work average and the variance @xcite , @xmath173_{n-1,n-1},\end{aligned}\ ] ] where @xmath174 is the lattice laplacian , @xmath175 , and @xmath176 . for the backward protocol , the pdf is also gaussian with the same variance , @xmath177 as given in eq .
( [ avesig ] ) , and the work average @xmath178 indicating that the free energy difference for the gaussian distribution is determined by the difference of the forward and backward work averages : @xmath179 . on the other hand , the hysteresis , defined by eq .
( [ hysteresis ] ) becomes @xmath180 . for a given final extension
the pulling speed is a crucial factor in determining the value of @xmath177 , as it appears in the prefactor . according to eq .
( [ avesig ] ) , for @xmath181 , i.e. for a sudden change of the end - to - end distance , the variance approaches a finite value which is almost independent of the chain length @xcite . with increasing duration of the protocol
the work variance becomes dependent on the chain length , indicating a collective response of the chain . in the limit of infinitely slow processes the variance asymptotically vanishes as @xmath182 , as one would expect for an isothermal quasi - static protocol .
`` slow '' and `` fast '' can be quantified relative to characteristic time scales of the system , for example , to its relaxation times . for gaussian chains considered here the relaxation time
is given by @xmath183 with @xmath184 being the minimum eigenvalue of the negative lattice laplacian @xmath185 : @xmath186 $ ] .
for @xmath187 , the relaxation time increases quadratically with the chain length . since we are interested in the efficiency of different estimators of free energy change we compared these for two different chain lengths and different protocols , in all cases leading to the same change of the free energy . according to eq .
( [ deltaf ] ) the final extension then depends on the chain length as @xmath188 for large @xmath189 .
the ratio of the two time scales , @xmath190 then quantifies the progress rate of the protocol : in particular , for @xmath191 , the protocol approaches a reversible processes with vanishing dissipated work . , and various progress rates of the protocol , @xmath150 , as a function of the sampling size @xmath114
. averages are performed over @xmath192 blocks .
the estimated free energy difference shows the monotonic behavior predicted by eq .
( [ jeba ] ) , and is larger than the true value for any finite @xmath114 in agreement with eq .
( [ mono ] ) .
the lines represent the large-@xmath114 asymptotic behavior , eq .
( [ asymp ] ) . only the protocols with slow pulling speeds for @xmath193 and @xmath194
display this asymptotic behavior . in order to see it for faster pulling speeds one has to go to even larger sample sizes . ] for given @xmath15 and @xmath150 , the exact value of the variance @xmath177 can be found using eqs .
( [ deltaf])-([workb ] ) , which completely characterize the gaussian work pdf given by eq .
( [ gauss ] ) @xcite .
this gaussian distribution indicates the ideal work pdf which one would obtain from an infinite number of measurements , leading to unbiased results of the estimators . in order to address the statistical bias for a finite number of data
, we investigated the dependence of various free energy estimators on @xmath150 and on the sample size by drawing data from the exactly known gaussian work distribution given by the eqs .
( [ gauss ] ) : in this way we could avoid time - consuming simulations of the langevin equations ( [ legc ] ) and yet obtain large amounts of data with low computational effort . first , in fig .
3(a ) , we present results for the hysteresis @xmath64 as a function of @xmath150 . from the value of @xmath64 at a given @xmath150
, one can estimate the required number of measurements for a reasonable estimate of @xmath15 by the jarzynski equality according to @xmath195 .
for example , for @xmath196 , @xmath197 .
if the chain is pulled faster , @xmath84 may become enormously large so that the free energy estimation from a finite number of measurements becomes totally unreliable .
we used the known work pdf of the gaussian chain as a test case and determined block averages over @xmath198 blocks of different sizes .
figure 3(b ) shows the errors in the free energy estimation from the forward work measurement , @xmath199 , where @xmath200 is obtained via eq .
( [ jeba ] ) for two different sample sizes with @xmath158 and @xmath161 . with @xmath150 the dissipated work increases giving rise to increasing deviations of the estimated free energy change from its true value .
with increasing time asymmetry the total dissipated work grows .
for the jarzynski estimator this growth leads to a systematic increase of its bias relative to the exact value as displayed in fig .
4 . in particular
, the bias diverges if the limiting value @xmath75 of the time asymmetry is approached .
this divergence persists for any finite number of data in accordance with the exponential scaling of the required number of data , @xmath201 , with the hysteresis , @xmath202 .
the bias of the jarzynski estimator is presented in fig . 5 as a function of the block size @xmath114 for different pulling speeds .
it asymptotically approaches zero at different rates depending on the pulling speed for large values of @xmath114 . according to ref .
@xcite the asymptotic approach is given by a power law reading : @xmath203 this behavior though can only be identified for the two cases with slowest pulling speeds @xmath204 and @xmath205 ( see the straight lines in fig .
5 ) . for the other , faster protocols ,
the sample sizes examined here are insufficient to enter the asymptotic regime being governed by the central limit theorem .
instead , we observe algebraic decay behavior as @xmath206 with @xmath207 . for the sample sizes studied here ,
the decay exponent @xmath208 becomes very small with increasing pulling speed , signaling a bad convergence of the finite sample average for fast protocols . and
each sample of size @xmath114 was independently generated @xmath210 times .
the symbols j , c , and b are assigned to those regions in the parameter plane spanned by the inverse box size and @xmath150 , where the jarzynski , crooks , and bennett methods , respectively , give an estimate of the free energy difference with an error being less than @xmath36 . along the line connecting the squares ,
the expected error of the jarzynski estimator equals @xmath211 , i.e. , @xmath212 . below this line , for smaller pulling speeds ,
the error is smaller .
the line connecting the circles indicates the corresponding curve for the crooks estimator .
( b ) estimated bias of three estimators of free energy difference as a function of @xmath150 with @xmath158 and @xmath198 .
the jarzynski error becomes larger than @xmath36 as soon as @xmath213 .
the crooks crossing criterion gives the correct value of @xmath15 up to @xmath214 , where the overlap between the forward and the backward pdf disappears so that the crooks criterion only delimits the range within which the free energy difference is located .
apart from an increase of the expected sampling error , bennett s method according to eq .
( [ bennettfinite ] ) is insensitive to the pulling speed , yielding precise estimates of @xmath15 .
the error bars indicate the magnitude of the variance of the free energy estimators based on @xmath198 blocks of size @xmath215 .
the variance of the jarzynski estimator increases rapidly with increasing speed , while the variances of the crooks and the bennett estimators grow much slower . ]
our next goal is to systematically compare the biases of three estimators , the jarzynski equality , crooks crossing criterion , and the bennett method , for the gaussian work pdf as exemplified by the pulling protocol of the gaussian chain . in fig 6(a ) , we display the absolute magnitude of the biases of the three estimators as functions of @xmath150 and sampling size @xmath114 .
as expected , the bennett method is always superior to the other estimators .
the symbols j , c and b in fig .
6 ( a ) indicate those parameter regions within which the jarzynski , crooks and bennett estimators , respectively , deviate by less than the thermal energy @xmath216 from the true free energy difference .
the jarzynski estimator soon becomes unreliable with increasing pulling speed , while the crooks estimator yields better values for larger pulling speeds provided that it is based on a sufficiently large number of data .
it is interesting to note that for small @xmath114 and very slow pulling speeds the jarzynski estimator performs better than the crooks estimator . to better understand this observation , note that already for @xmath217 , the block average ( [ avesig ] ) yields @xmath218 which is close to the correct value for weakly dissipative protocols , according to eq .
( [ jeba ] ) . on the other hand , for the crook s crossing criterion to properly work , the tails of the forward and backward pdfs need to be sampled with sufficient accuracy , necessitating a sample of reasonable size .
the bennett estimator for this model with a gaussian work pdf is found to be free of any bias in the whole investigated parameter region . yet with increasing pulling speed the variance of the bennett estimator increases as displayed in panel ( b ) of fig . 6 for a fixed @xmath114 .
this increase though is less pronounced than those of the respective variances of the crooks and the jarzynski estimators .
the pairing interactions are present between the @xmath131-th and @xmath219-th monomers where @xmath220 with @xmath221 denoting the number of pairs . in this work
, we considered a hairpin chain with @xmath222 . ] , @xmath223 and @xmath224 ; the maximal end to end distance of the chain reached at the end of the forward protocol is @xmath225 ; for further details see the text below eq .
( [ lehp ] ) .
each histogram was obtained as an average over @xmath226 independently generated histograms based on @xmath215 simulations of the langevin equation ( [ lehp ] ) .
the variances of the such estimated probabilities indicated by error bars are relatively small . on the right hand side the histograms of the forward protocols and on the left hand side those for negative work of the backward protocols
are depicted .
panel ( a ) displays the case when the protocol lasted @xmath227 representing a relatively slow pulling speed for which the forward and backward work histograms overlap .
note that here @xmath228 is the relaxation time of a three - dimensional gaussian chain , @xmath229 .
consequently a reliable estimate of @xmath230 indicated by the vertical dotted line was obtained from the crooks crossing criterion .
panel ( b ) gives the work distributions for a five times faster pulling speed ( @xmath231 ) . at this higher speed the histograms
become broader and more dissimilar and no longer overlap with each other . ] a particular feature of a gaussian work distribution given by eq .
( [ gauss ] ) , such as the one for the pulling process of a gaussian chain , is the mirror - symmetry of the forward and backward distributions with respect to @xmath15 relating @xmath21 and @xmath25 . as a consequence ,
the dissipated works are equal for the forward and the backward processes , and , moreover , the bennett estimator is unbiased ( see fig .
in order to demonstrate the dependence of the bennett estimator on the difference of the forward and backward dissipated works , we thus need to consider an asymmetric work pdf .
as shown in the first experimental realization of the jarzynski equality @xcite , pulling a hairpin molecule typically exhibits pronounced asymmetric work pdfs , depending upon the direction of the protocol .
therefore , we consider a chain in three dimensions consisting of @xmath189 monomers with hookean bonds along the chain and , additionally , pair - specific interactions leading to a hairpin structure in the mechanical equilibrium state , as depicted in fig 7 .
the potential energy of this chain is given by @xmath232 , \end{split}\ ] ] where @xmath233 denotes the position of the @xmath131-th monomer , and @xmath234 the distance between two monomers .
the first sum represents the contribution of the hookean springs connecting neighboring monomers along the chain . here
@xmath235 is the equilibrium bond length and @xmath236 the spring - constant .
the second sum describes the interaction between pairs of monomers @xmath131 and @xmath219 , given by the lennard - jones potential .
the summation index runs over the monomers constituting the pairs as depicted in fig 7 .
the total number of pairs is @xmath221 . as for the gaussian case , additionally , strong friction and random forces act on the beads of the chain resulting in an overdamped dynamics which is governed by the langevin equation @xmath237 where @xmath145 denotes the friction constant experienced by a monomer , @xmath238 the three dimensional gradient with respect to the position of the @xmath131th monomer , and @xmath239 thermal gaussian white noise forces satisfying @xmath240 and @xmath241 .
similarly as for the gaussian chain , the one end of a hairpin chain is fixed at @xmath242 and the other end @xmath243 is pulled with constant speed @xmath244 in the @xmath245-direction . as for the gaussian chain ,
the work performed in this way is given by @xmath246 with the hairpin potential , eq .
( [ hairpin ] ) . unlike the case of pulling a gaussian chain , for a pulled hairpin chain
, we do not know the exact work distribution from which we could sample the data . in order to study the effect of an asymmetric work pdf on the biases
, we had to perform direct simulations of the langevin dynamics .
for the numerical simulations , we rescaled all lengths by the bond length according to @xmath247 and discretized the langevin equation with a time step @xmath248 .
the iterative langevin equation then reads in terms of the discrete time variable @xmath249 as @xmath250 where @xmath251 is the rescaled potential , and @xmath252 = @xmath253 denotes the rescaled mobility with @xmath254 being a characteristic time - scale of the chain .
the variances of the dimensionless gaussian random forces @xmath255 are determined by @xmath256 .
the dimensionless potential @xmath257 is determined by the dimensionless spring constant @xmath258 and the dimensionless energy parameter of the lennard - jones potential @xmath259 . in the simulations these parameters
were set as @xmath223 and @xmath224 . for the efficiency as well as the numerical convergence of the simulations ,
the rescaled mobility was chosen as @xmath260 .
we considered a hairpin - like molecule with n=20 monomers having pairing potentials between the monomer pairs @xmath261 up to @xmath262 .
before the forward protocol was started , the chain had been prepared into a thermal equilibrium state having the form of a hairpin with constrained positions of the first and last monomers , @xmath242 , @xmath263 , respectively , where @xmath264 denotes the unit vector in @xmath245-direction .
this thermal equilibrium state was established by simulating the langevin equation ( [ dle ] ) for a sufficiently large time with clamped end positions .
upon equilibration , the last monomer was pulled at a constant speed @xmath244 in the @xmath245-direction until it had reached the distance @xmath265 .
the backward protocol was started with a thermal equilibrium distribution with the first and last monomer at the final positions of the forward protocol ; then the chain was compressed by moving the last monomer at the same absolute velocity @xmath244 in the @xmath266-direction until it had reached the initial position of the forward protocol , @xmath267 .
the histograms displayed in fig .
8 represent averages of @xmath226 raw histograms each based on @xmath215 simulations of the langevin equation for forward and backward protocols at two different pulling speeds .
the statistical uncertainty of these averages were estimated from the variance of the distributions of raw histograms and are indicated by error bars . for the slow protocol displayed in the upper panel , the forward and the backward histograms cross at @xmath268 indicated by the vertical dotted line , and appears as almost symmetric about the crossing point .
in contrast , for the fast protocol shown in the panel ( b ) , the two histograms do not overlap , and therefore , do not allow to extract @xmath15 from the crooks relation .
moreover , these histograms are no longer mirror symmetric ; the forward histogram is significantly more dispersed than the backward histogram .
we also present the biases of the free energy estimators in fig .
as the protocol speeds up , the bias of the jarzynski estimators ( open squares ) becomes more pronounced and , at the same time , the dissipated work ( triangles ) grows .
we note that the 1/2-formula ( open diamonds ) given by eq .
( [ half ] ) perfectly coincides with the bennett estimation ( filled red triangles ) .
both methods lead to a significantly smaller bias than the jarzynski estimate .
unlike the gaussian case , the dissipated work is larger during the forward than during the backward process ; see the curve for @xmath269 in comparison with the curve for @xmath270 in fig .
9(a ) . figure 9(b ) presents the bias of the bennett estimator as a function of the difference between these dissipated works .
the monotonic increase of both quantities as functions of the time - ratio @xmath271 leads to a proportionality between the bias of the bennett estimator and the difference between the dissipated works in the forward and backward protocols confirming the previous conjecture .
the influence of the difference between the dissipated forward and backward works on the bennett estimator can be understood qualitatively , at least in the limit of large dissipation . according to eq .
( [ half ] ) , in this limit , the bennett estimator can be expressed by half of the difference of the jarzynski estimators for the forward and the backward protocol . for a gaussian work distribution the forward and the backward distributions @xmath21 and @xmath25 are symmetric with respect to @xmath15 and consequently
, the biases of the two estimators are identical and compensate each other in the 1/2 formula . on the other hand , in all cases with different dissipated works of the forward and the backward processes ,
the symmetry of the forward and backward distributions is apparently lost and therefore the forward and the backward jarzynski estimators will have different biases , which then no longer compensate each other in the 1/2 formula . .
the statistical error bars are smaller than the size of symbols . with increasing speed
, the jarzynski bias and the dissipated work of the forward and the backward process grow fast . at the same time but at a much slower rate also the difference between the forward and the backward dissipated work as well as the bias of the bennett estimator increase .
( b ) the bennett bias is shown to be proportional to the difference between dissipated works , @xmath272 in the forward process and @xmath273 in the backward process . ]
in summary , we discussed the statistical behaviors of the free energy estimators based on the jarzynski equality , crooks crossing criterion , and bennett s acceptance ratio method , in relation to the amount of dissipated work and the time asymmetry .
in particular , we investigated the limiting behaviors of the solutions of eq .
( [ bennett ] ) and demonstrated that while both the jarzynski and the crooks method are hampered by large dissipated work , the finite sampling error of bennett s method is determined by the _ difference _ of the dissipated works of the forward and the backward processes . as a consequence , it is less severely influenced by the entropy production but rather by the asymmetry between the forward and the backward process .
the examples of a gaussian and a non - gaussian chain considered here demonstrate these features .
the finite sampling error of the jarzynski estimator rapidly increases with the amount of dissipated work .
the crooks estimator has a binary character : either it yields a reliable solution , as long as the forward and the backward work pdfs overlap , or provides no solution at all . at best an upper and a lower bound
can be estimated from the extension of the gap between @xmath21 and @xmath25 . for gaussian work
pdfs the bennett method leads to precise estimates over a remarkably wide range of pulling speeds and sampling sizes .
for non - gaussian work pdfs , the bias of the bennett estimate is still smaller than for other estimators .
it , however , increases with increasing differences between the forward and backward dissipated works .
finally we like to emphasize that in all cases where the jarzynski and crooks estimators fail , the determination of @xmath15 can be substantially simplified by means of the 1/2-formula , eq .
( [ half ] ) , expressing the free energy difference as half of the difference of the forward and backward jarzynski estimates .
y.w.k . acknowledges support from basic science research program through the national research foundation of korea ( nrf ) funded by the ministry of education , science and technology(grant no .
2010 - 0025196 ) .
j.y . acknowledges support from the national research foundation of korea ( nrf ) grant funded by the korea government ( mest ) ( grant no .
2011 - 0021296 ) . c. jarzynski , phys .
lett . * 78 * , 2690 ( 1997 ) .
j. liphardt , s. dumont , s. b. smith , i. tinoco jr .
, c. bustanamte , science * 296 * , 1832 ( 2002 )
. n. c. harris , y. song and c .- h .
kiang , phys .
lett . * 99 * , 068101 ( 2007 ) .
f. douarche , s. ciliberto , a. petrosyan , and i. rabbiosi , euro .
* 70 * , 593 ( 2005 ) .
v. blickle , t. speck , l. helden , u. seifert , and c. bechinger , phys .
. lett . * 96 * , 070603 ( 2006 ) . c. jarzynski , phys .
e * 73 * , 046105 ( 2006 ) .
r. f. fox , proc .
usa * 100 * , 12537 ( 2003 ) .
j. gore , f. ritort , and c. bustamante , proc .
usa * 100 * , 12564 ( 2003 ) .
m. palassini and f. ritort , phys .
lett . * 107 * , 060601 ( 2011 ) .
d. m. zuckerman and t. b. woolf , phys .
* 89 * , 180602 ( 2002 ) .
d. m. zuckerman and t. b. woolf , j. stat .
phys . * 114 * , 1303 ( 2004 ) .
g. e. crooks , phys .
e * 60 * , 2721 ( 1999 ) j. m. r parrondo , c. van den broeck and r. kawai , new j. phys . *
11 * , 073008 ( 2009 ) . c. h. bennett , j. comput .
* 22 * , 245 ( 1976 ) .
m. r. shirts , e. bair , g. hooker , and v. s. pande , phys .
lett . * 91 * , 140601 ( 2003 ) .
a. pohorille , c. jarzynski , and c. chipot , j. phys .
b * 114 * , 10235 ( 2010 ) .
f. ytreberg , r. h. swendsen , and d. m. zuckerman , j. chem .
phys . * 125 * , 184114 ( 2006 ) .
n. lu , j. k. singh , and d. a. kofke , j. chem .
* 118 * , 2977 ( 2003 ) .
s. s. n. chari , k. p. n. murthy , and r. inguva , phys .
e * 85 * , 041117 ( 2012 ) .
e. h. feng and g. e. crooks , phys .
* 101 * , 090602 ( 2008 ) . c. jarzynski , annu .
rev . condens .
matter phys .
* 2 * , 329 ( 2011 ) .
m. campisi , p. hnggi , entropy * 13 * , 2024 ( 2011 ) . c. jarzynski , phys .
e * 56 * , 5018 ( 1997 ) .
a. dhar , phys .
e * 71 * , 036126 ( 2005 ) . | the estimate of free energy changes based on bennett s acceptance ratio method is examined in several limiting cases and compared with other estimates based on the jarzynski equality and on the crooks relation . while the absolute amount of dissipated work , defined as the surplus of average work over the free energy difference , limits the practical applicability of jarzynski s and crooks methods , the reliability of bennett s approach is restricted by the difference of the dissipated works in the forward and the backward process .
we illustrate these points by considering a gaussian chain and a hairpin chain which both are extended during the forward and accordingly compressed during the backward protocol .
the reliability of the crooks relation predominantly depends on the sample size ; for the jarzynski estimator the slowness of the work protocol is crucial , and the bennett method is shown to give precise estimates irrespective of the pulling speed and sample size as long as the dissipated works are the same for the forward and the backward process as it is the case for gaussian work distributions . with an increasing dissipated work difference the bennett estimator also acquires a bias which increases roughly in proportion to this difference .
a substantial simplification of the bennett estimator is provided by the 1/2-formula which expresses the free energy difference by the algebraic average of the jarzynski estimates for the forward and the backward processes .
it agrees with the bennett estimate in all cases when the jarzynski and the crooks estimates fail to give reliable results . |
cosmological models with a cosmological term @xmath8 are currently serious candidates to describe the dynamics of our four dimensional universe .
the history of cosmological term dates back to einstein , and its original role was to allow static homogeneous solutions to einstein s equations in the presence of matter which turned out to be unnecessary when the expansion of the universe was discovered .
however , particle physicists then realized that the non - vanishing cosmological constant can be interpreted as a measure of the energy density of the vacuum which turned out to be the sum of a number of apparently disjoint contributions of quantum fields . in fact , a dynamical characteristic for the vacuum energy density ( cosmological term ) was attributed by quantum field theorists since the developments in particle physics and inflationary scenarios . according to modern quantum field theory
, the structure of a vacuum turns out to be interrelated with some spontaneous symmetry - breaking effects through the condensation of quantum ( scalar ) fields .
this phenomenon gives rise to a non - vanishing vacuum energy density of the form @xmath9 .
therefore , the observed ( or effective ) cosmological term receives an extra contribution from @xmath10 as follows : @xmath11 where @xmath12 is the bare cosmological constant and @xmath13 is the gravitational constant . from quantum field theory
we may expect @xmath14 ( @xmath15 is the planck mass ) , or another energy scale related to some spontaneous symmetry breaking effect such as @xmath16 or @xmath17 .
therefore , the bare cosmological constant receives potential contributions from these mass scales resulting in a large effective cosmological term .
however , the experimental upper bound on the present value of the cosmological term , @xmath8 , provided by measurements of the hubble constant , @xmath7 , reads numerically as @xmath18 which is too far from the expectation of quantum field theory .
the question of why the observed vacuum energy is so small in comparison to the scales of particle physics is known as the cosmological constant problem .
it is generally thought to be easier to imagine an unknown mechanism which would set @xmath8 exactly to zero than one which would suppress it by just the right amount to yield an observationally tiny cosmological constant . if @xmath8 is a dynamical variable ( or vacuum parameter ) , then it is natural to suppose that in an expanding universe the cosmological term relaxes to the present tiny value by some relaxation mechanism which may be provided by a time - varying vacuum with a rolling scalar field @xcite .
there are still other possibilities to be advocated . in recent years
, several attempts in these directions have been done , in the context of quantum cosmology @xcite .
one plausible explanation for a tiny cosmological term is to suppose that @xmath8 is dynamically evolving and not constant , i.e. , @xmath19 , where @xmath2 is the scale factor of the universe and @xmath20 is a parameter .
so , as the universe expands from its small size in the early universe , the initially large effective cosmological term evolves and reduces to its present small value@xcite .
the study of @xmath8-decaying cosmological models has recently been the subject of particular interest both from classical and quantum aspects .
the @xmath8 decaying models may serve as potential candidates to solve this problem by decaying the large value of the cosmological constant @xmath8 to its present observed value . also , there are strong ( astronomical ) observational motivations for considering cosmological models in which @xmath8 is dynamically decreasing as @xmath21 .
some models assume _ a priori _ a fixed value for the parameter @xmath20 .
the case @xmath6 , corresponding to the cosmic string matter , has mostly been taken based on dimensional considerations by some authors @xcite .
the case @xmath22 which resembles the ordinary radiation has also been considered by some other authors @xcite .
a third group of authors have also studied the case @xmath23 corresponding to the ordinary matter @xcite .
there are also some other models in which the value of @xmath20 is not fixed _ a priori _ and the numerical bounds on the value of @xmath20 is estimated by observational data or obtained by calculation of the quantum tunnelling rate @xcite
. other aspects of @xmath8-decaying models have also been discussed with no specific numerical bounds on @xmath20 @xcite .
it is clear that the functional dependence @xmath21 is phenomenological and does not result from the first principles of particle physics .
however , for some domain for example , @xmath24 , the decaying law @xmath25 deserves further investigation .
one important reason is that the age of the universe , in these models , is always larger than the age obtained in the standard einstein - de sitter cosmology , or the one we get in an open universe .
therefore , if we are interested in solving the age problem , the decaying @xmath8 term appears to be a good candidate .
in fact , according to the ansatz @xmath21 , one may suppose the natural value @xmath26 to be the value of @xmath8 at the planck time when @xmath2 was of the order of the planck length .
theoretically this ansatz does not directly solve the cosmological constant problem , but it relates this problem to the age problem of why our universe is so old and have a radius @xmath2 much larger than the planck length . in other words ,
this ansatz reduces two above problems to one problem of `` why our universe could have escaped the death at the planck time '' , which seems to be the most natural fate of a baby - universe in quantum cosmology ?
one may assume that the value of @xmath8 in the early universe might have been much bigger than its present value and large enough to drive some symmetry breakings which might have occurred in the early universe .
on the other hand , the idea that our 4-dimensional universe might have emerged from a higher dimensional space - time is now receiving much attention @xcite where the compactification of higher dimensions plays a key role .
however , the question of how and why this compactification occurs remains as an open problem . from string theory we know that the compactification may take place provided that the higher dimensional manifold admits special properties , namely if the geometry of the manifold allows , for example , the existence of suitable killing vectors .
however , it is difficult to understand why such manifolds are preferred and whether other possible mechanisms for compactification do exist . in cosmology , on the other hand , different kinds of compactifications could be considered .
for example , in an approach , called _ dynamical compactification _
, the extra dimensions evolve in time towards very small sizes and the extra - dimensional universe reduces to an effective four - dimensional one .
this type of compactification was considered in the context of modern kaluza - klein theories @xcite .
it is then a natural question that how an effective four dimensional universe evolve in time and whether the resulting cosmology is similar to the standard friedmann - robertson - walker four dimensional universe without extra dimensions .
meanwhile , the recent distance measurements of type ia _ supernova _
suggest strongly an accelerating universe @xcite .
this accelerating expansion is generally believed to be driven by an energy source called _ dark energy _ which provides negative pressure , such as a positive cosmological constant @xcite , or a slowly evolving real scalar field called _ quintessence _
moreover , the basic conclusion from all previous observations that @xmath27 70 percent of the energy density of the universe is in a dark energy sector , has been confirmed after the recent wmap @xcite . to model a universe based on these considerations
one may start from a fundamental theory including both gravity and standard model of particle physics . in this
regards , it is interesting to begin with ten or eleven - dimensional space - time of superstring / m - theory , in which case one needs a compactification of ten or eleven - dimensional supergravity theory where an effective 4-dimensional cosmology undergoes acceleration .
however , it has been known for some time that it is difficult to derive such a cosmology and has been considered that there is a no - go theorem that excludes such a possibility , if one takes the internal space to be time - independent and compact without boundary @xcite .
however , it has recently been shown that one may avoid this no - go theorem by giving up the condition of time - independence of the internal space ; and a solution of the vacuum einstein equations with compact hyperbolic internal space has been proposed based on this model .
similar accelerating cosmologies can also be obtained for sm2 and sd2 branes , not only for hyperbolic but also for flat internal space .
on the other hand , from cosmological point of view , it is not so difficult to find cosmological models in which the 4-dimensional universe undergoes an accelerating expansion and the internal space contracts with time , exhibiting the _ dynamical compactification _ @xcite , @xcite , @xcite . in @xcite , for instance , it is shown that using a more general metric , as compared to ref.@xcite , and introducing matter without specifying its nature , the size of compact space evolves as an inverse power of the radius of the universe .
the friedmann - robertson - walker equations of the standard four - dimensional cosmology is obtained using an effective pressure expressed in terms of the components of the higher dimensional energy - momentum tensor , and the negative value of this pressure may explain the acceleration of our present universe . to the author s knowledge the question of @xmath8-decaying cosmological model has not received much attention in higher dimensional kaluza - klein cosmologies .
moreover , the exotic matter has not been considered as an alternative candidate to produce the acceleration of the universe .
the purpose of the present chapter is to study a @xmath0-dimensional kaluza - klein cosmology , with an extended robertson - walker type metric , in this context @xcite . as we are concerned with cosmological solutions , which are intrinsically time dependent
, we may suppose that the internal space is also time dependent .
it is shown that by taking this higher dimensional metric and introducing a 4-dimensional exotic matter , a decaying cosmological term @xmath28 with @xmath29 is appeared as a type of dark energy , and for the case @xmath6 the resulting field equations yield the exponential solutions for the scale factors of the four - dimensional universe and the internal space .
these solutions may account for the accelerating universe and dynamical compactification of extra dimensions , driven by the negative pressure of the exotic matter .
remaining spatial dimensions . ]
it should be noted , however , that the solutions in principle describe typical inflation rather than the recently observed acceleration of the universe which is known to take place in an ordinary matter dominated universe .
nevertheless , regarding the fact that about 70 percent of the total energy density of the universe is of dark energy type with negative pressure , we may approximate the matter content of the universe with almost dark energy and consider the present model as a rather simplified model of a real accelerating universe . the quantum cosmology of this model is also studied by obtaining the wheeler - dewitt equation and finding its general exact solutions .
it is then shown that a good correspondence exists between the classical and quantum cosmological solutions , based on the interpretation of hartle of the classical limits of quantum cosmology .
the chapter is organized as follows : in section * 2 * , we introduce the classical cosmology model by taking a higher dimensional robertson - walker type metric and a higher dimensional matter whose non - zero part is a four - dimensional exotic matter . in section * 3 * , we obtain the einstein equations for the two scale factors . in section * 4 * , we solve the einstein equations and obtain the solutions . in section *
5 * , we study the corresponding quantum cosmology and derive the wheeler - dewitt equation . in section * 6 * , the exact solutions of the wheeler - dewitt equation is obtained .
finally , in section * 7 * , we show a good correspondence between the classical and quantum cosmology .
the chapter is ended with concluding remarks .
to begin with , we study the metric considered in @xcite in which the space - time is assumed to be of robertson - walker type having a ( 3 + 1)-dimensional space - time part and an internal space with dimension @xmath3 .
we adopt a real chart @xmath30 with @xmath31 , @xmath32 , and @xmath33 denoting the time , space coordinates and internal space dimensions , respectively .
we , therefore , take instead of taking @xmath34 . ]
@xmath35 where @xmath36 is the lapse function , @xmath37 and @xmath38 are the scale factor of the universe and the radius of internal space , respectively ; @xmath39 , and @xmath40 , reflecting flat , open or closed type of four - dimensional universe and @xmath3-dimensional space .
we assume the internal space to be flat with compact topology @xmath41 , which means @xmath42 .
this assumption is motivated by the possibility of the compact spaces to be flat or hyperbolic in `` _ accelerating cosmologies from compactification _ '' scenarios , as discussed in introduction .
the form of energy - momentum tensor is dictated by einstein s equations and by the symmetries of the metric ( [ 1 ] ) .
therefore , we may assume @xmath43 where @xmath44 and @xmath45 run over both the space - time coordinates and the internal space dimensions .
now , we examine the case for which the pressure along all the extra dimensions vanishes , namely @xmath46 . in
so doing , we are motivated by the _
brane world _
scenarios where the matter is to be confined to the 4-dimensional universe , so that all components of @xmath47 is set to zero but the space - time components and it means no matter escapes through the extra dimensions .
we assume the energy - momentum tensor @xmath48 of space - time to be an exotic @xmath49 fluid with the equation of state @xmath50 where @xmath51 and @xmath52 are the pressure and density of the fluid , respectively and the parameter @xmath20 is restricted to the range @xmath29 @xcite .
it is worth noting that the equation of state ( [ 4 ] ) with @xmath29 resembles a universe with negative pressure matter , violating the strong energy condition @xcite and this violation is required for a universe to be accelerated .. ] using standard techniques we obtain the scalar curvature corresponding to the metric ( [ 1 ] ) @xmath53 and then substitute it into the dimensionally extended einstein - hilbert action ( without higher dimensional cosmological term ) plus a matter term indicating the above mentioned exotic fluid .
this leads to the effective lagrangian ] @xmath54 where a dot represents differentiation with respect to @xmath31 .
we now take a closed @xmath55 universe .
although the flat universe @xmath56 is almost favored by observations , we will show an equivalence between @xmath55 and @xmath56 universes .
one may obtain the continuity equation by using the contracted bianchi identity in ( 4+@xmath3 ) dimensions , namely @xmath57 together with the assumption that the matter is confined to ( 3 + 1)-dimensional space - time as @xmath58 which gives rise to @xmath59 or @xmath60 it is easily shown that substituting the equation of state ( [ 4 ] ) into the continuity equation ( [ 6 ] ) leads to the following behavior of the energy density in a closed ( @xmath61 ) friedmann - robertson - walker universe @xcite @xmath62 where @xmath63 is the value of the scale factor at an arbitrary reference time @xmath64 .
now , if we believe that the cosmological term plays the role of vacuum energy density , we may define the cosmological term @xcite @xmath65 which leads to @xmath66 where the cosmological term is now decaying with the scale factor@xmath2 as @xmath67 note that @xmath8 is now playing the role of an evolving dark energy @xcite in 4-dimensions , because we did not consider explicitly a @xmath0 dimensional cosmological term in the action , and @xmath8 appears merely due to the specific choice of the equation of state ( [ 4 ] ) for the exotic matter .
the decaying @xmath8 term may also explain the smallness of the present value of the cosmological constant since as the universe evolves from its small to large sizes the large initial value of @xmath8 decays to small values .
this phenomenon may somehow alleviate the cosmological constant problem . of particular interest , to us , among the different values of @xmath20
is @xmath6 which has some interesting implications in reconciling observations with inflationary models @xcite , and is consistent with quantum tunnelling @xcite .
we take @xmath6 and set the initial values of @xmath63 and @xmath68 as @xmath69 leading to a positive cosmological term which , according to ( [ 8 ] ) , guarantees the weak energy condition @xmath70 .
the lapse function @xmath36 , in principle , is also an arbitrary function of time due to the fact that einstein s general relativity is a reparametrization invariant theory .
we , therefore , take the gauge @xmath71 now , the lagrangian becomes @xmath72 where eq.([11 ] ) has been used .
it is seen that the parameters @xmath73 and @xmath8 are effectively removed from the lagrangian and this implies that although @xmath73 and @xmath8 are not zero in this model the corresponding 4-dimensional universe is equivalent to a flat universe with a zero cosmological term . in other words , we do not distinguish between our familiar 4-dimensional universe , which seems to be flat and without any exotic fluid , and a closed universe filled with an exotic fluid .
+ we now define the new variables @xmath74 the lagrangian ( [ 13 ] ) is written as @xmath75 the equations of motion are obtained @xmath76 @xmath77 combining the equations ( [ 16 ] ) and ( [ 17 ] ) we obtain @xmath78 @xmath79
the solutions for @xmath80 and @xmath81 in eqs .
( [ 18 ] ) and ( [ 19 ] ) are obtained @xmath82 @xmath83 and the solutions for @xmath37 and @xmath38 are then as follows @xmath84 @xmath85 where the constants `` @xmath86 , @xmath87 , @xmath88 and @xmath89 '' or `` @xmath44 , @xmath45 , @xmath90 and @xmath91 '' should be obtained , in principle , in terms of the initial conditions .
it is a reasonable assumption that the size of all spatial dimensions be the same at @xmath92 .
moreover , it may be assumed that this size would be the planck size @xmath93 in accordance with quantum cosmological considerations .
therefore , we take @xmath94 so that @xmath95 , and @xmath96 @xmath97 it is important to note that the constants @xmath98 are not independent , and a relation may be obtained between them .
this is done by imposing the zero energy condition @xmath99 which is the well - known result in cosmology due to the existence of arbitrary laps function @xmath36 in the theory .
the hamiltonian constraint is obtained through the legender transformation of the lagrangian ( [ 15 ] ) @xmath100 which is written in terms of @xmath90 and @xmath91 as @xmath101 this constraint is satisfied only for @xmath102 or @xmath103 . for @xmath104 , the case @xmath105 or @xmath106 gives rise to time independent scale factors , namely @xmath107 , which is not physically viable since we know , at least based on observations , the scale factor of the universe is time dependent .
we , therefore , choose @xmath108 so that the universe and the internal space would expand and contract , respectively , in accordance with the present observations .
for the case @xmath109 , we find @xmath110 the former is not physically viable , since it predicts no time evolution for the universe .
the latter , however , may predict exponential expansion for @xmath37 , and exponential contraction for @xmath38 , both with the same exponent @xmath111 . for the general case @xmath112 ,
we find @xmath113 , \label{27}\ ] ] which gives two positive values for @xmath90 indicating two possible expanding universes provided @xmath114 which indicates the compactification of extra dimensions . moreover , the values of @xmath115 , for a given negative value of @xmath91 , become larger for higher dimensions .
therefore , the universe expands more rapidly in both possibilities . on the contrary , for a given positive value of @xmath90 , indicating an expanding universe , the parameter @xmath91 may take two negative values @xmath116^{-1},\ ] ] indicating two ways of compactification .
moreover , they become smaller for higher dimensions , exhibiting lower rates of compactification . to find the constants
@xmath98 we first obtain the hubble parameter for @xmath37 @xmath117 by which the constant @xmath90 is fixed .
the observed positive value of @xmath7 will then justify our previous assumption , @xmath111 .
we may , therefore , write the solutions ( [ 22 ] ) and ( [ 23 ] ) in terms of the hubble parameter @xmath7 as @xmath118 @xmath119 for @xmath109 , and @xmath120 @xmath121^{-1 } } , \label{32}\ ] ] and @xmath122 } , \label{0}\ ] ] @xmath123 for @xmath112 . for a given @xmath124 , it is seen that the solution corresponding to @xmath109 may predict an accelerating ( de sitter ) universe and a contracting internal space with exactly the same rates . for @xmath112 , in eqs.([31 ] ) and ( [ 32 ] ) , for a given @xmath124 in the exponent of @xmath37 the exponent in @xmath38 takes two negative values and becomes smaller for higher dimensions .
this means that while the 4-dimensional ( de sitter ) universe is expanding by the rate @xmath7 , the higher dimensions may be compactified in two possible ways with different rates of compactification as a function of dimension , @xmath3 . in eqs.([0 ] ) and ( [ 00 ] ) , on the other hand , for a given @xmath114 the exponent in @xmath37 takes two positive values which become larger for higher dimensions .
this also means that while the extra dimensions contract by the rate @xmath91 , the universe may be expanded in two possible ways with different expansion rates as a function of @xmath3 .
it is easy to show that the lagrangian ( [ 15 ] ) ( or the equations of motion ) is invariant under the simultaneous transformation @xmath125 which is consistent with the time reversal @xmath126 .
therefore , four different phases of `` _ expansion - contraction _ '' for @xmath37 and @xmath38 are distinguished , eqs.([31 ] ) - ( [ 00 ] ) .
one may prefer the `` _ expanding @xmath37 - contracting @xmath38 _ '' phase to `` _ expanding @xmath38 - contracting @xmath37 _ '' one , considering the present status of the 4d universe .
, both the lagrangian ( [ 15 ] ) and the hamiltonian constraint ( [ 24 ] ) are invariant under the transformation @xmath127 therefore , we have a dynamical symmetry between @xmath2 and @xmath1 , namely @xmath128 in this case there is no real line of demarcation between @xmath1 and @xmath2 to single out one of them as the real scale factor of the universe .
this is because the internal space is flat @xmath129 and according to ( [ 13 ] ) one may assume the 4d universe with @xmath130 to be equivalent to the one in which @xmath131 .
therefore , both have the same topology @xmath132 . ]
the deceleration parameter @xmath133 for the scale factor @xmath2 is obtained @xmath134 observational evidences not only do not rule out the negative deceleration parameter but also puts the limits on the present value of @xmath133 as @xmath135 @xcite .
therefore , this negative value seems to favor a cosmic acceleration in the expansion of the universe . in the expansion phase of the closed ( @xmath61 ) universe
the cosmological term @xmath8 decays exponentially with time @xmath31 as @xmath136 whereas in the contraction phase ( @xmath137 ) it grows exponentially to large values so that at @xmath92 it becomes extremely large , of the order of @xmath138 .
this huge value of @xmath8 may be extinguished rapidly by assuming a sufficiently large hubble parameter @xmath7 , consistent with the present observations , to alleviate the cosmological constant problem .
an appropriate quantum mechanical description of the universe is likely to be afforded by quantum cosmology which was introduced and developed by dewitt @xcite . in quantum cosmology the universe , as a whole , is treated quantum mechanically and is described by a single wave function , @xmath139 , defined on a manifold ( _ superapace _ ) of all possible three geometries and all matter field configurations .
the wave function @xmath139 has no explicit time dependence due to the fact that there is no a real time parameter external to the universe .
therefore , there is no schr@xmath140dinger wave equation but the operator version of the hamiltonian constraint of the dirac canonical quantization procedure @xcite , namely vanishing of the variation of the einstein - hilbert action @xmath141 with respect to the arbitrary lapse function @xmath142 @xmath143 which is written @xmath144 this equation is known as the wheeler - dewitt ( wdw ) equation .
the goal of quantum cosmology by solving the wdw equation is to understand the origin and evolution of the universe , quantum mechanically . as a differential equation
, the wdw equation has an infinite number of solutions . to get a unique viable solution , we should also respect the question of boundary condition in quantum cosmology which is of prime importance in obtaining the relevant solutions for the wdw equation . in principle
, it is very difficult to solve the wdw equation in the _ superspace _ due to the large number of degrees of freedom . in practice
, one has to _ freeze out _ of all but a finite number of degrees of freedom of the gravitational and matter fields .
this procedure is known as quantization in _
minisuperspace _ , and will be used in the following discussion .
the minisuperspace in our model is two - dimensional with gravitational variables @xmath80 and @xmath81 . to obtain the wheeler - dewitt equation , in this minisuperspace
, we start with the lagrangian ( [ 15 ] ) .
the conjugate momenta corresponding to @xmath80 and @xmath81 are obtained @xmath145 @xmath146 from which we obtain @xmath147 , \label{34'}\ ] ] @xmath148 .
\label{35}\ ] ] substituting eqs.([34 ] ) , ( [ 35 ] ) into the hamiltonian constraint ( [ 24 ] ) , we obtain @xmath149 now , we may use the following quantum mechanical replacements @xmath150 by which the wheeler - dewitt equation is obtained @xmath151\psi(x , y)=0 , \label{37}\ ] ] where @xmath152 is the wave function of the universe in the @xmath153 mini - superspace .
we introduce the following change of variables @xmath154 by which the wheeler - dewitt equation takes a simple form @xmath155 now , we can separate the variables as @xmath156 to obtain the following equations @xmath157 @xmath158 where we assume @xmath159 .
the solutions of eqs.([40 ] ) , ( [ 41 ] ) in terms of @xmath160 are as follows @xmath161 @xmath162 leading to the four possible solutions for @xmath163 as @xmath164 @xmath165 or alternative solutions in terms of @xmath166 as @xmath167 @xmath168 where @xmath169 are the normalization constants .
we may also write down the solutions in terms of @xmath2 and @xmath1 , there is a exchange symmetry @xmath170 under the exchange @xmath171 . ]
@xmath172 @xmath173 it is now important to impose the _ good _ boundary conditions on the above solutions to single out the physical ones . in so doing , we may impose the following condition @xmath174 which requires the wave function of the universe to be normalizable .
this means that our minisuperspace model has no classical solutions that expand simultaneously to infinite values of @xmath1 and @xmath2 , as eqs.([30])-([00 ] ) show .
then , one may take the following solutions @xmath175 where @xmath176 are the normalization constants and the exponents of @xmath2 and @xmath1 are negative for any value of @xmath3 .
, the exponent of `` @xmath1 '' corresponding to @xmath177 becomes zero so that @xmath177 depends only on @xmath2 with the condition @xmath178 . ]
one may obtain the solutions ( [ 49 ] ) in @xmath153 mini - superspace as @xmath179
one of the most interesting topics in the context of quantum cosmology is the mechanisms through which the classical cosmology may emerge from quantum theory . when does a wheeler - dewitt wave function predict a classical space - time ?
quantum cosmology is the quantum mechanics of an isolated system ( universe ) .
it is not possible to use the copenhagen interpretation , which needs the existence of an external observer , since here the observer is part of the system . indeed ,
any attempt in constructing a viable quantum gravity requires understanding the connections between classical and quantum physics .
much work has been done in this direction over the past decade .
actually , there is some tendency towards using semiclassical approximations in dividing the behaviour of the wave function into two types , oscillatory or exponential which are supposed to correspond to classically allowed or forbidden regions .
hartle @xcite has put forward a simple rule for applying quantum mechanics to a single system ( universe ) : _ if the wave function is sufficiently peaked about some region in the configuration space we predict to observe a correlation between the observables which characterize this region_. halliwell @xcite has shown that the oscillatory semiclassical wkb wave function is peaked about a region of the _ minisuperspace _ in which the correlation between the coordinate and momentum holds good and stresses that both _ correlation _ and _ decoherence _ are necessary before one can say a system is classical . using wigner functions , habib and laflamme @xcite
have studied the mutual compatibility of these requirements and shown that some form of coarse graining is necessary for classical prediction from wkb wave functions . alternatively , gaussian or coherent states with sharply peaked wave functions are often used to obtain classical limits by constructing wave packets . in the investigation of classical limits
, we first take @xmath109 and look for a correspondence between classical and quantum solutions . using eqs.([30 ] ) and ( [ 30 ] ) in the planck units , the corresponding classical locus in @xmath180 configuration space , is @xmath181 whereas in @xmath153 coordinates we have @xmath182 we now consider the wave functions ( [ 50 ] ) in @xmath153 mini - superspace for @xmath109 @xmath183 @xmath184 the above wave functions , in their present form , are not square integrable as is required for the wave functions to predict the classical limit .
however , one may take the absolute value of the exponents to make the wave functions square integrable @xmath185 @xmath186 we next consider the general case @xmath112 . eliminating the parameter @xmath31 in eqs .
( [ 31 ] ) and ( [ 32 ] ) the classical loci in terms of @xmath187 are obtained @xmath188^{-1 } } .
\label{57}\ ] ] the corresponding forms of these loci in terms of @xmath166 are @xmath189^{-1 } , \label{58}\ ] ] @xmath190^{-1}. \label{59}\ ] ] the wave functions ( [ 50 ] ) also are not square integrable , so we may replace the exponents by their absolute values @xmath191 to make them square integrable .
now , following hartle s point of view , we try to make correspondence between the classical loci and the wave functions .
figures 1 - 6 show respectively the 2d plots of the typical wave functions @xmath192 - @xmath193 in terms of @xmath153 for @xmath194 ; figures 23 - 28 show the corresponding 3d plots , respectively . on the other hand ,
figures 12 - 17 show the classical loci corresponding to @xmath195 , respectively .
it is seen that the 2d and 3d plots of the wave functions @xmath192 - @xmath193 are exactly peaked on the classical loci . in the same way ,
figures 7 - 11 show respectively the 2d plots of the wave functions @xmath196 - @xmath197 .
figures 29 - 33 show the corresponding 3d plots , respectively .
figures 18 - 22 show the classical loci for @xmath198 , respectively .
again , an exact correspondence is seen between the 2d and 3d plots of the wave functions @xmath196 - @xmath197 and the classical loci .
this procedure will apply for all @xmath3 .
first , we have studied a @xmath0-dimensional classical kaluza - klein cosmology with a robertson - walker type metric having two scale factors , @xmath2 for the universe and @xmath1 for the higher dimensional space . by introducing a typical exotic matter with the equation of state @xmath199 in 4-dimensions ,
a decaying cosmological term is obtained effectively as @xmath4 . by taking @xmath6 ,
the corresponding einstein field equations are obtained and we find exponential solutions for @xmath2 and @xmath1 in terms of the hubble parameter @xmath7 .
these exponential solutions indicate the accelerating expansion of the universe and dynamical compactification of extra dimensions , respectively .
it turns out that the rate of compactification of extra dimensions as well as expansion of the universe depends on the number of extra dimensions , @xmath3 .
the more extra dimensions , the less rate of compactification and the more rate of acceleration .
it is worth noting that the model is free of initial singularity problem because both @xmath2 and @xmath1 are non - zero at @xmath92 , resulting in a finite ricci scalar .
although the model describes in principle a closed universe with non - vanishing cosmological constant , it is equivalent to a flat universe with zero cosmological constant
. therefore , one may assume that we are really living in a closed universe with @xmath200 , but it effectively appears as a flat universe with @xmath201 .
note that we have not considered ordinary matter sources in the model except an exotic matter source which is to be considered as a source of dark energy .
therefore , it seems the solutions to describe typical inflation rather than the recently observed acceleration of the universe which is known to take place in an ordinary matter dominated universe .
however , if the large percent of the matter sources in the universe would be of dark energy type ( as the present observations strongly recommend ) , then one may keep the results here even in the presence of other matter source , keeping in mind that the relevant contribution to the total matter source of the universe is the dark energy .
a question may arise on the fact that no physics is supposed to exist below the planck length whereas for the contracting solution , the scale factor @xmath38 goes to zero starting from @xmath93 .
however , it is not a major problem because we have not considered elements of quantum gravity theory in this model and merely studied a model based on general relativity which is supposed to be valid in any scale without limitation .
the scale @xmath93 , in this paper , is not introduced within a quantum gravity model ( action ) ; it just appears as a typical initial condition , in the middle of a classical model , based on the quantum cosmological consideration .
one may choose another scale based on some other physical considerations .
we have also studied the corresponding quantum cosmology , through the wheeler - dewitt equation , and obtained the exact solutions .
based on hartlepoint of view on the correspondence between the classical and quantum solutions , we have shown by 2d and 3d plots of the wave functions a good correspondence between the classical and quantum cosmological solutions for any @xmath3 , provided that the wave functions vanish for the infinite scale factors .
there is no such a correspondence if another boundary condition , other than stated , is taken .
therefore , this correspondence guaranties that the chosen boundary condition is a _ good _ one .
+ fig . 1 .
2d plot of @xmath192 in terms of @xmath153 for @xmath194 + fig .
2d plot of @xmath202 in terms of @xmath153 for @xmath194 + fig .
2d plot of @xmath203 in terms of @xmath153 for @xmath194 + fig .
4 . 2d plot of @xmath204 in terms of @xmath153 for @xmath194 + fig .
2d plot of @xmath205 in terms of @xmath153 for @xmath194 + fig .
2d plot of @xmath193 in terms of @xmath153 for @xmath194 + fig .
2d plot of @xmath196 in terms of @xmath153 for @xmath194 + fig .
2d plot of @xmath206 in terms of @xmath153 for @xmath194 + fig . 9 .
2d plot of @xmath207 in terms of @xmath153 for @xmath194 + fig . 10 .
2d plot of @xmath208 in terms of @xmath153 for @xmath194 + fig .
2d plot of @xmath197 in terms of @xmath153 for @xmath194 + fig .
classical locus @xmath209 for @xmath109 + fig . 13 .
classical locus @xmath210^{-1}$ ] for @xmath211 + fig . 14 .
classical locus @xmath210^{-1}$ ] for @xmath212 + fig . 15 .
classical locus @xmath210^{-1}$ ] for @xmath213 + fig .
classical locus @xmath210^{-1}$ ] for @xmath214 + fig .
classical locus @xmath210^{-1}$ ] for @xmath215 + fig . 18 .
classical locus @xmath216^{-1}$ ] for @xmath211 + fig . 19 .
classical locus @xmath216^{-1}$ ] for @xmath212 + fig . 20 .
classical locus @xmath216^{-1}$ ] for @xmath213 + fig . 21 .
classical locus @xmath216^{-1}$ ] for @xmath214 + fig . 22 .
classical locus @xmath216^{-1}$ ] for @xmath215 + fig . 23 .
3d plot of @xmath192 in terms of @xmath153 for @xmath194 + fig .
3d plot of @xmath202 in terms of @xmath153 for @xmath194 + fig .
3d plot of @xmath203 in terms of @xmath153 for @xmath194 + fig . 26 .
3d plot of @xmath204 in terms of @xmath153 for @xmath194 + fig
3d plot of @xmath205 in terms of @xmath153 for @xmath194 + fig .
3d plot of @xmath193 in terms of @xmath153 for @xmath194 + fig .
3d plot of @xmath196 in terms of @xmath153 for @xmath194 + fig .
3d plot of @xmath206 in terms of @xmath153 for @xmath194 + fig .
3d plot of @xmath207 in terms of @xmath153 for @xmath194 + fig .
3d plot of @xmath208 in terms of @xmath153 for @xmath194 + fig .
3d plot of @xmath197 in terms of @xmath153 for @xmath194 | we study a @xmath0-dimensional kaluza - klein cosmology with a robertson - walker type metric having two scale factors @xmath1 and @xmath2 , corresponding to @xmath3-dimensional internal space and 4-dimensional universe , respectively . by introducing an exotic matter in the form of perfect fluid with an special equation of state , as the space - time part of the higher dimensional energy - momentum tensor
, a four dimensional effective decaying cosmological term appears as @xmath4 with @xmath5 , playing the role of an evolving dark energy in the universe . by taking @xmath6 , which has some interesting implications in reconciling observations with inflationary models and
is consistent with quantum tunneling , the resulting einstein s field equations yield the exponential solutions for the scale factors @xmath1 and @xmath2 .
these exponential behaviors may account for the dynamical compactification of extra dimensions and the accelerating expansion of the 4-dimensional universe in terms of hubble parameter , @xmath7 .
the acceleration of the universe may be explained by the negative pressure of the exotic matter .
it is shown that the rate of compactification of higher dimensions as well as expansion of 4-dimensional universe depends on the dimension , @xmath3 .
we then obtain the corresponding wheeler - dewitt equation and find the general exact solutions in @xmath3-dimensions .
a good correspondence between the solutions of classical einstein s equations and the solutions of quantum wheeler - dewitt equation in any dimension , @xmath3 , is obtained based on hartle s point of view concerning the classical limits of quantum cosmology .
-0.75 in |
four - dimensional black hole solutions of einstein gravity coupled to a scalar field have been an avenue of intense research for many years .
questions pertaining to their existence , uniqueness and stability were seeking answers over these years .
the kerr - newman solutions of four - dimensional asymptotically flat black holes coupled to an electromagnetic field or in vacuum , imposed very stringent conditions on their existence in the form of no - hair " theorems . in the case of a minimally coupled scalar field in asymptotically flat spacetime
the no - hair theorems were proven imposing conditions on the form of the self - interaction potential @xcite .
these theorems were also generalized to non - minimally coupled scalar fields @xcite . for asymptotically flat spacetime , a four - dimensional black hole coupled to a scalar field with a zero self - interaction potential
is known @xcite .
however , the scalar field diverges on the event horizon and , furthermore , the solution is unstable @xcite , so there is no violation of the no - hair " theorems . in the case of a positive cosmological constant with a minimally coupled scalar field with a self - interaction potential
, black hole solutions were found in @xcite and also a numerical solution was presented in @xcite , but it was unstable . if the scalar field is non - minimally coupled , a solution exists with a quartic self - interaction potential @xcite , but it was shown to be unstable @xcite . in the case of a negative cosmological constant
, stable solutions were found numerically for spherical geometries @xcite and an exact solution in asymptotically ads space with hyperbolic geometry was presented in @xcite and generalized later to include charge @xcite .
this solution is perturbatively stable for negative mass and may develop instabilities for positive mass @xcite .
the thermodynamics of this solution were studied in @xcite where it was shown that there is a second order phase transition of the hairy black hole to a pure topological black hole without hair .
the analytical and numerical calculation of the quasi - normal modes of scalar , electromagnetic and tensor perturbations of these black holes confirmed this behaviour @xcite .
recently , a new exact solution of a charged c - metric conformally coupled to a scalar field was presented in @xcite .
a schwarzschild - ads black hole in five - dimensions coupled to a scalar field was discussed in @xcite , while dilatonic black hole solutions with a gauss - bonnet term in various dimensions were discussed in @xcite . from a known black hole solution coupled to a scalar field other solutions
can be generated via conformal mappings @xcite . in all black hole solutions in the einstein frame the scalar field
is coupled minimally to gravity . applying a conformal transformation to these solutions ,
other solutions can be obtained in the jordan frame which are not physically equivalent to the untransformed ones @xcite . the scalar field in the jordan frame
is coupled to gravity non - minimally and this coupling is characterized by a dimensionless parameter @xmath10 .
there are strong theoretical , astrophysical and cosmological arguments ( for a review see @xcite ) which fix the value of this conformal coupling to @xmath11 .
if the scalar potential is zero or quartic in the scalar field , the theory is conformally invariant ; otherwise a non - trivial scalar potential introduces a scale in the theory and the conformal invariance is broken . in this work we present a new class of black hole solutions of four - dimensional einstein gravity coupled to a scalar field and to vacuum energy .
we analyse the structure and study the properties of these solutions in the einstein frame . in this frame
, the scalar self - interaction potential is characterised by a dimensionless parameter @xmath0 .
if this parameter vanishes , then the known solutions of black holes minimally coupled to a scalar field in ( a)ds space are obtained @xcite .
transforming these solutions to the jordan frame , the parameter @xmath0 can be interpreted as giving the measure of departure from conformal invariance .
this breakdown of conformal invariance allows the back - scattering of waves of the scalar field off of the background curvature of spacetime , and the creation of `` tails '' of radiation . this effect may have sizeable observation signatures in cosmology @xcite .
following @xcite , we perform a perturbative stability analysis of the solutions .
we find that the hairy black hole is stable near the conformal point if the mass is negative and may develop instabilities in the case of positive mass .
we also study the thermodynamics of our solutions . calculating the free energy we find that there is a critical temperature above which the hairy black hole loses its hair to a black hole in vacuum .
this critical temperature occurs at a point where the black hole mass flips sign , as in the case of the mtz black hole @xcite .
interestingly , another phase transition occurs at a higher critical temperature which is of first order and involves a different branch of our solution .
this new critical temperature diverges as the coupling constant in the potential @xmath2 .
these exact hairy black hole solutions may have interesting applications to holographic superconductors @xcite , where new types of holographic superconductors can be constructed @xcite .
our discussion is organized as follows . in section [ sec2 ]
we introduce the self - interaction potential and we present the hairy black hole solution . in section [ sec4 ]
we discuss the thermodynamics of our solution . in section [ secst ]
we perform a stability analysis .
finally , in section [ sec6 ] we summarize our results .
to obtain a black hole with scalar hair , we start with the four - dimensional action consisting of the einstein - hilbert action with a negative cosmological constant @xmath12 , along with a scalar , @xmath13,\label{action11}\ ] ] where @xmath14 is newton s constant and @xmath15 is the ricci scalar .
the corresponding field equations are @xmath16 where the energy - momentum tensor is given by @xmath17 the potential is chosen as @xmath18\nonumber\\\label{potentialnew}\end{aligned}\ ] ] and it is given in terms of a coupling constant @xmath0 . setting @xmath1
we recover the action that yields the mtz black hole @xcite .
this particular form of the potential is chosen so that the field equations can be solved analytically .
the qualitative nature of our results does not depend on the detailed form of the potential .
a similar potential was considered in a different context in @xcite ( see also @xcite for the derivation of a potential that yields analytic solutions in the case of a spherical horizon ) .
if one goes over to the jordan frame , in which the scalar field obeys the klein - gordon equation [ kleingordon ] -r -=0 , with @xmath11 , the scalar potential has the form v ( ) = - ^4 & - & .
[ potentialnm ] evidently , the scalar field is conformally coupled but the conformal invariance is broken by a non - zero value of @xmath0 .
the mass of the scalar field is given by m^2 = v(0 ) = - where we defined @xmath19 .
notice that it is independent of @xmath0 and coincides with the scalar mass that yields the mtz black hole @xcite .
asymptotically ( @xmath20 ) , the scalar field behaves as @xmath21 where @xmath22 . in our case @xmath23 and @xmath24 .
both boundary conditions are acceptable as both give normalizable modes .
we shall adopt the mixed boundary conditions ( as @xmath20 ) ( r ) = + + ,
c = - < 0 .
[ eqbcphi]this choice of the parameter @xmath25 coincides with the mtz solution @xcite .
solutions to the einstein equations with the boundary conditions ( [ eqbcphi ] ) have been found in the case of spherical horizons and shown to be unstable @xcite . in that case , for @xmath26 , it was shown that @xmath27 always and the hairy black hole had positive mass .
on the other hand , mtz black holes , which have hyperbolic horizons and obey the boundary conditions ( [ eqbcphi ] ) with @xmath27 , can be stable if they have negative mass @xcite .
this is impossible with spherical horizons , because they always enclose black holes of positive mass .
the numerical value of @xmath25 is not important ( except for the fact that @xmath28 ) and is chosen as in ( [ eqbcphi ] ) for convenience .
the field equations admit solutions which are black holes with topology @xmath29 , where @xmath30 is a two - dimensional manifold of constant negative curvature .
black holes with constant negative curvature are known as topological black holes ( tbhs - see , e.g. , @xcite ) .
the simplest solution for @xmath31 reads @xmath32 where @xmath33 is a constant which is proportional to the mass and is bounded from below ( @xmath34 ) , @xmath35 is the line element of the two - dimensional manifold @xmath30 which is locally isomorphic to the hyperbolic manifold @xmath36 and of the form @xmath37 with @xmath38 a freely acting discrete subgroup ( i.e. , without fixed points ) of isometries .
the line element @xmath35 of @xmath30 can be written as @xmath39 with @xmath40 and @xmath41 being the coordinates of the hyperbolic space @xmath36 or pseudosphere , which is a non - compact two - dimensional space of constant negative curvature .
this space becomes a compact space of constant negative curvature with genus @xmath42 by identifying , according to the connection rules of the discrete subgroup @xmath38 , the opposite edges of a @xmath43-sided polygon whose sides are geodesics and is centered at the origin @xmath44 of the pseudosphere .
an octagon is the simplest such polygon , yielding a compact surface of genus @xmath45 under these identifications .
thus , the two - dimensional manifold @xmath30 is a compact riemann 2-surface of genus @xmath46 .
the configuration ( [ linel ] ) is an asymptotically locally ads spacetime .
the horizon structure of ( [ linel ] ) is determined by the roots of the metric function @xmath47 , that is @xmath48 for @xmath49 , this equation has two distinct non - degenerate solutions , corresponding to an inner and to an outer horizon @xmath50 and @xmath51 respectively . for @xmath52
, @xmath47 has just one non - degenerate root and so the black hole ( [ linel ] ) has one horizon @xmath51 .
the horizons for both cases of @xmath33 have the non - trivial topology of the manifold @xmath30 .
we note that for @xmath53 , @xmath47 has a degenerate root , but this horizon does not have an interpretation as a black hole horizon .
the boundary has the metric [ eqnew ] ds_^2 = -dt^2 + l^2 d^2 , so spatially it is a hyperbolic manifold of radius @xmath54 ( and of curvature @xmath55 ) . the action ( [ action11 ] ) with a potential as in ( [ potentialnew ] ) has a static black hole solution with topology @xmath56 and with scalar hair , and it is given by @xmath57~ , \ee where \be\label{psiconf } f(r ) = \frac{r^2}{l^2 } - g\frac{r_0}{l^2 } r - 1 + g \frac{r_0 ^ 2}{l^2 } - \left ( 1 - 2 g \frac{r_0 ^ 2}{l^2 } \right ) \frac{r_0}{r } \left ( 2 + \frac{r_0}{r } \right ) + g \frac{r^2}{2l^2 } \ln\left ( 1 + \frac{2r_0}{r}\right)\ , \ee and the scalar field is \be \phi(r ) = \sqrt{{\frac}{3}{4 \pi g}}~ { \rm arctanh } \left({\frac}{r_0}{r+r_0}\right)~ , \label{scafield } \ee obeying the boundary conditions ( \ref{eqbcphi } ) by design .
\section{thermodynamics } \label{sec4 } to study the thermodynamics of our black hole solutions we consider the euclidean continuation ( $ t \rightarrow i\tau$ ) of the action in hamiltonian form \begin{equation } i=\int \left [ \pi ^{ij}\dot{g}_{ij}+p\dot{\phi}-n\mathcal{h}-n^{i}\mathcal{h}_{i}\right ] d^{\,3}xdt+b , \label{hamiltaction}\ ] ] where @xmath58 and @xmath59 are the conjugate momenta of the metric and the field respectively ; @xmath60 is a surface term .
the solution reads : @xmath61 where n(r ) = r(r+2r_0)(r+r_0)^2 ,
f^2(r ) = ( r+r_0)^2r(r+2r_0)f(r ) , r^2(r ) = r^3(r+2r_0)(r+r_0)^2 , with a periodic @xmath62 whose period is the inverse temperature , @xmath63 .
the hamiltonian action becomes @xmath64 where @xmath65 is the area of @xmath66 and = nr^2 .
the euclidean solution is static and satisfies the equation @xmath67 .
thus , the value of the action in the classical limit is just the surface term @xmath60 , which should maximize the action within the class of fields considered .
we now compute the action when the field equations hold .
the condition that the geometries which are permitted should not have conical singularities at the horizon imposes t = f^(r_+)4 [ eqtemperature ] . using the grand canonical ensemble ( fixing the temperature ) , the variation of the surface term reads @xmath68 where @xmath69 } _ { r_{+}}^{\infty } \ ; , \label{delg}\ ] ] and the contribution from the scalar field equals @xmath70 for the metric , the variation of fields at infinity yields @xmath71 so @xmath72 the surface term at infinity is @xmath73 the variation of the surface term at the horizon may be found using the relations @xmath74 we observe that @xmath75 vanishes , since @xmath76 , and @xmath77 thus the surface term at the horizon is @xmath78 therefore , provided the field equations hold , the euclidean action reads @xmath79 the euclidean action is related to the free energy through @xmath80 .
we deduce @xmath81 where @xmath82 and @xmath83 are the mass and entropy respectively , m = ( 3 - 8gr_0 ^ 2/l^2)12 g r_0 , s= 4 gr^2(r_+)=a_h4 g [ massentropy ] it is easy to show that the law of thermodynamics @xmath84 holds . for @xmath85 ,
these expressions reduce to the corresponding quantities for mtz black holes @xcite . alternatively
, the mass of the black hole can be found by the ashtekar - das method @xcite .
a straightforward calculation confirms the expression ( [ massentropy ] ) for the mass . in the case of the topological black hole ( [ ftbh ] ) the temperature , entropy and mass
are given by respectively , t= ( - ) , s_tbh= , m_tbh= ( - 1 ) , [ relations2 ] and also the law of thermodynamics @xmath84 is obeyed .
we note that , in the limit @xmath86 @xmath87 from eq .
( [ psiconf ] ) and the corresponding temperature ( [ eqtemperature ] ) reads @xmath88 which equals the temperature of the topological black hole ( [ relations2 ] ) in the limit @xmath89 ( @xmath90 ) .
the common limit @xmath91 dt^{2}+\left [ \frac{r^{2}}{l^2}-1\right ] ^{-1}dr ^{2}+r ^{2}d\sigma ^{2}\ ; \label{muzero}\ ] ] is a manifold of negative constant curvature possessing an event horizon at @xmath92 .
the tbh and our hairy black hole solution match continuously at the critical temperature [ eqtcr ] t_0 = , which corresponds to @xmath93 , with ( [ muzero ] ) a transient configuration . evidently , at the critical point ( [ eqtcr ] ) a scaling symmetry emerges owing to the fact that the metric becomes pure ads . at the critical temperature ( [ eqtcr ] ) a higher - order phase transition occurs as in the case of the mtz black hole ( with @xmath1 ) . introducing the terms with @xmath94 in the potential
do not alter this result qualitatively .
next we perform a detailed analysis of thermodynamics and examine several values of the coupling constant @xmath0 .
henceforth we shall work with units in which @xmath95 .
we begin with a geometrical characteristic of the hairy black hole , the horizon radius @xmath96 ( root of @xmath97 ( eq . ( [ psiconf ] ) ) . in figure [ horizons ] we show the @xmath98 dependence of the horizon for representative values of the coupling constant , @xmath99 and @xmath100 we observe that , for @xmath101 the horizon may correspond to more than one value of the parameter @xmath102 for @xmath103 we see that , additionally , there is a maximum value of the horizon radius .
we note that one may express the radius of the horizon @xmath104 in terms of the dimensionless parameter = as r_+ = [ rhm3 ] .
the temperature reads t = , [ tm3 ] or equivalently t= , a third order equation in @xmath105 showing that , for a given temperature , there are in general three possible values of @xmath10 .
thus we obtain up to three different branches of our hairy black hole solution .
we start our analysis with a relatively large value of the coupling constant @xmath0 , namely @xmath99 and calculate the horizon radius , temperature and euclidean action for various values of @xmath98 . in figure
[ mass_action_q3 ] , left panel , we depict @xmath98 versus @xmath106 and it is clear that there is a @xmath106 interval for which there are really three corresponding values for @xmath102 outside this interval , there is just one solution .
the corresponding graph for the euclidean actions may be seen in the right panel of the same figure .
the action for the topological black hole with the common temperature @xmath106 is represented by a continuous line , while the actions for the hairy black holes are shown in the form of points .
we note that equation ( [ ftbh ] ) yields for the temperature of the topological black hole @xmath107 the largest euclidean action ( smallest free energy ) will dominate .
there are three branches for the hairy black hole , corresponding to the three different values of @xmath102 in particular , for some fixed temperature ( e.g. , @xmath108 ) the algebraically lowest @xmath98 corresponds to the algebraically lowest euclidean action ; similarly , the medium and largest @xmath98 parameters correspond to the medium and largest euclidean actions .
the medium euclidean action for the hairy black hole is very close to the euclidean action for the topological black hole .
in fact , it is slightly smaller than the latter for @xmath109 and slightly larger after that value .
if it were the only branch present , one would thus conclude that the hairy black hole dominates for small temperatures , while for large temperatures the topological black hole would be preferred .
this would be a situation similar to the one of the mtz black hole .
however , the two additional branches change completely our conclusions .
the upper branch shows that the hairy black hole dominates up to @xmath110 when the coupling constant @xmath0 decreases , equation ( [ tm3 ] ) along with the demand that the temperature should be positive , show that the acceptable values of @xmath98 are two rather than three , as may be seen in figure [ mass_action_q00005 ] , left panel .
the lowest branch of the previous corresponding figure [ mass_action_q3 ] shrinks for decreasing @xmath0 and it finally disappears .
an interesting consequence of this is that the temperature has an upper limit .
the graph for the euclidean actions ( figure [ mass_action_q00005 ] , right panel ) is influenced accordingly .
there are just two branches for the hairy black hole , rather than three in figure [ mass_action_q3 ] and the figure ends on its right hand side at @xmath111 the continuous line represents the euclidean action for the topological black hole with the same temperature .
similar remarks hold as in the previous case , e.g. , the phase transition moves to @xmath112 in addition , the largest value of @xmath98 corresponds to the upper branch of the hairy black hole . to understand the nature of this phase transition it is instructive to draw a kind of phase diagram , so that we can spot which is the dominant solution for a given pair of @xmath0 and @xmath113 we depict our result in figure [ phd ] .
the hairy solution dominates below the curve which shows the critical temperature as a function of the coupling constant @xmath0 .
the most striking feature of the graph is that the critical temperature diverges as @xmath2 .
thus , it does not converge to the mtz value @xmath114 at @xmath1 . for even the slightest nonzero values of @xmath0
the critical temperature gets extremely large values !
this appears to put the conformal point ( mtz black hole ) in a special status within the set of these hairy black holes . in other words ,
the restoration of conformal invariance is not a smooth process , and the mtz black hole solution can not be obtained in a continuous way as @xmath115 .
in fact , it seems that ( even infitesimally ) away from the conformal point @xmath1 black holes are mostly hairy !
to perform the stability analysis of the hairy black hole it is more convenient to work in the einstein frame .
henceforth , we shall work in units in which the radius of the boundary is @xmath95 .
we begin with the hairy black hole line element , @xmath116 , \ee which can be written in the form \begin{equation } ds^{2}=-\frac{f_0}{h_0^{2}}dt^{2}+\frac{dr^{2}}{f_0}+r^{2}d\sigma^{2}\ ] ] using the definitions f_0(r ) = f ( ) ( 1 + ) ^2 , + h_0(r ) = ( 1 + ) , under the change of coordinates r=^3/2(+2r_0)^1/2+r_0 . the scalar field solution reads _ 0 ( r ) = ^-1 , obeying the boundary conditions ( [ eqbcphi ] ) with [ eqxc ] = _ 0 = |r_0| .
we are interested in figuring out when the black hole is unstable ( losing its hair to turn into a tbh ) and discuss the results in the context of thermodynamic considerations . to this end
, we apply the perturbation @xmath117 which respects the boundary conditions ( [ eqbcphi ] ) with @xmath118 for an instability to develop .
the field equations read : -1-f - r f^+r f h^h+8 g r^2 v()=0 , + r f ^=0 , 2 h^+r h = 0 , ( f ) -1r^2(r^2 fh ^)^+ 1h v^()=0 .
the field equations give a schrdinger - like wave equation for the scalar perturbation , @xmath119 where we defined the tortoise coordinate = , and the effective potential is given by = .
the explicit form of the schrdinger - like equation reads : -f ( ) dd + v _ 1 = -^2 _ 1 , where the functional form of the function @xmath120 has been given in equation ( [ psiconf ] ) and . ]
= r_0 ^ 2 f()^2 ( 1 + 2 r_0)^2 ( 1 + 3 r_0 + 3 r_0 ^ 2 ^ 2)^2\{5 + 2+(11 g+54)r_0 ^ 2r_0 + 29+(47 g+189 ) r_0 ^ 2 ^ 2 . .
+ r_0 ( 150+(-3 g+270 ) r_0 ^ 2)^3 + r_0 ^ 2 ( 396+(-351 g+135 )
r_0 ^ 2)^4 + r_0 ^ 3 ( 612 -873 g r_0 ^ 2)^5 . .
+ r_0 ^ 4 ( 582 - 1047 g r_0 ^ 2)^6 + 324 r_0 ^ 5 ( 1 - 2 g r_0 ^ 2)^7 + 81 r_0 ^ 6 ( 1 - 2 g r_0 ^ 2)^8 .
. + g2 ( 5 + 54 r_0 + 189 r_0 ^ 2 ^ 2 + 270 r_0 ^ 3 ^ 3 + 135 r_0 ^ 4 ^ 4 ) ( 1 + 2 r_0 ) } . near the horizon
the schrdinger - like equation simplifies to -[f^(_+)]^2 dd = -^2 _ 1 , = -_+ and its acceptable solution reads _ 1 ~^ , = 1f^(_+ ) , > 0 .
regularity of the scalar field at the horizon ( @xmath121 ) requires the boundary conditions _ 1 = 0 , ( r - r_+ ) _
1 = _ 1 , > 0 . for a given @xmath118
, they uniquely determine the wavefunction . at the boundary @xmath122 the wave equation
is approximated by - + 5r_0 ^ 2 _ 1 = -^2 _ 1 , with solutions _ 1 = e^e
r _ * , e = , where @xmath123 .
therefore , for large @xmath124 , _ 1 = a + + [ fitab]to match the boundary conditions ( [ eqbcphi ] ) , we need [ eqx33 ] = 2 c _ 0 = - 2 r_0 . since the wavefunction has already been determined by the boundary conditions at the horizon and therefore also the ratio @xmath125 , this is a constraint on @xmath126 if ( [ eqx33 ] ) has a solution , then the black hole is unstable .
if it does not , then there is no instability of this type ( however , one should be careful with non - perturbative instabilities ) . in figure [ stable_q0_q00005 ] ( left panel )
we show the ratio @xmath125 for the standard mtz black hole ( corresponding to @xmath1 ) versus @xmath127 at a typical value of the mass parameter , namely @xmath128 it is obvious that the value of the ratio lies well below the values @xmath129 it is clearly impossible to have a solution to this equation , so the solution is stable .
in fact this value of the mass parameter lies in the interesting range for this black hole , since thermodynamics dictates that for negative values of @xmath98 mtz black holes are favored against topological black holes .
we find that the mtz black holes turn out to be stable .
next we examine the case @xmath130 for which we have presented data before in figure [ mass_action_q00005 ] . as we have explained there
, the most interesting branch of the graphs is the upper branch , on the right panel , which dominates the small @xmath106 part of the graph and corresponds to large values of @xmath98 ( typically around @xmath131 on the left panel of the same figure ) .
thus we set @xmath132 and plot @xmath125 versus @xmath127 in the right panel of figure [ stable_q0_q00005 ] .
it is clear that the curve lies systematically below the quantity @xmath133 and no solution is possible , so the hairy black hole with these parameters is stable .
finally we come to the case with @xmath134 which has three branches . in the left panel of figure [ stable_q3 ]
we show the results for @xmath135 which corresponds to the upper branch of figure [ mass_action_q3 ] ; the curve again lies below the quantity @xmath136 and no solution is possible , so this hairy black hole is stable . in the right panel of figure [ stable_q3 ]
we show the results for @xmath137 which corresponds to the lowest branch of figure [ mass_action_q3 ] , which disappears for decreasing @xmath138 in this case we find something qualitatively different : the curve cuts the line @xmath139 around @xmath140 and a solution is possible , signaling instability .
thus , for @xmath99 the hairy black hole may be stable or unstable , depending on the value of @xmath102
we presented a new class of hairy black hole solutions in asymptotically ads space . the scalar field
is minimally coupled to gravity with a non - trivial self - interaction potential .
a coupling constant @xmath0 in the potential parametrizes our solutions .
if @xmath1 the conformal invariant mtz black hole solution , conformally coupled to a scalar field , is obtained . if @xmath141 a whole new class of hairy black hole solutions is generated .
the scalar field is conformally coupled but the solutions are not conformally invariant .
these solutions are perturbative stable near the conformal point for negative mass and they may develop instabilities for positive mass .
we studied the thermodynamical properties of the solutions . calculating the free energy we showed that for a general @xmath0 , apart the phase transition of the mtz black hole at the critical temperature @xmath142 , there is another critical temperature , higher than the mtz critical temperature , which depends on @xmath0 and where a first order phase transition occurs of the vacuum black hole towards a hairy one .
the existence of a second critical temperature is a manifestation of the breaking of conformal invariance . as @xmath143 the second critical temperature diverges , indicating that there is no smooth limit to the mtz solution .
the solutions presented and discussed in this work have hyperbolic horizons .
there are also hairy black hole solutions with flat or spherical horizons of similar form . however , these solutions are pathological . in the solutions with flat horizons ,
the scalar field diverges at the horizon , in accordance to the no - hair " theorems . in the case of spherical horizons , calculating the free energy we find that always the vacuum solution is preferable over the hairy configuration .
moreover , studying the asymptotic behaviour of the solutions , we found that they are unstable for any value of the mass .
g. s. was supported in part by the us department of energy under grant de - fg05 - 91er40627 .
j. e. chase , `` event horizons in static scalar - vacuum space - times , '' , 276 ( 1970 ) ; + j. d. bekenstein , `` transcendence of the law of baryon - number conservation in black - hole physics , '' , 452 ( 1972 ) ; + j. d. bekenstein , `` nonexistence of baryon number for static black holes , '' , 1239 ( 1972 ) ; + j. d. bekenstein , `` nonexistence of baryon number for black holes .
ii , '' , 2403 ( 1972 ) ; + m. heusler , `` a mass bound for spherically symmetric black hole spacetimes , '' , 779 ( 1995 ) [ arxiv : gr - qc/9411054 ] ; + m. heusler and n. straumann , `` scaling arguments for the existence of static , spherically symmetric solutions of self - gravitating systems , '' , 2177 ( 1992 ) ; + d. sudarsky , `` a simple proof of a no - hair theorem in einstein - higgs theory , '' , 579 ( 1995 ) ; + j. d. bekenstein , `` novel ` no - scalar - hair ' theorem for black holes , '' , r6608 ( 1995 ) .
a. e. mayo and j. d. bekenstein , `` no hair for spherical black holes : charged and nonminimally coupled scalar field with self interaction , '' phys
. rev .
d * 54 * , 5059 ( 1996 ) [ arxiv : gr - qc/9602057 ] ; + j. d. bekenstein , `` black hole hair : twenty - five years after , '' arxiv : gr - qc/9605059 .
n. bocharova , k. bronnikov and v. melnikov , vestn .
* 6 * , 706 ( 1970 ) ; + j. d. bekenstein , annals phys.*82 * , 535 ( 1974 ) ; + j. d. bekenstein , `` black holes with scalar charge , '' annals phys . *
91 * , 75 ( 1975 ) .
k. a. bronnikov and y. n. kireyev , `` instability of black holes with scalar charge , '' , 95 ( 1978 ) .
t. torii , k. maeda and m. narita , `` no - scalar hair conjecture in asymptotic de sitter spacetime , '' phys .
d * 59 * , 064027 ( 1999 ) [ arxiv : gr - qc/9809036 ] .
c. martinez , r. troncoso and j. zanelli , `` de sitter black hole with a conformally coupled scalar field in four dimensions , '' , 024008 ( 2003 ) [ arxiv : hep - th/0205319 ] .
t. torii , k. maeda and m. narita , `` scalar hair on the black hole in asymptotically anti - de sitter spacetime , '' phys .
d * 64 * , 044007 ( 2001 ) .
e. winstanley , `` on the existence of conformally coupled scalar field hair for black holes in ( anti-)de sitter space , '' found .
* 33 * , 111 ( 2003 ) [ arxiv : gr - qc/0205092 ] . c. martinez , r. troncoso and j. zanelli , `` exact black hole solution with a minimally coupled scalar field , '' phys .
d * 70 * , 084035 ( 2004 ) [ arxiv : hep - th/0406111 ] . c. martinez , j. p. staforelli and r. troncoso , `` topological black holes dressed with a conformally coupled scalar field and electric charge , '' , 044028 ( 2006 ) [ arxiv : hep - th/0512022 ] ; + c. martinez and r. troncoso , `` electrically charged black hole with scalar hair , '' , 064007 ( 2006 ) [ arxiv : hep - th/0606130 ] .
g. koutsoumbas , e. papantonopoulos and g. siopsis , `` exact gravity dual of a gapless superconductor , '' jhep * 0907 * , 026 ( 2009 ) [ arxiv:0902.0733 [ hep - th ] ] .
g. koutsoumbas , s. musiri , e. papantonopoulos and g. siopsis , `` quasi - normal modes of electromagnetic perturbations of four - dimensional topological black holes with scalar hair , '' jhep * 0610 * , 006 ( 2006 ) [ arxiv : hep - th/0606096 ] ; + g. koutsoumbas , e. papantonopoulos and g. siopsis , `` phase transitions in charged topological - ads black holes , '' jhep * 0805 * , 107 ( 2008 ) [ arxiv:0801.4921 [ hep - th ] ] ; + g. koutsoumbas , e. papantonopoulos and g. siopsis , `` discontinuities in scalar perturbations of topological black holes , '' , 105004 ( 2009 ) [ arxiv:0806.1452 [ hep - th ] ] .
a. anabalon and h. maeda , `` new charged black holes with conformal scalar hair , '' arxiv:0907.0219 [ hep - th ] .
k. farakos , a. p. kouretsis and p. pasipoularides , `` anti de sitter 5d black hole solutions with a self - interacting bulk scalar field : a potential reconstruction approach , '' phys .
d * 80 * , 064020 ( 2009 ) [ arxiv:0905.1345 [ hep - th ] ] .
n. ohta and t. torii , `` black holes in the dilatonic einstein - gauss - bonnet theory in various dimensions iv - topological black holes with and without cosmological term , '' arxiv:0908.3918 [ hep - th ] .
v. faraoni , e. gunzig and p. nardone , fund . cosmic phys .
* 20 * , 121 ( 1999 ) [ arxiv : gr - qc/9811047 ] .
v. faraoni and s. sonego , `` on the tail problem in cosmology , '' phys .
a * 170 * , 413 ( 1992 ) [ arxiv : astro - ph/9209004 ] ; + t. w. noonan , `` huygen s principle in conformally flat spacetimes , '' class .
* 12 * , 1087 ( 1995 ) .
t. hertog and k. maeda , `` stability and thermodynamics of ads black holes with scalar hair , '' phys .
d * 71 * , 024001 ( 2005 ) [ arxiv : hep - th/0409314 ] .
s. a. hartnoll , c. p. herzog and g. t. horowitz , `` building a holographic superconductor , '' phys . rev
. lett .
* 101 * , 031601 ( 2008 ) [ arxiv:0803.3295 [ hep - th ] ] .
s. a. hartnoll , c. p. herzog and g. t. horowitz , `` holographic superconductors , '' jhep * 0812 * , 015 ( 2008 ) [ arxiv:0810.1563 [ hep - th ] ] .
d. f. zeng , `` an exact hairy black hole solution for ads / cft superconductors , '' arxiv:0903.2620 [ hep - th ] . | we present a new class of black hole solutions with a minimally coupled scalar field in the presence of a negative cosmological constant .
we consider an one - parameter family of self - interaction potentials parametrized by a dimensionless parameter @xmath0 .
when @xmath1 , we recover the conformally invariant solution of the martinez - troncoso - zanelli ( mtz ) black hole .
a non - vanishing @xmath0 signals the departure from conformal invariance .
thermodynamically , there is a critical temperature at vanishing black hole mass , where a higher - order phase transition occurs , as in the case of the mtz black hole .
additionally , we obtain a branch of hairy solutions which undergo a first - order phase transition at a second critical temperature which depends on @xmath0 and it is higher than the mtz critical temperature . as @xmath2 ,
this second critical temperature diverges .
+ * theodoros kolyvaris @xmath3 * , * george koutsoumbas @xmath4 * , + * eleftherios papantonopoulos @xmath5 * + department of physics , national technical university of athens , + zografou campus gr 157 73 , athens , greece + department of physics and astronomy , the university of tennessee , + knoxville , tn 37996 - 1200 , usa + @xmath6 e - mail address : [email protected] + @xmath7 e - mail address : [email protected] + @xmath8 e - mail address : [email protected] + @xmath9 e - mail address : [email protected] |
the modification of in - medium properties of mesons is one of the crucial concepts explored in experimental studies of hot and dense matter in created in heavy ion collisions .
it is the hope that the creation of a dense matter and eventually also the quark gluon plasma can be unambiguously demonstrated through the study of modifications generated in the spectrum of heavy ( @xmath0-suppression ) as well as light quark bound states ( shift of the @xmath1-meson mass , resonance broadening , thermal dilepton enhancement , .. ) .
lattice calculations can contribute to the theoretical discussion of these burning experimental questions through the calculation of thermal spectral functions , which control the structure of hadron correlation functions at finite temperature .
the reconstruction of spectral functions from hadron correlation functions calculated on the lattice at finite temperature is based on a statistical tool , the maximum entropy method ( mem ) @xcite which has been explored and tested over recent years .
we will discuss here results of an analytic calculation , which becomes feasible in the infinite temperature limit which corresponds to a free fermion theory .
in addition to calculations on isotropic and anisotropic lattices based on hadronic currents constructed from point sources we will present a new calculation of spectral functions performed for extended quark sources .
the calculation of hadronic spectral functions ( spf ) on the lattice starts from the euclidian correlation function of hadronic currents , @xmath2 , with given quantum numbers , @xmath3 . in the infinite temperature limit these correlation functions ( cf )
will approach the free fermion cf , which can be expressed in terms of the free lattice quark propagator @xmath4 .
we will discuss here analytic calculations for spectral functions determined from correlation functions defined with currents constructed from point - like ( p ) as well as extended ( exponentially smeared @xmath5 or fuzzed @xmath6 ) sources @xcite .
furthermore , we will present results on isotropic as well as anisotropic lattices .
the anisotropy in the spatial ( @xmath7 ) and temporal ( @xmath8 ) lattice spacings is controlled by the anisotropy parameter @xmath9 . while the calculations on anisotropic lattices have been performed for the case of a wilson action with @xmath10-parameter being set to unity as well as for the case of wilson fermions where the time - like @xmath10-parameter is linked to the anisotropy @xmath11 .
this latter choice is often used in heavy quark simulations and has also been used in the mem analysis of heavy quark bound states @xcite .
in addition we present results from calculations with a truncated fixed - point action @xcite . in momentum space the euclidean time correlation functions for free fermions is then given by , @xmath12{\rm d}^3k\quad\nonumber\\ & = & \hspace*{-10pt}\int \;\sigma_h^l(\omega , t ) \frac{{\rm
ch}(\omega(\tau-1/2t))}{{\rm sh}(\omega/2 t ) } \;{\rm d}\omega \;,\label{eq : frecf2}\end{aligned}\ ] ] @xmath13 { l } 1 \\
\exp({\bf k}^2/4 ) \\ \exp({\bf k}^2/2 ) \\ \cos({\bf kr } ) \end{array } } \right.\;\;\mbox{for } s= \left\ { { \begin{array}[h ] { l } pp\\ ep \\ ee \\
r=1\end{array } } \right.,\end{aligned}\ ] ] where @xmath14 is a product of @xmath15-matrices that fixes the quantum numbers @xmath3 .
the crucial step in eq.([eq : frecf2 ] ) is the second equality where the 3-dimensional momentum integral has been replaced by a 1-dimensional integral over the energy @xmath16 .
two further integrations are hidden in the definition of the spectral function @xmath17 .
we note that a separation of the dependence on euclidean time @xmath18 in the integration kernel , @xmath19 which kept its continuum form , and the lattice spectral function @xmath17 , which is @xmath18-independent and carries the entire cut - off dependence of the cfs , could be achieved .
this makes the introduction of a lattice kernel , which has been used in some numerical studies based on the mem analysis @xcite questionable .
the analytic calculations are presented in @xcite for point - like sources and the truncated fixed - point action .
the resulting spectral functions obtained in these cases on lattices with different anisotropies are shown in fig .
[ fig : pspf ] . = 6.8 cm a closer look at the resulting spfs shows that they all agree quite well with the continuum result , @xmath20 in the ir regime .
this agreement extends to larger energies for the wilson action on anisotropic lattices and even more so for the truncated fixed - point action @xcite .
the choice @xmath11 apparently leads to a very abrupt deviation from this continuum like behavior and leads to large cut - off effects for @xmath21 .
this makes its use for the heavy quark spectroscopy at high temperature questionable .
the other actions also show the influence of lattice artefacts in the uv region .
lattice artefacts show up as peaks and cusps which can directly be related to the sudden restriction of the available momentum phase space near the corners of the brillouin zone for the quark dispersion relation .
a close relation between the quark dispersion relation and the structure of spfs thus is obvious in the free theory .
to enhance the overlap with the ground state excitation in a given quantum number channel , extended operators are often used . above @xmath22 , the existence of a well separated ground state is , however , not guaranteed and it thus is questionable whether extended operators can be of any use at all .
nevertheless , extended operators have been used in the mem analysis of meson correlation functions@xcite .
as they reduce to point like operators in the continuum limit , when their extend is kept fixed in lattice units , they may indeed also be suitable for studies of spectral functions .
we thus have analyzed some of them also in the free field limit .
the spfs of extended operators can be easily obtained using the so called `` binning procedure '' ( see @xcite for details ) .
some results are shown in fig .
[ fig : extmes ] . as can be seen the uv regime
indeed is strongly suppressed and a good agreement with the continuum spectral function can be achieved of a wide range of energies up to the region where the point - like currents lead to large contributions from the `` wilson doublers '' .
this suppression is more effective for operators with larger extend , as can be observed with the ( @xmath23 ) smeared operator which in the considered limit can also be viewed as a broader smeared exponential ( @xmath24 ) operator .
we note , however , that the use of extended sources requires tuning of parameters and care has to be taken that the low energy part does not get to be suppressed too , which happen easily with the ( @xmath23 ) operator . some versions of extended operators also violate the positivity of the spectral functions , which enters as a basic assumption in the mem analysis @xcite .
the fuzzed ( @xmath6)-source has this problem for @xmath25 . moreover , the exponentially smeared sources are not gauge invariant and gauge fixing becomes necessary . for the application in simulations this additional computational overhead has to be compared with the improvement which could be achieved by using improved fermion actions or by increasing @xmath26 .
the analysis of the free lattice spf provides basic information on the structure of spectral functions reconstructed in the interacting case from a statistical analysis based on mem .
it also can provide guidance for a suitable choice of the default model in the mem analysis and the choice of an appropriate integration kernel in the spectral representation . in the free field limit lattice artefacts only lead to a modification of the spf
, the integration kernel keeps its continuum form .
extended meson operators lead to an efficient suppression of the uv part of the spfs .
this , however , requires a careful choice of parameters in order not to influence also the ir part of the spectral functions which at finite temperature is not well separated from the short distance part of euclidean correlation functions . | we present results from an analytic calculation of thermal meson spectral functions in the infinite temperature ( free field ) limit .
we compare spectral functions for various lattice fermion formulations used at present in studies of in - medium properties of hadrons based on the maximum entropy method ( mem ) .
in particular , we will present a new calculation of spectral functions performed with extended quark sources .
bi - tp 2003/28 + bnl - nt-03/29 |
the new generation of cosmic microwave background radiation ( cmbr ) experiments has opened a new era in astrophysics , the precision cosmology era .
recent observations , especially those of wmap ( wilkinson anisotropy microwave probe ) , acbar ( arcminute cosmology bolometer array receiver ) and cbi ( cosmic background imager ) , brought a new light to cmbr fluctuations studies in large , intermediate and small scales .
this new generation of experiments made possible the characterization of the power spectrum of temperature fluctuations up to the third acoustic peak ( hinshwaw et al . 2003
; pearson et al . 2002 ; kuo et al . 2004 ) . indeed , the combination of cmbr and large scale structure ( lss ) data allows cosmologists to constrain cosmological parameters for a given set of scenarios and also determine the nature of the primordial fluctuations . since cmbr carries the intrinsic statistical properties of cosmological perturbations , it is considered the most powerful tool to investigate the nature of cosmic structure . the most accepted model for structure formation assumes initial quantum fluctuations created during inflation and amplified by gravitational effects .
the standard inflation model predicts an adiabatic uncorrelated random field with a nearly flat , scale - invariant spectrum on scales larger than @xmath1 1 - 2@xmath5 ( guth 1981 ; salopek , bond & bardeen 1981 ; bardeen , steinhardt & turner 1983 ) .
simple inflationary models also predict the random field follows a nearly - gaussian distribution , where only small deviations from gaussianity are allowed ( e.g. gangui et al .
1994 ) . in hybrid inflation models ( battye & weller 1998 ;
battye , magueijo & weller 1999 ) , structure is formed by a linear combination of ( inflation - produced ) adiabatic and ( topological defects induced ) isocurvature density fluctuations .
the topological defects are assumed to appear during the phase transition that marks the end of the inflationary epoch .
correlated mixed field with adiabatic and isocurvature fluctuations has been considered by many authors , as bucher et al .
( 2000 ) , gordon ( 2001 ) and amendola et al ( 2002 ) .
andrade et al . (
2004 ) have suggested a new mixed scenario describing a correlated mixture of two fields , one of them being a dominant gaussian adiabatic process to which is added a small contribution of an isocurvature field . in a previous work , ribeiro , wuensche & letelier ( 2000 ) used such mixed scenario to probe the galaxy cluster abundance evolution in the universe and found that even a very small level of non - gaussianity in the mixed field may introduce significant changes in the cluster abundance rate .
andrade , wuensche and ribeiro ( 2004 ) showed the effects of mixed models in the cmbr power spectrum , combining a gaussian ( adiabatic ) field with a second isocurvature field to produce a positive skewness mixed density fluctuation field .
this approach adopted a scale dependent mixture parameter and a power - law initial spectrum to simulate the cmbr temperature and polarization power spectra for a flat @xmath2-cdm model , generating quite a large grid of cosmological parameters combination .
the results show how the shape and amplitude of the fluctuations in cmbr depend upon such mixed fields and how we can easily distinguish a standard adiabatic gaussian field from a mixed non - gaussian one and easily quantify the contribution of the second component . in this work , we apply a statistical test to both the mixed power spectrum simulations and recent cmbr data obtained by wmap ( spergel et al .
2003 ) , acbar ( kuo et al . 2004 ) and cbi ( pearson et al . 2002 ) in order to compare how well the standard and mixed models describe the most recent observations .
we also estimate the possible contribution of an isocurvature field to the primordial density fluctuation field in the mixed model scenario .
the experiments acbar and cbi present the highest sensitivity and highest signal to noise observations of cmbr temperature distribution in small angular scales ( @xmath1 4 -5 arcminutes ) . in scales
larger than the above mentioned ( @xmath1 20 arcminutes ) the wmap satellite made a set of all - sky maps measuring both the temperature and polarization anisotropies , opening a new window for cosmological investigations .
the wmap observations , combined with lss observations has made possible to constrain a precise picture of the cosmos . using the three - dimensional power spectrum p(k ) from over 200,000 galaxies in the sloan digital sky survey ( sdss ) in combination with wmap and other data , tegmark et al .
( 2003 ) show that recent observations are consistent with a flat adiabatic @xmath2-cdm model .
specifically , the cmb power spectrum exhibits a first acoustic peak at l = 220.1 @xmath6 0.8 , with amplitude of the 74.7 @xmath6 0.5 @xmath7k ; and a second acoustic peak at l = 546 @xmath6 10 , with amplitude of 48.8 @xmath6 0.9 @xmath7k , a picture consistent with inflation predictions ( peiris et al .
the best fit model compared with the wmap plus acbar and cbi is for a @xmath2-cdm universe with the following cosmological parameter combination : @xmath8@xmath9h@xmath10 = 1.02 @xmath6 0.002 ; @xmath11 ; @xmath12 ; @xmath13 ; @xmath14 and a slope running spectral index with @xmath15 ( spergel et al .
the present cmbr data provides a strong support for adiabatic field domination .
however , the wmap data show several peculiarities at various values of l. wmap observations show that the fluctuations in large scale present a lower amplitude than the standard inflation model predicts .
the temperature power spectrum is almost 30% suppressed on @xmath1 1 degree angular scales ( hinshaw et al . , 2003 ) , specially those scales related to the quadrupole ( l = 2 ) and the octopole ( l = 3 ) , when compared with the predictions of the standard gaussian @xmath2-cdm models .
another impressive wmap conclusions is the evidence of an optical depth to re - scattering by electrons with a value of @xmath16 = 0.17 @xmath6 0.04 , describing a reionization scenario at redshift 11 @xmath17 z @xmath17 30 ( kogut et al .
2003 ) . in this paper
, we try to reproduce the main feature of the cmbr power spectrum , in large , intermediate and small scales , invoking just the primordial physics , by considering a correlated mixed field in combination with six cosmological parameters : @xmath8@xmath18 , @xmath8@xmath19 , @xmath8@xmath20 , n , h@xmath21 , the amplitude of the power spectrum , a , and the mixed coefficient , @xmath22@xmath21 .
since we are interested in just cosmological principles , no secondary effects are considered in cmb fluctuations , like reionization models , gravitational lenses nor sunyayev - zeldovich effects .
in the mixed scenario , we suppose that the field has a probability density function of the form : @xmath23~\propto ~(1-\alpha)f_1(\delta_k ) + \alpha f_2(\delta_k)\ ] ] the first field will always be the gaussian component and a possible effect of the second component is to modify the gaussian field to have a positive tail .
the parameter @xmath22 in ( 1 ) allows us to modulate the contribution of each component to the resultant field .
like the hybrid inflation models ( battye & weller 1998 ; battye , magueijo & weller 1999 ) , the mixture models consider the scenario in which structure is formed by both adiabatic density fluctuations produced during inflation and active isocurvature perturbations created by cosmic defects during a phase transition which marks the end of inflationary epoch
. nevertheless , the mixed scenario considers a possible correlation between the adiabatic and the isocurvature fields on the post - inflation universe .
so , the fluctuations in super - horizon scales ( inflated during the exponential expansion ) are strictly uncorrelated and the mixing effect acts only inside the hubble horizon , in sub - degree scales . to allow for this condition and keep a continuous mixed field , a scale dependent mixture parameter , @xmath22 @xmath24 @xmath22(k ) was defined .
we assume the simplest choice of @xmath22(k ) , a linear function of @xmath25 : @xmath26 to constrain the two component random field , we take @xmath27 , where @xmath28 is a random number with distribution given by ( 1 ) .
so , we consider the primordial power spectrum of the mixed field in the form : @xmath29 where the @xmath30 represents a primordial conventional power - law spectrum and m@xmath31(@xmath22 ) is the mixture term , a functional of @xmath32 and @xmath33 , which account for the statistics effect of a new component : @xmath34\nu^2\ ; d\nu\ ] ] the effective mixed primordial power spectrum is expressed as : @xmath35 where m(@xmath22@xmath21 ) represents only the coefficient dependence , @xmath22@xmath21 . in the case of a pure gaussian field , m(@xmath22@xmath21 )
@xmath36 @xmath22@xmath21 @xmath36 0 , the mixed power spectrum will be a simple power - law spectrum , @xmath37 , where n is the conventional spectral index predicted by inflationary models . in the case of a mixed field ,
the phase correlations between both fields are estimated by the integral in eq.(4 ) , on mixture scales defined by eq .
( 2 ) . to estimate the cmbr anisotropy we need to evaluate the evolution of fluctuations field generated in the early universe through the radiation - dominated era and recombination . in the mixed model , the evolution of both adiabatic and isocurvature components of the mixed density field
is considered an independent process , when only their effective amplitude correlation is considered on the decoupled surface of the cmbr photons . to compute the independent evolution of the adiabatic @xmath2-cdm mode and the isocurvature @xmath2-cdm mode , we have used the most precision code , the linger function of the cosmics code package ( bertschinger 1999 ; ma & bertschinger 1995 ) .
linger does generate the most accurate results for the photon density field , being able to compute the cmbr anisotropy with integrations errors less than 0.15% .
however , linger routine is very time consuming , but for investigation of small deviations from gaussianity , the cosmics package seems to be the most precise code for mixed cmb simulations .
once estimated the photon density evolution , the multipole moments , c@xmath38 , of the cmbr temperature power spectra were estimated by a mixed photon density function incorporated to the original cosmics package : @xmath39 the function @xmath40 represents the mixed photon density field in the last scattering surface defined by : @xmath41 where @xmath42 and @xmath43 are the photon density function estimated by cosmics , respectively , for an adiabatic and an isocurvature seed initial conditions . inserting ( 7 ) in ( 6 ) , we obtain a mixed term in the c@xmath38 estimation .
this condition suggests that the amplitude of both fields are cross - correlated at the last scattering surface , with a mixing ratio defined by @xmath22@xmath21 and in a characteristic scale defined by @xmath22@xmath21_k_. the power spectra estimated by eq.(3 ) consider a flat @xmath2-cdm universe slightly distorted by a non - gaussian statistics with constant spectral index .
a number of realizations of cmbr temperature power spectra were made and the mean temperature fluctuations combining gaussian , exponential , lognormal , rayleigh , maxwellian and a chi - squared distributions was estimated ( andrade et al . 2004 ) .
the simulations show that the influence of the specific statistics of the second component in the mixed field is not so important as the cross - correlation between the amplitudes of both fields .
however , some important results were obtained .
it is possible to directly assess and quantify the mixture of a correlated adiabatic and isocurvature non - gaussian field .
this behavior points to another possibility to extract information from a cmbr power spectrum : the possibility of detecting weakly mixed density fields , even if we can not exactly identify the mixture components distribution .
cmbr temperature mixed angular power spectrum estimated for a @xmath2-cdm model in different mixing degrees . ] in figure 1 , we see how the shape and amplitude of the spectrum change even for small values of @xmath22@xmath44 ( @xmath110@xmath45 - 10@xmath46 ) .
the peaks intensities are clearly susceptible to the existence of mixed fields , although distinguishing peaks of higher order ( 2@xmath47 , 3@xmath48 , etc . ) in the mixed context is not a straightforward task , since their intensities , compared to the first peak , are very low .
the effect of increasing @xmath22@xmath21 is a power transfer to smaller scales ( l ` > ` 1000 ) , while the super - degree scales are less affected .
however , for large values of @xmath22@xmath21 ( @xmath22@xmath21 ` >
` 3x10@xmath46 ) there is a fast increase in the temperature fluctuations , probably caused by correlation excess between the mixed fields , resulting in more power in small scales .
therefore , an acceptable range for @xmath49 is set to be @xmath50 .
in order to quantify a possible non - gaussian isocurvature contribution to the fluctuations field , we applied a maximum likelihood approach to the power spectrum estimated by the most recent cmbr observations .
we ran a set of cmbr realizations for mixed and pure density fields , considering a flat @xmath2-cdm universe consistent with inflation predictions and lss observations . instead of the usual cmbr analysis approach , which considers tensors contribution and reionization effects , we consider only the main six cosmological parameters ( @xmath51 , @xmath52 , @xmath53@xmath20 , n , h@xmath21 , @xmath49 ) , and also a , described in the range of possibilities : 0.8 ` < ` n ` < ` 1.2 ; 0.015 ` < ` @xmath8@xmath18 ` < ` 0.03 ; 0.6 ` < ` @xmath8@xmath20 ` < ` 0.8 ; 60 ` < ` h@xmath21 ` < ` 80 ; cold dark matter density set as 1-(@xmath8@xmath18 + @xmath8@xmath20 ) , ranging of 0.170 ` < ` @xmath8@xmath19 ` < ` 0.385 ; and the range values of @xmath22@xmath21 set in 0.0 ` < ` @xmath22@xmath21 ` < ` 3x10@xmath46 . with this considerations ,
we ensure the simulations are in agreement with the basic preferences of standard inflation scenarios : flat geometry ( @xmath8@xmath9 @xmath1 1 ) and a nearly scale invariant primeval spectrum ( n @xmath1 1 ) .
@xmath8@xmath18 = ( 0.019 @xmath6 0.01)h@xmath54 is consistent with the mass density of baryons determined by big bang nucleossynthesis and the large scale structure observations , which suggests that the hubble constant h@xmath21 assumes values in the range ( 66 ` < ` h@xmath21 ` < ` 75 ) km / sec.mpc ; and a positive cosmological constant value in the range ( 0.065 ` < ` @xmath8@xmath20 ` < ` 0.75 ) , as estimated recently by tegmark et al . ( 2003 ) .
the amplitude of the primordial power spectrum , a , was estimated by the best fit value for each model simulated .
we applied a reduced @xmath4 test ( bevington 1969 ) between the cmb data and the power spectra simulations in the mixed context in four steps .
first , we consider the wmap , cbi and acbar data independently , estimate the reduced @xmath4 for all three data sets and then add all data to obtain a new estimation of the reduced @xmath4 .
so we estimate the residual difference for c@xmath38 estimation in @xmath7k@xmath55 , and consider the power spectrum data points ( 899 for wmap ; 24 for cbi and 14 for acbar ) minus 7 degree of freedom .
we create a grid of nearly one hundred simulated values of @xmath22@xmath21 for each combination of standard cosmological parameters .
since cosmics is very time consuming , the variations in the standard cosmological parameters were set in order to give just the main direction for the @xmath4 face variations in cosmological parameters .
the most precise models , as defined by the @xmath4 test , are summarized on table 1 , which also contains the best @xmath22@xmath21 value estimated for each parameter combination listed with 68% confidence limit . [ cols="^,^,^,^,^,^,^,^,^,^,^ " , ] the simulations clearly show that wmap data tends to favor mixed models .
the @xmath0 estimated for the fit between mixed models and wmap data are significatively lower for mixing degrees of order @xmath1@xmath56 , in all combination of cosmological parameters tested .
however , cbi and acbar data in small angular scales tends to favor the adiabatic standard models .
the combination of all data are more sensible to wmap influence ( more signal in large and intermediate scales with low error bars ) resulting in best fits models in a mixed scenario .
all data favor a @xmath2-cdm universe with contents of barions , dark matter and dark energy coherent with that estimated by the wmap and lss analysis .
figure 2 shows the powers spectrum of the standard adiabatic and mixed models for the three best fit models obtained by the @xmath0 test .
it also shows the behavour of the all data @xmath0 while varying the @xmath22@xmath21 value .
it is clearly seen that the mixed model adjust the data better than the standard model in basic combination of cosmological parameters . in this figure , is possible to observe that the relation between the amplitudes of the acoustic peaks in the mixed scenario is more adequate to describe the cmbr power spectrum feature without consideration of any model of reionization or tensors contributions .
although the mixed model predicts higher quadrupole and octopole contributions that predicts the standard adiabatic @xmath2-cdm model , the plots clearly show that the mixed predictions are consistente with acoustic peaks amplitude estimated by wmap .
indeed , a model estimation in a good fit is obtained for a strictly flat universe , rather than an open universe or , instead , a slope running spectral index ( kosowsky & turner , 1995 ) as suggested by the wmap team .
one possible explanation of the low amplitude in large scales and the high amplitude of the secondary acoustic peaks predicted by the standard model ( pure adiabatic ) simulated in this work is the lack of the high optical depth , as considered by the wmap team ( hinshaw et al .
2003 ; spergel et al .
reionization affects the cmbr in several manners , through thomson scattering of photons from free electrons in the intergalactic medium , which suppress the primordial anisotropies of the cmbr imprinted on the decoupled surface . in a reionizing scenario
, the acoustic peaks in the cmbr power spectrum would have the amplitude lowered by @xmath57 and the rescattering would generate large angle temperature fluctuations through the doppler effect ( hu & white , 1997 ) . in this work , rather than considering reionization effects and invoking just primordial physics , we can explain the main feature of the cmbr power spectrum in large , intermediate and small scales in the context of mixed density field .
we applied a modified statistical @xmath0 test to the most recent cmbr observations .
the results allow us to compare the performance of power spectra simulated with mixed and standard density fields to describe the mean feature of cmbr temperature fluctuations .
the results showed that the wmap data tends to favor mixed density fields with isocurvature fluctuation contribution of order @xmath1@xmath56 ( @xmath58@xmath1 1.5 for mixed model against @xmath58@xmath1 2.4 in standard models ) without the need of considering any model of reionization .
also , large scale fluctuations are in good agreement with wmap measurements .
our result may looks discrepant with the reduced @xmath4 of 1.07 obtained with the same model in spergel et al .
( 2003 ) for temperature only , however , our work is restricted to a region of the parameter space that does not include the best fit model for tt only of the wmap team .
we can conclude that the mixed scenario offers a good alternative to describe the shape of tt cmbr power spectrum .
indeed , the data favors a @xmath2-cdm universe with barions , dark matter and dark energy contents which are coherent with a strictly flat universe , with no need to include tensor contributions nor a slope running spectral index . for a @xmath2-cdm universe with contents of @xmath8@xmath18= 0.03 ; and the best value indicated for the spectral index , by the @xmath4 test , [email protected] ; the contribution of the isocurvature field is estimated as @xmath22@xmath21= 0.00042 @xmath6 0.0003 with 68% confidence limit .
although the standard @xmath2-cdm model fits well a small number of data points of cbi ( @xmath59@xmath1 1 for standard models against @xmath59@xmath1 2.3 in mixed models ) and acbar ( @xmath60@xmath1 2.4 for standard models against @xmath60@xmath1 6.8 in mixed models ) , when considering all data points , cbi , acbar and wmap , the mixed model clearly shows ability to describe the mean feature of tt cmbr observation in large , intermediate and small scales . with the upcoming generation of experiments which will probe temperature fluctuations in small scales , such as planck satellite
, it will be possible to set new , more stringent , estimates of the parameter spaces and isocurvature fluctuations contributions .
also , we will try to constrain the mixed model predictions upon the temperature - polarization cross correlated power spectrum , in order to consider reinonizations models in the mixed scenario .
presently , we conclude that isocurvature fluctuations can not be ruled out in cosmological studies . also , conflicting with the results of wmap map team ( komatsu et al .
2003 ) , the results of some independent searches for non - gaussianity in wmap maps evidence that wmap data are consistent with gaussian condition for l ` < ` 250 , marginally gaussian for 224`<`l ` < ` 350 , and non - gaussian for l ` > ` 350 , as revealed , for instance , by the phase mapping technique ( chiang et al . 2003 ; coles et al . 2003 ) .
this picture points an extra expectation in the study of the mixed model predictions , altough , in order to constrain a better estimative of @xmath22@xmath21 , reionizations models must be investigated in the context of mixed fields .
the next step in the investigation of the mixed models in cmbr features will be the predictions of the reionization effects , even in a low redshifts , the degeneraices between cosmological parameters in mixed context , the predictions of the angular correlation function and the temperature - polarization cross correlated power spectrum .
the authors acknowledge edmund bertschinger for the use of the cosmics package , funded by nsf under grant ast-9318185 .
apaa thanks the financial support of cnpq and fapesb , under grant 1431030005400 .
caw was partially supported by cnpq grant 300409/97 - 4 .
albr thanks the financial support of cnpq , under grant 470185/2003 - 1 . | this work presents a reduced @xmath0 test to search for non - gaussian signals in the cmbr tt power spectrum of recent cmbr data , wmap , acbar and cbi data sets , assuming a mixed density field including adiabatic and isocurvature fluctuations .
we assume a skew positive mixed model with adiabatic inflation perturbations plus additional isocurvature perturbations possibly produced by topological defects .
the joint probability distribution used in this context is a weighted combination of gaussian and non - gaussian random fields .
results from simulations of cmbr temperature for the mixed field show a distinct signature in cmb power spectrum for very small deviations ( @xmath1 0.1% ) from a pure gaussian field , and can be used as a direct test for the nature of primordial fluctuations .
a reduced @xmath0 test applied on the most recent cmbr observations reveals that an isocurvature fluctuations field is not ruled out and indeed permits a very good description for a flat geometry @xmath2-cdm universe , @xmath3@xmath1 1.5 , rather than the simple inflationary standard model with @xmath3@xmath1 2.3 .
this result may looks is particular discrepant with the reduced @xmath4 of 1.07 obtained with the same model in spergel et al .
( 2003 ) for temperature only , however , our work is restricted to a region of the parameter space that does not include the best fit model for tt only of spergel et al .
( 2003 ) . |
the tully - fisher relation ( tfr ) between the luminosity of galaxies and their rotation velocity is a fundamental scaling relation that constrains galaxy formation and evolution models ( e.g. , * ? ? ? * and references therein ) .
recent local studies focus on stellar or baryonic masses ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , which are easier to model than luminosity .
distant stellar - mass tfrs ( m@xmath0tfrs ) have also been studied ( conselice et al . 2005 ; flores et al .
neither found evidence for evolution to redshifts @xmath8 or 0.6 , respectively , but each pruned their data to exclude morphologically odd galaxies .
this letter presents a study of the m@xmath0tfr to redshift @xmath9 using a sample of 544 galaxies with strong emission lines .
following @xcite , we exclude few galaxies and adopt a new velocity measure that includes both velocity gradients ( usually rotation ) and velocity spread ( dispersion ) .
we adopt a @xmath10cdm cosmology with @xmath11 , @xmath12 , and @xmath13 .
galaxy kinematics and morphologies at @xmath8 are possible with high resolution spectra and images . to track different morphologies and redshifts , samples should be large enough to be subdivided .
the all - wavelength extended groth strip international survey ( aegis ) provides such data .
single - orbit images were taken by the hubble space telescope advanced camera for surveys ( hst / acs ) in the f606w ( @xmath14 ) and f814w ( @xmath15 ) band - passes , and spectra were obtained by the deep extragalactic evolutionary probe 2 survey ( deep2 ) ; see details in @xcite .
observations and reduction of kinematics are described in @xcite for data similar to deep2 .
kinematics are measured from the strongest emission line(s ) among h@xmath16 , h@xmath17 , [ oii ] @xmath18 , and [ o iii ] @xmath19 .
the spectral resolution is fwhm=@xmath20 , or 56 km @xmath21 at @xmath22 ( gaussian @xmath23 km s@xmath24 ) .
emission line widths can be measured down to @xmath25 of the @xmath26 or @xmath27 km s@xmath24 , and velocity centroids for rotation curves are good to 0.4 of @xmath26 or @xmath28 km s@xmath24 .
rotation velocities can be detected down to @xmath29 km s@xmath24 , if present .
we chose a sample of galaxies that span a wide range in redshift , stellar mass , and morphology .
the deep2 survey had magnitude limits of @xmath30 , bright enough to yield good hst / acs imaging .
a redshift limit of @xmath31 ensures that galaxies of all colors on the blue sequence " @xcite are well - sampled . to yield good quality kinematics ,
emission lines were required to have integrated intensities
@xmath32 e@xmath33 in the summed one - dimensional spectrum , and spectrographic slits had to be aligned to within 40 of the major axes of the galaxies ( see fig .
13 of @xcite ) .
also , to reduce the effects of dust , edge - on systems with @xmath34 were excluded , and to allow a reliable rotation measure , nearly face on galaxies with @xmath35 were removed .
a total of 14 galaxies were excluded because they had emission lines with disturbed morphologies or that had contamination by emission from a nearby galaxy on the sky .
after additional cosmetic cuts were made @xcite , the final sample became 544 galaxies .
our final sample is primarily selected on emission line strength .
in order to interpret dynamical measurements made from the spectra , accurate position angles and ellipticities of galaxies must be obtained from hst imaging . to measure these parameters ,
the sextractor galaxy photometry software @xcite is used , as discussed in @xcite . to obtain quantitative morphologies for galaxies
, we use the gini / m@xmath36 method of @xcite .
this system of objective structural parameters has been shown to reliably divide galaxies into bins analogous to hubble types ; namely , early type spirals , late type spirals and irregulars , and major mergers .
galaxies are classified in the @xmath14 image for @xmath37 and in @xmath15 for @xmath38 .
such quantitative morphologies are obtainable only for galaxies with petrosian radii @xmath39 , images that have a signal - to - noise ratio of @xmath40 , and that do not fall near the edge of the hst / acs ccd chip .
galaxies without quantitative morphologies are not removed from the sample for completeness sake , and do not differ statistically from the rest of the sample .
all photometric parameters , including quantitative morphologies , will appear in a forthcoming paper , @xcite . in a separate exercise ,
we visually classified all galaxies and found examples that were fairly normal according to gini / m@xmath36 , but looked to our eye to be more disturbed or compact ; we discuss these galaxies separately .
spatially extended emission lines in the galaxy spectra are used to measure gas rotation and dispersion profiles .
since the spatial extent of the line emission is at most only a few times the seeing ( typically @xmath41 fwhm ) , the effect of seeing must be modeled .
we do this with the rotcurve fitting procedure of @xcite .
the kinematic model we fit has two parameters : the velocity on the flat part of the rotation velocity profile , @xmath2 , and the velocity dispersion from a spatially resolved fit to the two - dimensional spectra , @xmath6 .
the best - fit @xmath2 values are corrected for inclination with the measured galaxy ellipticities . for galaxies without a disk - like geometry ( based on visual inspection of hst / acs images ) , @xmath2 is not inclination - corrected .
these galaxies are flagged in the following analysis , but fits do not change if all galaxies are inclination - corrected .
the @xmath6 we measure does not have to correspond to a literal gas velocity dispersion like that of stars in an elliptical galaxy .
instead , for @xmath42 km s@xmath24 , @xmath6 likely measures the relative motions of individual regions , as in spiral arms or a thick disk ; for @xmath43 km s@xmath24 , @xmath6 likely represents an effective velocity dispersion caused by the blurring of velocity gradients on scales at or below the seeing limit that may not even have a preferred plane .
an uncertainty of 0.1 dex is adopted in both @xmath6 and the inclination - corrected @xmath2 to account for random errors and the dependence of the model parameters on the assumed seeing and scale radius of the rotation curve .
results do not change significantly if uncertainties of 0.2 are chosen , as discussed in 6 .
a few methods of estimating the stellar masses ( @xmath44 ) of galaxies have been investigated in a separate study @xcite .
these methods are : a color-@xmath45 relation from @xcite for @xmath46 and @xmath47 , a color-@xmath45 relation from @xcite for @xmath46 and @xmath48 , @xmath49 , broad - band spectral energy distribution ( sed ) fits of @xcite which incorporate observed optical and @xmath50-band data , and the method of @xcite .
we choose to adopt the method of @xcite .
these authors calibrated rest - frame @xmath51 photometry and redshifts to the full sed fits of bundy et al .
( 2005 ) to obtain @xmath44 estimates for very blue galaxies that are not detected at @xmath50 , which bundy et al .
( 2005 ) require to derive a @xmath44 .
while the evolution of seds with redshift is included and errors of 0.2 dex are assumed , the results of this paper , as discussed next , change insignificantly if we do not take into account evolution or increase the errors to 0.3 dex @xcite .
the quantity @xmath52 , where @xmath50 is a constant , combines dynamical support from ordered motion with that from disordered motions @xcite . for a spherically symmetric tracer population with isotropic velocity dispersion and density @xmath53 , @xmath54 ( * ? ? ?
* @xmath55 ) , where @xmath56 . therefore ,
if galaxies are virialized systems , and these assumptions about the tracer population are at least approximately correct , then @xmath57 should at least approximately trace the global @xmath26 of galaxy - halo systems . in this case
, the tracer population is the gas producing the emission lines , which we assume follows the @xmath44 . for disk galaxies with exponential @xmath44 profiles
, @xmath58 brackets a range of reasonable values ( 5.2 of @xcite ) .
a satisfactory approximation is adopted for all morphologies , @xmath59 ; if @xmath60 is adopted , the overall results do not change .
we henceforth use @xmath4 for our analysis .
the top panel of figure 1 shows @xmath2 versus @xmath44 ( the @xmath61tfr ) for a range of redshifts . the majority of spirals ( green triangles and blue squares not outlined in black ) form a clear ridge - line that compares well to a local relation from @xcite , consisting of well - ordered spirals , and to a @xmath62 relation from @xcite , who included only elongated disks . "
in contrast , almost all of the other classes ( red circles and symbols outlined in black ) lie to lower @xmath2 , causing a large scatter ( @xmath3 dex in velocity ) . this morphological dependence of tully - fisher scatter is well known . among local studies , e.g. ,
larger scatter is found for close pairs with kinematic disorders @xcite and when peculiar galaxies are included @xcite . at higher redshifts ,
larger scatter is found for @xmath63 galaxies with complex kinematics " @xcite .
moreover , @xmath64% of galaxies found with kinematics unrelated to rotation are excluded from tully - fisher studies at @xmath65 @xcite and at @xmath66 @xcite . in summary , because of morphological pruning , no study except that of @xcite incorporates the population of tully - fisher outliers that we detect in a measurement of tully - fisher evolution , and the conclusions of such studies will only apply for the subsample of galaxies that are morphologically well - ordered . for @xmath67 , the percentage of galaxies that scatter to low @xmath2 from the tully - fisher ridge - line ( defined as the left - most dotted line in figure 1 ) increases with redshift ( 18% , 35% , 42% , and 62% for the four redshift bins plotted ) .
similarly , @xcite find bright galaxies with @xmath68 at @xmath69 that are rare at low redshift .
the bottom panel of fig.1 uses @xmath4 instead of @xmath2 in the m@xmath0tfr .
the resulting relation has scatter that is not much greater than our measurement uncertainties and does not evolve significantly with @xmath70 .
dex ) decrease in intercept from @xmath22 to @xmath71 due simply to the predicted increase in @xmath44 of @xmath72 for a typical spiral galaxy ( @xmath73 and an exponentially decreasing star formation rate with @xmath74 gyr ) , but it would not be detectable given our errors .
we do not attempt to determine @xmath50 by minimizing the scatter in the @xmath75tfr because @xmath50 is sensitive to the morphological mix of the sample and the distribution of measurement errors .
] evidently , galaxies with low @xmath2 compared to the @xmath2/m@xmath0tfr ridge - line have a significant @xmath6 component which causes them to lie on the @xmath4/m@xmath0tfr .
lines are fit to the @xmath4/m@xmath0tfr in the four redshift bins in fig.1 with a maximum likelihood method detailed in @xcite .
we fit log@xmath76 and all excess scatter is allocated to the @xmath4-coordinate .
the fits do not vary significantly for the different redshift bins .
average values for the slope , intercept at @xmath77 , and intrinsic scatter over all redshifts are @xmath78 , and 0.10 dex in @xmath4 . the average rms scatter is 0.16 dex in @xmath4 , which corresponds to 0.47 dex in @xmath44 ( @xmath79 magnitudes ) .
the fitted relations are not very sensitive to the error estimates for @xmath4 and @xmath44 ; changing either error estimate by @xmath80 dex changes fit intercepts within the 1-@xmath26 error , and slopes by not more than @xmath81 dex .
however , our limited @xmath44 range at higher redshift makes slope measurements more uncertain than the formal fit errors .
furthermore , different estimates of @xmath44 can yield a range of @xmath82 dex in intercept , and @xmath44 estimators with non - evolving seds can cause the @xmath83tfr to show @xmath84 dex evolution in intercept with redshift , but this reflects problems of these estimators rather than actual evolution in the @xmath83tfr .
since no tully - fisher sample at low or high redshift has as much scatter as our @xmath2/m@xmath0tfr due to morphological pruning , it is rather remarkable that the @xmath4/m@xmath0tfr scatter for the same sample of galaxies manages to approach as close as it does to the scatters found for the @xmath2/m@xmath0tfrs for galaxies with relatively undisturbed morphologies ( @xmath25 dex in @xmath44 for @xcite ) .
in addition , fig.1 shows that the @xmath44 faber - jackson relation for low redshift ( @xmath85 ) early type galaxies from @xcite is consistent with the @xmath4/m@xmath0tfr in terms of slope and intercept .
if those galaxies with @xmath86 m@xmath87 that are off the @xmath61tfr ridge - line with low @xmath2 at higher redshift eventually form into well - ordered disk galaxies on the ridge - line , this trend implies that more massive disk galaxies may be moving onto the @xmath2/m@xmath0tfr with time as their stars and gas settle " into more circular / planar - dominated orbits .
the possible mechanisms behind this settling can be circularization of gas orbits , gaseous dissipation to the disk plane , and/or mergers ending with the growth of a gaseous disk .
the scatter for @xmath88 , which we can observe at lower redshift , is also large , suggesting that lower @xmath44 galaxies may currently be settling .
a scenario of galaxy formation that is consistent with our results is one that begins with matter assembling in dark halo potential wells with random orbital kinematics .
the baryonic components form proto - disks that are initially supported by a combination of ordered and random motions .
the material in these proto - disks has been settling since then , unless they undergo major mergers . over the last half or so of their lives , proto - disks and their descendants lie on nearly the same @xmath4/m@xmath0tfr . at first glance ,
the similarity between the faber - jackson relation and the @xmath4/m@xmath0tfr in slope and intercept is reassuring , given that merging proto - disks on parabolic binary orbits should put the merger products close to the faber - jackson relation in fig.1 ( e.g. , * ? ? ?
this suggests that the origin of the faber - jackson relation could in fact be the pre - existing @xmath4/m@xmath0tfr for disk - like galaxies , if merging is the process that creates ellipticals .
however , additional studies are needed to confirm this speculative scenario .
if the framework we adopt is correct , halo kinematic properties should be related to measurable galaxy kinematics . in particular ,
n - body models show that the slope of the relation between halo mass ( @xmath89 ) and halo @xmath26 ( @xmath90 ) , both measured within the scale radius @xcite , is 0.33 : log@xmath91 = 0.33 log@xmath92 + constant .
remarkably , this slope is nearly consistent with _ both _ galaxy relations .
in addition , neither the halo @xmath93 relation , nor @xmath89 or @xmath90 individually , is predicted to evolve to within @xmath94 from @xmath31 to now ( @xcite ; see @xcite for evolution in terms of virial quantities ) .
so , if @xmath4 is linearly related to halo @xmath90 and @xmath95constant , then the @xmath4/m@xmath0tfr should also evolve little from @xmath31 , in agreement with observations .
furthermore , that neither @xmath89 nor @xmath90 individually evolves might imply that absolute changes in inner baryon structures are also small .
however , when comparing galaxies to halos , there are at least three major complications .
first , since @xmath44 is subject to the star - formation history of a galaxy , it is not as fundamental a quantity as baryonic mass ( @xmath96 ) .
locally , low-@xmath44 galaxies have a greater gas - to - star fraction than more massive galaxies ( e.g. , * ? ? ?
* ) , so the low - redshift @xmath4/m@xmath97tfr ought to have a steeper slope than the @xmath4/m@xmath0tfr . indeed ,
if we assume that @xmath98 ( roughly consistent with data from @xcite using relations from @xcite to obtain @xmath44 , which we later convert to a @xcite initial mass function ) , then the @xmath4/m@xmath0tfr slope in the lowest redshift bin steepens to @xmath99 .
secondly , galaxies should convert gas into stars over time , causing evolution in @xmath100 .
the third issue is the complicated relation between @xmath4 and @xmath90 , which is expected to be affected by baryon dissipation and the response of the dark halo to it ( e.g. , * ? ? ?
* ) , neither of which are completely understood . with a better understanding or measurement of these factors and their time evolution , the relationship between galaxy and halo scaling laws will ultimately become clearer and the effects of baryonic processes during galaxy formation may even be elucidated .
we acknowledge the following grants : nsf ast-0507483 , hst go-10134 , and hst ar-09936 and the very significant cultural role and reverence that the summit of mauna kea has always had within the indigenous hawaiian community and appreciate the opportunity to conduct observations from this mountain .
sak would like to thank a. faltenbacher , b. holden , and z. zheng for helpful conversations .
barton , e. j. et al .
2001 , , 121 , 625 bell , e. f. & de jong , r. s. 2001 , , 550 , 212 bell , e. f. , mcintosh , d. h. , katz , n. , & weinberg , m. d. 2003 , , 149 , 289 bertin , e. & arnouts , s. 1996 , , 117 , 393 binney , j. & tremaine , s. 1987 , galactic dynamics ( princeton : princeton univ . press ) blumenthal , g. r. , faber , s. m. , flores , r. , & primack , j. r. 1986 , , 301 , 27 bhm , a. 2004 , , 420 , 97 bundy , k. , ellis , r. s. , & conselice , c. j. 2005 , , 625 , 621 chabrier , g. 2003 , pasp , 115 , 763 conselice , c. j. , et al .
2005 , , 628 , 160 dalcanton , j. j. , spergel , d. n. , & summers , f. j. 1997 , , 482 , 659 davis , m. , et al .
2006 , , accepted , astro - ph/0607355 davis , m. , et al .
2003 , proc .
spie , 4834 , 161 diemand , j. , et al .
2007 , in preparation flores , h. , et al . 2006 , , 455 , 107 gallazzi , a. , charlot , s. , brinchmann , j. , & white , s. d. m. 2006 , , 370 , 1106 kannappan , s. , fabricant , d. g. , & franx , m. 2002 , , 123 , 2358 kannappan , s. j. & gawiser , e. 2007 , , accepted , astro - ph/0701749 kassin , s. a. , et al .
, in preparation lin , l. , et al .
2006 , , accepted , astro - ph/0607272 lotz , j. m. , primack , j. , & madau , p. 2004
, , 128 , 163 lotz , j. m. , et al .
2006 , astro - ph/0602088 lotz , j. m. , et al .
2007 , in preparation mcgaugh , s. s. 2005 , , 632 , 859 navarro , j. f. , frenk , c. s. , & white , s. d. m. 1997 , , 490 , 493 pizagno , j. et al . , 2005 , , 633 , 844 robertson , b. et al .
2006 , , 645 , 986 simard , l. & pritchet , c. j. 1998 , , 505 , 96 wechsler , r. h. et al .
2002 , , 568 , 52 weiner , b. j. , et al .
2006a , , 653 , 1027 weiner , b. j. , et al .
2006b , , 653 , 1049 willmer , c. n. a. 2006 , , 647 , 853 | we combine newly measured rotation velocities , velocity dispersions , and stellar masses to construct stellar mass tully - fisher relations ( m@xmath0tfrs ) for 544 galaxies with strong emission lines at @xmath1 from the all wavelength extended groth strip international survey ( aegis ) and the deep extragalactic evolutionary probe 2 survey ( deep2 ) .
the conventional m@xmath0tfr using only rotation velocity ( @xmath2 ) shows large scatter ( @xmath3 dex in velocity ) .
the scatter and residuals are correlated with morphology in the sense that disturbed , compact , and major merger galaxies have lower velocities for their masses .
we construct an m@xmath0tfr using the kinematic estimator @xmath4 which is defined as @xmath5 and accounts for disordered or non - circular motions through the gas velocity dispersion ( @xmath6 ) .
the new m@xmath0tfr , termed @xmath4/m@xmath0tfr , is remarkably tight over @xmath1 with no detectable evolution of its intercept or slope with redshift .
the average best fit relation has 0.47 dex scatter in stellar mass , corresponding to @xmath7 ` magnitudes , ' assuming a constant mass - to - light ratio .
interestingly , the @xmath4/m@xmath0tfr is consistent with the absorption - line based stellar mass faber - jackson relation for nearby elliptical galaxies in terms of slope and intercept , which might suggest a physical connection between the two relations . |
conformational dynamics is essential for a protein s ability to exhibit allostery .
the coupling between two distant binding sites is frequently accomplished by a conformational change between a `` closed '' ( apo ) to an `` open '' ( holo ) conformation upon ligation.@xcite although the end point conformations often give valuable insight into protein function , a detailed description of the allosteric mechanism for a particular protein requires one to consider a broader conformational ensemble .
the landscape theory of binding@xcite acknowledges that a folded protein is inherently dynamic and explores the thermally accessible conformational states in its native basin.@xcite this conformational ensemble comprises the protein s `` functional landscape''.@xcite while only a small subset of the states comprising the folding energy landscape@xcite , the functional landscape determines how a protein responds to the changes in its local environment such as ligand interactions . due to the heterogeneous nature of the conformational ensemble ,
a ligand preferentially stabilizes some conformations more than others , causing the protein s thermal population to redistribute to a ligated ensemble which in general has distinct equilibrium properties.@xcite the ensemble nature of allostery accommodates a rich and diverse set of regulatory strategies and provides a general framework to understand binding thermodynamics and kinetics of specific proteins.@xcite even simple landscapes with a small number of well defined basins separated by kinetic barriers can have subtle binding mechanisms because they depend on ligand interactions to short - lived transient states .
experimental progress on this challenging kinetics problem has appeared only very recently.@xcite in principle , affinities of metastable states can also be obtained from thermodynamic binding measurements , although such analysis may not always be practical . in this paper
, we focus on the cooperative binding of two ca@xmath0ions to the binding loops of the domains of calmodulin ( cam ) through equilibrium coarse - grained simulations . in this minimal model , the conformational transition between the open and closed ensembles are simulated explicitly and the dynamic shift in population due to ligand binding and unbinding is approximated by discrete jumps between a ligated and unligated free energy surfaces.@xcite the protein dynamics are governed by a native - centric potential that couples the open and closed conformational basins while ligation is represented implicitly through ligand mediated protein contacts .
this model , developed by takada and co - workers , has been used to investigate the kinetic partitioning of induced fit and conformational selection binding pathways@xcite as well as mechanical unfolding of calmodulin in the presence of ca@xmath0.@xcite here , we assume that the ligands bound to the protein are in equilibrium with a dilute solution and calculate binding thermodynamics as a function of ligand concentration .
the model is parameterized so that the closed basin is more stable than the open basin in the unligated ensemble .
ligands interact with all conformations in the ensemble , but the affinity is largest for conformations within the open basin due to their high structural compatibility with the ligand .
thus , the population shifts towards the open ensemble with increasing ligand concentration ( see _ si
_ fig.s1 ) .
the simulated ensembles have significant molecular fluctuations which modulate ligand affinities and affect the coupling between the binding sites .
when binding thermodynamics are dominated by the open and closed ensembles , this model provides a molecular realization of the celebrated monod - wyman - changeux ( mwc ) model of allostery.@xcite for binding a single ligand the mwc model has four states : unligated - open , unligated - closed , ligated - open , and ligated - closed . appealing to this simple four - state model allows us to extract binding free energies of the isolated sites in the simulated open and closed ensembles and to calculate the free energy associated with the cooperative coupling between the sites .
the simulations connect the conformational ensemble underlying the protein s dynamics with the mwc phenomenological binding parameters.@xcite early work on binding thermodynamics of cam has revealed that the affinities and cooperativities of the n - terminal domain ( ncam ) and the c - terminal domain ( ccam ) are distinct despite their structural similarity.@xcite although some experimental data has been reanalyzed recently@xcite , the traditional analysis of thermodynamic binding data has not used a dynamic landscape ( or mwc ) framework.@xcite nuclear magnetic resonance experiments@xcite and all atom molecular dynamics simulations@xcite that show a dynamic equilibrium between the open and closed conformations of cam s domains in the absence of ca@xmath0support our approach .
calmodulin ( cam ) is a small , 148 amino acid long protein consisting of two topologically similar domains .
each domain consists of four @xmath1-helices and a pair of ef - hand ca@xmath0-binding loops .
the n - terminal domain ( ncam ) has helices labeled a d with binding loops i and ii , and the c - terminal domain ( ccam ) has helices labeled e
h with binding loops iii and iv .
we simulate open / closed allosteric transitions of the isolated domains of cam using a native - centric model implemented in the @xmath2 simulation package developed by takada and co - workers.@xcite this model couples two energy basins , one biased to the open ( pdb : 1cll@xcite ) reference structure and the other biased to the closed ( pdb : 1cfd@xcite ) reference structure .
the energy of a conformation , specified by the @xmath3 position vectors of the c-@xmath1 atoms of the protein backbone , @xmath4 , is given by @xmath5 where @xmath6 is the single basin potential defined by the open structure and @xmath7 is the single basin potential defined by the closed structure .
the interpolation parameters , @xmath8 and @xmath9 , control the barrier height and the relative stability of the two basins
. parameters defining the single energy basins are set to their default values with uniform contact strength .
the simulation temperature is set below the folding transition temperature of each of the four conformations .
specifically , the simulation temperature is set to @xmath10 where @xmath11 is the folding transition temperature corresponding to the closed ( apo ) state of ncam , the lowest transition temperature among the open and closed states of ncam and ccam . equilibrium trajectories of length @xmath12 steps are simulated using langevin dynamics with a friction coefficient of @xmath13 and a timestep of @xmath14 ( in coarse - grained units).@xcite calcium binding to the two ef - hand loops of each domain of cam is modeled implicitly by adding a potential defined from the ligand - mediated contacts in the ef - hand loops of the open ( holo ) conformation @xmath15.\ ] ] here , the sum is over pairs of residues that are each within 4.5 @xmath16 of a ca@xmath0ion and closer than 10.0 @xmath16 in the open ( holo ) conformation .
the binding energy parameters @xmath17 , @xmath18 , and @xmath19 are taken to be the same for each ligand - mediated contact for simplicity .
binding cooperativity is influenced by the relative stability of the unligated open and closed states determined by @xmath9 and the binding free energy determined by @xmath20 . in principle , these parameters can be adjusted to match measured binding properties . in the absence of clear measured constraints , we choose parameters so that the relative stability between the open and closed states are the same for each domain .
the transition barrier height is determined by @xmath8 which is set to @xmath21 for ncam and @xmath22 for ccam .
adjusting @xmath23 for ncam and @xmath24 for ccam while keeping other parameters fixed gives an energy difference between the unligated open and closed states , @xmath25 for both domains .
experimentally , the folding temperatures of the n - terminal and c - terminal domains in intact cam are approximately @xmath26 and @xmath27 , respectively.@xcite connecting to the domain opening kinetics in the intact protein , our simulation temperature corresponds to approximately @xmath28 which is 95% of ncam s simulated folding temperature , and 98% of ccam s simulated folding temperature . for the results reported in this paper , the binding energy parameters
are set to @xmath29 ( default value in @xmath2 ) , @xmath30 and @xmath31 where @xmath32 is the corresponding separation distance in the open ( holo ) reference conformation .
we have performed additional simulations to explore the dependence of binding thermodynamics on the ligand - mediated contact strength and interaction range . at higher values of @xmath17 and @xmath33 ,
the affinities of ligand binding to individual loops increase .
nevertheless , the slope of the titration curve at the midpoint of the transition ( a measure of binding cooperativity ) remains the same ( _ si _
fig.s2 ) .
the simulated conformational ensembles are characterized structurally in terms of local and global order parameters based on the contacts formed in each sampled conformation .
the set of native contacts in the open and closed conformations are separated into three groups : those that occur exclusively in either the open or the closed native structures , and those that are common to both states .
a native contact in a given conformation is considered to be formed provided the distance between the two residues is closer than 1.2 times the corresponding distance in the native conformation .
local order parameters @xmath34 and @xmath35 are defined as the fraction of native contacts involving the @xmath36 residue that occur exclusively in the open and closed native structures , respectively .
overall native similarity is monitored by corresponding global order parameters , @xmath37 and @xmath38 , where the average is taken over the residues of the protein .
we identify metastable conformational basins from minima in the free energy computed through the population histogram parameterized by @xmath39 and @xmath40 .
ligand binding / unbinding events coupled with a conformational change of the protein is modeled within the grand canonical ensemble . throughout the protein s
conformational transitions , the ligation state of each loop is determined stochastically through a monte carlo step attempted every 1000 steps in the langevin trajectory . if the loop is unligated , a ligand is introduced to the binding loop ( @xmath41 ) with probability @xmath42.\ ] ] if the loop is ligated , the ligand dissociates from the binding loop ( @xmath43 ) with probability @xmath44.\ ] ] here , @xmath45 is the chemical potential of a bound ligand . at equilibrium
, @xmath45 equals the chemical potential of the ligand in solution , @xmath46 where @xmath47 is the ligand concentration , and @xmath48 and @xmath49 are the reference concentration and reference chemical potential , respectively . to compute binding curves , a series of simulations
are preformed , each at a different value of the ligand chemical potential .
these simulated titration curves are reported as function of the chemical potential , or equivalently , in terms of the relative ligand concentration defined through @xmath50 where @xmath51 .
this approach with monte carlo acceptance rates given in eq .
[ eq : metropolis1 ] and eq .
[ eq : metropolis2 ] is oriented towards binding thermodynamics from the outset .
takada and co - workers present a different choice motivated by ligand binding kinetics.@xcite instead of introducing a chemical potential , ligand concentration enters their model through a variable binding attempt rate , while the attempt rate of unbinding is fixed .
binding titration curves can also be calculated in this model , but as a function of the binding attempt rate rather than the concentration directly.@xcite
we first consider ca@xmath0binding exclusively to each individual loop by simulating the conformational change of the entire domain while permitting binding only to a single site . as shown in fig.[fig : ncam_ccam_affinity ] ( a ) and fig.[fig : ncam_ccam_affinity ] ( b ) , the bound population as a function of ligand concentration , @xmath52 , follows a typical sigmoidal profile connecting a fully unbound population at low concentration and a fully bound population at high concentration .
the overall binding strength of the individual loops is reflected in the dissociation constant , @xmath53 , shown in table.[tab : single_loop_data ] .
binding affinities of ncam s loops are nearly the same , whereas the affinities of ccam s loops are significantly different , with @xmath54 .
comparing the binding strength of cam s loops , our simulations predict that @xmath55 .
.number of ligand - mediated contacts , dissociation constants , and binding free energies for the loops of cam . [ cols="<,^,^,^,^",options="header " , ] [ tab : equil_const ] although the right hand side of eq.[eq : cab ] can be evaluated directly from simulated populations , it is convenient to take advantage of the parameterization provided by the mwc model since it accurately describes the simulated equilibrium populations . using eq.([eq : prob_only_1_bound][eq : prob_both_bound ] ) with @xmath56 leads to @xmath57 \left [ 1 + \exp(-\beta ( \epsilon + \delta\epsilon^{\mathrm{a } } + \delta\epsilon^{\mathrm{b}}))\right ] } { \left [ 1 + \exp(-\beta ( \epsilon + \delta \epsilon^{\mathrm{a}}))\right ] \left [ 1 + \exp(-\beta ( \epsilon + \delta \epsilon^{\mathrm{b}}))\right ] } , \ ] ] with @xmath58 and @xmath59 .
the computed equilibrium constants , shown in table.[tab : equil_const ] , indicate that ca@xmath0-binding to ccam ( with @xmath60 ) is more cooperative than ca@xmath0-binding to ncam ( with @xmath61 ) in qualitative agreement with experiment.@xcite the cooperative free energy is estimated to be @xmath62 for ccam and @xmath63 for ncam .
the cooperative free energy for ccam is 1.8 times that of ncam in agreement with the experimental measured range of relative free energies of 1.2 3 reported in ref . and ref .. binding thermodynamics determined from experiments that can not distinguish between binding to individual sites
are often reported through the macroscopic equilibrium constants @xmath64 and @xmath65.@xcite the macroscopic equilibrium constants describing the simulated binding thermodynamics are shown in table.[tab : equil_const ] .
the value of @xmath66 for ccam is greater than @xmath66 for ncam by a factor of 1.5 in agreement with the experimentally reported range of 1.2
2.2.@xcite the free energy of binding two ca@xmath0ions can be estimated from the macroscopic binding constants summarized in table.[tab : equil_const ] , @xmath67 .
the simulated relative values of @xmath68 for ccam is approximately 1.5 times the value of @xmath68 for ncam , which is in agreement with experimentally reported value of approximately 1.1 1.3.@xcite taken together , the simulated values of the macroscopic binding constants for cam are in qualitative agreement with those reported from experiments .
the simulations offer a detailed molecular description of ca@xmath0binding as well as insight into the conformational ensembles underlying the binding free energies , @xmath69 and @xmath70 .
fig.[fig : ncam_ccam rms variation ] shows the root mean square fluctuations ( rmsf ) of each residue for the unligated ( closed ) ensemble at low ligand concentration and the fully saturated ( open ) ensemble at high ligand concentration .
focusing on ncam , we see that helix a , the n - terminal end of helix b , and the b - c linker become more flexible upon ca@xmath0-binding , while helix c and helix d show little change in flexibility .
the temperature factors of the corresponding regions in ccam show qualitatively similar behavior .
all four binding loops , on the other hand , become more rigid upon ca@xmath0coordination .
the difference in flexibility upon binding is largest for loop iv due to its large fluctuations in the unligated ensemble .
greater entropic stabilization of loop iv in the unligated state explains its relatively small binding affinity.@xcite furthermore , accounting for differences in loop entropy completes the rationalization of the binding free energies to the loops of cam : while the value of @xmath70 is dominated by the energetic stabilization of binding to the open state , the value of @xmath69 reflects the degree of conformational entropy of the loop in the unligated ensemble .
( black curve ) .
the rmsf curves are calculated for each ensemble after aligning to the open native conformation .
( aligning to the closed conformation give similar curves . ) also shown is the reference fluctuations given in eq.[eq : rms avg ] ( green curve ) . ]
the flexibility of individual residues are local order parameters that characterizes residue - specific conformational changes upon ca@xmath0binding.@xcite to qualitatively understand cam s structural changes along the binding curve , we compare the fluctuations of the @xmath71 residue to a two state reference rmsf , @xmath72 , given by average @xmath73 where the rmsf of the open ensemble , @xmath74 , and the closed ensemble , @xmath75 , are weighted by the fractional occupancy of the binding sites @xmath76 .
the structural ordering of a residue at any concentration can be characterized as early or late compared to mean flexibility @xmath72 evaluated at the corresponding value of @xmath77 .
for example , fig.[fig : ncam_ccam rms variation ] shows the simulated rmsf of each residue at @xmath53 , as well as the reference fluctuations evaluated at @xmath78 . although the ca@xmath0occupancy of the binding loops is only 50% , the local environment of helix a and the b - c linker of ncam as well as corresponding helix e and f - g linker of ccam is already similar to that of the open state ensemble .
this `` early '' transition to the open ensemble is a reflection of the allosteric cooperativity .
in contrast , the average structural order of the binding loops is similar to the weighted average of the open and closed state flexibility .
the exception is the @xmath79-sheet in the c - terminal end of loop iv which takes on the open state structure at higher ligand concentrations .
this `` late '' transition is in harmony with its lower binding affinity .
in this paper , we introduce a method to simulate binding curves involving a protein that undergoes a conformational change upon binding .
this approach allows us to identify the structural origins of binding affinity and to quantify allosteric cooperativity within a simple coarse - grained description of the protein dynamics . in this implicit ligand model ,
the protein conformation modulates the protein - ligand interactions through effective ligand - mediated contacts among residues in the binding site .
the influence of ligand concentration on the effective binding strength is described through its uniform chemical potential . applying this approach to cam
, we find that this model can distinguish the binding properties of the two domains of cam : binding loops i and ii of ncam have similar affinities , while in ccam , binding loop iii has significantly greater affinity than loop iv . the broader range of binding affinities in ccam accounts for its greater cooperativity .
simulated populations of the ligation states as a function of concentration are accurately described by the mwc model with appropriate binding free energies for the individual loops .
these binding free energies are average properties of the simulated ensemble and are not obvious solely from the open and closed structures .
while the simulated binding thermodynamics is well - described by the mwc model , this simple analysis can obscure complexities in the free energy landscape . in separate publication , we describe how subtle differences in the topology and stability of the two domains lead to distinct simulated mechanisms for ca@xmath0-free domain opening for ncam and ccam ( submitted ) .
in particular , we find that ccam unfolds more readily than ncam during the domain opening transition under similar conditions , a result consistent with the lower thermal stability of the c - terminal domain in the intact protein.@xcite although the unfolded conformations play a minor role in the binding thermodynamics described in this paper ( aside from modifying the binding free energies to the open and closed states ) , global folding and unfolding in the domain opening transition likely has a significant qualitative influence on the binding kinetics .
this is a problem we plan to address in future work .
we would like to thank daniel gavazzi for help in figure preparation .
financial support from the national science foundation grant no .
mcb-0951039 is gratefully acknowledged .
47ifxundefined [ 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 10.1016/j.sbi.2004.01.005 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1038/nchembio.232 [ * * , ( ) ] link:\doibase 10.1038/nature06522 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1002/prot.340210302 [ * * , ( ) ] link:\doibase 10.1110/ps.9.1.10 [ * * , ( ) ] link:\doibase 10.1126/science.1169377 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1073/pnas.0802524105 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1110/ps.03259908 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1110/ps.9.8.1519 [ * * , ( ) ] link:\doibase 10.1073/pnas.0804672105 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1006/jmbi.1999.2770 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1016/s0006 - 3495(01)76182 - 6 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1016/j.jmb.2013.03.013 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1006/jmbi.1999.3188 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1021/ja067791a [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1073/pnas.0806872106 [ * * , ( ) ] * supporting information : coarse - grained molecular simulations of allosteric cooperativity *
and the y - axis represents simulated free energy in units of @xmath80 . ] simulated free energy in terms of one - dimensional progress coordinate , @xmath81 , as shown in fig.s1 , illustrates that for both domains , the closed state is more stable in the unligated ensemble .
binding of first ligand stabilizes both the closed and open states but the high affinity open state is stabilized to a greater extent due to its structural compatibility with the ligand . in the fully saturated ensemble
, the open state has greater stability .
for the results presented in the paper , we made specific choices for the ligand - mediated contact strength , @xmath82 , and interaction range , @xmath83 , respectively .
as shown in fig.s2 , with the increase of @xmath84 and @xmath33 , the value of @xmath53 for individual loops decreases . however , the slope of the binding transition curve at a concentration for which @xmath85 remains the same . here , we only show results for binding loop i as an illustration . | interactions between a protein and a ligand are often accompanied by a redistribution of the population of thermally accessible conformations .
this dynamic response of the protein s functional energy landscape enables a protein to modulate binding affinities and control binding sensitivity to ligand concentration . in this paper , we investigate the structural origins of binding affinity and allosteric cooperativity of binding two ca@xmath0ions to each domain of calmodulin ( cam ) through simulations of a simple coarse - grained model . in this model ,
the protein s conformational transitions between open and closed conformational ensembles are simulated explicitly and ligand binding and unbinding is treated implicitly within the grand canonical ensemble .
ligand binding is cooperative because the binding sites are coupled through a shift in the dominant conformational ensemble upon binding .
the classic monod - wyman - changeux model of allostery with appropriate binding free energy to the open and closed ensembles accurately describes the simulated binding thermodynamics .
the simulations predict that the two domains of cam have distinct binding affinity and cooperativity .
in particular , c - terminal domain binds ca@xmath0with higher affinity and greater cooperativity than the n - terminal domain . from a structural point of view , the affinity of an individual binding loop depends sensitively on the loop s structural compatibility with the ligand in the bound ensemble , as well as the conformational flexibility of the binding site in the unbound ensemble . |
et al_. [ 1 ] , introduced the first hierarchical clustering algorithm for wsns , called leach .
it is one of the most popular protocols in wsns .
the main idea is to form clusters of the sensor nodes .
leach outperforms classical clustering algorithm by using adaptive clustering and rotating chs .
this saves energy as transmission will only be performed on that specific ch rather than all the nodes .
leach performs well in homogeneous environment however , it s performance deteriorates in heterogeneous environment .
threshold - sensitive energy efficient network ( teen ) [ 2 ] is a reactive protocol for time critical applications .
the ch selection and cluster formation of nodes is same as that of leach . in this scheme , ch broadcasts two threshold values i.e. hard threshold ( ht ) and soft threshold ( st ) .
ht is the absolute value of an attribute to trigger a sensor node .
ht allows nodes to transmit the event , if the event occurs in the range of interest .
therefore , this not only reduces transmission to significant numbers but also increases network lifetime .
georgios _ et al_. [ 3 ] , proposed a two level heterogeneous aware protocol , consisting of normal and advance ( high energy ) nodes .
it is based on the weighted election probabilities of each node according to their respective energy to become a ch . intuitively , advance nodes have more probability to become a ch than normal nodes , which seems logical according to their energy consumption .
stable election protocol ( sep ) does not require any global knowledge of the network .
the drawback of sep is that it does not consider the changing residual energy of the node hence , the probability of advanced nodes to become ch remains high irrespective of the residual energy left in the node .
moreover , sep performs below par if the network is more than two levels . in [ 4 ] ,
authors proposed distributed energy efficient clustering ( deec ) protocol for wsns .
deec is a clustering protocol for two and multilevel heterogeneous networks .
the probability for a node to become ch is based on residual energy of the nodes and average energy of network .
the epoch for nodes to become chs is set according to the residual energy of a node and average energy of the network .
the node with higher initial and residual energy has more chances to become a ch than the low energy node .
a number of routing protocols have been proposed in the area of wsns .
most of them are based on ch selection . however , not much attention has been devoted towards time critical applications .
most of the routing protocols are for proactive networks .
deec , being a proactive heterogeneous network protocol is not well suited for time critical applications .
teen is a reactive protocol which guarantees that in homogeneous environment , unstable region will be short .
after the death of the first node , all the remaining nodes are expected to die on average within a small number of rounds as a consequence of the uniform remaining energy due to the well distributed energy consumption . on the other hand teen in the presence of high energy nodes
yields a large unstable region . the reason being , all high energy nodes are equipped with almost the same energy however , the ch selection process is unstable and as a result most of the time these nodes are idle , as there is no ch to transmit .
hence , in our research paper we focus on developing a protocol that gives us better results for time critical applications in both environments i.e. homogeneous and heterogeneous environment .
we define following parameters which evaluate and compare the performance of clustering protocols .
it is the time interval when the network starts its operation till the death of the first node .
it is also referred to as stable region or steady state .
it is the time interval from the death of the first node until the death of the last node or till the time the network is dead .
it is referred as unstable region .
it is the time interval from the start of operation of the network till the death of last alive node : @xmath0 this instantaneous measure shows the total number of nodes ( advanced , normal ) alive i.e. the nodes having energy greater than zero .
it is the total rate of data sent over the network from nodes to their respective chs and from chs to base station .
in this section , we describe the functionality and characteristics of our model of a wsns .
we particularly present the setting , the ch selection and how transmissions occur .
teen is the first reactive protocol . in this scheme ,
closer nodes form clusters with a ch to transmit the collected data to one upper layer .
this is same as leach protocol however , at every cluster change time , the ch broadcasts two threshold values i.e hard and st .
ht is the absolute value of an attribute to trigger on its transmitter and report to its respective ch .
ht allows nodes to transmit data , if the data occurs in the range of interest .
therefore , a significant reduction of the transmission delay occurs . moreover , st is the small change in the value of the sensed attribute .
next transmission occurs when there is a small change in the sensed attribute once it reaches the ht .
so , it further reduces the number of transmissions .
deec is a proactive protocol designed for two and multi level heterogeneous networks .
all the nodes use the initial and residual energies to select a ch . the node with higher initial and residual
energy has greater probability to become a ch . in a two - level heterogeneous network , we have two categories of nodes , m.n advanced nodes with initial energy equal to @xmath1 and ( 1-m).n normal nodes , where @xmath2 is the initial energy .
moreover , @xmath3 and @xmath4 are two variables which control the nodes ( advanced or normal ) percentage types and @xmath5 is the total energy in the network
. the value of @xmath5 is given as : @xmath6 the value of the total initial energy of the multi - level heterogeneous networks is given as : @xmath7 the probability of a node to become ch in a two level heterogeneous network is : @xmath8 for normal nodes @xmath8 for advanced nodes this model can be easily extended to multi - level heterogeneous networks : @xmath9 as we are assuming uniformly distributed node , so distance of cluster members from ch is : @xmath10 average distance between base station and ch is : @xmath11 in this section , we describe heer , which improves the stable region for clustering hierarchy process for a reactive network in homogeneous and heterogeneous environment .
we use the initial and residual energies of the nodes to become ch similar to that of deec . it does not require any global knowledge of energy at any election round.when cluster formation is done , the ch transmits two threshold values , i.e @xmath12 and @xmath13 .
the nodes sense their environment repeatedly and if a parameter from the attributes set reaches its @xmath12 value , the node switches on its transmitter and transmits data .
the current value ( @xmath14 ) on which first transmission occurs , is stored in an internal variable in the node called sensed value ( @xmath15 ) .
this reduces the number of transmissions .
now the nodes will again transmit the data in same cluster period when @xmath16 .
that is , if @xmath14 differs from @xmath15 by an amount equal to or greater than @xmath13 , then it further reduces the number of transmissions . figure .
1 shows different states of a cluster i.e. from data sensing to data transmitting .
every node selects itself as a ch on the basis of its initial energy and residual energy . in state
( 1 ) a cluster is formed the node senses its environment continuously until the parameter ( cv ) reaches its ht value . when cv reaches ht value , the nodes become green as shown in the figure in state ( 2 ) .
the node then switches on its transmitter and sends the data to the ch .
the ch aggregates and transmits data to base station .
the cv on which first transmission occurs is stored in sv .
the node , then again starts sensing its environment as shown in state ( 3 ) until the cv differs from sv by an amount equal to or greater than st . when this condition becomes true , the node again switches on its transmitter and sends data to ch.the ch then transmits data to base station as shown in state ( 4 ) of figure . 1 .
* heer performs best for time critical applications in both homogeneous and heterogeneous environment . *
it reduces the number of transmissions resulting in the reduction of energy consumption .
* it increase the stability period and network lifetime .
proactive protocols sense their environment and transmit data periodically .
they consume energy continuously with time due to periodic transmission .
our main focus in proactive protocols is on increasing lifetime , throughput and to decrease energy consumption .
contrary to proactive protocol , reactive protocol is application dependent .
it senses the environment periodically but transmits data only when its @xmath14 reaches to absolute value of the attribute . as data transmission consumes more energy than data sensing , so , in reactive network the throughput can be minimized or maximized as per its application .
the throughput in reactive networks is inversely proportional to the network lifetime or its stability period .
if transmissions are less the stability period and network lifetime will be prolonged as @xmath14 does not reach the absolute value .
however , if the @xmath14 reaches @xmath12 value ( absolute value ) repeatedly then maximum number of transmissions will occur and nodes will die quickly .
in this section , we simulate an environment with varying temperature in different regions .
our field has dimensions of @xmath17 square units .
the number of nodes in the field is @xmath18 .
we assume that base station is in the center of sensing nodes . to evaluate the performance of heer
, we simulate it with teen and deec .
the parameters used in our simulation are listed in table 1 .
our goals in conducting the simulation are as follows .
* we examine the performance of teen and heer for the prolonging of stability period and network life time . *
we also observe the throughput of both the protocols .
we observe the performance of teen , deec and heer under homogeneous environments .
we also examine the sensitivity of our protocol to the degree of heterogeneity in the sensor network .
.parameters used in our simulations [ cols="<,<",options="header " , ] = 100 and @xmath13=2,width=340,height=264 ] = 70 and @xmath13=10,width=340,height=264 ] = 70 and @xmath13=10,width=340,height=264 ] in this section , we compare heer , teen and deec protocols in homogeneous environment .
we observe from figure .
3 that in teen after the death of first node , all the remaining nodes die within a small number of rounds .
this is due to the reason that all the nodes have same probability to become a ch .
deec prolongs the stability period and network life time due to the election of ch on the basis of residual energy .
high residual energy nodes have greater probability of becoming a ch .
the stability period of heer ( hard and soft ) is much longer than that of teen and deec .
heer introduces ht value which decreases the number of transmissions to base station .
this increases the stability period and network lifetime .
st further reduces the number of transmissions resulting in the reduction of energy consumption .
this prolongs the network life time .
heer outperforms deec in terms of stability period and network life time by a factor of 1.78 and 1.60 respectively .
moreover , both stability period and network life time of heer also outperforms teen by a factor of 2.0 . from figure .
5 , we can observe that by changing the values of thresholds , the stability period and network life time of heer ( hard ) changes noticeably .
the stability period and network lifetime decreases if we decrease the value of @xmath19 as shown in figure .
moreover , figure.6 shows that the throughput difference between heer ( hard ) and deec decreases by decreasing the difference between the two thresholds .
st also effects the network lifetime .
if number of transmissions increase , we observe a decrease in network lifetime and vice versa .
in this paper , we present a hybrid reactive protocol of teen and deec for homogeneous environment .
heer minimizes the energy consumption by first distributing load to all high energy nodes and then on to low energy nodes . like teen , it is well suited for time critical applications and is more efficient than teen and deec .
w.heinzelman , a. chandrakasan , and h. balakrishnan , `` energy - efficient communication protocol for wireless sensor networks , '' in the proceeding of the hawaii international conference system sciences , hawaii , january 2000 .
a. manjeshwar and d. p. agarwal , `` teen : a routing protocol for enhanced efficiency in wireless sensor networks , '' in 1st international workshop on parallel and distributed computing issues in wireless networks and mobile computing , april 2001 .
g. smaragdakis , i. matta , a. bestavros , sep : a stable election protocol for clustered heterogeneous wireless sensor networks , in : second international workshop on sensor and actor network protocols and applications ( sanpa 2004 ) , 2004 . | wireless sensor networks ( wsns ) consist of numerous sensors which send sensed data to base station .
energy conservation is an important issue for sensor nodes as they have limited power .
many routing protocols have been proposed earlier for energy efficiency of both homogeneous and heterogeneous environments .
we can prolong our stability and network lifetime by reducing our energy consumption . in this research paper
, we propose a protocol designed for the characteristics of a reactive homogeneous wsns , heer ( hybrid energy efficient reactive ) protocol . in heer
, cluster head(ch ) selection is based on the ratio of residual energy of node and average energy of network .
moreover , to conserve more energy , we introduce hard threshold ( ht ) and soft threshold ( st ) .
finally , simulations show that our protocol has not only prolonged the network lifetime but also significantly increased stability period .
wireless , sensor , networks , energy , hybrid , cluster , reactive |
superclusters are the largest systems of galaxies , composed of multiple clusters of galaxies , extending over 10 - 20 mpc ( e.g. , bahcall , soneira 1984 ; postman , geller , huchra 1988 ; quintana et al
. 1995 ; small et al . 1998 ) .
according to the n - body simulations which successfully reproduce the observed filamentary structures at the local universe ( e.g. , peacock et al .
2001 ) , clusters of galaxies at @xmath10 are still in the process of formation ( e.g. , moore et al .
therefore , clusters of galaxies at this cosmological distance , especially superclusters , are important sites where we can directly see the process of structure formation and evolution and mass assembly to cluster cores . despite the importance in the cosmological context
, however , only a few superclusters with the scale of @xmath910 mpc have been identified so far at high redshifts , especially beyond @xmath10 ( connolly et al .
1996 ; lubin et al .
2000 ; tanaka et al .
the limitation comes largely from the small field of view of large telescopes . the advent of the suprime - cam ( miyazaki et al .
2002 ) , a huge format optical camera with a 30@xmath2 field of view on the prime focus of the 8.2 m subaru telescope , has made it possible for us to view a @xmath910 mpc region all at once at high redshifts .
kodama et al .
( 2001 ) have taken this unique advantage of the suprime - cam on subaru to map out the large scale structure around the a851 cluster at @xmath11 .
they have found many filamentary structures and subclumps spreading over 5 mpc from the dense cluster core , on the basis of photometric redshift ( photo - z ) analysis , an approach similar to that taken in this _
letter_. by utilizing the great light - collection power of the subaru telescope , we now target the most distant supercluster ever firmly identified with spectroscopy : ie . , the lynx supercluster region at @xmath5 composed of two known clusters , rxj 0848.9 + 4452 at @xmath0 and rxj 0848.6 + 4453 at @xmath1 , which are spectroscopically confirmed x - ray emitting clusters firstly reported by rosati et al .
( 1999 ) and stanford et al .
the bolometric x - ray luminosities of rxj 0848.9 and rxj 0848.6 are @xmath12 ergs s@xmath13 and @xmath14 ergs s@xmath13 , respectively , which are derived with the chandra observation ( stanford et al .
2001 ) .
throughout this paper , we use the ab magnitude system ( oke , gunn 1983 ) . the adopted cosmological parameters are @xmath15 and @xmath16 , which gives a physical scale of 8.37@xmath17 kpc arcsec@xmath13 at the cluster redshift .
.log of the observations . [ cols="<,^,^,^,^ " , ] @xmath18 @xmath19 aperture + @xmath20 quoted from stanford et al .
( 2001 ) + @xmath21 the tabulated numbers for gr1 may be significantly affected by the contamination of the blue galaxies in the @xmath22 ( arcmin ) region , and hence + unreliable .
+ @xmath23 the flux ( corrected for the galactic absorption ) and luminosity are given in the rest - frame 0.52 kev band .
they are converted from the 0.52 kev + count rate in the observed frame by assuming an optically - thin thermal plasma model at @xmath1 with a temperature of 6 kev and elemental + abundances of 0.3 solar .
+ @xmath24 3@xmath7 upper limits table [ table2 ] summarizes the properties of the known clusters and the cluster / group candidates .
the j2000.0 coordinates of the each cluster / group candidate are given in columns 2 and 3 .
columns 4 and 5 show the surface number density of all the galaxies and the red galaxies only , respectively , within a circle of @xmath25 mpc radius .
we also only use galaxies at @xmath26 .
the remaining field contamination has been subtracted using the sdf data ( see above ) , and the errors indicate the poisson statistics based on the number of cluster galaxies and that of the subtracted field galaxies . in column 6 , we show the estimated richness of each cluster / group candidate , using the @xmath27 indicator introduced by hill & lilly ( 1991 ) .
this indicates the number of galaxies within a 0.5 mpc radius from the cluster center and within the magnitude range between @xmath28 and @xmath29 measured in the @xmath30 band at @xmath31 .
since the @xmath30 band at @xmath31 roughly corresponds to the @xmath32-band at @xmath5 , we can make a direct comparison with their measurements .
the cluster richness estimated based on @xmath27 is 1 - 2 for cl1 and cl2 , while 0 - 1 for gr1-gr7 ( see table 4 of hill & lilly 1991 ) .
therefore we find that the candidates we newly find are poorer systems than the two known clusters .
this is also suggested from deep x - ray data around the lynx field taken with chandra ( stanford et al .
2001 ; stern et al .
2002 ) . using the chandra image in the 0.52 kev band
, we measure the flux of excess diffuse emission in the circular region around each candidate with a radius of 35@xmath33 .
the background ( the unresolved x - ray background and non x - ray background ) is estimated from a surrounding annular region .
the result is shown in column 8 of table [ table2 ] .
note that gr6 and gr7 are outside of the chandra field .
we can not detect any significant x - ray excess at more than the @xmath34 level for our cluster / group candidates except for the two known clusters .
it should , however , be noted that the @xmath34 upper limit luminosity of each region ( see column 9 of table [ table2 ] ) is comparable to a typical luminosity of nearby groups of galaxies .
thus , the non - detection in the chandra data does not necessarily imply that the regions we find have unusually low x - ray luminosities compared with their optical richnesses .
we argue therefore that our ` suprime - cam imaging + photo - z ' method is a very powerful technique in finding clusters or groups in the distant universe even beyond @xmath35 out to @xmath36 .
we note that this is practically the highest redshift structures of ` normal ' galaxies that can be traced by optical images with our scheme ( bracketing the 4000 - break region in the rest frame ) .
we have presented the results of deep panoramic imaging of the lynx supercluster field at @xmath5 taken with the suprime - cam on subaru telescope .
our multicolour image covers an area of @xmath37 ( @xmath38 mpc ) , allowing us for the first time to investigate large scale structure spreading around the known cluster regions at this high redshift . by applying the photometric - redshift technique ,
we have mapped out the spatial distribution of galaxies near the redshift of the supercluster down to @xmath8 .
we have newly discovered seven cluster / group candidates .
these candidates show red colour - magnitude sequences and centrally concentrated profiles , similar to those of the two known clusters in this field , indicating that they are likely to be real clusters / groups of galaxies around the lynx supercluster , comprising a large - scale structure over @xmath3913 @xmath17mpc scales .
the clumpiness of the candidates suggests that they are in early dynamical stages , being in the process of assembly to a massive cluster
. the next key step will be to confirm with deep spectroscopic observations that the cluster / group candidates we discovered in this study are physically associated with one another and forming a real large - scale structure at @xmath40 .
if confirmed , this will be the highest redshift structure of ` normal ' galaxies ever identified , which will have a large impact on structure formation theories and will bring us invaluable information on the formation and evolution of clusters of galaxies at their early stages .
we thank r. ellis , r. bower , m. balogh , t. yamada and w. couch for useful discussion .
fn acknowledge the financial support from uk pparc .
fn , tk and mo acknowledge support from japan society for the promotion of science through its research fellowships for young scientists .
so and tk acknowledge the daiwa - adrian prize 2001 given by the daiwa anglo - japanese foundation .
this work was financially supported in part by a grant - in - aid for the scientific research ( no.15740126 ) by the japanese ministry of education , culture , sports and science . | we report the discovery of a probable large scale structure composed of many galaxy clumps around the known twin clusters at @xmath0 and @xmath1 in the lynx region .
our analysis is based on deep , panoramic , and multi - colour imaging ( 26.4@[email protected]@xmath2 ) in @xmath4-bands with the suprime - cam on the 8.2 m subaru telescope .
this unique , deep and wide - field imaging data - set allows us for the first time to map out the galaxy distribution in the highest redshift supercluster known .
we apply a photometric redshift technique to extract plausible cluster members at @xmath5 down to @xmath6=26.15 ( 5@xmath7 ) corresponding to @xmath8 at this redshift . from the 2-d distribution of these photometrically selected galaxies , we newly identify seven candidates of galaxy groups or clusters where the surface density of red galaxies is significantly high ( @xmath95@xmath7 ) , in addition to the two known clusters .
these candidates show clear red colour - magnitude sequences consistent with a passive evolution model , which suggests the existence of additional high density regions around the lynx superclusters .
[ firstpage ] galaxies : clusters : individual ( rxj 0848.6 + 4453 , rxj 0848.9 + 4452 ) galaxies : evolution galaxies : formation cosmology : large - scale structure of universe |
the dust content in galaxies is intimately linked to their evolutionary history
. nevertheless , the mass of the dust in the interstellar medium ( ism ) , and its ratio with the gas mass ( dust - to - gas ratio , dgr ) , are critical parameters for establishing the evolutionary state of a galaxy .
however , the complex interplay between dust destruction and dust formation mechanisms ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , makes it difficult to infer evolutionary trends from dust alone .
another signature of evolutionary status is a galaxy s metal abundance , generally quantified by the nebular abundance of oxygen , o / h . although dust grains consist mainly of metals , grains are not the dominant contributor to the metal budget of galaxies @xcite .
nevertheless , the metals in dust and the metals in gas are expected to be coupled through the ism energy cycle , and much work has been focused on comparing dust content with metallicity . indeed , the dgr and o / h seem to be fairly well correlated both in the milky way and in nearby star - forming galaxies .
observations of the dgr compared with the gas - phase metallicity suggest approximate linearity between the two ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , a trend which is generally true even for gradients of the dgr within galaxies , which tend to follow the radial changes in metallicity ( e.g. , * ? ?
* ; * ? ? ?
since the seminal work by @xcite , many galaxy evolution models and models of the ism now _ assume _ that dust content is proportional to metallicity ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in the ism models , this assumption
is directly related to the capability of the gas to self - shield from intense ultraviolet radiation , and is thus crucial for the formation of h@xmath3 . however , new results with _ herschel _ and
alma challenge the assumption of a direct ( linear ) correspondence of the dgr with metal abundance . based on dust masses calculated with _
herschel _
data from the dwarf galaxy survey @xcite , @xcite found that the dgr is linearly proportional to o / h only to 12@xmath4log(o / h)@xmath08 ( 20%@xmath1 ) ; at lower metallicities the dependence of dgr is steeper , implying relatively less dust ( or more gas ) at lower abundances . by including new alma data at 870 , (
* hereafter h14 ) derive vastly different dgrs even at the same metallicity , in particular for the two most metal - poor star - forming galaxies in the local universe , sbs0335@xmath2052 and izw18 , both at @xmath03%@xmath1 . as shown in fig .
[ fig : dgroh ] ( adapted from @xcite , see also sects .
[ sec : atomicgas ] [ sec : moleculargas ] ) the dgrs of these two galaxies differ by almost two orders of magnitude , despite their similar metallicity . in this paper , we examine possible reasons for this dramatic difference in dust content at similar metallicity in these two galaxies .
we first discuss their basic properties in sect .
[ sec : obs ] .
then , in sect .
[ sec : model ] , we explore the origin of the observed dust in sbs0335@xmath2052 and izw18 using a semi - analytic chemical evolution model with dust . the model has been first introduced by @xcite and then applied by @xcite and @xcite within the context of a hierarchical model for the evolution of cosmic structures , as predicted by the concordance @xmath5cdm model .
here we do not investigate the hierarchical evolution of sbs0335@xmath2052 and izw18 , but rather limit the discussion to their chemical evolution assuming that they evolve in isolation with a constant star formation rate ( sfr ) .
below , we only briefly describe the model , referring interested readers to the above papers for a more detailed description .
we discuss implications of our results for high - redshift dust formation in star - forming galaxies in sect .
[ sec : discussion ] , and our conclusions are given in sect .
[ sec : conclusions ] .
the observed properties of the two metal - poor dwarf galaxies that we have adopted for our analysis are presented in table [ tab : data ] .
the observational data and the methods used to infer the physical properties listed in the table have been thoroughly described in h14 and references therein ( see in particular tables 1 and 3 in h14 ) .
h14 analyzed dust and gas surface densities and did not include ionized gas .
here we revise the estimates of gas masses in both galaxies relative to h14 in order to take into account the ionized gas component and the spatial extent of cool dust emission .
the values obtained for the individual ism components and the total gas masses are given in table [ tab : data ] and shown graphically in fig .
[ fig : dgroh ] .
both izw18 and sbs0335@xmath2052 are embedded in vast hi envelopes which include other galaxies or galaxy components .
sbs0335@xmath2052 has a western component sbs0335@xmath2052w at a distance of @xmath022kpc within a 64kpc hi cloud @xcite ; izw18 ( main body ) lies about 2kpc away from the `` c '' component , or zwicky s flare , within a diffuse hi cloud extending over @xmath019kpc @xcite .
thus it is difficult to determine exactly where the hi of the galaxy ends , and the more extended hi envelope begins .
the large beams with which hi is typically observed exacerbate the problem , both because of beam dilution which causes hi surface density to be underestimated , but also because the tiny dimensions of the galaxies compared with their hi envelope make it difficult to assess the hi content of the galaxy itself .
because our interest here is in dust production , and in the cool gas reservoir that provides the fuel for star formation , we have taken hi surface densities from the highest - resolution observations available ( as in h14 ) , and calculated hi mass over the optical extent of the galaxy as determined from ( @xmath6-band ) surface - brightness profiles .
the @xmath6-band brightness profile of izw18 falls to 25magarcsec@xmath7 at a radius of @xmath08arcsec ( 700pc , * ? ? ?
* ) , and a radius @xmath032 for sbs0335@xmath2052 ( 838pc , * ? ? ?
* ) . with mean hi surface densities @xmath8=56@xmath9pc@xmath7 for sbs0335@xmath2052 @xcite and 64@xmath9pc@xmath7 for izw18 @xcite
, we thus estimate total hi masses of 3.1@xmath10@xmath9 and 2.5@xmath10@xmath9 for sbs0335@xmath2052 and izw18 , respectively .
the hi mass for sbs0335@xmath2052 is about 10 times lower than the total hi mass by @xcite in a 40arcsec beam ( about 6 times larger than the optical extent of the galaxy ) , and @xmath04 times lower than the total for izw18 given by @xcite .
using the total hi masses from the literature would lower the resulting dgr by roughly these amounts ( e.g. , * ? ? ?
052 and izw18 , including ionized gas , atomic and ( putative , as yet undetected ) molecular gas ; open circles give dgrs including only the ionized and atomic gas components .
galaxies from @xcite are shown as ( green ) open stars , those from @xcite as ( red ) crosses , and from @xcite as ( magenta ) open circles .
the short dashed line is the linear prediction by @xcite , and the vertical dotted line illustrates solar metallicity 12@xmath4log(o / h)=8.69 @xcite .
the long dashed line ( starting at 12@xmath4log(o / h)=8.0 ) is the empirical fit given by @xcite for low metallicities . ] in low - metallicity star - forming galaxies such as izw18 and sbs0335@xmath2052 , ionized gas constitutes an important part of the total gas mass budget .
hence , we have attempted to determine the mass of the ionized component in the two galaxies .
radio continuum observations of the free - free emission of ionized gas give emission measures , and thus mean densities and source size . the radio spectrum for izw18
is flat , consistent with optically thin emission @xcite , while that of sbs0335@xmath2052 falls at low frequency , indicative of free - free absorption and consequently high ionized - gas density @xcite .
as discussed in detail in sect .
[ sec : densities ] , the density of ionized gas in izw18 is @xmath05@xmath11 , and @xmath03200@xmath11 in sbs0335@xmath2052 , corresponding to emitting regions of @xmath0 390pc radius for izw18 @xcite and @xmath07.7pc radius for sbs0335@xmath2052 @xcite . because ionized gas tends to be clumped @xcite , we need to estimate a volume filling factor over the optical extent of the galaxy .
we have done this by comparing the source size inferred from the radio emission measure to the optical size ( see sect . [
sec : atomicgas ] ) , assuming a spherical geometry for the hii region .
this comparison gives a volume filling factor of 0.17 for izw18 and @xmath12 for sbs0335@xmath2052 .
although the value for izw18 is within those observed for local hii regions ( e.g. , * ? ? ?
* ) , the value for sbs0335@xmath2052 is extremely low .
this could be a consequence ( e.g. , * ? ? ?
* ) of the optically thick gas in this galaxy as indicated by the radio spectrum , so for sbs0335@xmath2052 we have used conservatively a filling factor of @xmath13 , roughly the lowest value found by @xcite . with these densities and filling factors , and the optical size of the galaxy as for hi , we estimate ionized gas masses of @xmath14@xmath9 and @xmath15@xmath9 for izw18 and sbs0335@xmath2052 , respectively .
the mass of the ionized gas is comparable to that of the atomic gas in both galaxies .
if , instead of the radio - determined value of 3200@xmath11 , we consider the ionized gas density of 500@xmath11 for sbs0335@xmath2052 inferred from the optical [ s ii ] lines @xcite , and a larger filling factor of @xmath16 , we would estimate an ionized gas density of @xmath17@xmath9 , 50% of our former estimate .
this can be considered as a rough measure of the uncertainty inherent in this calculation .
co emission has never been detected in either galaxy ; thus it is difficult to measure h@xmath3 masses , independently of the unknown co luminosity - to - h@xmath3 mass conversion factor @xmath18 .
thus , h14 measured the distance from the gas scaling relations in order to estimate the missing ( undetected ) h@xmath3 gas surface density ( for more details , see h14 ) .
using the densities from h14 ( @xmath19=94@xmath9pc@xmath7 and @xmath19=342@xmath9pc@xmath7 , for izw18 and sbs0335@xmath2052 , respectively ) and the optical size of the galaxy as above , we infer h@xmath3 masses for izw18 and sbs0335@xmath2052 of @xmath20@xmath9 and @xmath21@xmath9 , respectively .
these masses are highly uncertain , and assume that sfr surface density at low metallicity follows the same scaling relations as more metal - enriched galaxies .
dust masses for sbs0335@xmath2052 and izw18 were measured by h14 by fitting the optical - to - mm spectral energy distributions ( seds ) with spherical _ dusty _ models @xcite .
three different grain populations were included for determination of the best - fit model ( for more details see h14 ) .
the resulting dust masses for both galaxies differ from those measured by other groups ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
h14 discussed differences relative to @xcite for sbs0335@xmath2052 and @xcite for izw18 ; here we compare our measurements to more recent work although we are unable to compare our dust - mass estimates with @xcite because there is no tabulation of their dust masses .
@xcite use a multiple - temperature modified blackbody ( mbb ) approach , and fix the emissivity index for all components .
however , they only discuss the warm and cold dust ( not the hot ) because , as they note , the hot dust ( for @xmath22 m ) is not in thermal equilibrium .
considering their two - component fits , the seds shown in their fig .
6 almost never pass through the longer wavelength points .
this is a consequence of the assumption of single - temperature mbbs , and is expected to bias dust - mass estimates toward lower masses .
although seemingly a small deviation , the constraints offered by longer wavelengths significantly raise dust mass estimates , especially at low metallicity ( e.g. , * ? ? ?
it is well known that including short wavelengths in a single - temperature mbb fit gives unrealistically low dust masses because the mean temperature of the dust radiating at these short wavelengths is higher than the bulk of the cooler dust which dominates the mass .
the warm dust dominates the light but the cool dust dominates the mass and a single temperature ( or even two temperatures as in * ? ? ?
* ) are not able to accommodate the temperature gradients .
the cold - dust mass found by @xcite for izw18 is 70% of that found by h14 ; compensating for the differences in dust emissivities worsens slightly the discrepancy to @xmath065% .
this value is also a factor of 2 below the lower limit of the range of dust mass estimated for izw18 by @xcite who use models by @xcite , even after compensating for the different emissivities adopted by the two groups .
more physically realistic models which contemplate dust temperature gradients through variations in the interstellar radiation field that heats the dust tend to give larger dust masses than single - temperature mbb fits . for sbs0335@xmath2052 , @xcite find a cold dust mass of dex3.06@xmath9 , roughly a factor of 30 below h14 . however , they estimate a cold " temperature of 57k , very close to the temperature of 59k discussed by h14 under the assumption of single - temperature dust , and thus do not take into account any cool dust .
moreover , the dust in sbs0335@xmath2052 is optically thick at short wavelengths @xcite ; thus the assumption of optically thin dust emission implicit in mbb fitting is incorrect .
the differences found in dust masses for sbs0335@xmath2052 and izw18 of roughly a factor of 100 are consistent with the differences in their integrated ir luminosity : @xmath23 for sbs0335@xmath2052 is @xmath24@xmath25 , while for izw18@xmath23=@xmath26@xmath25 .
assuming the same mass - to - light ratio ( similar overall mean temperatures ) for the dust , this would give a factor of 80 in dust mass , inconsistent with the difference found by @xcite of roughly a factor of 5 for the cold dust masses in the two galaxies .
this illustrates one of the difficulties of the mbb approach adopted by @xcite ; they use two temperatures and thus apply two different mass - to - light ratios ( roughly inverse temperature ) to the two ir integrals . as mentioned above , in galaxies like sbs0335@xmath2052 and izw18 most of the dust _ emission
_ comes from warm dust , while most of the dust _ mass _ is cold .
more emissive grain mixtures , as suggested by @xcite , would decrease the inferred dust mass by a factor 3 - 4 in both galaxies .
a comparable reduction of the dust mass of sbs0335@xmath2052 ( @xmath27@xmath9 ) would be obtained by artificially imposing a low 870 dust flux ( 1@xmath28 below the reported h14 flux ) .
we will discuss the implication of this reduction in dust mass for our results in sect .
[ sec : model ] .
figure [ fig : dgroh ] shows the dgrs of izw18 and sbs0335@xmath2052 , together with samples of galaxies taken from the literature ( with dust masses obtained through sed fitting rather than mbb approximations ) . because the dust - mass estimates for these samples are based on the @xcite models , the dust emissivities are comparable among the samples . figure [ fig : dgroh ] also shows the steeper - than - linear slope found by @xcite for galaxies with oxygen abundance 12@xmath4log(o / h)@xmath298.0 .
the dgr of sbs0335@xmath2052 is roughly consistent with a linear slope of dgr with o / h ( e.g. , * ? ? ?
* ) , while izw18 follows the steeper slope found by @xcite .
stellar masses in star - forming dwarf galaxies such as sbs0335@xmath2052 and izw18 are notoriously difficult to determine .
the main problem is contamination by nebular continuum emission which affects both the optical ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and the near - infrared emission ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . in sbs0335@xmath2052 ,
the contamination from free - free emission at 3.4 is 27% @xcite and @xmath050% at 2.2 @xcite ; even more extreme contamination is observed in izw18 , with @xmath3050% of the irac 4.5flux due to nebular emission @xcite .
hot dust is also a problem , especially in sbs0335@xmath2052 where it comprises @xmath067% of its 4 emission @xcite . while nebular continuum levels can be estimated from sfrs ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , it is difficult to ascertain hot - dust levels without detailed multi - wavelength photometry .
thus stellar masses of low - metallicity dwarf starbursts such as sbs0335@xmath2052 and izw18 are prone to large uncertainties .
table [ tab : data ] reports in parentheses the values of stellar mass , m@xmath31 , and age inferred from _ dusty _ sed fitting ( h14 ) , and sfrs that correspond to the constant values needed to produce the _ dusty _ sed best - fit m@xmath31when integrated over the best - fit age .
however , for our analysis , we require a _
age for the stellar populations producing the dust currently observed . a mean age of stellar clusters of 6 myr has been calculated by averaging values from @xcite and @xcite for izw18 and from @xcite for sbs0335@xmath2052 .
thus , we have also computed the stellar mass accumulated over the timespan of the mean ages of the clusters at a constant rate given by the observed values of sfr derived from radio free - free emission @xcite .
all these values ( not in parentheses ) are also reported in table [ tab : data ] .
the resulting m@xmath31 values are within 40% of previous estimates obtained by fitting the optical - nir seds of individual star clusters with single stellar population models ( sbs0335@xmath2052 : @xmath32@xmath9 , izw18 : @xmath33@xmath9 , * ? ? ?
* ; * ? ? ?
* respectively ) .
only if age priors ( 3 - 6myr ) are imposed for the fit are the _ dusty _ stellar masses obtained by h14 consistent with these values for sbs0335@xmath2052 .
there is a similar discrepancy for the _ dusty _ stellar mass of izw18 at an age of 18.3myr , almost a factor 2 larger than the value obtained with a constant sfr of 0.17@xmath9yr@xmath34 .
part of the reason for this is that , at a given luminosity , the mass - to - light ratios are smaller for younger stellar populations ( see h14 for details ) .
another reason is that there is an age gradient in both galaxies , and the sed fitting relies on global photometry that encompasses both young and old clusters . in sbs0335@xmath2052 ,
the northern super - star clusters ( sscs ) are older with a maximum age of @xmath012myr , compared to @xmath293myr for the southern ones @xcite ; in izw18 , the se cluster and c component are older ( @xmath35myr ) compared to the younger nw cluster ( @xmath03myr , * ? ? ?
such a difference in the stellar ages can have important implications for the chemical evolution of the system . in fig .
[ fig : stellartime ] we show the stellar mass - lifetime relation for stars with initial metallicity @xmath36=0.03 , using the @xcite formulation .
the shaded regions illustrate the ranges of stellar masses that contribute to metal enrichment by means of core - collapse sn explosions and stellar winds from intermediate mass stars .
the vertical grey lines indicate the values of stellar ages shown in the table ( dashed lines are the _ dusty _ stellar ages ) .
the figure shows that part of the metals and dust that we presently observe may have originated _ in situ _ , from supernova explosions ( _ self - enrichment _ scenario ) .
however , if we adopt the mean age of the stellar clusters ( solid line ) , only the most massive supernovae with 34@xmath9@xmath3740@xmath9 have evolved to their metal production stage .
hence the ism of the galaxies must have achieved most of its metal content prior to the current star - formation episode ( _ pre - enrichment _ scenario ) .
if , instead , the stellar population age for sbs0335@xmath2052 ( izw18 ) is 13.1myr ( 18.3myr ) , as inferred from _ dusty _ sed fit , then the mass range of the stars that can contribute to the ism enrichment extends to 16@xmath9@xmath3740@xmath9 ( 13@xmath9@xmath3740@xmath9 ) .
[ fig : cosmicyields ] quantifies this difference in terms of the mass of metals and dust that can be produced by self - enrichment .
it shows that depending on the adopted stellar age the mass fraction of metals ( dust ) relative to the stellar mass is in the range @xmath38 ( @xmath39 ) for sbs0335@xmath2052 and in the range @xmath40 ( @xmath41 ) for izw18 . using the corresponding stellar masses reported in table [ tab : data ] , we find that even assuming that all the newly formed dust injected in the ism is conserved the dust mass produced by self - enrichment is always smaller than observed , being @xmath42 for sbs0335@xmath2052 and @xmath43 for izw18 . as explained below , these values
have been obtained using the dust and metal yields presented by @xcite and assuming that the stars form at a constant rate with a salpeter initial mass function ( imf ) in a mass range 0.1@xmath9@xmath44 100 @xmath9 with a metallicity of @xmath45 .
= 0.03 ( solid black line ) .
the orange and blue regions indicate the progenitor mass ranges of core - collapse sn and agb stars , respectively . the solid and dashed vertical lines
indicate the estimated stellar ages for the two galaxies , based on individual super stellar clusters fits and on _ dusty _ sed fits , respectively ( see text ) . ]
the remaining quantities required for our models are the mean density and temperature of the cold neutral medium ( cnm ) and molecular clouds in which dust grains form and grow by accretion ( see sect.[sec : model ] ) .
we can derive the number densities of the molecular phase by assuming thermal pressure balance at the atomic - molecular interface ( e.g. , * ? ? ?
observationally and from a theoretical point of view , the mean number density of molecular clouds , @xmath46 , is expected to be @xmath47 times that in the cnm of the ism : @xmath46=@xmath48 .
we can therefore calculate approximate values of @xmath49 in sbs0335@xmath2052 and izw18by first considering the observed hi column densities524pc ( 2 ) for sbs0335@xmath2052and @xmath0436pc ( 5 ) for izw18 . ] : @xmath50cm@xmath7 and @xmath52cm@xmath7 for sbs0335@xmath2052 and izw18 , respectively @xcite .
these can be converted to mean volume densities over the regions of interest ( the area subtended by the massive star clusters ) by considering the diameters of the star - forming region : @xmath5315.8pc for sbs0335@xmath2052 @xcite and @xmath53170pc for izw18 @xcite . we thus obtain a mean molecular density @xmath46@xmath541435@xmath11 and @xmath46@xmath5491@xmath11 for sbs0335@xmath2052 and izw18 , respectively .
despite the similar metallicities of the two galaxies , the inferred molecular densities differ by more than an order of magnitude .
nevertheless , these estimates are roughly consistent , given the uncertainties , with the densities that would be inferred from the putative ( unobserved ) molecular component discussed by h14 .
assuming that the kennicutt - schmidt law relating gas and star - formation surface densities holds also for these two galaxies , and with the same sizes as above , we would derive @xmath46@xmath551800@xmath11 for sbs0335@xmath2052 and @xmath46@xmath5540@xmath11 for izw18 .
the radio spectrum also shows evidence for such a difference in gas densities in the two galaxies .
while in sbs0335@xmath2052 there is the clear signature of a strongly absorbed thermal component , both globally @xcite and around the individual southern sscs @xcite , izw18 shows a typical flat bremtrahlung spectrum @xcite .
based on fits of the radio spectrum , the ionized - gas densities inferred for the two objects are @xmath03200@xmath11 for sbs0335@xmath2052 @xcite , and @xmath5610@xmath11 for izw18 @xcite .
finally , even the optical spectra of the two objects show a large difference in the densities measured from the [ sii ] lines : @xmath57500@xmath11 for sbs0335@xmath2052and @xmath29100@xmath11 for izw18 @xcite . given the uncertainties in the above estimates , in what follows we take the reference values of 1500@xmath11and 100@xmath11 for the molecular gas densities in sbs0335@xmath2052 and izw18 , respectively .
following @xcite , we can use the temperature - density relation for the cnm predicted by @xcite ism model to find @xmath58 and then derive the temperature of the molecular gas implied by thermal pressure balance , @xmath59 .
the resulting values for the two galaxies are shown in table [ tab : data ] .
the equations describing the chemical evolution of a galaxy that evolves in isolation ( closed - box approximation ) , can be summarized as follows : @xmath60 @xmath61 @xmath62 @xmath63 where @xmath64 is the stellar mass , @xmath65 is the total mass in the ism ( the sum of the gas and dust masses ) , @xmath66 is the total mass in heavy elements ( both in the gas phase and in dust grains ) , @xmath67 is the dust mass ( so that the mass of gas phase elements is given by @xmath68 ) , @xmath69 is the ism metallicity , and @xmath70 is the total dust abundance in the ism .
the terms @xmath71 , @xmath72 and @xmath73 are the rates at which the mass of gas , heavy elements and dust is returned to the ism after stellar evolution , respectively .
these time - dependent terms depend on the adopted model grids and stellar imf .
we compute them as follows : @xmath74
@xmath75 @xmath76 where
the lower limit of integration , @xmath77 , is the mass of a star with a lifetime @xmath78 ; @xmath79 and @xmath80 are respectively the remnant mass , the metal and dust mass yields , which depend on the stellar mass and metallicity ; and @xmath81 is the stellar imf , which we assume to be a salpeter law in the mass range 0.1@xmath9@xmath82100@xmath9 ( e.g. , * ? ? ?
* ) . for stars with @xmath83@xmath9
, we adopt metal yields from @xcite and dust yields for intermediate - mass stars on the agb phase of the evolution by @xcite . for massive stars ( 12@xmath9@xmath8440@xmath9 ) metal and dust yields have been taken from @xcite and from @xcite , using the proper mass - and metallicity - dependent values and including the effect of the reverse shock on dust survival in sn ejecta . for stars in the intermediate mass range 8@xmath9@xmath8512 @xmath9 , we interpolate between the agb yields for the largest - mass progenitor and sn yields from the lowest - mass progenitor . above 40 @xmath9 , stars are assumed to collapse to black hole without contributing to the enrichment of the ism . finally , the last two terms in the right - hand side of eq .
( [ eq : dustevo ] ) represent the effects of dust destruction by interstellar shock waves and grain growth in the dense phase of the ism . here
we simplify the treatment of the two - phase ism model described in @xcite , and quantify the fraction of ism in the cold phase with the parameter @xmath86 .
this is taken to be time - dependent and rescaled from the sfr , @xmath87 where @xmath88 is the star formation efficiency @xcite , @xmath89 is the free - fall timescale at the mean density of molecular clouds @xmath90 . finally , the timescale for grain destruction , @xmath91 and grain growth @xmath92 are computed as in @xcite .
the latter timescale depends on the gas phase metallicity , temperature and density of molecular clouds , @xmath93 where we have assumed that grains which experience grain growth have a typical size of @xmath94 @xcite . if @xmath95 and @xmath96 , the accretion timescale for gas at solar metallicity is only 2 myr @xcite .
below we apply the chemical evolution model to each of the two galaxies under investigation . throughout the following ,
we adopt a solar metallicity of @xmath97 @xcite .
the observed mass of metals in sbs0335@xmath2052 is @xmath98 \ , \ , m_\odot$ ] ( the lower limit corresponds to pure atomic gas and the upper one to the total gas mass , see table [ tab : data ] ) .
the observed dust mass is @xmath99 .
hence , the total mass in heavy elements @xmath100 ranges from @xmath101 to @xmath102 . as described in sect .
[ sec : obs ] , the maximum dust mass that can be produced by self - enrichment from observed stellar populations in sbs0335@xmath2052 is @xmath103 , @xmath104 times lower than the estimated dust mass reported in table [ tab : data ] @xcite and @xmath105 lower than the lowest limit on the dust mass reported by @xcite .
it is therefore unlikely that the origin of the dust in sbs0335@xmath2052 is self - enrichment .
in fact , the same stars responsible for metal pre - enrichment may have also formed dust . alternatively , the dust mass may have formed _ in situ _ by means of grain growth .
we examine each of these two possibilities in turn .
we first assume that the stars observed in sbs0335@xmath2052 , with a total mass of @xmath106 , have a mean age of 6 myr and have formed at a constant rate of @xmath107/yr with a salpeter imf ( see the values in table [ tab : data ] ) . under these conditions , only stars with masses @xmath108 had the time to evolve and the total mass of heavy elements and dust injected in the ism by sne is @xmath109 and @xmath110 , respectively , much smaller than observed . from eq .
( [ eq : metevo ] ) , we can estimate the mass of metals that was originally present in sbs0335@xmath2052 , @xmath111 , where @xmath112 .
this is equal to the mass presently observed , corrected for astration and self - enrichment , @xmath113 , \label{eq : premetsbs}\ ] ] and we find @xmath114 . assuming the salpeter imf - averaged dust and metal yields for a fully - evolved stellar population with a metallicity @xmath45 ( @xmath115 and @xmath116 ) , the maximum mass of dust that sbs0335@xmath2052 could have achieved by pre - enrichment can be estimated as @xmath117 which is only @xmath118 of the observed value .
we can repeat the same calculation but assuming that the observed stars in sbs0335@xmath2052 have a mean age of 13.1 myr , a total stellar mass of @xmath119 and have formed with a constant sfr of @xmath120/yr with a salpeter imf ( see the values in parenthesis in table [ tab : data ] ) . in this case , newly formed stars with masses @xmath121 produce @xmath122 of heavy elements and @xmath123 of dust , too small to account for the observed dust mass . using eq .
( [ eq : premetsbs ] ) , we find that the mass of heavy elements achieved by means of pre - enrichment at @xmath124 is @xmath125 , which corresponds to a maximum dust mass of @xmath126 ( from eq .
[ eq : predustsbs ] ) , @xmath127 of the observed value .
it is important to stress that the above values of @xmath128 should be regarded as upper mass limits .
in fact , if we were to consider only the atomic gas rather than the total ( including the putative molecular component ) , the dust masses allowed by pre - enrichment would be even lower , @xmath129 in both cases . moreover , eq . ( [ eq : predustsbs ] )
is based on the implicit assumption that all the dust produced by previous stellar generations has been conserved in the ism , without undergoing any destruction by interstellar shocks .
hence , pre - enrichment can not account for the observed dust masses , even if we were to consider a factor 3 - 4 reduction in the observed dust mass , either by using more emissive grain mixtures @xcite , or by artificially lowering the observed 870 dust flux in the sed ( see sect .
[ sec : dustmass ] ) . independently of the adopted stellar ages , a major fraction of the existing mass of dust in sbs0335@xmath2052
must have formed by means of grain growth in the dense phase of the ism . in figure
[ fig : sbss0335evo ] we show the results of chemical evolution models .
since it is impossible to constrain the initial value of dust mass that sbs0335@xmath2052 has inherited from previous stellar generations , we have explored two limiting cases : in the first one , we take the observed gas density of @xmath46@xmath01500@xmath11 ( see table [ tab : data ] ) and we start from the minimum initial dust mass that allows the model to reproduce the observations ( red shaded region between the two solid lines ) .
this model shows that if the stellar age is 6 myr , sbs0335@xmath2052 must start with the maximum possible initial dust mass predicted by eq .
( [ eq : predustsbs ] ) , grain growth accounts for @xmath130 of the existing dust mass , with pre - enrichment providing the remaining @xmath118 .
if the stellar age is 13.1myr , due to the longer time available for grain growth , sbs0335@xmath2052 can start with a dust mass that is @xmath131 of the maximum value achieved by pre - enrichment and grain growth accounts for @xmath132 of the observed dust mass . in the second model ( blue shaded region between the two shaded lines )
, we assume an initial dust mass of only @xmath133 , as if the chemical initial conditions inherited from previous stellar generations were not favourable to dust pre - enrichment . under this pessimistic scenario , a gas density of @xmath134 ( @xmath135 ) would allow grain growth to increase the dust mass , reaching the observed value in 6myr ( 13.1myr ) .
the above results have been obtained solving the system of equations ( 1 ) - ( 4 ) assuming an initial gas mass @xmath136 ( closed - box approximation ) and fixing the free parameters @xmath86 and @xmath88 to reproduce the mass of molecular gas component at @xmath137 . for @xmath46@xmath01500@xmath11
this constrains the star formation efficiency to be @xmath138 and the fraction of molecular gas to be @xmath139 .
hence we conclude that , provided that a major fraction of the dense gas in sbs0335@xmath2052 is at densities @xmath140@xmath141@xmath11 , the observed dust mass in sbs0335@xmath2052 can be reproduced by means of grain growth .
+ we next analyse the evolution of izw18 .
the observed mass of metals and dust are @xmath142 \times 10 ^ 4 \ , m_\odot$ ] ( the lower limit corresponds to pure atomic gas and the upper one to the total gas mass , see table [ tab : data ] ) and @xmath143 , resulting in a total mass of heavy elements of @xmath144 \times 10 ^ 4 \ , m_\odot$ ] . depending on the adopted stellar ages ( @xmath145 ) , between 4 and 60% of the observed metal mass can be achieved by self - enrichment , the remaining fraction must have come from previous stellar generations .
the same is true for the dust mass : even assuming that all the dust injected by sne in the ism is conserved , self - enrichment can produce between @xmath146 and @xmath147 of the existing dust mass . using eqs .
( [ eq : premetsbs ] ) and ( [ eq : predustsbs ] ) , the initial mass of heavy elements and dust can be estimated to be : @xmath148 where the upper ( lower ) initial values have been obtained assuming a stellar age of 6 ( 18.3)myr and considering only atomic versus total gas mass .
because the initial maximum dust masses are a factor 2 4 larger than the observed value , all of the existing dust mass in izw18 could originate from dust pre - enrichment .
figure [ fig : izw18evo ] shows the results of chemical evolution models .
similarly to the case of sbs0335@xmath2052 , we solve the system of equations ( 1 ) - ( 4 ) in the closed - box approximation with parameters @xmath86 and @xmath88 fixed to reproduce the mass of molecular gas component at @xmath137 .
we first fix the gas density to its observed value , @xmath46=100@xmath11 , and we run the model starting from the minimum initial dust mass that allows the observed dust mass to be reproduced ( red shaded region between the two solid lines ) .
this implies a star formation efficiency @xmath149 and a molecular gas fraction of @xmath150 .
it is clear that grain growth is negligible and that the required initial dust mass is smaller when the stellar age is 18.3myr , due to the larger contribution given by self - enrichment .
according to this scenario , the system starts with an initial dust mass of @xmath151 \ , m_\odot$ ] , and then evolves very slowly , with dust injected by newly - formed stars , partly compensated by grain destruction by interstellar shocks and astration . as a result ,
more than @xmath152 of the observed dust mass is inherited from pre - enrichment , with grain growth making up the rest .
this is different from the solid lines shown in fig .
[ fig : sbss0335evo ] , and a consequence of the lower gas density observed in izw18 , which causes grain growth to be inefficient .
the dashed lines in fig .
[ fig : izw18evo ] show the evolution when by construction
we impose grain growth to provide the dominant contribution to the final dust mass . in this case
, we can start from a negligible initial dust mass but the required gas densities are in the range @xmath153@xmath11@xmath154@xmath155@xmath11 , values which are not supported by observations of any of the gas phases .
052 is compared with the observed value ( grey data point ) .
the upper grey shaded region indicate the range of variation of the inferred stellar ages . the maximum initial dust mass achieved by pre - enrichment is illustrated by the solid red line in the lower grey shaded region
the red region illustrate the range of models where the observed dust mass is reproduced adopting @xmath46@xmath01500@xmath11 to compute the grain growth timescale and requiring an initial dust mass from pre - enrichment .
the blue region illustrate models where dust pre - enrichment is negligible and we adjust the density of dense gas so that the observed dust mass is reproduced by means of grain growth ( the required range of @xmath156 is also shown , see text ) . ]
but for izw18 . ]
the strong dependence of grain growth and the assembly of dust mass on ism density has profound implications for early galaxy evolution . to date , only one star - forming galaxy at @xmath157 , a1689-zd1 with @xmath158 , has a clear detection of dust emission @xcite .
this galaxy has a ( lensing - corrected ) stellar mass @xmath159@xmath9 , and sfr@xmath160@xmath9yr@xmath34 . with an estimated dust mass of @xmath161@xmath9 ,
the dust - to - stellar mass ratio for this galaxy is relatively high , @xmath00.02 , comparable to the highest values found for local star - forming galaxies at similar masses ( e.g. , * ? ? ?
the age of a1689-zd1 is estimated to be @xmath080myr , at a redshift when the universe was @xmath29500myr old . despite extensive efforts , dust emission in other star - forming galaxies at comparable redshifts ( @xmath162 )
has not yet been detected ( e.g. , iok-1 , z8-gnd-5296 : * ? ? ?
* ; * ? ? ?
thus , we predict that a1689-zd1 will be found to have a dense ism , and thus able to assemble a considerable dust mass over the short times available at these redshifts .
our findings suggest that dust mass can be a sensitive indicator of the physical conditions in the ism , and that metallicity alone is insufficient to determine the amount of dust .
in this paper we have investigated the origin of the observed dust masses in the two most metal - poor local dwarf galaxies , sbs0335@xmath2052 and izw18 . despite their comparable metallicities , gas and stellar masses
, these two galaxies show a huge variation in their dust content , with a dust mass of @xmath163@xmath9 in sbs0335@xmath2052 , as inferred by recent alma observations ( h14 ) , and a dust mass of only 340@xmath9 in izw18 . by means of a chemical evolution model with dust
, we find that : * the observed stellar population in sbs0335@xmath2052 can not account for the existing metal and dust masses , hence previous stellar populations must have pre - enriched the ism of the galaxy . * even assuming the maximum possible stellar dust yields , the same stars which have pre - enriched the ism of sbs0335@xmath2052 could have injected a dust mass which is at most 15% of the observed value .
hence , a major fraction of the observed dust mass must originate _ in situ _ through grain growth . *
the observed gas density of sbs0335@xmath2052 is large enough to activate efficient grain growth .
if @xmath46= 1500@xmath11 , grain growth can account for more than 85% of the existing dust mass , with dust pre - enrichment making up the rest .
* despite the longer age spread estimated for the stellar population in izw18 , only between 4 and 60% of the observed metals are produced by self - enrichment through sn explosions of massive stellar progenitors ( @xmath164@xmath9 ) .
hence , the metallicity has been mostly inherited by previous stellar generations . * due to the smaller gas density of izw18 , grain growth is very inefficient .
since dust injected by newly - formed stars is partly compensated by grain destruction by interstellar shocks and astration , more than @xmath165 of the existing dust mass is inherited from pre - enrichment , with grain growth making up the rest .
since dust grains can be efficiently destroyed by interstellar shocks , it is very hard to predict the inital mass of dust that can be assembled by pre - enrichment .
for this reason , we have also explored a limiting case where the galaxies start with a negligible dust content and achieve all of their dust mass by means of grain growth .
this requires the gas density to be larger than inferred from observations : @xmath166@xmath11 for sbs0335@xmath2052 and @xmath153@xmath11@xmath167@xmath11 for izw18 . while for sbs0335@xmath2052
@xmath168 is within the values estimated by johnson et al .
( 2009 ) for the ionized gas densities ( see their table 4 ) , no observations for izw18 suggest similarly high gas densities .
however , independently of these considerations , our study suggests that the widely different dust masses in sbs0335@xmath2052 and izw18 reflect the different efficiencies of grain growth in their ism , which given the comparable metallicity is likely to originate from the different gas densities of the two galaxies .
we thank luca graziani for his insightful comments and robert nikutta for a careful analysis of the sed fitting .
the research leading to these results has received funding from the european research council under the european union s seventh framework programme ( fp/2007 - 2013 ) / erc grant agreement n. 306476 .
lh is grateful to funding from inaf - prin 2012/2015 . | although dust content is usually assumed to depend uniquely on metallicity , recent observations of two extremely metal - poor dwarf galaxies have suggested that this may not always be true . at a similar oxygen abundance of @xmath03%@xmath1 , the dust - to - gas and dust - to - stellar mass ratios in sbs0335@xmath2052 and izw18differ by a factor 40 - 70 according to including molecular gas or excluding it . here
we investigate a possible reason for this dramatic difference through models based on a semi - analytical formulation of chemical evolution including dust .
results suggest that the greater dust mass in sbs0335@xmath2052 is due to the more efficient grain growth allowed by the high density in the cold interstellar medium ( ism ) , observationally inferred to be almost 20 times higher than in izw18 .
our models are able to explain the difference in dust masses , suggesting that efficient dust formation and dust content in galaxies , including those with the highest measured redshifts , depend sensitively on the ism density , rather than only on metallicity .
[ firstpage ] ism : abundances ism : evolution galaxies : starburst galaxies : dwarf galaxies : evolution galaxies : individual : sbs0335 - 052 galaxies : individual : izw18 |
the interaction between matter and the gravitational field has evaded a complete quantum description since the first attempts to formulate a quantum theory of gravity more than 60 years ago .
lacking a satisfactory quantum description of the gravitational interaction , quantum field theory in curved spacetimes ( which links general relativity with quantum field theory ) is thus far the most satisfactory framework to describe the interaction of quantum fields with the space - time curvature . as of today
none of its predictions has been experimentally confirmed beyond analogue gravity @xcite , and bringing those effects within experimental reach is a matter of great interest @xcite .
one of the chief predictions of quantum field theory ( qft ) in curved spacetimes is the well - known unruh effect @xcite .
it dictates that a detector with constant acceleration @xmath0 in free space , in which the field is in the minkowski vacuum , will experience a response equivalent to its submersion into a heat bath with a temperature proportional to its acceleration .
this phenomenon is intrinsically related to the so - called hawking effect @xcite , and understanding it is essential in order to investigate more complex phenomena such as black hole dynamics and possible quantum corrections to relativistic gravity .
the first derivations of the unruh effect ( based on the characterization of the minkowski vacuum in a unitarily inequivalent rindler quantization scheme via bogoliubov transformations ) are not above criticism .
a number of very strong assumptions have to be made in order to justify the observation of a thermal bath by an accelerated observer in the minkowski vacuum .
for example , the infinite amount of energy required to sustain the eternal rindler trajectory .
it has also been argued that difficulties arise in defining a minkowski vacuum when boundary conditions are specified on the scalar field on a manifold @xcite . despite these criticisms ,
later results in the context of axiomatic quantum field theory @xcite were used provide a model - independent derivation of the unruh effect .
furthermore , derivations based on accelerated particle detectors @xcite that appeared soon after the original derivation reassert the importance of the unruh effect .
the standard approach in grappling with the difficult computational problem of an arbitrary system interacting with a scalar field on a curved spacetime manifold is to use an unruh - dewitt detector @xcite .
this simplified detector model considers a two - level system coupled to a scalar field with a monopole interaction of the form @xmath1,\ ] ] where @xmath2 is the monopole - moment of the detector , @xmath3 $ ] is the field operator evaluated along the worldline of the detector , and @xmath4 is a number - valued function that represents the strength and time - dependence of the coupling .
it has been shown that , although simple , this hamiltonian is a good model of the light - matter interaction when no exchange of angular momentum is involved @xcite .
typically the unruh - dewitt model is used within the framework of perturbation theory , and very often restricted to lowest - order calculations ( see , for example , @xcite ) .
however , perturbation theory is not always applicable and breaks down when analyzing scenarios involving strong coupling , large energy , or large time scales . while it is relatively easy to perturbatively
show that the response of an accelerated detector to the vacuum state is planckian @xcite , perturbation theory is not the most appropriate approach to study the thermalization properties of the detector . in practice
this is mainly because higher orders of perturbation theory would be required , increasing the calculational complexity even beyond those of non - perturbative methods . more importantly
, thermalization is an equilibrium result achieved over the course of long time scales . in general
, such time scales will not be accessible to perturbation theory since the perturbative parameter , i.e. @xmath5 becomes larger as time increases .
since thermalization is , in general , an equilibrium process that requires analysis in the limit @xmath6 , reasonable criticism may be raised about a perturbative claim of thermalization . indeed , to check whether or not the detector evolves to an exactly thermal state
, we will consider long time - scale evolution combined with adiabatic switching ( to perturb the system the least when the interaction is switched on ) .
concretely , one should not only check that the probability of excitation of the detector has a planckian response , one should also check to what extent the state of the detector becomes thermal if the detector is carefully switched on and if the interaction lasts for long enough times .
this requires a complete calculation of the detector s density matrix ; it is a common misconception that a detector s planckian response implies that the detector thermalizes .
for instance , the detector could evolve to a squeezed thermal state which may exhibit the same probability of excitation as some thermal state , but which is not actually a thermal state . by the use of non - perturbative methods we can make sure that thermalization is achieved and that it is not an artifact of the use of perturbation theory in regimes beyond its applicability .
such non - perturbative methods were recently developed and applied to examine the response of a detector within a cavity containing a scalar field @xcite .
the cavity was a wave - guide with periodic boundary conditions in which the detector was allowed to entirely cycle several times during its evolution . whereas this is physically reasonable for the case of periodic boundary conditions
, it is not the correct setting to compare with more general boundary conditions .
we consider in this paper the thermality of accelerated detectors in optical cavities with different boundary conditions .
extending the work of brown et al .
@xcite ( see also @xcite ) , we will demonstrate non - perturbatively that an accelerated unruh - dewitt detector coupled to the vacuum state of a scalar field thermalizes to a temperature proportional to its acceleration , regardless of the boundary conditions imposed .
the scenarios we consider here also differ from previous work in the way that the detector trajectories are defined with respect to the cavity .
for example , we modify the cavity length such that the detector remains inside a single cavity during its interaction with the field , which is of capital importance for physicality in the case of non - periodic cavities .
we do note that there has been an effort to understand how imposing different boundary conditions modifies the response of detectors in non - inertial scenarios in free space .
for example , work has been done in a very different context to examine the continuum rindler case @xcite , and a number of boundary conditions in hartle - hawking vacua have been studied @xcite . however
, these studies do not tell us if the boundary effects of a cavity will prevent a uniformly accelerating particle detector from thermalizing due to the unruh effect . to our knowledge the only work that addresses this issue is the aforementioned paper by brown et al .
@xcite in the periodic cavity case .
in addition to the study of the universality of the unruh effect and the existence of thermalization for different sets of boundary conditions , we shall also briefly comment on the effect of different methods of coupling the detector to the field . typically , we couple the detector locally to the field through the detector s monopole moment , @xmath7 . here , we will also explore the effects of a different form of coupling , namely the coupling of the detector s monopole moment to the momentum of the field .
our findings indicate that in all of the scenarios under consideration , the unruh effect occurs .
we observe that the detector achieves thermalization with temperature proportional to acceleration .
thus , not only does the unruh effect occur inside a cavity ( which imposes an ir - cutoff on the field and , furthermore , isolates the field in the cavity from the rest of the spacetime ) , it appears to occur independently of the details of this ir cutoff and of the spatial distribution of the cavity modes .
this demonstrates that the unruh effect , which many have argued relies on idealized details and thus can not lead to thermalization @xcite , is in fact a very general and universal phenomenon and that thermalization of particle detectors can be computed non - perturbatively .
not only is this a remarkable result from a fundamental point of view , it also gives hope to the possibility of an experimental realization of the unruh effect in quantum optical settings , where it has been shown that general relativistic scenarios like the one we study here can already be simulated @xcite .
our paper is organized as follows . in sect .
[ settingsect ] we discuss the physical setup of our system , elaborating on the differences between the various scenarios and boundary conditions considered . in sect .
[ gaussiansect ] we explain the oscillator - detector model that we will be using in our study , as presented in @xcite , and go on to discuss how we solved for the evolution of the detector - field system . in sect .
[ mainsect ] we present our results on the thermalization of the accelerating detector along with the linear dependence of its temperature on acceleration .
furthermore , we demonstrate that these results are largely independent of the boundary conditions imposed on the field . in sect .
[ consect ] we finish with some concluding remarks .
we will consider a uniformly accelerated point - like detector in its ground state going through a cavity prepared in the vacuum state
. the trajectory will be such that the detector starts moving inside the cavity with a given initial speed , with a constant acceleration in a direction opposite to its initial motion .
hence the atom will be decelerated while crossing the cavity .
the detector reaches the center of the cavity exactly when it reaches zero speed , and then travels back to the initial point with increasing speed until it reaches the position in which it started , having the same speed as when it entered the cavity but in the opposite direction .
this trajectory of the atom within the cavity as a function of the detector s proper time @xmath8 is shown in fig .
[ figtraj ] . for larger values of the acceleration
the detector will exit the cavity . in principle
this is an issue as we mirror the field modes outside of the cavity .
however , over the range of accelerations that we will consider the coupling decays so quickly past the edges that these tails will not contribute significantly to the observed final state .
in addition , we find that the linearity of the temperature plots is preserved even when the detector escapes from the cavity . .
the gaussian switching function is plotted on the right axis with a jagged magenta line . ] to have a clean signature one must be careful with the way in which the detector is switched on @xcite since a sudden switching stimulates strong quantum fluctuations that may overcome the unruh effect . in order to reduce switching noise
we apply the same approach as @xcite : the interaction is smoothly switched on following a gaussian time profile so that switching quantum noise effects are reduced . in particular , the switching function that we use has the form @xmath9 we prepare the ground state of the detector and the vacuum of the field at a time @xmath10 where the interaction is switched on following the gaussian profile above , the atom has some initial speed and starts decelerating until it reaches the centre of the cavity at time @xmath11 .
the atom continues accelerating until the time @xmath12 when it reaches the initial point again . in the settings that we shall analyze
, we will consider as parameters @xmath13 , @xmath14 and @xmath15 . with these parameters
we find that the switching is more than smooth enough for our purpose ; namely the detector s response from the switching noise is negligible compared with the response from acceleration .
we are going to consider different scenarios that correspond to different cavity field settings .
let us rewrite the interaction hamiltonian in the interaction picture in the following general form @xmath16 + \hat a^\dagger_n u^*_n[x(\tau),t(\tau ) ] \big)\,,\end{aligned}\ ] ] where we have expanded the field operator in terms of an orthonormal set of field mode functions @xmath17 $ ] which will depend on the boundary conditions that are imposed upon the field .
the normalization of the field modes is performed with respect to the klein - gordon inner product @xcite .
if we choose a cavity of length @xmath18 and we impose the dirichlet boundary conditions @xmath19=\phi[0,t]=0 $ ] ( the two walls of the cavity are ideal mirrors ) we find that the mode functions are the stationary waves
@xmath20=\frac{1}{\sqrt{k_n l}}e^{-{\mathrm{i}}\omega_n t}\sin(k_n x)\ ] ] where @xmath21 and @xmath22 .
in the case of a periodic cavity of length @xmath18 ( which would correspond to physical settings such as closed optical fibres or microwave guides or any other setting with a torus topology ) , the modes are the set of right and left - moving waves @xmath20=\frac{1}{\sqrt{2 |k_n| l}}e^{-{\mathrm{i}}(\omega_n t - k_n x)}\ ] ] where @xmath23 is an integer , @xmath24 ( negative ( positive ) @xmath23 corresponds to left - moving ( right - moving ) modes ) and @xmath25 . in the case of a neumann cavity of length @xmath18 ( this is to say , @xmath26=\partial_x\phi[0,t]=0 $ ] ) , the modes become @xmath20=\frac{1}{\sqrt{k_n l}}e^{-{\mathrm{i}}\omega_n t}\cos(k_n x)\ ] ] where @xmath21 and @xmath22 . finally ,
as we described above , the worldline of the detector inside the cavity parametrized in terms of its proper time will be given by @xmath27\ ] ] and the interaction will be smoothly switched on following the curve from time @xmath28 to @xmath29 with suitable values of @xmath18 and @xmath29 such that the atom always remains in the cavity while the interaction is `` on '' ( see fig .
[ figtraj ] ) .
we will introduce the non - perturbative oscillator - detector model . replacing the usual two - level system in the unruh - dewitt model with a harmonic oscillator
is somewhat common in the literature @xcite .
however , in almost all of these cases , a perturbative approach was used , and not many practical non - perturbative results have been obtained in the past . here
we will use the powerful non - perturbative gaussian formalism developed in @xcite to analyze the thermalization properties of the detector .
the only restrictions that apply to this formalism in a cavity scenario are that the initial state of the system is a gaussian state ( such as the vacuum , a coherent or squeezed state or a thermal state ) and that the interaction hamiltonian is quadratic in the quadrature operators ( in order to preserve the state s gaussianity through time evolution ) .
restricting our consideration to quadratic hamiltonians is quite reasonable since the interaction between matter and light is of this nature @xcite .
in addition , as the unruh effect has almost solely been studied in the context of free space , this provides an excellent excuse for us to tread into unknown waters by considering different cavity settings .
furthermore , any experimental verification of the unruh effect is likely to be more easily implementable in the context of optical cavities , so it is important to understand the phenomenon in such a scenario .
let us summarize the tools that we are going to employ in this paper to obtain , non - perturbatively , the response of a particle detector to the field vacuum . instead of the interaction picture , it will be more convenient to work in the heisenberg picture , following @xcite .
let us form a vector from the detector s and field s annihilation and creation operators in the heisenberg picture : @xmath30 where the subscript @xmath31 corresponds to the detector and the others correspond to the modes of the field .
the commutators of the components of this vector generate a symplectic form @xmath32 \label{eqn : sympform}\ ] ] similarly , we can form a vector of quadrature operators of the form @xmath33 these operators are related to the creation and annihilation operators of each mode by @xmath34 in order to ensure that gaussian states will remain gaussian over the course of their evolution , we also need to evolve by a quadratic hamiltonian : @xmath35 where @xmath36 will in our case be given by ( [ eq : inth1 ] ) in the heisenberg picture . here
@xmath37 is the frequency of the oscillator - detector and @xmath38 is the frequency of the @xmath39 field mode . in our notation
we use @xmath8 to denote the proper time of the detector ( which is generally moving with respect to the cavity ) and @xmath40 to denote the lab time , which is the time with respect to which the field evolves . when computing the system s evolution as generated by some hamiltonian we must be careful to choose a specific time parameter and construct the hamiltonian accordingly . in the above hamiltonian
we choose to evolve with respect to the detector s proper time @xmath8 .
since the field evolves with respect to @xmath40 , this means that we must include a `` blue - shift factor '' on the free field s hamiltonian .
a more complete and rigorous explanation of this can be found in @xcite . since it is gaussian , the state of the detector - field system can be completely described by a covariance matrix consisting of the first and second moments of the quadrature operators . in our scenario
we need not consider states with first moments other than zero because of the absence of hamiltonian terms linear in the quadrature operators , and so we determine the state by a covariance matrix of the form @xmath41 where @xmath42 is the vector formed by the dimensionless position and momentum operators from equation .
thus the state of our single detector can be completely described using the @xmath43 covariance matrix of the form @xmath44 the time evolution of the entire covariance matrix , including both the detector and the field , is governed by the equation of unitary evolution @xcite @xmath45 where @xmath46 is a symplectic matrix : @xmath47 .
in addition , the symplectic matrix generated by a ( generally time - dependent ) hamiltonian @xmath48 satisfies the equation @xmath49 with initial condition @xmath50 .
here @xmath51 , where @xmath52 is a phase - space matrix encoding the form of the hamiltonian via @xmath53 keeping in mind that we will continue working in the heisenberg picture ( our operators are fully time dependent ) , we will make use of some computational techniques beyond what was indicated in @xcite that are inspired by the principle of the interaction picture . here , we make use of an exact solution of the free symplectic time - evolution matrix to speed up the computation .
namely , we split the evolution matrix into an exactly solvable part ( we could call it the free part ) and a non - exact part ( that we could call the interaction part ) . from equation ( [ eqn : timeevol ] )
, we take @xmath54 where @xmath55 is exactly solvable . for our circumstances , we choose @xmath56 here @xmath57 and @xmath58 are the symmetrized matrices corresponding to the free hamiltonians of the detector and field , respectively .
that is , their non - symmetrized versions satisfy @xmath59 and @xmath60 .
although , strictly speaking , @xmath55 is time - dependent due to the @xmath61 factor , the fact that this is a total derivative and that we are integrating over @xmath8 to solve the dynamics means that we are still able to solve for the free evolution exactly , so that if @xmath62 , equation ( [ eqn : timeevol ] ) has the exact solution @xmath63 \label{eqn : s0exact}\ ] ] applying this interaction - picture - like approach , we define @xmath64 it is then easily seen that ( [ eqn : timeevol ] ) becomes @xmath65 the evaluation of @xmath66 can then be accomplished by standard numerical techniques .
the full heisenberg evolution matrix is then simply @xmath67 , and the evolved state of the detector - field system is given by @xmath68 .
here we present the results obtained by applying the formalism of sect .
[ gaussiansect ] to the scenario outlined in sect .
[ settingsect ] . our goal is to test the universality of the unruh effect with respect to a change of boundary conditions in the cavity .
we begin by considering an oscillator detector uniformly accelerating through a cavity field that is initially in the vacuum state . it should be noted that when considering accelerating detectors it is necessary to include many modes in the field expansion .
this is due to the fact that the detector will experience an exponentially changing time dilation with respect to the cavity frame which translates into a modulating blueshift of the mode frequencies as seen by the detector .
thus , the field modes with which the detector is resonant will rapidly change and , if the acceleration or time of evolution is large enough , very high mode numbers can make significant contributions to the evolution of the detector . in our work
we have been vigilant to ensure that enough field modes were included such that further additions do not modify the results obtained for the detector . because the non - perturbative gaussian formalism presented in sect .
[ gaussiansect ] slows down considerably for a large number of field modes , we use the standard perturbative formalism up to the first order to ensure complete convergence with respect to the number of field modes .
for example , in the periodic case this method involves evaluating the integral @xmath69 so , for periodic boundary conditions we therefore need to evaluate @xmath70 which is the contribution to the excitation probability amplitude for a given field mode , where @xmath71 is used to sum over left- and right - moving modes and the last term in the exponential is the gaussian switching function ( see below for more details ) .
we can use them ( after normalization ) in a partial sum over the number of field modes .
that is , @xmath72 this gives the first - order perturbation theory result for the probability of transition of the detector from the ground state to the first excited state .
now , when beginning its evolution the detector is initially in its ground state , but through its interaction with the field it will generally become excited .
after the evolution of the detector - field system is complete we will examine the state of the detector , which will be fully specified by its @xmath73 covariance matrix @xmath74 . in order to conclude that the detector has experienced a thermal unruh bath during its acceleration , we look for two things : first , that the detector has evolved to a thermal state and , second , that the corresponding temperature grows linearly with the acceleration experienced by the detector .
the non - perturbative approach from @xcite is very well suited for testing the thermality of our detector .
not only is thermality easily tested , but because we have the exact ( non - perturbative ) state of the detector we are able to make a definitive statement about thermalization . in order to make conclusions about thermalization using perturbation theory
, one would at the least need to expand to higher orders , and at the worst it would be impossible due to the long time scales typically required for thermalization to occur where perturbation theory may break down .
once the detector has completed its evolution it will be , up to phase - space rotation ( i.e. free evolution ) , in a squeezed thermal state of the form @xmath75 where @xmath76 is the covariance matrix s symplectic eigenvalue and @xmath77 is its squeezing parameter .
these diagonal entries are just the eigenvalues of @xmath74 , @xmath78 , from which the symplectic eigenvalue and squeezing parameter follow as @xmath79 and @xmath80
. we must now determine whether the amount of thermality introduced by @xmath76 is a much greater contributor to the energy of the detector s state compared to the amount of squeezing .
if so , then the state can be said to be nearly thermal .
if they are comparable , or if squeezing has the greater contribution , then we can not claim that the detector thermalizes . to compare these two effects
we study how they contribute to the free energy of the detector . for small
squeezing ( which is satisfied in our scenario ) the energy above the ground state energy , to leading order in @xmath77 , a power series expansion straightforwardly yields @xmath81,\end{aligned}\ ] ] where @xmath37 is the detector frequency . since @xmath76 is of order unity ( and is in fact remains very close to unity for our situation ) a good test for thermality is @xmath82 if this inequality is satisfied then the detector can be said to be very nearly thermal .
equivalently , if @xmath83 is very small , the detector is said to be thermal .
we find that the detector thermalizes very well in all of the three boundary conditions considered : periodic , dirichlet , and neumann .
we find numerically that for the parameters given in sect .
[ settingsect ] , @xmath84 is on the order of @xmath85 in all three cases .
that is , the squeezing experienced by the oscillator is extremely minute compared to its thermality , and thus the detector can be said to be very nearly thermal .
the first of our two conditions to verifying the unruh effect ( thermality and temperature proportional to acceleration ) is satisfied for all three boundary conditions .
we are now in a position to compute the temperature of the evolved detector @xmath74 . for a single oscillator of frequency @xmath37
the form of an exactly thermal state is @xmath86 , for which the temperature is @xcite @xmath87^{-1}.\end{aligned}\ ] ] since in our scenario we have already confirmed that our detector thermalizes to an excellent approximation we are able to use this equation to compute the temperature of our detector with negligible error , where @xmath88 as above . for each of the three boundary conditions ( periodic , dirichlet , and neumann ) we compute this temperature for various values of acceleration .
these results are displayed in figure ( [ boundarycomp ] ) .
notice that our least - squares fit is performed on first - order _ perturbative _ results .
for small coupling strength the perturbative result for the transition probability of the accelerated detector is in close agreement with the result obtained by the non - perturbative approach .
specifically , the probability computed up to leading order ( as explained above ) is of @xmath89 and it is easily shown that the next relevant order , and thus the difference between the perturbative and non - perturbative answers , is @xmath90 . for small temperatures , therefore , one need not use the non - perturbative approach to estimate the temperature of the oscillator detector , assuming that the detector is in a thermal state .
when including a very large number of field modes it is computationally more convenient to plot the first order perturbative results with three objectives in mind : 1 ) from the probability ( [ eqn : perturbative2 ] ) and assuming thermality , compute the temperature using the standard boltzmann distribution and check that it varies linearly with acceleration , 2 ) check that the non - perturbative results ( using ( [ tempnon ] ) ) are computed with enough numerical accuracy to reproduce the perturbative plot up to @xmath90 , ensuring that had we used an infinite number of modes both methods would converge , and 3 ) use the non - perturbative results to ensure thermality of the system . note
that this last step can not be easily carried out with a perturbative calculation ( this was done and discussed in the previous section ) . from the plot of the trajectory ( fig .
[ figtraj ] ) , notice that very high acceleration data points involve the atoms exiting the cavity for a small part of their trajectories . to prevent switching noise from damaging the results , we keep the field continuous beyond the cavity , but
because the switching function is effectively zero when the detector crosses the boundary , the interaction will be negligible .
notice that the high acceleration results are not qualitatively different from lower acceleration results , justifying our negligibility assumption .
remarkably we find that for all three boundary conditions the temperature grows linearly with acceleration , demonstrating that the qualitative features expected of the unruh effect are very much independent of the details of the cavity .
this settles any doubt regarding the existence of the unruh effect when an ir - cutoff for the field is introduced ( i.e. when inside a cavity )
. there has been some skepticism @xcite stemming from the large number of technical assumptions that go into the canonical derivation of the unruh effect @xcite and how the presence of a cavity might alter or even eliminate its existence .
we have demonstrated not only thermality and the existence of the effect in a cavity ( first shown in @xcite and reaffirmed here ) , but also that the boundary conditions ascribed to this cavity are all but irrelevant ( see fig .
[ boundarycomp2 ] ) .
indeed the numerical similarity between the different cases is striking .
we note that the slope of the detector temperature with respect to acceleration is not equal to the value of @xmath91 predicted by the canonical free - space derivation .
this is not overly surprising since we are working in a cavity setting rather than free space ; significant border effects should be expected when studying such phenomena .
our results demonstrate , however , that the inclusion of an ir cutoff does not destroy the unruh effect understood as the thermal response of a particle detector with a temperature proportional to the acceleration , and what is more , that the detector actually thermalizes to that particular temperature .
if we were to take our cavity to the continuum limit we would expect ( at least in the case of periodic boundary conditions ) the slope to converge to the usual value of @xmath91 .
we leave such a study for future work . in this portion of the paper
we shall briefly discuss the effect of varying the form of coupling between the field and the detector .
in particular , the question we ask is : what happens when the monopole coupling of equation ( [ eq : inth1 ] ) is modified such that the detector couples to the conjugate momentum of the field instead of the field itself ? in this form , we can no longer maintain the pointlike coupling assumption for the detector - field coupling .
after a fourier transform , it is apparent that a point - like @xmath92 coupling is akin to coupling the detector s monopole moment to the field in an extremely delocalized way , such that the detector couples to the field everywhere .
thus this coupling is ill - defined and , without the introduction of spatial smearing , will yield divergences .
we will therefore regularize the interaction assuming that the detector s coupling strength to the canonical momentum of the field varies with the field frequency .
a frequency - dependent effective coupling appears naturally when considering spatially smeared detectors @xcite , although in our case we would make the simpler assumption that the coupling strength is inversely proportional to the frequency of the mode . in this way we can analyze the following coupling for periodic boundary conditions : @xmath93 \label{eqn : xpcoupling}\end{aligned}\ ] ] where @xmath94 so that the energy density falls off with high energy modes at the same rate as in the @xmath95 coupling case .
we note here that in first order perturbation theory , the negative sign does not contribute to the probability of transition and this scenario is exactly the same as the @xmath92 coupling .
in the non - perturbative case , we produced the same plot as the periodic curve of figure [ boundarycomp ] with the modified coupling ; all data points were exactly the same .
this tells us that the periodic @xmath92 coupling sign change is still a symmetry of the non - perturbative case .
we believe that this should be observable when examining the equations ( [ hamilgood ] ) and ( [ eqn : xpcoupling ] ) but we do not prove this here .
in this work we have non - perturbatively solved for the evolution of an oscillator detector undergoing uniform acceleration through a cavity field . we have confirmed recent previous work @xcite demonstrating that the unruh effect does indeed occur inside a cavity , and furthermore we have demonstrated that this result is independent of the boundary conditions applied to the field , implying that the unruh effect is a very universal phenomenon .
specifically , we have considered vacuum cavity fields with periodic , dirichlet , and neumann boundary conditions . in all three cases
we have observed that an accelerating oscillator detector evolves to a thermal state and that the temperature obtained by the detector increases linearly with its acceleration .
furthermore the results between the three cases are numerically very similar .
this indicates that not only is the phenomenon qualitatively universal but , with respect to the case of boundary conditions , also quantitatively universal .
we have also made some conclusions regarding the use of different detector - field couplings that further strengthens these claims .
moreover , our use of the non - perturbative oscillator model has allowed us to make significantly stronger claims regarding the thermality experienced by the detector than can be made using the standard perturbative framework typically employed in the literature .
that is , we have concluded that in our scenario an accelerating detector in fact evolves to a thermal state , rather than merely exhibiting a thermal response function .
questions of thermalization can not be made in perturbation theory without resorting to higher order expansions , and in some scenarios may actually be impossible due to the large time scales often required for thermalization where perturbation theory breaks down . more generally , the results of this paper suggest that the unruh effect and similar phenomena such as hawking radiation may be largely independent of the details of the system @xcite .
in addition to theoretical interest , such universality bodes well for an eventual experiment where the unruh effect could be measured .
the authors would like to thank very much william donnelly for his helpful comments and insight on the interaction picture approach to the gaussian formalism .
this work was supported in part by the national sciences and engineering research council of canada .
was partially funded by the banting postdoctoral fellowship programme .
w. b. was funded by the vanier cgs award . | we explore the effects of different boundary conditions and coupling schemes on the response of a particle detector undergoing uniform acceleration in optical cavities .
we analyze the thermalization properties of the accelerated detector via non - perturbative calculations .
we prove non - perturbatively that if the switching process is smooth enough , the detector thermalizes to the unruh temperature regardless of the boundary conditions and the form of the coupling considered . |
cosmology deals with the large - scale structure of the universe .
a global description of this structure employs an averaging and smoothing of all the objects of astrophysics from solar systems to galaxies and clusters of galaxies .
the result is a fluid model of the cosmos goverened by gravitation , as it was proposed by einstein @xcite , see @xcite pp .
160 - 164 and @xcite pp .
704 - 710 .
this fluid model is governed by einstein s differential equations which relate the space - time metric to the energy - momentum tensor .
to device such a model one has to make assumptions about the global structure and topology of the underlying space - time manifold .
einstein in his initial analysis of 1917 assumed cosmic 3-space in form of a 3-sphere .
his assumption implies an average curvature @xmath0 .
the topology of cosmic 3-space has found new attention in relation to the cosmic microwave background ( cmb ) radiation , which is supposed to originate from an early stage of the universe .
the fluctuations of the incoming cmb radiation are well described by a standard model , except for rather low amplitudes at the lowest multipole orders .
this raised the question if multipole selection rules could be due to a particular topology of 3-space .
* genesis of the mbius strip . * left : as rectangular cell of the mbius crystal * cm * with broken glide reflection line , see fig .
[ fig : cm+8cellb ] , middle : twisted by @xmath1 along the glide line , right : bended into a circle and glued by homotopy.,scaledwidth=100.0% ] * mbius crystal and 8-cell . *
left : the red mbius strip tiles its cover , the plane , into the cells of a mbius crystal with crystallographic symbol * cm * , with a broken glide line and two pointed mirror reflection lines .
the vertical edges are glued by homotopy , fig.[fig : moebi ] , right : a spherical red cube tiles its cover , the 3-sphere , into the 8 spherical cubes of the 8-cell , shown in a projection.,scaledwidth=90.0% ] a familiar paradigm for topology is the mbius strip . in @xcite
we explain its topological properties and relate it to the crystallographic space group * cm*. a cell of this crystal is shown on the left - hand side of fig .
[ fig : cm+8cellb ] . any spherical topological 3-manifold @xmath2 can be viewed as a prototile on einstein s 3-sphere with pairs of faces glued according to the prescriptions of a group of homotopies . among the spherical 3-manifolds
are the platonic ones .
their homotopies were derived by everitt @xcite . in @xcite and papers @xcite , @xcite quoted therein we derived from the homotopy or fundamental group of a platonic manifold @xmath2 the corresponding groups deck(@xmath2 ) of deck operations acting on the 3-sphere .
these groups generate by fixpoint - free action the tiling from the prototile .
each group of deck operations for a platonic 3-manifold @xmath2 is constructed in @xcite as a subgroup of a coxeter group @xmath3 . finally from the unimodular subgroups of the coxeter groups we construct three new spherical 3-manifolds .
to study actions on the 3-sphere we pass from the set of four cartesian coordinates @xmath4 in euclidean 4-space to a matrix description , @xmath5,\ ; z_1= x_0-ix_3,\ : z_2=-x_2-ix_1,\ : z_1\overline{z}_1+z_2\overline{z}_2=1.\ ] ] with the help of the pauli matrices @xmath6 and @xmath7 and the trace @xmath8 we recover the cartesian coordinates in the form @xmath9 the wigner polynomials @xcite , @xcite are homogeneous polynomials @xmath10 in the four complex variables @xmath11 of total degree @xmath12 , see @xcite appendix a. in the familiar euler half - angles @xmath13 , the wigner polynomials take the form @xmath14 we look at the set of harmonic wigner polynomials by starting from the integer or half - integer pairs of numbers @xmath15 .
these pairs of numbers can be viewed as points from two nested lattices on a plane , see fig.[fig : gridn2n3 ] . for given integer or half - integer values of @xmath16 , one finds @xmath17 we say that any lattice point @xmath15 in this plane carries a _ tower _ , labelled by @xmath18 , of wigner polynomials @xmath19 according to eq . [ eq5 ] .
the wigner polynomials form a complete orthogonal system of polynomial functions on the 3-sphere .
moreover they are harmonic , that is , vanish under the application of the laplacian acting on functions on euclidean 4-space , see @xcite .
therefore they are a basis for harmonic analysis on the 3-sphere .
the isometric rotations of the group @xmath20 can be written in the form @xmath21 and act on the coordinates @xmath22 as @xmath23 the elements of the form @xmath24 generate a subgroup @xmath25 acting on @xmath22 by conjugation .
the 3-sphere can be written as the homogeneous space @xmath26 .
we write the action eq .
[ eq6 ] of @xmath27 on the wigner polynomials as @xmath28 here we used the representation property of the wigner polynomials .
a general pair @xmath29 can always be brought to diagonal form @xmath30 we interprete the transformation @xmath31 as a transformation to new coordinates @xmath32 . in these new coordinates ,
@xmath33 are diagonal with diagonal entries @xmath34 , and the action eq . [ eq6a ] with eq .
[ eq7 ] takes the form @xmath35 now we can go to a lattice description of the harmonic analysis for the two spherical cubic 3-manifolds : the basis for the harmonic analysis consists of the towers of wigner @xmath19 polynomials on top of a sublattice in the @xmath15-plane as given in fig . [
fig : gridn2n3 ] .
as examples we shall choose the two spherical cubic manifolds @xmath36 .
the first homotopy groups of the platonic spherical polyhedra were given in everitt @xcite . in @xcite and
work cited therein we construct from everitt s results the groups of deck transformations .
these act on the 3-sphere and tile it into platonic polyhedra .
in @xcite one finds by an enumeration the homotopic face and edge gluings for this manifold .
the tiling of the 3-sphere is the 8-cell shown on the right in fig [ fig : cm+8cellb ] .
the cubic 3-manifold we take as the central spherical cube in the 8-cell .
the two different spherical cubic 3-manifolds differ in their face and edge gluing . in fig .
[ fig : twocubetwist ] we have marked the faces with numbers @xmath37 by triangles with the colors yellow , blue , and red .
the homotopic self - gluing of the initial 3-manifold is converted in the 8-cell tiling into a gluing of neighbour cubes sharing a face .
this is illustrated in colors in fig .
[ fig : cubeglueb ] . in fig .
[ fig : twocubetwist ] we use the same color coding to illustrate the two different next neighbour cubes for the two 3-manifolds @xmath36 .
conversion of homotopic face gluing @xmath38 into a deck operation of two cubic tiles for the manifold n2 .
the half - faces of @xmath39 are marked by the colors yellow , blue , red respectively .
for gluings of other faces see the left of fig .
[ fig : twocubetwist],scaledwidth=50.0% ] * the cubic manifolds @xmath40 and @xmath41*. the cubic prototile and three neighbour tiles sharing its faces * f1 , f2 , f3*. the four cubes are replaced by their euclidean counterparts and separated from one another .
visible half - faces * f1 , f2 , f3 * are marked by the colors yellow , blue and red .
the actions transforming the prototile into its three neighbours generate the deck transformations and the 8-cell tiling of @xmath42.,scaledwidth=100.0% ]
algebraically , the deck operations , being rotations , contain an even number of weyl reflections and can be written in terms of elements @xmath43 .
we now construct by projection the linear combinations of wigner polynomials that span the harmonic analysis on the two cubic 3-manifolds . for @xmath40 ,
the group @xmath44 of deck transformations of the 8-cell tiling from @xcite is a cyclic group @xmath45 .
its generator @xmath46 is given in eq .
we start from the set of wigner polynomials and use their representation under @xmath27 , in terms of irreducible representations @xmath19 of @xmath47 .
this allows to apply to them the projection operator to the identity representation of the group @xmath48 , @xmath49.\ ] ] in general , eq .
[ eq10 ] gives a linear combination of wigner polynomials . in case of the platonic manifold @xmath40
, @xmath48 is a cyclic group . by a transformation @xmath50 of coordinates
we can reduce the action of this cyclic group to diagonal form .
we start from the generator @xmath51 of the group @xmath52 , @xmath53,\ : g_r=\left [ \begin{array}{ll } 0&\overline{a}\\ -a&0\\ \end{array } \right],\ : a=\exp(i\frac{\pi}{4}).\ ] ] we diagonalize @xmath54 as @xmath55 c^{-1},\ : c= \frac{1}{\sqrt{2 } } \left [ \begin{array}{ll } -1&-a\\ \overline{a}&-1\\ \end{array } \right],\ : c^{-1}=c^{\dagger},\ : { \rm det}(c)=1.\ ] ] next we define new coordinates @xmath32 by @xmath56 so that @xmath51 acts on @xmath32 by diagonal matrices from left and right as @xmath57 \ : u'\ : \left [ \begin{array}{ll } a^2&0\\ 0&-a^2\\ \end{array } \right ] = \delta_l^{-1}u'\delta_r.\ ] ] these relations allow to apply eq .
it follows that we can reduce the representation of @xmath48 to diagonal form .
invariance under @xmath48 now gives a certain linear relation between @xmath58 and @xmath59 .
this relation singles out on the full lattice of points @xmath15 all the points of the sublattice with points @xmath60 a basis of this sublattice is @xmath61 , compare fig .
[ fig : gridn2n3 ] . the harmonic analysis on @xmath40
is now given by the towers of wigner polynomials on top of the black sublattice points .
for the cubic 3-manifold @xmath41 we construct three glue generators @xmath62 in table [ tablen3a ] from the homotopy group @xcite . the result for three faces
is shown on the right of fig .
[ fig : twocubetwist ] .
the deck operations corresponding to the gluings generate the quaternionic group @xmath63 , @xcite p. 134 .
it acts exclusively by left action on the 3-sphere .
@xmath64=-\mathbf{k}&e\\ \hline 2 & ( x_2,-x_3,-x_0,x_1 ) & \left [ \begin{array}{ll } 0&-1\\ 1&0\\ \end{array } \right]=-\mathbf{j}&e\\ \hline 3 & ( x_3,x_2,-x_1,-x_0 ) & \left [ \begin{array}{ll }
-i&0\\ 0&i\\ \end{array } \right]=-\mathbf{i}&e\\ \hline \end{array}$ ] for the harmonic analysis on @xmath41 we employ the projection eq .
[ eq10 ] for the quaternion group @xmath65 , @xmath66 \left[1+(-1)^{m_1}\right ] \left[d^j_{m_1,m_2}(u)+(-1)^jd^j_{-m1,m_2}(u)\right].\ ] ] * lattice representation of @xmath40 and @xmath41 basis : * any lattice point @xmath15 carries the countable tower @xmath10 , @xmath67 of wigner polynomials . harmonic analysis on cubic manifolds , left @xmath40 , right @xmath41 ( with vertical mirror line ) , selects the towers @xmath68 on sublattices marked by black points with sublattice bases @xmath69 .
only these obey the homotopic boundary conditions.,scaledwidth=80.0% ]
in fig . [ fig : gridn2n3 ] , we display the sublattices for the two cubic spherical platonic 3-manifolds . for the manifold @xmath41 we put the symmetry eq .
[ eq16 ] , @xmath70 as a vertical mirror line . for the harmonic analysis
it follows that an orthogonal basis is given by the collection of all towers of wigner polynomials on top of the sublattice points .
any basis function can be characterized by a sublattice point @xmath15 and by a number @xmath71 .
in this section we discuss the algebraic tools for analysing incoming cmb radiation in terms of the harmonic bases for a chosen topology . for the harmonic analysis on spherical 3-manifolds we use the spherical harmonics in the form of wigner polynomials .
these polynomials in the coordinates @xmath72 are often expressed in terms of euler angle coordinates eq .
[ eq4 ] .
an alternative system of polar coordinates is used by aurich et al .
@xcite . here
@xmath73\\ \\
\hline \end{array}\ ] ] we shall see in eq . [ eq21 ] that these polar coordinates are adapted to the analysis of incoming radiation in terms of its direction .
the cmb radiation as observed is given as a function of polar angles @xmath74 for its direction . to compare with an expansion of the harmonic basis of a given 3-manifold , we must rewrite the wigner polynomials in terms of polar angles . in terms of representation theory ,
this can be achieved by reducing the representations of @xmath27 into irreducible representations of its subgroup @xmath25 , see eq.[eq19 ] .
we relate our analysis algebraically to this description . to adapt the wigner polynomials to a multipole expansion , we transform them for fixed degree @xmath75 by use of wigner coefficients of @xmath47 , @xcite pp . 31 - 45 , into the new harmonic polynomials @xmath76 whereas the index @xmath77 of the wigner polynomials can be integer or half - integer , the multipole index @xmath78 takes only integer values . for fixed @xmath78
we have @xmath79 , and for fixed @xmath75 : @xmath80 .
using representation theory of @xmath47 it can be shown from eq .
[ eq6a ] that the conjugation action @xmath81 of the group @xmath25 acts by a rotation @xmath82 only on the coordinate triple @xmath83 , and the new polynomials eq .
[ eq18 ] transform as @xmath84 like the spherical harmonics @xmath85 .
we therefore adopt eq .
[ eq19 ] as the action of the usual rotation group for cosmological models covered by the 3-sphere , and eq . [ eq19 ] qualifies @xmath78 as the multipole index of incoming radiation .
the basis transformation eq . [ eq18 ] is inverted with the help of the orthogonality of the wigner coefficients @xcite to yield @xmath86 the result eq .
[ eq20 ] can be further elaborated by use of the alternative coordinates @xmath87 eq . [ eq17 ] . from @xcite ,
. 9 - 17 we find @xmath88 where @xmath89 is a gegenbauer polynomial .
the alternative coordinates admit the separation of the new basis into a part depending on @xmath90 and a standard spherical harmonic as a function of polar coordinates @xmath74 .
@xmath91 @xmath92
any platonic spherical 3-manifold is distinguished by a specific point symmetry group m which stabilizes its center point .
there arises the following enigma : the point group stabilizes the center point , but the deck group acts fixpoint - free .
so the two groups can never mix .
can they nevertheless be brought together , and what happens to the harmonic analysis ? to examine this question we turn to the cubic spherical manifolds @xmath36 .
their coxeter group from @xcite has the coxeter diagram @xmath93 the point group is the full cubic rotation group @xmath94 of order @xmath95 .
the group of deck transformations for @xmath41 is @xmath96 , the quaternion group with eight elements .
the cubic tiling of the 3-sphere is the 8-cell tiling of einstein s 3-sphere , see @xcite p. 178 .
the relation between the deck and point groups is addressed in appendix c of @xcite . in @xcite one
finds selection rules from point symmetry for the multipole orders @xmath97 . for the cubic spherical manifold @xmath41 we found there : * prop 1 * : the cubic point group @xmath98 of @xmath41 under conjugation leaves invariant the group h = deck(n3)=q , the quaternion group , and with q forms a semidirect group @xmath99 , which turns out to be @xmath100 , the rotational unimodular subgroup of the coxeter group , generated by an even number of weyl reflections .
the relation between point and deck group resembles the case of symmorphic space groups in euclidean crystallography .
there the commutative infinite translation group acts fixpoint - free , and a cubic cell has again the cubic group as its point group .
the difference is that , when going from euclidean 3-space to the 3-sphere , the deck group is finite and no longer commutative .
if we consider first of all only the cubic point group @xmath94 as a subgroup of the rotation group @xmath101 in euclidean 3-space , there are well - known results from molecular physics for the multiplicity of its representation in a given representation of @xmath101 with angular momentum @xmath78 , see @xcite p. 438 .
for the identity representation denoted by @xmath102 of @xmath94 , the lowest non - zero angular momentum is @xmath103 .
the cubic invariant linear combination of standard spherical harmonics for lowest values of @xmath78 are given in table [ table : cubicl ] .
now we wish to include the quaternion group @xmath63 of deck transformations . from the semidirect product property it follows that the projectors on the identity representation for @xmath94 and @xmath65 commute with one another .
this allows for the following procedure : we take a cubic @xmath94-invariant linear combination of spherical harmonics and combine it according to eq .
[ eq21 ] with the lowest possible function of the angle @xmath90 .
then we transform this linear combination back by eq .
[ eq20 ] into wigner polynomials and apply the projector eq .
[ eq16 ] to @xmath65-invariant form .
next we transform back with eq .
[ eq18 ] to the basis adapted to the multipole analysis .
the resulting linear combination must still be @xmath94-invariant but may contain new @xmath94-invariant linear combinations of spherical harmonics . by use of the cubic invariants from table [ table : cubicl ] we obtain the fully @xmath104-invariant polynomials of table [ table : cubicinv ] .
the construction requires only the wigner coefficients of @xmath47 and can easily be continued to higher polynomial degree . for the physics on the cubic spherical manifold with point symmetry
, there follows from table [ table : cubicinv ] a special and observable property : different multipole orders of spherical harmonics must be linearly combined to assure the overall invariance . what happens with the first cubic spherical manifold @xmath40 under cubic point symmetry ?
here we have from @xcite the following universality : * prop 2 * : any particular homotopy of a regular polyhedron with fixed geometric shape implies a pairwise homotopic boundary condition on its faces . if full rotational symmetry is applied , all the faces and also all their edges are on the same footing .
this implies that any particular homotopic boundary condition is automatically fulfilled .
for the two cubic spherical manifolds @xmath36 it follows that the same rules apply to their @xmath100-invariant basis whose lowest part we give in table [ table : cubicinv ] .
the order of the semidirect product group @xmath99 is @xmath105 .
this is half the order @xmath106 of the coxeter group .
it means that we are projecting to the identity representations of @xmath100 .
the new 3-manifold @xmath107 as fundamental domain of the unimodular coxeter group s@xmath108 , glued from two coxeter simplices.,scaledwidth=50.0% ] the new 3-manifold @xmath109 as fundamental domain of the unimodular coxeter group s@xmath110 , glued from two coxeter simplices.,scaledwidth=60.0% ] the new 3-manifold @xmath111 as fundamental domain of the unimodular coxeter group s@xmath112 , glued from two coxeter simplices.,scaledwidth=60.0% ]
we give a re - interpretation of the results of the previous section , we recall that in @xcite we constructed the spherical 3-manifolds from four coxeter groups generated by weyl reflections .
table [ table : table2 ] gives the data . in the last section we found that under inclusion of the cubic point group
, the group @xmath113 extends into the unimodular subgroup s@xmath3 of the cubic coxeter group @xmath3 , generated by an even number of weyl reflections .
when we introduce in addition to topology the point symmetry of the spherical cube , we can define a fundamental subdomain on the cube under the cubic point group .
this subdomain may be taken as the cone , shown in euclidean form in yellow on fig .
[ fig : cubus1b ] .
the cone is formed as a double simplex from two coxeter simplices of the cubic coxeter group @xmath3 , with the second simplex the image of the first one under reflection in the weyl plane perpendicular to weyl vector @xmath114 . in the 8-cell tiling that covers the 3-sphere , the double simplex is a fundamental domain with respect to the unimodular coxeter group s@xmath3 , of volume fraction @xmath115 . this double simplex on the 3-sphere forms a new topological 3-manifold @xmath109 with the group @xmath116 . with its small volume fraction
it is an attractive candidate for cosmic topology .
the first polynomials invariant under @xmath117 are the entries of table [ table : cubicl ] . turning to the tretrahedral and octahedral 3-manifolds discussed in @xcite ,
their unimodular coxeter groups admit the analogeous construction . the unimodular subgroup for the tetrahedron is the even subgroup @xmath118 .
its analysis in @xcite takes up work with marcos moshinsky @xcite on permutational symmetry .
we name the new 3-manifolds @xmath119 , show their double simplices in figs .
[ fig : tetra1b ] , [ fig : cubus1b ] , and [ fig : octa1b ] , and give their main data in table [ table : table1 ] , extended from table 1 in @xcite .
the double simplices are spherical counterparts to the notion of asymmetric units as used in classical euclidean crystallography .
since we know the geometry and the deck groups for these new manifolds , it should not be hard to determine the corresponding homotopies .
the harmonic analysis for @xmath120 still has to be done .
the harmonic bases will be invariant in particular under the point group of the tetrahedron and octahedron respectively .
the lowest non - zero multipole index is @xmath121 for the tetrahedron and @xmath103 for the cubes .
if we wish to accomodate lower multipole order we must reduce the point symmetry of the manifold .
we conclude : * prop 3 * : the harmonic analysis on the three new 3-manifolds @xmath122 strictly obeys the multipole selection rules given in @xcite table 3 for the tetrahedral and cubic point groups respectively . @xmath123@xmath124||@xmath125m@xmath124h = deck(m)@xmath124|h|@xmath126 - -- @xmath127120@xmath128n1@xmath124c_5@xmath1295@xmath130 --@xmath124384@xmath131n2@xmath132c_8@xmath1298@xmath133n3@xmath132q@xmath1348@xmath135 - - @xmath1241152@xmath136n4@xmath137c_3q@xmath12424@xmath138n5@xmath137b@xmath13424@xmath138n6@xmath137t^*@xmath13724@xmath139 - - @xmath124120120@xmath140n1@xmath124j^*@xmath137120@xmath141 - -- @xmath142120@xmath143n8@xmath144@xmath14260@xmath145 --@xmath142384@xmath146n9@xmath144@xmath142192@xmath145 -- @xmath1421152@xmath146n10@xmath144@xmath142576@xmath147
on the example of the cubic spherical 3-manifolds , we have explained the construction of the harmonic analysis from topology and its transformation into an expansion for the cmb radiation , ordered by the multipole index @xmath78 .
we implemented the additional assumption of point symmetry for spherical manifolds .
this assumption yields strong selection rules , including a lowest non - trivial multipole order @xmath78 .
similar rules apply to the other platonic spherical manifolds analyzed in @xcite .
these strong selection rules are easier to test from the fluctuation spectrum of the cmb radiation .
moreover we have shown that the inclusion of the unimodular coxeter groups s@xmath3 yields @xmath149 new topological 3-manifolds which cover rather small fractions of the 3-sphere .
this meeting is devoted to the memory of our great teacher and good friend , professor marcos moshinsky . for over five decades
i had the chance to share with him his insight into groups , their representations and applications in physics .
the initial steps of the present analysis were discussed with him in 2008 in mexico .
we all miss marcos , and we shall never forget him . | low multipole amplitudes in the cosmic microwave background cmb radiation can be explained by selection rules from the underlying multiply - connected homotopy .
we apply a multipole analysis to the harmonic bases and introduce point symmetry .
we give explicit results for two cubic 3-spherical manifolds and lowest polynomial degrees , and derive three new spherical 3-manifolds .
address = institut fr theoretische physik der universitt tbingen , + germany |
according to dijkgraaf and vafa @xcite , the effective glueball superpotential of @xmath0 supersymmetric @xmath1 gauge theory has an asymptotic expansion given by the planar part of the topological expansion of a matrix model . to give the effective potential for various vacua of the gauge theory , dijkgraaf and vafa
propose a formula obtained by saddle point expansion of the matrix integral around different critical points .
the filling fractions , i.e. , the fraction of the eigenvalues sitting close to each of the critical points of the potential are the parameters selecting the vacua . in this paper we give a way to define non - perturbatively ( i.e. , beyond the saddle point expansion ) the matrix integrals .
different integrals are integrals over different cycles in the space of normal matrices . in the case of generic polynomial potentials
it is possible to construct a cycle for each of the critical points , so that all effective potentials considered by dijkgraaf and vafa arise in the asymptotic expansion of our integrals .
we note that the idea of integrating over eigenvalues along suitable contours in the complex plane appears in special situations in @xcite .
david s contours were also more recently considered in @xcite , where a ( non - holomorphic ) modification of the integral was proposed to obtain arbitrary filling factors .
a second result of this paper is the description of the asymptotic distribution of eigenvalues for each critical point .
the eigenvalues , as predicted by dijkgraaf and vafa , lie asymptotically along arcs connecting branch points of a two - fold cover of the complex plane .
we give a reality condition which specifies the shape of the arcs and the density of eigenvalues . before considering the case of general @xmath2 matrices , it is instructive to consider the case @xmath3 of ordinary integrals .
consider the airy integral @xmath4 this integral gives a formal solution of the airy differential equation @xmath5 . indeed
the saddle point expansion around the two critical points of the integrand give the two linear independent formal power series solutions .
the non - perturbative solutions can be obtained by integrating along paths in the complex plane connecting any two of the three directions at infinity where the integrand goes to zero .
more generally , the integrals of the form @xmath6 for a generic polynomial @xmath7 of degree @xmath8 are solutions of a differential equation of order @xmath9 .
we show that a system of @xmath9 linearly independent solutions are obtained by choosing @xmath8 integration contours in the complex plane going to infinity in directions where the integrand decays exponentially .
moreover the contours can be chosen so that each of them passes through exactly one of the @xmath8 critical points .
it follows that the @xmath9 saddle point expansions at each of the critical points are the asymptotic expansions of true solutions given by convergent integrals . in the case of matrices
we consider integrals of the form @xmath10 over real @xmath11-dimensional cycles @xmath12 in the space of complex @xmath2 matrices .
the potential @xmath13 is a polynomial with complex coefficients , @xmath14 and the normalization factor @xmath15 is such that @xmath16 . the observables @xmath17 are holomorphic functions invariant under conjugation . in the well - studied case of integrals over hermitian matrices , with a polynomial @xmath13 with real coefficients bounded below ,
the saddle point approximation becomes exact in the large @xmath18 limit with fixed @xmath19 and the relevant critical point is described by an asymptotic _ density of eigenvalues _ which has support on a union of intervals on the real axis . if @xmath19 is small , these intervals are small neighborhoods of the minima of the potential @xmath13 . to obtain the densities of eigenvalues needed to make contact to gauge theory
one considers the variational problem for critical points subject to the side condition that the fraction of eigenvalues in the vicinity of each of the critical points ( not just minima ) are given numbers .
we consider here the case of a generic polynomial @xmath13 of degree @xmath8 with complex coefficients , with distinct critical points @xmath20 , and propose to consider integrals over cycles in the space of normal matrices ( a matrix is normal if it commutes with its adjoint or , equivalently , if it is conjugated to a diagonal matrix by a unitary matrix ) .
the cycles we consider are parametrized by integers @xmath21 whose sum is @xmath18 and are characterized by the condition that @xmath22 eigenvalues belong to a path @xmath23 in the complex plane going through @xmath24 and going to infinity in a direction where @xmath25 . in the limit @xmath26 with
@xmath27 fixed , the eigenvalues in the saddle point approximation ( supposed to be exact in the limit ) are distributed along @xmath9 arcs in the complex plane . for small @xmath19
the arcs are close to the critical points and the @xmath28th arcs contains the fraction @xmath29 of the eigenvalues .
let @xmath30 be a polynomial of degree @xmath31 with complex coefficients .
we want to consider integrals of the form @xmath32 for polynomials @xmath33 . before considering the question of integration cycles we may evaluate such integrals as asymptotic series as @xmath34 by formal application of the saddle point method at each of the critical points @xmath20 , which we assume to be distinct . in this way we get @xmath9 asymptotic series of the form @xmath35 and
the question is whether these are asymptotic expansions of our integral for suitable cycles @xmath23 . the cycles which we should consider here are ( linear combinations of ) paths for which the integral converges .
as the integrand is holomorphic , homotopic paths will give the same answer and what matters is the behavior at infinity .
as @xmath36 , @xmath37 , so there are @xmath8 directions in the complex plane for which @xmath38 as @xmath39 tends to infinity in these directions .
let us call these asymptotic directions valleys as in these directions the integrand decays exponentially .
neighboring valleys are separated by hills , which are directions of exponential increase of the integrand .
so the cycles one needs to consider are linear combinations of infinite paths connecting pairs of distinct valleys . as paths connecting two valleys can be deformed into sums of paths connecting the two valleys with any third one , there are only @xmath9 linearly independent cycles .
let us assume for simplicity that the critical values @xmath40 have distinct imaginary parts .
then there is a canonical way to associate to each critical point @xmath24 a path @xmath23 in such a way that the asymptotic expansion of the integral over @xmath23 as @xmath34 is obtained by the saddle point expansion at @xmath24 .
namely , we take the _ steepest descent _
paths(see , e.g. , @xcite and @xcite ) emerging from @xmath24 , defined by the condition that the tangent vector at each point points in the direction of the gradient of @xmath41 . as @xmath42 is holomorphic , the gradient of @xmath43 is orthogonal to the gradient of @xmath44 by the cauchy riemann equations .
thus steepest descent paths are level lines for the imaginary part of @xmath42 .
each non - degenerate critical point @xmath24 is at the intersection of two such level lines .
one of these two lines , the one along which @xmath41 takes its minimum at @xmath24 , is the steepest descent path @xmath23 . along the other line @xmath45 ,
the real part takes its maximum at @xmath24 . we claim that @xmath23 is a smooth path going to infinity in both directions and connecting two valleys
indeed , @xmath23 is ( in suitable parametrization ) given by a solution of the differential equation @xmath46 .
it follows that latexmath:[$\frac d{dt}\mathrm{re}(p(z(t)))= imaginary parts , so the steepest descent path passing through @xmath24 may not come close to any other critical point .
thus @xmath48 is bounded below so that , as we go away from @xmath24 , @xmath49 must go to infinity and the path @xmath23 connects two valleys .
similarly @xmath50 connects two hills .
as @xmath23 and @xmath50 cross at @xmath24 , the two valleys connected by @xmath23 are separated by hills and are thus different .
we have thus shown that for each critical point @xmath24 there is a steepest descent path @xmath23 going through @xmath24 and connecting pairs of different valleys .
on @xmath23 the real part of @xmath42 is minimal at @xmath24 so that the saddle point expansion at @xmath24 indeed gives the asymptotic expansion of the integral over @xmath23 .
we consider matrix integrals of the form for @xmath42 a polynomial of degree @xmath8 .
they are integrals of holomorphic differential forms over @xmath11 dimensional cycles . for each set of natural numbers
@xmath51 summing up to @xmath18 , we have a cycle in the normal matrices , characterized by the condition that @xmath22 eigenvalues run over the path @xmath23 of the previous section .
more precisely , the cycle @xmath12 is parametrized by @xmath52 : @xmath53 the first @xmath54 diagonal elements @xmath55 are parametrizations of @xmath56 , the next @xmath57 are parametrizations of @xmath58 and so on . the usual argument to reduce the integral to an integral over the eigenvalues ( see @xcite ) gives @xmath59 the integral is over @xmath60 and the function @xmath61 , a function on matrices invariant under conjugation , is regarded as a symmetric function of the eigenvalues . to study the large @xmath18 limit , it is useful to introduce the trace of the resolvent : @xmath62 as products of such traces are generating functions of polynomial functions invariant under conjugation .
the `` loop equation '' @xcite for this quantity is @xmath63 where @xmath64 is a polynomial of degree @xmath65 with leading coefficient @xmath66 .
this equation can be derived from the identity @xmath67 in the limit @xmath68 , the matrix integral is supposed to be dominated by an integral over a region where the eigenvalues are close ( for small @xmath19 ) to critical points of @xmath13 . with our choice of integration cycles and keeping @xmath69 fixed as @xmath68 , there will be a fraction @xmath70 of the eigenvalues close to @xmath71 . in the limit @xmath68 one
expects that the saddle point approximation becomes exact and thus @xmath72 converges to @xmath73 for some probability measure @xmath74 with support in regions around the critical points of @xmath13 and so that @xmath29 is the measure of the region around @xmath24 .
technically one assumes that the limit @xmath68 exists and that @xmath75 .
the function @xmath76 is defined and holomorphic for @xmath39 outside the support of the measure . setting @xmath77 we finally obtain @xmath78 for some polynomial @xmath79 of degree @xmath65 with leading coefficient @xmath66 .
thus the function @xmath80 , which is a priori defined on the complement of the support of the measure , has an analytic continuation to a two - fold covering of the complex plane .
the original function @xmath81 is the branch of the function defined by which behaves at infinity as @xmath82 . with the analogy with the case of hermitian matrices in mind ,
it is reasonable to assume that the measure @xmath74 has support on a collection of arcs @xmath83 which for small @xmath19 are close to the critical points of @xmath13 , and that @xmath84 for some _ density of eigenvalues _ @xmath85 defined on the arcs : namely , the measure @xmath86 of a set @xmath87 intersecting one of the arcs , say @xmath88 , in a piece @xmath89 is @xmath90 note that for the right - hand side to be defined we need to fix an orientation on @xmath88 .
then @xmath80 is a holomorphic function outside on the complement of the arcs and the density on @xmath88 is related to the discontinuity of @xmath81 : @xmath91 here @xmath92 ( @xmath93 ) denotes the limit of @xmath80 as @xmath39 tends to @xmath94 from the left ( from the right ) of the oriented curve @xmath88 . from this information
we deduce that the arcs @xmath88 connect pairs of zeros of @xmath81 , the branch points of the hyperelliptic curve @xmath95 of genus @xmath65 .
the measure of the @xmath96th arc is then the period @xmath97 over a cycle @xmath98 enclosing the pair of branch points in counterclockwise direction .
it follows from the condition on the leading coefficient of @xmath99 that @xmath100 .
we also note that since @xmath101 , we have the formula @xmath102 .
there remains to determine the precise form of the arcs and the coefficients in @xmath99 as functions of the filling fractions @xmath70 .
first of all , the relation between the @xmath65 free complex coefficients @xmath103 of @xmath79 ( recall that the leading coefficient is fixed to be @xmath104 ) and the periods @xmath70 ( @xmath105 ) subject to @xmath100 is , locally around any point where the branch points are distinct , a holomorphic diffeomorphism , since the jacobian matrix @xmath106 is non - degenerate , being the matrix of @xmath107-periods of a basis of abelian differentials on a smooth curve .
it follows that locally there exists a real @xmath108-dimensional submanifold in the complex space @xmath109 of coefficients @xmath110 which maps to real positive @xmath70 .
the condition that fixes the shape of the arcs is the reality and positivity condition for the density : if @xmath111 is a parametrization of the arc @xmath88 respecting its orientation , the condition is @xmath112 using @xmath113 , we may parametrize the arcs ( away from the endpoints ) to be solutions of the differential equation @xmath114 connecting branch points . alternatively
, arcs may be described as level lines of a function : introduce the hyperelliptic integral @xmath115 it is a holomorphic many - valued function on the complement of the support of the measure . as we go around an arc @xmath88 , @xmath116 increases by @xmath70 so @xmath117
is single valued .
the measure of a piece between two points @xmath118 , @xmath119 on a curve @xmath88 is @xmath120 which is real .
thus the arcs @xmath88 are level lines of the imaginary part of @xmath116 . around a branch point
@xmath121 which is a simple zero of @xmath122 , @xmath123 .
therefore there are three smooth level lines of @xmath61 emerging from every simple branch points . in the most general situation the support of the measure may then be a graph consisting of level lines of @xmath124 joining branch points .
a more precise description is possible in the case of small @xmath19 to which we turn . for small @xmath125
we claim that the arcs and the density of eigenvalues are determined completely by the filling fractions @xmath70 through and . to show this , notice first that as @xmath34
, pairs of branch points @xmath126 converge to the critical points @xmath71 and the periods @xmath70 ( eq . ) , regarded as functions of @xmath19 and the coefficients @xmath110 of @xmath99 are holomorphic at @xmath127 .
we have @xmath128 since there is a bijective holomorphic correspondence between values @xmath129 at the distinct points @xmath71 and coefficients @xmath110 of @xmath99 , we have at @xmath127 , and by analyticity also for small @xmath19 , a biholomorphic map @xmath130 , @xmath100 .
in particular , we can invert this map and find a unique @xmath99 for each set of @xmath131 , such that @xmath100 .
it remains to show that for all small @xmath125 there is a level line @xmath88 of @xmath117 connecting @xmath132 to @xmath133 . as @xmath34 , @xmath134
the level lines of @xmath135 at @xmath127 are thus the level lines of @xmath43 . in the neighborhood of a non - degenerate critical point
@xmath71 they look like the left picture in fig . [ f-1 ] : on any small circle around @xmath71 each value is taken on at most four times .
( 150,130)(60,0 ) for positive small @xmath19
the critical point @xmath71 splits into two branch points @xmath132 , @xmath133 from each of which three level lines emerge .
the condition that the period @xmath70 is real implies that @xmath136 .
the function @xmath137 is defined up to a sign in a neighborhood of @xmath126 , so that its zero level line is uniquely defined .
it follows that there is a level line @xmath88 of @xmath124 , namely the zero set of @xmath138 , joining @xmath132 to @xmath133 as on the right picture in fig .
[ f-1 ] : if none of the level lines emerging from @xmath132 and @xmath133 were to join , @xmath139 would take the value zero six times on any small circle encircling @xmath126 , which can not be , as this does not happen at @xmath127 . also , a level line can not go from a point @xmath132 or @xmath133 to itself as the real part of @xmath61 is monotonic along level lines .
we conclude that for any small @xmath125 , and any given filling fractions @xmath140 summing up to 1 , there is a unique polynomial @xmath141 , so that the curve @xmath142 has @xmath107-periods @xmath70 .
the zeros of @xmath81 are connected in pairs by arcs @xmath88 obeying the reality condition . according to the discussion above ,
these arcs are are the support of the measure and the density of eigenvalue is @xmath143 , @xmath144 .
we have given a non - perturbative definition of the matrix integrals that in the large @xmath18 limit give the superpotentials considered in @xcite .
we considered the case of a generic polynomial potential with complex coefficients . for small t
hooft coupling @xmath19 , the density of eigenvalues was shown to be given by arcs connecting pairs of branch points of a hyperelliptic curve .
the shape of the arcs is uniquely determined by a reality condition . for larger @xmath19 or for potentials with degenerate critical points
, one expect the arcs to combine into graphs in the complex plane . it would be interesting to understand what kind of graphs can arise in this way .
r. dijkgraaf , c. vafa , `` matrix models , topological strings , and supersymmetric gauge theories '' ( arxiv : hep - th//0206255 ) , nucl.phys . b644 ( 2002 )
320 r. dijkgraaf , c. vafa , `` a perturbative window into non - perturbative physics '' ( arxiv : hep - th/0208048 ) , 2002 f. david , `` non - perturbative effects in matrix models and vacua of two dimensional gravity '' phys.lett .
b302 ( 1993 ) 403410 c. i. lazaroiu `` holomorphic matrix models , '' hep - th/0303008 , jhep 0305 ( 2003 ) 044 n.g .
de bruijn , `` asymptotic methods in analysis , '' north - holland publishing , amsterdam , 1958 a. erd ' elyi , `` asymptotic expansions , '' dover publications , new york , 1956 s. wadia , `` dyson - schwinger equation approach to the large - n limit : model systems and string representation of yang
mills theory '' phys .
d 24 ( 1981 ) 970978 a. a. migdal , `` loop equations and 1/n expansion '' phys .
102 ( 1983 ) 199290 d. bessis , c. itzykson , j. b. zuber , `` quantum field theory techniques in graphical enumeration , '' advances in applied mathematics , 1:109157 , 1980 . | we study a class of holomorphic matrix models .
the integrals are taken over middle dimensional cycles in the space of complex square matrices . as the size of the matrices tends to infinity
, the distribution of eigenvalues is given by a measure with support on a collection of arcs in the complex planes .
we show that the arcs are level sets of the imaginary part of a hyperelliptic integral connecting branch points .
_ department of mathematics , eth zurich _ _ 8092 zurich , switzerland _ |
the alpha magnetic spectrometer @xcite ( ams ) is a particle detector to be installed in the international space station ( iss ) for at least three years .
the spectrometer will be able to measure the rigidity ( @xmath0 ) , the charge ( @xmath1 ) , the velocity ( @xmath2 ) and the energy ( @xmath3 ) of cosmic rays from some mev up to @xmath41tev within a geometrical acceptance of @[email protected] .
shows a schematic view of the ams spectrometer . at both ends of the ams spectrometer
exist the transition radiation detector ( trd ) ( top ) and the electromagnetic calorimeter ( ecal ) ( bottom ) .
both will provide ams with capability to discriminate between leptons and hadrons . additionally the calorimeter will trigger and detect photons .
the trd will be followed by the first of the four time - of - flight ( tof ) system scintillator planes .
the tof is composed of four roughly circular planes of 12 cm wide scintillator paddles , one pair of planes above the magnet , the upper tof , and one pair bellow , the lower tof
. there will be a total of 34 paddles .
the tof will provide a fast trigger within 200ns , charge and velocity measurements for charged particles , as well as information on their direction of incidence .
the tof operation at regions with very intense magnetic fields forces the use of shielded fine mesh phototubes and the optimization of the light guides geometry , with some of them twisted and bent .
moreover the system guarantees redundancy , with two photomultipliers on each end of the paddles and double redundant electronics .
a time resolution of 140ps for protons is .
the tracking system will be surrounded by veto counters and embedded in a magnetic field of about 0.9tesla produced by a superconducting magnet .
it will consist on a silicon , made of 8 layers of double sided silicon sensors with a total area of @xmath46.7m@xmath6 .
there will be a total of @xmath42500 silicon sensors arranged on 192 ladders .
the position of the charged particles crossing the tracker layers is measured with a precision of @xmath410@xmath7 m along the bending plane and @xmath430@xmath7 m on the transverse direction . with a bending power ( bl@xmath6 ) of around 0.9t.m@xmath6 ,
particles rigidity is measured with an accuracy better than 2% up to 20gv and the maximal detectable rigidity is around 1 - 2tv .
electric charge is also measured from energy deposition up to z@xmath426 .
the ring imaging erenkov detector ( rich ) @xcite will be located right after the last tof plane and before the electromagnetic calorimeter .
it will be described in detail in next section .
[ fig : ams ] the long stay of ams in space and its large acceptance will allow the accumulation of a large statistic of events increasing in several orders of magnitude the sensitivity of the proposed physical measurements . with an average collection rate of 1000 events per second
, a total of @xmath8 protons per year and around @xmath9 antiprotons will be accumulated .
the main goals of the ams-02 experiment are : * a precise measurement of the charged cosmic ray spectrum between and @xmath41tev , and the detection of photons up to a few hundred gev ; * a search for heavy antinuclei ( @xmath10 ) , which if discovered would signal the existence of primordial antimatter ; * search for non - baryonic dark matter through the detection of annihilation products appearing as anomalies of the cosmic - ray spectra ( e@xmath11 , @xmath12 , @xmath13 and @xmath14 ) ;
the rich is a proximity focusing device with a dual radiator configuration on the top made of 92 aerogel 25 mm thick tiles with a refractive index 1.050 and sodium fluoride ( naf ) tiles with a thickness of 5 mm in the center covering an area of 34@xmath1534cm@xmath6 .
the naf placement prevents the loss of photons in the hole existing in the center of the readout plane ( @xmath17 cm@xmath6 ) , in front of the ecal calorimeter located below .
the radiator tiles are supported by a 1 mm thick layer of methacrylate ( n=1.5 ) free of uv absorbing additives .
the detection matrix is composed of 680 multiplixelized photon readout cells each consisting of a photomultiplier coupled to a light guide , hv divider plus front - end ( fe ) electronics , all housed and potted in a plastic shell and then enclosed in a magnetic shielding with a thickness varying from 0.8 to 1.2 mm .
the photon detection is made with an array of multianode hamamatsu tubes ( r7600 - 00-m16 ) with a spectral response ranging from 300 to 650 nm and a maximum quantum efficiency at @xmath18@xmath4420 nm . to increase the photon collection efficiency ,
a light guide consisting of 16 solid acrylic pipes glued to a thin top layer ( 1 mm ) was produced .
it is optically coupled to the active area of phototube cathode through a @xmath19 mm flexible optical pad . with a total height of @xmath20 mm and a collecting surface of @xmath21 mm@xmath6
, it presents a readout pixel size of @xmath22 mm and a pitch of 37 mm . the light guide is mechanically attached through nylon wires to the photomultiplier polycarbonate housing .
a high reflectivity conical mirror surrounds the whole set .
the mirror was included to increase the device acceptance since around 33% of the aerogel generated photons impact on the mirror .
it consists of a carbon fiber reinforced composite substrate with a multilayer coating made of aluminium and sio@xmath23 deposited on the inner surface .
this ensures a reflectivity higher than 85% for 420 nm wavelength photons .
the rich has a truncated conical shape with an expansion height of 46.3 cm , a top radius of 60 cm and a bottom radius of 67 cm .
the total height of the detector is 60.5 cm and it covers 80% of the ams magnet acceptance . shows a schematic view of the rich detector .
rich was designed to measure the velocity ( @xmath24 ) of singly charged particles with a resolution @xmath25 of 0.1@xmath26 , to extend the charge separation ( z ) up to iron ( z=26 ) , to contribute to @xmath27 separation and to albedo rejection . in order to validate the rich design ,
a prototype with an array of 9@xmath1511 cells filled with 96 photomultiplier readout units similar to part of the matrix of the final model was constructed .
the performance of this prototype has been tested with cosmic muons and with a beam of secondary ions at the cern sps produced by fragmentation of a primary beam in 2002 and 2003 .
the light guides used were prototypes with a slightly smaller collecting area ( 31@xmath1531mm@xmath6 ) .
different samples of the radiator materials were tested and placed at an adjustable supporting structure .
different expansion heights were set in order to have fully contained photon rings on the detection matrix like in the flight design .
a segment of a conical mirror with 1/12 of the final azimuthal coverage , which is shown in left picture of , was also tested .
the rich assembly has already started at ciemat in spain and is foreseen to be finished in july 2007 .
a rectangular grid has already been assembled and has been subject to a mechanical fit test , functional tests , vibration tests and vacuum tests .
the other grids will follow .
the refractive index of the aerogel tiles is being measured and the radiator container was subjected to a mechanical test .
the final integration of rich in ams will take place at cern in 2008 .
[ cols="^,^ " , ] runs with a mirror prototype were also performed and its reflectivity was derived from data analysis .
the obtained value is in good agreement with the design value .
ams-02 will be equipped with a proximity focusing rich detector based on a mixed radiator of aerogel and sodium fluoride , enabling velocity measurements with a resolution of about 0.1% and extending the charge measurements up to the iron element .
velocity reconstruction is made with a likelihood method .
charge reconstruction is made in an event - by - event basis .
evaluation of both algorithms on real data taken with in - beam tests at cern , in october 2003 was done .
the detector design was validated and a refractive index 1.05 aerogel was chosen for the radiator , fulfilling both the demand for a large light yield and a good velocity resolution .
the rich detetector is being constructed and its assembling to the ams complete setup is foreseen for 2008 .
c. lechanoine - leluc , _ proc .
29th icrc ( pune ) _ * 3 * , 381 - 384 ( 2005 ) .
f. baro , l. arruda , _ et al .
_ , _ proc .
29th icrc ( pune ) _ * 9 * , 299 - 302 ( 2005 ) . c. lechanoine - leluc , _ proc .
29th icrc ( pune ) _ * 9 * , 299 - 302 ( 2005 ) .
d. casadei et al . , nuclear physics b ( proc .
suppl . ) * 113*,133 ( 2002 ) . . | the alpha magnetic spectrometer ( ams ) to be installed on the international space station ( iss ) will measure charged cosmic ray spectra of elements up to iron , in the rigidity range from 1gv to 1tv , for at least three years .
ams is a large angular spectrometer composed of different subdetectors , including a proximity focusing ring imaging cherenkov ( rich ) detector .
this will be equipped with a mixed radiator made of aerogel and sodium fluoride ( naf ) , a lateral conical mirror and a detection plane made of 680 photomultipliers coupled to light guides .
the rich detector allows measurements of particle s electric charge up to iron , and particle s velocity .
two possible methods for reconstructing the erenkov angle and the electric charge with the rich will be discussed . a rich prototype consisting of a detection matrix with 96 photomultipliers , a segment of a conical mirror and samples of the radiator materials was built and its performance was evaluated using ion beam data .
results from the last test beam performed with ion fragments resulting from the collision of a 158gev / c / nucleon primary beam of indium ions ( cern sps ) on a lead target are reported . the large amount of collected data allowed to test and characterize different aerogel samples and the naf radiator .
in addition , the reflectivity of the mirror was evaluated .
the data analysis confirms the design goals . |
recent detections of near - infrared excesses around a number of metal - rich white dwarfs ( wd ) imply existence of warm circumstellar material reprocessing stellar radiation ( zuckerman & becklin 1987 ; graham 1990 ; farihi 2010 ) .
spectral modeling suggests that this material resides in an extended , compact , optically thick and geometrically thin disk ( jura 2003 ; jura 2007 ) , similar to the saturn s rings ( cuzzi 2010 ) .
disks have inner edges at several tens of the wd radii @xmath5 , roughly consistent with being set by particle sublimation at these locations .
their outer radii lie close to the roche radius of the wd @xmath6 r@xmath7 , supporting the suggestion by jura ( 2003 ) that such compact debris disks are produced by tidal disruption of asteroid - like bodies scattered into low - periastron orbits by gravitational perturbations of massive planets , which have survived the agb phase of stellar evolution .
availability of large reservoir of high - z elements in the form of debris disk in the immediate vicinity of some wds naturally led to the suggestion ( jura 2003 ) that metal enrichment of these stars is caused by accretion from such disks .
this scenario provides a promising alternative to the previously proposed _ interstellar _ accretion model of metal pollution of wds ( dupuis 1993 ) , which is not consistent with observations of wds with he atmospheres .
theoretical estimates of settling time of heavy elements in wds imply that their observed atmospheric abundances can be maintained against gravitational settling by accretion of metals at rates @xmath8 g s@xmath1 ( farihi 2009 , 2010 ) .
if the _ circumstellar _ accretion hypothesis is correct , an evolving disk of debris must be able to supply such @xmath9 to the wd .
the actual transfer of metals from the disk of solids truncated at sublimation radius @xmath10 to the wd must be accomplished in this picture via the gas disk extending from the wd surface to @xmath10 and beyond .
observational evidence of such gaseous component around several metal - rich wds hosting compact debris disks has been found by gnsicke ( 2006 , 2007 , 2008 ) in the form of double - peaked emission lines of ca ii and fe ii .
these spectroscopic signatures are naturally explained as arising in a disk of metallic gas ( no h or he emission lines have been detected around these wds ) in keplerian rotation around wds and spatially coincident with dusty disks ( melis 2010 ) . even though metals are passed to the wd through the gaseous disk the rate of mass transfer @xmath9
is ultimately controlled by evolution of the particulate disk . by analogy with saturn s rings
one expects that the collisional viscosity in debris disk is too low to drive the non - negligible @xmath9
. however , radiation of the wd can be quite important and rafikov ( 2011 ; hereafter r11 ) demonstrated that poynting - robertson ( pr ) drag on the disk naturally drives mass accretion at the rate @xmath11 g s@xmath1 .
while this @xmath9 is high it still falls short of explaining the highest observed @xmath4 g s@xmath1 .
here we propose a new picture of the debris disk evolution , which naturally combines several physical ingredients present in close vicinity of disk - bearing wds .
our model of the wd - disk system is described below and is schematically illustrated in figure [ fig : f1 ] .
we assume a disk of solid particles to extend from the roche radius @xmath12 all the way in to the sublimation radius r_s ( ) ^222 r_t_,4 ^ 2 ( ) ^2 , [ eq : r_s ] where @xmath13 is the sublimation temperature of solid particles ( which we take to be @xmath14 k if particles are silicate ) , @xmath15 k ) is the normalized stellar temperature @xmath16 , @xmath5 is the wd radius , and @xmath17 is the ratio of particle emissivities for starlight and for its own thermal radiation ( in the following we assume macroscopic particles and set @xmath18 ) . for @xmath19
( ehrenreich 2011 ) one finds @xmath20 r@xmath7 , in agreement with observations ( jura 2007 , 2009 ) . at present the size of particles @xmath21 constituting
the disk is rather poorly constrained .
graham ( 1990 ) suggested @xmath22 cm . at the same time high - resolution infrared spectroscopy with irs onboard of _ spitzer _ reveals strong @xmath23 m emission feature in disk spectra indicative of the existence of a population of small , micron - sized silicate dust particles ( jura 2009 ) .
but the fraction of total disk mass locked up in such fine dust is unknown .
the actual particle size is not very important for our treatment as long as the disk is optically thick .
then it is similar to saturn s rings and most likely behaves as a granular flow . in the following
we will treat particulate disk as if it were a solid plate .
solid particles brought to @xmath10 sublimate feeding a disk of metallic gas at this location .
viscous torques cause it to spread all the way to the wd surface providing means of metal transport from @xmath10 to the star .
however , because of angular momentum conservation part of the newly produced metallic gas has to move _ outwards _ of @xmath10 ( lynden - bell & pringle 1974 ) naturally explaining the existence of gaseous disks spatially coexisting with dusty debris disks in some systems ( melis 2010 ) .
jura ( 2008 ) proposed collisional sputtering of small asteroids as another source of metallic gas ; for the sake of clarity we do not consider this mechanism here .
we emphasize here that the two disks have an overlap in _ radial _ distance , while in vertical direction gaseous disk is much more extended than the disk of particles , see figure [ fig : f1 ] .
the latter must be very thin because inelastic collisions between particles rapidly damp any vertical random motions .
external gas disk outside @xmath10 has temperature different from ( higher than ) that of the dust disk at the same radii ( melis 2010 ) because of the different balance of heating and cooling for the two disks .
this keeps gas temperature @xmath24 above the sublimation point even outside @xmath10 , although some condensation of metallic gas on the particle surfaces may be happening there ( we neglect it in this work ) .
simultaneous existence of the two disks of high - z elements in different phases drives their mutual evolution in the following way .
any coupling between the outer portion of the gas disk and particulate disk acts to transfer angular momentum from the faster rotating particulate disk to the slower spinning gaseous disk , causing inward motion of particles in the disk of solids .
if the coupling is strong enough positive feedback becomes possible in the system : increasing mass of gaseous disk leads to the increase of @xmath9 through the disk of solids ( as described below ) , which in turn reinforces evaporation at @xmath10 and increases gaseous mass even further .
coupling between the gaseous and particulate high - z disks arises because gaseous disk orbits wd at angular speed @xmath25 slightly lower than the keplerian speed @xmath26 at which the disk of solid particles rotates .
this difference is caused by the pressure support present in gaseous disk , @xmath27 , and is known to cause a variety of important effects in protoplanetary disks , such as inward migration of solids ( weidenschilling 1977 ) .
relative azimuthal velocity between the gaseous and particulate disks at distance @xmath28 from the wd is v_=c_s10 ^ 2 ^ -1 ( m_,1 ^ -1 ) ^1/2 , [ eq : v_rel ] where @xmath29 is a constant , @xmath30 is the normalized wd mass @xmath31 , @xmath32 is the mean molecular weight of the metallic gas normalized by @xmath33 ( value of @xmath34 for pure si ) , @xmath35 k ) is the normalized gas temperature @xmath24 , and @xmath36 km s@xmath37 is the gas sound speed ( clearly @xmath38 ) .
this azimuthal drift gives rise to aerodynamic drag between the disks .
azimuthal drag force @xmath39 per unit surface area of a particulate disk with surface mass density @xmath40 causes inward radial migration of the disk material at speed @xmath41 .
this inward particle drift gives rise to mass transport at the rate m_z=2r v_r_d=. [ eq : dotmgen ] note that if @xmath39 is independent of @xmath40 then the same is true for @xmath9 .
external force @xmath39 per unit area can be generally represented in the form f_=a_g+b , [ eq : f_e_decomp ] where the first term describes the coupling between the gas disk with surface density @xmath42 and particulate disk , and constant @xmath43 determines the strength of coupling .
it is natural to expect that drag scales with gas density @xmath42 ; this expectation is confirmed in [ sect : aero ] .
second term @xmath44 represents forces acting even in the absence of coupling to gas disk .
pr drag is an example of such force and one can easily show ( r11 ) that for pr drag @xmath45 , where @xmath46 is the speed of light , @xmath47 is the efficiency of radiative momentum absorption by the disk surface , and @xmath48 ( friedjung 1985 ) is the incidence angle of stellar radiation . from equation ( [ eq : dotmgen ] ) the rate of mass transport due to pr drag alone ( when @xmath49 ) at @xmath10 is ( r11 ) m_pr&=&= [ eq : mdot_pr ] + & & 10 ^ 8 ^ -1_r . in the following
we will assume that @xmath50 .
particle sublimation at @xmath10 increases the mass of gaseous disk at the rate @xmath9 . assuming that surface density of gaseous disk @xmath42 varies on scale @xmath51 we can write that sublimation increases @xmath42 at the rate @xmath52 . at the same time , viscous spreading reduces @xmath42 at the rate @xmath53 , where @xmath54 is the characteristic viscous time in the disk t_~10 ^-1 ( m_,1 ) ^1/2 , [ eq : t_nu ] assuming @xmath55-parametrization of viscosity @xmath56 ( shakura & sunyaev 1973 ) .
this timescale can be very short if @xmath55 is not very small .
we can now describe the evolution of @xmath42 in the vicinity of @xmath10 with the following heuristic equation : = _ + -_-= ( a_g+b)- .
[ eq : heuristic ] a solution of this equation satisfying the initial condition @xmath57 ( no gas disk initially ) is _ g(t)&= & ( 1-)^-1 + & & \{-1 } , [ eq : sol ] where we used equation ( [ eq : mdot_pr ] ) and t_s= , [ eq : tau_s ] is the _ sublimation time
_ @xmath58 on which the mass of gaseous disk increases if @xmath59 . from equations ( [ eq :
dotmgen ] ) , ( [ eq : f_e_decomp ] ) and ( [ eq : sol ] ) the rate at which particles sublimate is = ( 1-)^-1 \{- } , [ eq : mzsol ] and the total mass lost by the debris disk to sublimation @xmath60 can be easily obtained by integrating this expression . the behavior of @xmath9 and @xmath60 is shown in figure [ fig : f2 ] for different values of @xmath61 , which clearly demonstrates that for @xmath62 ( i.e. as long as @xmath63 ) debris disk evolution is insensitive to @xmath61 and @xmath64 . this is because initially the disk of metallic gas is not dense enough for the drag it produces on the debris disk to compete with the pr drag it takes certain time to accumulate enough gas mass by particle sublimation .
this means , in particular , that if the debris disk starts with mass @xmath65 , which is lower than the critical mass @xmath66 , then its evolution is determined only by the pr drag and coupling to the gaseous disk is never effective the debris disk is eroded before gas disks grows massive enough .
lifetime of the debris disk is then t_pr=3 ( ) ^-1 , [ eq : t_pr ] independent of the relation between @xmath67 and @xmath54 . in the opposite case of a massive initial disk @xmath68
evolution does depend on @xmath61 . whenever @xmath69 the action of viscosity is so effective at removing gas released by particle sublimation at @xmath10 that the gas does not accumulate there
. then @xmath42 simply saturates at the constant low level , accretion rate tends to @xmath70 for @xmath71 , and disk gets exhausted on timescale @xmath72 given by equation ( [ eq : t_pr ] ) .
this evolutionary path clearly can not explain the highest observed values of @xmath9 .
however , in the case of a massive disk and @xmath73 sublimation supplies gas to the @xmath74 region faster than viscous diffusion removes it , and @xmath42 grows exponentially on timescale @xmath75 ( as long as particle disk has enough mass to provide the source ) .
this means that @xmath9 also increases exponentially for @xmath76 ( see figure [ fig : f2 ] ) .
this runaway stops and gaseous disk wanes only when the mass of particulate disk @xmath65 is exhausted .
the latter happens on the _ runaway timescale _
t_runt_s(1-)^-1 ( ) , [ eq : t_end ] determined by the condition @xmath77 .
clearly , @xmath78 if @xmath79 .
thus , the ratio = [ eq : feedback ] is the critical parameter determining the strength of positive feedback for massive disks : if @xmath80 a particulate disk gets rather rapidly ( within several @xmath67 ) converted into metallic gas at @xmath10 , with its subsequent viscous accretion onto the wd on timescale of several @xmath54 . whereas for @xmath81 feedback is not strong enough to reinforce gas supply at @xmath10 and disk slowly evolves on timescale @xmath72 due to pr drag . as equation ( [ eq : feedback ] ) shows , runaway evolution and high @xmath9 require ( 1 ) strong coupling between the gaseous and particulate disks and ( 2 ) low viscosity .
relative azimuthal motion between gaseous and particulate disks at speed @xmath82 induces _ aerodynamic drag _ between them .
drag produces force per unit area of the disk ( schlichting 1979 ) f_=_^-1_g v_^2 , [ eq : aerodrag ] where @xmath83 is the gas density and re@xmath84 is a proportionality constant .
there is a significant spread of opinions regarding the value of re@xmath85 characterizing drag by the turbulent flow on a smooth solid plate , with numbers ranging between re@xmath86 ( dobrovolskis 1999 ) to re@xmath87 ( goldreich & ward 1973 ) .
it is also likely that a smooth plate approximation underestimates drag ( overestimates the value of re@xmath85 ) for the particulate debris disk , which does not have continuous surface and may interact with gas more like a rough plate ( schlichting 1979 ) or even as a combination of individual particles , in which case smaller re@xmath85 is more appropriate .
according to equation ( [ eq : aerodrag ] ) aerodynamic drag force can be written in the form @xmath88 with a== , [ eq : a_a ] where equation ( [ eq : v_rel ] ) was used . from equation ( [ eq : tau_s ] )
sublimation time at @xmath89 is t_s=10 ^ 3 , [ eq : t_s_a ] where @xmath90 km s@xmath91 .
the critical mass separating low- and high - mass disks ( see [ sect : coupled ] ) is m_prt_s310 ^ 18 .
[ eq : crit_mass ] the feedback parameter for aerodynamic coupling is = ( ) ^1/2 [ eq : f_a ] and depends on both the viscosity parameter @xmath55 and the strength of aerodynamic coupling , parameterized by re@xmath85 . equation ( [ eq : f_a ] ) shows that runaway evolution with @xmath92 requires rather low viscosity in the gaseous disk , at the level of @xmath93 , and for higher re@xmath85 smaller @xmath55 is needed .
viscosity is most likely provided by the magnetorotational instability ( mri ) , which requires certain level of ionization to operate effectively . in the ideal mhd limit simulations with no net flux
typically produce @xmath94 ( hawley 1995 ) .
however , in our scenario gaseous disk exists in immediate contact with the particulate disk , which is observationally known to contain a population of micron size dust grains ( jura 2009 ) .
mri - driven turbulence will mix some of this fine dust with the gas , lowering ionization fraction ( small grains have large surface area and are very efficient charge absorbers ) , and giving rise to the non - ideal mhd effects , e.g via the increased resistivity ( balbus 2009 ) .
the latter are known from simulations to decrease effective @xmath55 substantially , down to @xmath95 or lower ( fleming 2000 ; bai & stone , in preparation ) . as a result , it is conceivable that @xmath55 in our model can be much lower than in the ideal mhd limit of mri .
thus , presence of dusty debris disk in close contact with the gaseous disk can quite naturally lengthen viscous timescale @xmath54 and facilitate gas accumulation , making possible runaway evolution due to aerodynamic drag .
we note that after the particulate disk completely sublimates the value of @xmath55 in the resultant gaseous disk will go up since the source of small dust particles lowering ionization has disappeared .
closer to the wd , in the dust - free region @xmath96 , ionization and @xmath55 can _ always _ be higher than in the external disk .
faster viscous evolution in this region ( resulting in lower @xmath42 ) may explain lack of line emission from the inner part of the gas disk ( gnsicke 2006 , 2007 , 2008 ; melis 2010 ) .
our calculations clearly demonstrate that compact debris disks around wds are self - destructive whenever there is strong coupling between the gaseous and solid components . adopting for illustration re@xmath97 and setting all other dimensionless constants to unity we find from equations ( [ eq : t_s_a])-([eq : f_a ] ) @xmath98 yr , @xmath99 g and that @xmath100 is needed for @xmath80 .
then the typical lifetime of a massive disk with @xmath101 g ( mass of a 200 km asteroid ) in the runaway scenario ( if the viscosity is low ) is on the order of several sublimation timescales , i.e. about @xmath102 yr . according to equations ( [ eq : mzsol ] ) and ( [ eq : t_end ] ) in this case the _ maximum _
@xmath9 for this @xmath65 , achieved right before the debris disk completely disappears , is max@xmath103 g s@xmath1 .
these numbers agree with observational inferences ( farihi 2009 , 2010 ) quite well . however , if positive feedback is not strong enough to drive runaway ( e.g. because @xmath104 or higher re@xmath85 ) , then @xmath105 and it takes 3 myr for the same disk to be exhausted , as equation ( [ eq : t_pr ] ) demonstrates . in this case gas surface density at @xmath10 saturates at the low level @xmath106 g @xmath107 , see equation ( [ eq : sol ] ) .
equation ( [ eq : t_pr ] ) also shows that a low - mass disk with @xmath108 g ( mass of a 20 km asteroid ) is destroyed by pr drag very rapidly , within several thousand years ( as long as the disk is optically thick to incoming stellar radiation , see r11 ) .
our model of debris disk evolution naturally explains coexisting gaseous and particulate debris disks reported in melis ( 2010 ) .
systems with reported ir excesses but lacking emission lines of high - z elements in gas phase may simply possess lower mass gaseous disks , which have not had enough time to develop by sublimation of solids at @xmath10 .
compositional variations between different wds may also explain such systems .
calculations presented in this work are rather simple and local in nature .
we studied only one coupling mechanism aerodynamic drag , while other possibilities may be available as well ( e.g. induction interaction ( drell 1965 ; gurevich 1978 ) between the mri - generated b - field in the gaseous disk and the debris disk particles ) .
also , here we did not consider low surface density disks ( we use only the thin plate approximation ) , non - trivial initial radial distribution of debris surface density , fate of angular momentum lost by debris disk and absorbed by gas disk , and so on .
future global models of coupled evolution of gaseous and particulate debris disks ( bochkarev & rafikov , in preparation ) will take these issues into account to provide a more accurate description of wd pollution with circumstellar high - z material .
rrr is grateful to xue - ning bai and konstantin bochkarev for useful discussions .
the financial support of this work is provided by the sloan foundation , nasa via grant nnx08ah87 g , and nsf via grant ast-0908269 . | it was recently proposed that metal - rich white dwarfs ( wds ) accrete their metals from compact debris disks found to exist around more than a dozen of them . at the same time , elemental abundances measured in atmospheres of some wds imply vigorous metal accretion at rates up to @xmath0 g s@xmath1 , far in excess of what can be supplied solely by poynting - robertson drag acting on such debris disks .
to explain this observation we propose a model , in which rapid transport of metals from the disk onto the wd naturally results from interaction between this particulate disk and spatially coexisting disk of metallic gas .
the latter is fed by evaporation of debris particles at the sublimation radius located at several tens of wd radii . because of pressure support gaseous disk orbits wd slower than particulate disk .
resultant azimuthal drift between them at speed @xmath2 m s@xmath1 causes aerodynamic drag on the disk of solids and drives inward migration of its constituent particles . upon reaching the sublimation radius particles evaporate , enhancing the density of metallic gaseous disk and leading to positive feedback . under favorable circumstances ( low viscosity in the disk of metallic gas and efficient aerodynamic coupling between the disks )
system evolves in a runaway fashion , destroying debris disk on time scale of @xmath3 yr , and giving rise to high metal accretion rates up to @xmath4 g s@xmath1 , in agreement with observations . |
topological insulators have attracted a lot of interests due to their novel properties and potential applications @xcite .
one of the prototype of topological insulators is the haldane model of chern insulators @xcite , which describes spinless fermions hopping on a 2d honeycomb lattice , within a staggered magnetic field . with the shaking - lattice technique @xcite , the haldane model has been realized directly in a recent cold atom experiment @xcite , and the topological phase diagram is investigated using the bloch oscillation method .
it is worth mentioning that , although the haldane model is realized in the cold atom setup , the topological ground state for this system is generally hard , if not impossible , to be reached .
as the system is generally prepared in a topologically trivial state before the shaking is applied , by driving the system crossing the phase boundary to the topological phase , excitations are inevitable as the band gap closes at the phase boundary and adiabaticity could not be maintained .
the idea of a quantum quench , which starts with the ground state of a topologically trivial hamiltonian @xmath2 supplemented by the evolution of a topologically trivial / nontrivial hamiltonian @xmath3 , provides an alterative way to studying the properties of the haldane model ( and haldane - like topological models ) @xcite .
although the chern number @xcite of the time - dependent state remains unchanged as the state evolution governed by @xmath3 is unitary @xcite , the edge states @xcite as well as nontrivial hall response @xcite can be built up dynamically .
it is interesting to see that , quantum quenches can also be used to perform bloch - state tomography , as proposed in ref .
@xcite , and then realized experimentally for a haldane - like model on the honeycomb lattice by the hamburg group @xcite . in this experiment , they prepared the initial ( topologically trivial ) state adiabatically , and then measured the bloch state for each quasimomentum @xmath8 parameterized on the bloch sphere as @xmath1 .
the berry curvature of the corresponding topologically trivial state is also mapped out .
more recently , generalizing their previous scheme in ref .
@xcite , the hamburg group performed a double - quench experiment and observed emergent phase vortices in the azimuthal phase @xmath9 @xcite . to be more specific
, they prepare the initial state as all the @xmath10-dependent bloch vectors pointing to the north pole regime @xmath11 ( i.e. , @xmath12 ) , then quench the hamiltonian to @xmath3 giving rise to the time - dependent bloch state @xmath13 . a second quench to a flat band hamiltonian
is performed to measure the time- and quasimomentum- dependent state parameterized by @xmath14 , and @xmath15 .
they find that , two different types of vortices emerge in the phase profile of @xmath15 .
the first type is static vortices , which indicate the dirac points of @xmath3 .
the locations of this kind of vortices do not move as the name _
static _ implies .
the second type is dynamical vortices .
this kind of vortices emerge for particular quasimomentum points at particular times for some particular parameters of @xmath3 .
this kind of vortices can be pairwise created at some @xmath0 points .
they then move along some trajectories , and finally pairwise annihilate each other , making the trajectories closed .
the appearance of dynamical vortices is identified as the dynamical order parameter for a dynamical phase transition @xcite , which describes the singularities of the loschmidt amplitude of the time - dependent state at a critical time .
in addition to the identification of phase vortices , the dynamical phase diagram , which is defined as the parameter regime of the final hamiltonian @xmath3 that support the appearance of dynamical vortices , is also explored .
they find that , the region of the dynamical phase diagram is larger than the one of the equilibrium phase diagram of @xmath3 .
the dynamics of the phase vortices and the phenomena of enlarged dynamical phase diagram are not quantitatively discussed in ref .
@xcite .
it is the purpose of the current work to explore the dynamics of the phase vortices for a quenched topological model . as the effective hamiltonian ( of a haldane - like model ) in ref .
@xcite forbids a simple analytical expression , we study instead the phase vortices of the quenched haldane model .
we give explicitly the expression of the trajectories of the dynamical vortices , and give the possible reason for the enlarged dynamical phase diagram .
we will show that , the trajectories of the dynamical vortices tell exactly the chern number of the hamiltonian @xmath3 : when the dynamical vortices wind around static vortices , the lower - band chern number of @xmath3 is @xmath16 ; while when the dynamical vortices do not wind around static vortices , we have @xmath17 . more generally , as is proved in detail in ref .
@xcite , the linking number between these two kinds of trajectories in the @xmath18 space equals to the chern number of the lower band .
thus the quench protocol in ref .
@xcite can be used to directly identify the topological phase boundaries of the equilibrium hamiltonian .
the paper is organized as follows . in sec .
[ sec:2 ] , we describe the quench protocol of the haldane model , and study the state evolution of this system for the case of infinite large initial energy offset @xmath19 in the initial hamiltonian @xmath2 ( i.e. , @xmath20 ) . in sec .
[ sec:3 ] , the evolution of the time - dependent azimuthal phase @xmath15 is explored .
we show that , once the south pole ( @xmath21 ) is reached , the dynamical vortices are pairwise created .
the trajectories of the dynamical vortices as well as the dynamical phase diagram are identified . in sec .
[ sec : finite_mi ] , we study the case of a finite initial energy offset @xmath22 , and the enlarged dynamical phase diagram is achieved .
we conclude in sec .
[ sec : conclusion ] .
we consider the haldane model @xcite on the honeycomb lattice [ fig .
[ fig1](a ) ] with the following hamiltonian : @xmath23 where @xmath24 and @xmath25 are the annihilation operators for the @xmath26 and @xmath27 sublattices with coordinates @xmath28 respectively , @xmath29 and @xmath30 are respectively the nearest - neighboring ( nn ) and next - nearest - neighboring ( nnn ) hopping amplitude , @xmath31 ( @xmath32 ) with @xmath33 are the vectors pointing to the nn ( nnn ) sites , @xmath34 is the sublattice energy offset . here
@xmath35 are the nnn hopping phases , and the positive sign is taken for the nnn hopping along the arrows as shown in fig .
[ fig1](a ) .
we set @xmath36 as the length unit .
the fourier transformed hamiltonian in the quasimomentum space gives the following compact form @xmath37 where @xmath38 is the quasimomentum , @xmath39 and @xmath40 are the sublattice annihilation operators . in the above expression
, @xmath41 is a @xmath42 matrix with the form @xmath43 where @xmath44 is the @xmath42 unit matrix , @xmath45 is the vector of pauli matrices , @xmath46 , and @xmath47 with @xmath48 the spectrum of the system is @xmath49 , where @xmath50 .
the topology of the system can be characterized by the ( first ) chern number @xcite of the lowest band as follows @xcite @xmath51 where @xmath52 .
the nontrivial winding of @xmath53 [ see fig .
[ fig1](b ) ] exists in the parameter regime @xmath54 , which gives the phase diagram of the system [ fig .
[ fig1](c ) ] . and @xmath27 take an energy offset @xmath34 .
the arrows indicate the tunneling coefficients with positive hopping phases .
( b ) the normalized vector @xmath55 in the quasimomentum space for the topologically nontrivial case .
the hexagon shows the first brillouin zone .
( c ) the topological phase diagram of the haldane model and the quench protocol .
sublattice energy offset @xmath34 is quenched from @xmath22 to @xmath56 for a particular @xmath57 .
( d ) the state evolution visualized on the bloch sphere for a particular quasimomentum point @xmath8 .
( e ) the state evolution for quasimomentum points in the first brillouin zone , viewed from the north pole .
the south pole region is reached for time @xmath58 . here
we take @xmath59 , @xmath60 , @xmath61 , @xmath62 , and @xmath63 . , title="fig : " ] + we now consider the case that the sublattice energy offset @xmath34 is time dependent : @xmath64 suppose the system is initially prepared in the ground state of the haldane hamiltonian @xmath65 with offset @xmath22 , after the quench at @xmath66 the state evolves under a new hamiltonian @xmath67 with offset @xmath56 [ fig .
[ fig1](d ) ] .
we then want to ask the following question : could the time - dependent state give any insight of the topological property of the system ?
the answer is _
yes_. the time - dependent state , albeit always is topologically trivial , can indeed be used to identify the topological phase diagram of the equilibrium hamiltonian @xcite . for conceptual simplicity ,
we first consider @xmath68 , and defer the more complex case of finite @xmath22 to sec .
[ sec : finite_mi ] . for the @xmath68 case
, the initial single - particle state for each quasimomentum point reads @xmath69 under the hamiltonian @xmath70 , we get ( we always take the reduced planck constant @xmath71 ) @xmath72 in the above , as the initial state is @xmath73-independent , the subscript @xmath74 is omitted to keep notation concise .
this state can be parameterized on a bloch sphere as @xmath75 then we have @xmath76 .
\end{gathered } \right.\ ] ] the time- and quasimomentum - dependent bloch vectors are then given by @xmath77 the typical results for the state evolution visualized by the bloch vectors are given in fig .
[ fig1](e ) . as indicated in fig .
[ fig1](e ) , some of the bloch vectors can reach the south pole region as time goes by . as the effective magnetic field [ @xmath78 in eq .
( [ h_eff ] ) ] that drives the rotation of bloch vector is orthogonal to the latter , it should have a vanishing @xmath79 component for this particular case , i.e. , we should have @xmath80 . according to eq .
( [ h0_hxyz_haldane ] ) , we rewrite @xmath81 as @xmath82 , where @xmath83 is given by @xmath84 we plot @xmath83 in fig . [ fig2](b ) . we see that , @xmath83 peaks at @xmath85 points , and thus is bounded as @xmath86 . we thus see that , only when the hamiltonian is in the topological phase ( i.e. , @xmath87 ) does the @xmath88 function features zeros .
we now focus on the evolution of the time- and quasimomentum - dependent azimuthal phase @xmath89 , as shown in fig .
[ fig2](a ) .
the main features of the azimuthal phase dynamics is as follows : ( i ) there are always static vortices in the phase pattern at the corners of the brillouin zone ( k and k points ) in fig .
[ fig2](a ) .
( ii ) there are dynamical vortices that emerge [ fig
. [ fig2](a)iv ] , propagate [ fig .
[ fig2](a)(v - vi ) ] , and annihilate [ fig . [ fig2](a)vii ] in the quasimomentum space .
( iii ) the trajectories of the dynamical vortices are just the zeros of @xmath88 .
the locations of the static vortices correspond to the dirac points of @xmath90 , i.e. , the @xmath91 and @xmath92 points . now that we have @xmath93
, the bloch vectors of these points always point to the north pole .
we have shown that , only for the points satisfying @xmath94 , the south pole is reachable at a suitable time .
an accompanied question is that : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ for a particular quasimomentum point @xmath0 , must there be a dynamical vortex once the south pole is reached ? _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ this question can be also reformulated quantitatively as follows : consider a particular quasimomentum point @xmath95 at a particular time @xmath96 that satisfy @xmath97 what is the vorticity @xmath98 for this point ?
the answer is ( see appendix [ app : south - pole ] for details ) @xmath99.\ ] ] thus we see that , as long as the jacobian @xmath100 is nonvanishing , there is always a vortex at @xmath101 at time @xmath96 . and we can also show ( see appendix [ app : south - pole ] ) that the vortex dynamics features a time periodicity @xmath102 , where @xmath103 is the first time that the dynamical vortex emerge at the point @xmath101 .
it is also checked that , at the particular points that the vortices emerge or annihilate , the corresponding jacobians do vanish .
for instance , we consider the system shown in fig .
[ fig2 ] [ i.e. , in fig .
[ fig1](e ) ] . at the point @xmath104
, we have @xmath105 and @xmath106 .
this is also the point that the vortices are created at the time @xmath107 .
similarly , at the point @xmath108 ( located at the edge of the brillouin zone ) , we also have eq .
( [ eq : condition_for_south_pole ] ) satisfied with zero vorticity at the vortex annihilation time @xmath109 . between time
@xmath110 and @xmath111 , we have three pairs of dynamical vortices as shown in fig .
[ fig2](c ) .
it is also interesting to see the vortex trajectories in the @xmath7 space [ fig .
[ fig2](d ) ] which is related to the hopf mapping and linking numbers @xcite : the trajectories of dynamical vortices always wind around the dirac points where the static vortices locate .
the arguments on the vortex creation and annihilation time for the example can be extended to the whole phase diagram of the system .
for instance , we fix @xmath59 and @xmath112 , by varying the sublattice imbalance @xmath56 , we can get the vortex creation time @xmath110 and annihilation time @xmath111 as a function of @xmath113 , as shown in fig .
[ fig3](a ) . the primary information in fig . [ fig3](a )
is that , the dynamical vortices only appear when the evolution hamiltonian is topologically nontrivial , as @xmath114 does not possess nodal lines otherwise .
it is seen that , @xmath110 ( and hence @xmath111 ) diverges as the parameters approach the phase boundary of the static phase diagram [ @xmath115 .
this statement is also true for a general @xmath57 , and we thus get the dynamical phase diagram ( defined as the parameter region that dynamical vortices can appear @xcite ) as shown in fig .
[ fig3](b ) .
this dynamical phase diagram coincides exactly with the phase diagram of the equilibrium hamiltonian [ fig .
[ fig1](c ) ] .
( @xmath111 ) as a function of @xmath56 , for @xmath59 , @xmath60 , and @xmath112 .
they diverge at @xmath116 , which indicates that for @xmath117 case , no dynamical vortex can be created .
( b ) the dynamical phase diagram with the shaded regions hosting dynamical vortices for the @xmath60 case .
, title="fig : " ] +
we now briefly discuss the case of preparing the initial state as the hamiltonian with a finite ( positive ) @xmath22 in the topologically trivial phase . in the bloch sphere representation ,
the initial bloch vectors now point to a finite region lies in the northern hemisphere .
we can use @xmath118 and @xmath119 to parameterize the hamiltonian vector @xmath120 as @xmath121 then it can be shown ( see appendix [ app : finite_mi ] ) that the above discussion of vortex dynamics with @xmath19 in sec .
[ sec:3 ] and appendix [ app : south - pole ] is still valid as long as we replace the vortex trajectory @xmath122 by @xmath123 it recovers the previous results for @xmath19 , as @xmath124 and @xmath125 for this case . .
( a ) enlarged dynamical phase diagram for the @xmath126 case . apart from the colored parameter region shown in fig .
[ fig3](b ) , there exists a new gray region which is bounded by a @xmath57-independent critical final energy offset @xmath127 . for the @xmath128 case , a second type of dynamical vortices emerges .
( b ) the critical final energy offset @xmath127 for different initial offset @xmath22 .
( c , d ) the vortex trajectories in the @xmath7 space with parameters marked by the red crosses in ( a ) .
they give rise to linking numbers @xmath129 and @xmath130 respectively .
, title="fig : " ] + for this finite @xmath22 case , the vortex dynamics is summarized in fig .
the dynamical phase diagram is enlarged with an additional parameter region bounded by a @xmath57-independent critical @xmath127 [ fig .
[ fig4](a ) ] .
two types of vortex trajectories are possible as shown in fig .
[ fig4](c ) and ( d ) , with different linking numbers . the trajectory of dynamical vortices as shown in fig .
[ fig4](d ) shrinks to a point at @xmath131 point at critical @xmath127 , and the relation between @xmath22 and @xmath127 is given by @xmath132 or , written explicitly , @xmath133,\ ] ] as shown in fig .
[ fig4](b ) .
we see in fig .
[ fig4](b ) that , the critical value @xmath127 can be larger than @xmath134 as indicated by the dashed line , and it turns to zero for extremely large initial offset @xmath19 , which returns to the situation in fig .
[ fig3](b ) .
indeed , the parameter regime @xmath135 just indicates the fact that the dynamical vortices around @xmath131 point can emerge for geometric reasons .
such type of dynamical vortices can also coexist with the ones winding around static vortices in the topological phase of the equilibrium hamiltonian , as the former always give null contribution to linking numbers .
inspired by the recent hamburg group experiment on the creation of dynamical vortices in a quenched haldane - like model on a driven honeycomb lattice @xcite , we have studied the dynamics of the phase vortices of the quenched haldane model in detail .
we have shown that , two types of vortices can appear in the azimuthal phase profile : the static vortices correspond to the bloch states pointing to the north pole , while the dynamical vortices correspond to the bloch states pointing to the south pole .
the trajectories of the dynamical vortices also constitute of two types : one winds around dirac point ( @xmath91 or @xmath91 ) ; the other winds around @xmath131 point , which could give rise to the enlarged dynamical phase diagram .
the former gives linking number one , while the later gives linking number zero .
as the linking number equals to the chern number exactly as proved in ref .
@xcite , the trajectories of the dynamical vortices provide an alternative method to determine the topological phase diagram of the haldane model .
the discussion in this paper can also be applied to the topological square - lattice model as realized recently by the ustc group @xcite .
+ the author would like to thank xin chen , ce wang , hui zhai , and pengfei zhang for valuable discussions .
in this appendix , we give the reduction of the vorticity for a particular quasimomentum point @xmath95 at the time @xmath96 when the south pole is reached . with the conditions stated in eq .
( [ eq : condition_for_south_pole ] ) , the corresponding bloch state , according to eq .
( [ eq : psi_t ] ) , is given by ( apart from an irrelevant overall phase ) @xmath136 such a state exactly points to the south pole . by solving the second equation in eq .
( [ eq : condition_for_south_pole ] ) , we get @xmath137 where @xmath138 is a non - negative integer .
we then consider , at the time @xmath139 , the states nearby the quasimomentum point @xmath95 . for simplicity , we take @xmath140 , and thus set @xmath141 . at the point
@xmath142 , where @xmath143 is an infinitesimal real variable , we expand @xmath144 and @xmath145 as follows : @xmath146 where the partial derivatives are evaluated at the point @xmath101 .
we thus have @xmath147 where we have defined @xmath148 away from the @xmath85 points , @xmath149 is a slow - varying function of @xmath143 , thus we have the azimuthal phase given by @xmath150 + { \text{const}}.\ ] ] and the phase accumulation around @xmath101 is given by @xmath151.\ ] ] thus we see that , as long as the jacobian @xmath100 is nonvanishing , there is always a vortex at @xmath101 at time @xmath96 .
and the vorticity of this point is just the winding phase in unit of @xmath152 : @xmath153 , giving rise to eq .
( [ eq : vorticity ] ) .
it is easy to see that the above arguments also hold for the general @xmath154 cases .
thus the vortex dynamics features a time periodicity @xmath102 .
for the case the initial offset @xmath22 is finite , the corresponding initial state is not polarized to the north pole .
we use capital greek letters @xmath118 and @xmath119 to parameterize the hamiltonian vector @xmath120 : @xmath155 .
the initial state is the ground state of the hamiltonian @xmath156 ( the term proportional to identity is irrelevant to the relative phase , and is thus not considered here ) , or written explicitly , @xmath157 after quench , the state evolves under the new hamiltonian @xmath158 , as @xmath13 .
now that the quench just changes the @xmath88 term through changing @xmath34 , we see that @xmath159 and @xmath160 are not changed .
it thus leaves the azimuthal phase @xmath119 invariant through the quench , i.e. , we have @xmath161 we then have @xmath162 it is then clear that , if the conditions @xmath163 are satisfied , the south pole is reached .
comparing eqs .
( [ eq : condition_for_south_pole ] ) and ( [ eq : south - pole_mi ] ) , we see that , as long as we replace the vortex trajectory @xmath122 for the @xmath19 case by the first equation in eq .
( [ eq : south - pole_mi ] ) [ i.e. , eq .
( [ eq : hz_tilde ] ) in the main text ] , the analysis of the vortex dynamics for the finite @xmath22 case takes the same steps as the one with @xmath19 .
+ 22ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1103/revmodphys.82.3045 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.83.1057 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.61.2015 [ * * , ( ) ] link:\doibase 10.1126/science.1207239 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.109.145301 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.108.225304 [ * * , ( ) ] http://dx.doi.org/10.1038/nphys2750 [ * * , ( ) ] link:\doibase 10.1103/physreva.89.061603 [ * * , ( ) ] http://dx.doi.org/10.1038/nature13915 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.115.236403 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.117.126803 [ * * , ( ) ] link:\doibase 10.1103/physreva.94.053604 [ * * , ( ) ] https://arxiv.org/abs/1603.01621 [ ( ) ] https://arxiv.org/abs/1611.03304 [ ( ) ] link:\doibase 10.1103/physrevlett.49.405 [ * * , ( ) ] http://dx.doi.org/10.1038/ncomms9336 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.113.045303 [ * * , ( ) ] link:\doibase 10.1126/science.aad4568 [ * * , ( ) ] https://arxiv.org/abs/1608.05616 [ ( ) ] link:\doibase 10.1103/physrevlett.110.135704 [ * * , ( ) ] link:\doibase 10.1103/physrevb.74.085308 [ * * , ( ) ] link:\doibase 10.1126/science.aaf6689 [ * * , ( ) ] | using the recently developed bloch - state tomography technique , the quasimomentum @xmath0-dependent bloch states @xmath1 of a two - band tight - binding model with two sublattices can be mapped out .
we show that , if we prepare the initial bloch state as the lower - band eigenstate of a topologically trivial haldane hamiltonian @xmath2 , and then quench the haldane hamiltonian to @xmath3 , the time - dependent azimuthal phase @xmath4 supports two types of vortices .
the first type of vortices are static , with the corresponding bloch vectors pointing to the north pole ( @xmath5 ) .
the second type of vortices are dynamical , with the corresponding bloch vectors pointing to the south pole ( @xmath6 ) . in the @xmath7 space ,
the linking number between the trajectories of these two types of vortices equals exactly to the chern number of the lower band of @xmath3 , which provides an alternative method to directly map out the topological phase boundaries of the haldane model . |
the formation of a qgp ( quark gluon plasma ) is expected to be realized in high - energy heavy - ion collisons , _
e.g. _ at rhic and soon at the lhc .
one of the main tools for the description of such a formation is the relevance of relativistic hydrodynamic equations in some intermediate stage of the collisons , see fig.[qgp ] .
[ hb ] the problem of the hydrodynamic description is the uneasiness of the relation with the underlying fundamental theory .
indeed , the experimental observations seem to indicate an almost perfect - fluid behaviour with small shear viscosity , which naturally leads to consider a theory at strong coupling and thus within the yet unknown nonperturbative regime of qcd
. moreover the qgp formation appears to be fast , which may also point towards strong coupling properties .
another key point of the standard description is the approximate boost - invariance of the process as predicted in the seminal paper of ref.@xcite .
the goal of the string theoretic approach is to make use of the gauge / gravity correspondence as a way to tackle the problem of the hydrodynamic behaviour from the fundamental theory point of view .
it allows to draw quantitative relations between a strongly coupled gauge field theory and a weakly coupled string theory more specifically , the ads / cft correspondence between the @xmath1 supersymmetric @xmath2 gauge theory and superstrings in 10 dimensions can be used as a calculational laboratory for this kind of approach , at least as a first stage before a more realistic application to qcd .
the unconfined character of the qgp gives some hope that the explicit ads / cft example could be useful despite the lack of asymptotic freedom and other aspects specific of qcd .
as well - known , superstring theory is defined as the quantum embedding of a 2-dimensional world sheet describing the moving string into a @xmath3-dimensional space . in a qualitative way , the gauge - gravity correspondence can be given a geometric interpretation @xcite .
[ hb ] considering the cylinder stretched between the two @xmath4 in fig.[open ] , it may be described as a propagating closed string ( fig.[open ] left ) or an open string with ends on the branes performing a loop ( fig.[open ] right ) . in terms of fields ,
a gravity interaction between branes ( left ) becomes equivalent to a 1-loop gauge interaction ( right ) .
this gauge / gravity @xmath5 is also a weak / strong coupling one . indeed , considering a long distance between branes , it corresponds to a weak and thus @xmath6 gravitational interaction between branes ( left ) , while in gauge theorical terms , it is expected to imply a long - distance gauge field interaction including many excitation modes of the 1-loop open string , and thus a @xmath7 field theory at strong - coupling ( right ) .
this duality property can be given a quantitative framework , in particular in the case of ads / cft correspondence .
one typical and fascinating aspect of the gauge / gravity duality is the property of @xmath8 it states that the amount of information contained in the boundary gauge theory ( on the brane ) is the same as the one contained in the bulk string theory . in our problem , we shall make use in a quantitative way of this property by taking advantage of one of the remarkable relations due to the holographic renormalization @xcite . one can write @xmath9 where @xmath10 is the bulk metric in 5 dimensions , @xmath11 the boundary metric in physical ( 3 + 1 ) minkowski space and @xmath12 the v.e.v . of the physical energy - momentum tensor
. the higher coefficients of the expansion over the fifth dimension @xmath13 can be obtained by the einstein equations in the bulk provided the boundary energy - momentum tensor fulfils the zero - trace and continuity equations .
the interesting observation on which we shall elaborate , namely that there is a nontrivial dual relation between a perfect fluid at rest in ( 3 + 1 ) dimensions and a static 5d black hole in the bulk @xcite can be proven using holographic renormalization . indeed , implementing the perfect fluid behaviour with diagonal elements @xmath14 one can resum @xcite the holographic expansion and get ( @xmath15 a change of variables @xmath16 ) the following bulk metric @xmath17 where one recognizes the black hole ( in fact a black brane ) with a static horizon at @xmath18 in the 5th dimension .
in fact there exists a one - to - one correspondence between the thermodynamic properties of the black hole ( bh ) and those of the perfect fluid ( pf ) , namely its temperature ( @xmath19 ) and entropy ( @xmath20 ) .
it is in this context of a static black hole configuration that one can go further than the perfect fluid approximation and derive the viscosity @xcite , using the kubo formula .
indeed , the duality properties extend to a relation between the correlators of the energy - momentum tensor in two space - time points at zero frequency @xmath21 and the absorption cross section of a graviton by the static black hole in the bulk .
one writes @xmath22}\right\rangle } } \rightarrow { \frac}{\eta}{s}\equiv { \frac}{\sigma_{abs}(0)/16 \pi g } { a/4 g}={\frac}1{4\pi}\ , \ ] ] where @xmath23 is the famous entropy - area relation of a black hole . from this relation , and putting numbers , it appears that the viscosity is weak , much weaker than the one computed in the weak coupling regime and eventually realizing an absolute viscosity lower bound .
[ ht ] the previous results were obtained for static configurations , @xmath24 for a thermalized qgp at rest . in order to take into account , as much as possible , the actual kinematics of a heavy - ion collision , it is required to introduce the proper - time expansion of the plasma . on the gravity side , it calls for studying non - equilibrium geometries , eventually of 5d black hole configurations , which represent in itself a nontrivial and interesting issue . in ref.@xcite , it was proposed to build the dual geometries of the standard bjorken flow @xcite , that is the description of a boost - invariant expansion of the qgp , which is expected to correspond to the physical situation in the central rapidity region of the collision . in this context the questions why the qgp fluid appears to be nearly perfect ( small viscosity ) and why its thermalization time can be short have been discussed among other problems .
one starts from a family of proper - time dependent , boundary energy - momentum tensors @xmath25 where the function @xmath26 corresponds to an interpolation between different relevant regimes , namely perfect fluid @xmath27 @xmath28 free streaming @xmath29 @xmath30 fully anisotropic @xmath31 @xmath32 using the holographic renormalization to compute the coefficients of the corresponding metrics in the expansion on the fifth dimension and after resummation , it was possible to solve the dual geometry for given @xmath33 at asymptotic proper - time @xmath34 it reveals the existence of a scaling property of the solutions in terms of the proper - time dependent variable @xmath35 analyzing the family of solutions as a function of @xmath36 it appears that the only nonsingular solution for invariant scalar quantities ( here the square of the ricci tensor @xmath37 see fig.[ricci ] ) , is obtained for @xmath38 this solution is the only one of the family corresponding to a black hole moving away in the fifth dimension .
hence the perfect - fluid case is singled out and the moving black hole in the bulk corresponds through duality to the expansion of the qgp taking place in the boundary .
consequently , the bh horizon moves as @xmath39 the temperature as @xmath40 and the entropy stays constant since @xmath41 note that again the physical thermodynamical variables of the qgp are the same as those one may attribute to the bh in the bulk ( with the reservation that thermodynamics of a moving bh may rise nontrivial interpretation problems ) .
hence one finds a concrete realization of the idea @xcite of a duality between the qgp formation and a moving a moving black hole .
there has been a lot of activity along the lines of the ads / cft correspondence and its extensions to various geometric configurations .
dual studies of quark energy loss @xcite , jet quenching @xcite , quark dragging @xcite , etc ... have been and are being performed .
sticking to the configurations corresponding to an expanding plasma and going beyond the first order terms in proper - time , one has obtained results @xcite on the viscosity , confirming the universal bound of ref.@xcite , on the relaxation time of the plasma and very recently on the inclusion of flavor degrees of freedom @xcite .
let us finally focus on the thermalization problem , which can be usefully taken up using the gauge / gravity duality in the strong coupling hypothesis .
the problem is to give an explanation to the strikingly small thermalization time required for the formation of a qgp as can be abstracted from the experimental observations .
already in ref.@xcite , it has been found that if one performs a small deviation from the black hole metric by coupling with a scalar field and analyze the so - called quasi - normal modes defining the way how the system relaxes to its initial state , one finds a numerically small value of the relaxation time in units of the local ( and evolving ) temperature . even if a definite value of this relaxation time can not be inferred at this stage due to scale - invariance ,
this result was suggestive of a stability of the qgp in the strong coupling regime with respect to perturbations out of equilibrium . in order to go further
, one has to deal with the problem of the qgp evolution at small proper - times . in ref@xcite ,
the holographic renormalization program has been pursued for the small proper - time expansion .
relaxing the selection of the appropriate metric by requiring only the metric tensor to be a real and single - valued function of the coordinates everywhere in the bulk , one finds an unique solution corresponding to the `` fully anisotropic case '' , @xmath42 [ h ] in the same paper , an evaluation of the range of the isotropization time has been proposed , by extrapolation of realistic estimates abstracted from experiments to the supersymmetric case .
the idea is to match the large and small proper - time regimes at some value of the proper - time @xmath43 this proper - time is mathematically defined as the crossing value for the branch - point singularities of both regimes .
physically , it is expected to give an estimate of the proper - time range during which which the medium evolves from the full anisotropic regime ( small @xmath44 ) to the perfect fluid one ( large @xmath44 ) .
cd in order to give an idea of the possible physical implications of this strong coupling scheme , let us shortly reproduce the estimate made in ref.@xcite . implementing the estimated physical value of the energy density at some proper - time ( @xmath45 ) one finds @xmath46 this short isotropization time thus seems a characteristic feature of the strong coupling scenarios .
it is clear that more realistic estimates should take into account less idealized dual models , corresponding to qcd , such as the lack of supersymetry and the finite numbers of colors .
however , the non confined character of the qgp and the robustness of some predictions ( such as the @xmath47 ratio ) may give some confidence that this short isotropization time could be a reasonable estimate at strong coupling .
from the present rapid ( and partial ) survey of some of the results obtained in the ads / cft approach to the formation and expansion of the quark - gluon plasma in heavy - ion collisions , it appears that the gauge / gravity correspondence is a promising way to explore some features of qcd at strong coupling .
indeed some general features of this correspondence , relating at long distances the closed and open string geometries ( see fig.[open ] ) are expected to be valid in principle for various dual schemes and thus , hopefully , qcd . in practice , the quantitative dual schemes have been more precisely elaborated for the specific ads / cft case , @xmath24 the gauge theory with @xmath1 supersymmetries . among the results , it gives a calculable link between the hydrodynamic quasi - perfect fluid behaviour on the gauge theory side with a bh geometry in the higher dimensional gravity side in and ads background .
this relation can be extended from the static case to a dynamical regime reflecting ( within the ads / cft framework ) the relativistic expansion of the corresponding quark - gluon plasma .
this , and many other applications , some of them using more complex geometries , less supersymmetric backgrounds and examining other observables , gives hope on the fruitful possibilities of the gauge / gravity approach to the qgp formation .
as an outlook , it is worth mentionning some of the possible new directions of study one is led to consider . starting with the more technical ones , it is known that the bjorken flow is not exactly verified in heavy - ion collisions , since the observed distribution of particles is nearly gaussian in rapidity and thus not reflecting exactly the boost - invariance of the bjorken flow .
it would be interesting to investigate dual properties for non - boost invariant flows , such as the landau solution @xcite . on a more general ground ,
the whole approach still concerns only the hydrodynamical stage of the qgp expansion
. it would be important to attack both the initial ( partonic ) and final ( hadronic ) stages of the reaction in the same framework and thus the problem of _ phase transitions _ during the collision .
finally , one would like to have more realistic dual frameworks including a finite number of colors , flavor degrees of freedom and no ( or broken ) supersymmetry .
i would like to address special thanks to romuald janik for our stimulating and fruiful collaboration .
99 j. d. bjorken , phys . rev .
d * 27 * , 140 ( 1983 ) .
v. schomerus , `` strings for quantumchromodynamics , '' arxiv:0706.1209 [ hep - ph ] .
k. skenderis , class .
* 19 * , 5849 ( 2002 ) [ arxiv : hep - th/0209067 ] .
v. balasubramanian , j. de boer and d. minic , phys .
d * 65 * , 123508 ( 2002 ) [ arxiv : hep - th/0110108 ] ; r. c. myers , phys .
d * 60 * , 046002 ( 1999 ) [ arxiv : hep - th/9903203 ] .
r. a. janik and r. peschanski , phys .
d * 73 * , 045013 ( 2006 ) [ arxiv : hep - th/0512162 ] .
@xmath48 : g. policastro , d. t. son and a. o. starinets , phys .
lett . * 87 * , 081601 ( 2001 ) [ arxiv : hep - th/0104066 ] .
h. nastase , `` the rhic fireball as a dual black hole , '' arxiv : hep - th/0501068 .
e. shuryak , s. j. sin and i. zahed , j. korean phys .
soc . * 50 * , 384 ( 2007 ) [ arxiv : hep - th/0511199 ] .
@xmath48 : c. p. herzog , a. karch , p. kovtun , c. kozcaz and l. g. yaffe , jhep * 0607 * , 013 ( 2006 ) [ arxiv : hep - th/0605158 ] .
@xmath48 : h. liu , k. rajagopal and u. a. wiedemann , phys .
lett . * 97 * , 182301 ( 2006 ) [ arxiv : hep - ph/0605178 ] .
@xmath48 : s. s. gubser , phys .
d * 74 * , 126005 ( 2006 ) [ arxiv : hep - th/0605182 ] .
r. a. janik , phys .
lett . * 98 * , 022302 ( 2007 ) [ arxiv : hep - th/0610144 ] ; s. nakamura and s. j. sin , jhep * 0609 * , 020 ( 2006 ) [ arxiv : hep - th/0607123 ] ; m. p. heller and r. a. janik , phys .
d * 76 * , 025027 ( 2007 ) [ arxiv : hep - th/0703243 ] .
j. groe , r. a. janik and p. surowka , `` flavors in an expanding plasma , '' arxiv:0709.3910 [ hep - th ] .
r. a. janik and r. peschanski , phys .
d * 74 * , 046007 ( 2006 ) [ arxiv : hep - th/0606149 ] .
y. v. kovchegov and a. taliotis , phys .
c * 76 * , 014905 ( 2007 ) [ arxiv:0705.1234 [ hep - ph ] ] .
l. d. landau , `` on the multiparticle production in high - energy collisions , '' izv .
nauk ser .
* 17 * , 51 ( 1953 ) ; for a recent discussion of bjorken and landau flows : a. bialas , r. a. janik and r. peschanski , `` unified description of bjorken and landau 1 + 1 hydrodynamics , '' arxiv:0706.2108 [ nucl - th ] and phys.rev.c , to appear . | the quark - gluon plasma ( qgp ) is the qcd phase of matter expected to be formed at small proper - times in the collision of heavy - ions at high energy .
experimental observations seem to favor a strongly coupled qcd plasma with the hydrodynamic properties of a quasi - perfect fluid , _ i.e. _ @xmath0 rapid thermalization ( or isotropization ) and small viscosity .
the theoretical investigation of such properties is not obvious , due to the the strong coupling .
the gauge / gravity correspondence provides a stimulating framework to explore the strong coupling regime of gauge theories using the dual string description .
after a brief introduction to gauge / gravity duality , and among various existing studies , we focus on challenging problems of qgp hydrodynamics , such as viscosity and thermalization , in terms of gravitational duals of both the static and relativistically evolving plasma .
we show how a black hole geometry arises naturally from the dual properties of a nearly perfect fluid and explore the lessons and prospects one may draw for actual heavy ion collisions from the gauge / gravity duality approach . |
classical spin models have called the attention of statistical - mechanics and magnetism researchers throughout many years @xcite .
many techniques have been used in their study , both analytical and numerical , in such a way that a reasonable knowledge of their equilibrium thermodynamics has been achieved , and specially , of their critical properties . among those models
, one could mention the classical spin models , which present the xy ( @xmath17 ) and heisenberg ( @xmath18 ) as particular cases .
an interesting formulation of a @xmath19-vector classical model comes when one adds a kinetic term to its hamiltonian , i.e. , the spin variables may be interpreted as classical rotators ( see , e.g. , refs .
@xcite ) , so that the terminology `` inertial model '' is currently used .
this additional term does not pose difficulties in the calculation of equilibrium properties within a canonical - ensemble approach , but it turns possible to derive equations of motion for each rotator , which can be integrated by means of a molecular - dynamics procedure . in this way
, the dynamical behaviour of these models can be investigated numerically without the need of introducing any particular type of probabilistic transition rules for changing the microscopic states .
the inertial ferromagnetic xy model was introduced in ref .
@xcite , within a fully - coupled framework ( i.e. , infinite - range interactions ) , a limit where the mean - field approach becomes exact . mostly referred to as `` hamiltonian mean field '' ( hmf ) model , it became paradigmatic in the study of the dynamical behaviour of classical many - body hamiltonian systems , and it has given rise to a large amount of works @xcite .
one of the most interesting features in the hmf model concerns the appearance of metastable states , for some particular initial conditions , whose lifetime grows by increasing the total number of rotators , that could be interpreted as a breakdown of ergodicity in the thermodynamic limit .
a generalization of the hmf model was proposed in ref .
@xcite , by introducing a distance @xmath1 between rotators at sites @xmath2 and @xmath3 of a given lattice .
moreover , the two - body interaction was assumed to decay with @xmath1 , like a power - law , @xmath4 ( @xmath5 ) .
the index @xmath6 controls the range of the interactions , in such a way that one recovers both the hmf and nearest - neighbour - interaction models in the particular limits @xmath7 and @xmath8 , respectively . in between these two limits
, one finds an important change of behaviour in the thermodynamic quantities , yielding two physically distinct regimes , namely , the long- ( @xmath20 ) and short - range ( @xmath21 ) interaction regimes @xcite . in the latter
, one has the usual extensive and intensive quantities , whereas in the former , one may find also nonextensive thermodynamic quantities @xcite .
this generalization is usually referred to as @xmath6-xy or @xmath6-hmf model , and it also has been studied by several groups @xcite .
the dynamics of the fully - coupled inertial version of the ferromagnetic heisenberg model has been much less investigated in the literature @xcite .
metastable , or quasi - stationary states ( qsss ) , also occur for this system , similar to those that appear in the hmf model ; their lifetime diverge by increasing the total number of rotators @xmath0 , which implies that the order in which the thermodynamic limit ( @xmath22 ) and the infinite - time limit ( @xmath23 ) are considered , becomes important .
more specifically , if we let @xmath22 first , the system remains trapped in these qsss , never reaching the final boltzmann - gibbs ( bg ) equilibrium state , most probably being the qss itself the final state in such a case . in these metastable states ,
thermodynamical quantities , like temperature and magnetization , do not coincide with the canonical - ensemble predictions .
herein , we will modify the fully - coupled inertial ferromagnetic heisenberg model , by introducing a distance @xmath1 between rotators at sites @xmath2 and @xmath3 of a given lattice .
analogously to the @xmath6-hmf model @xcite , the rotator - rotator interaction will be taken to decay with the distance like a power - law , @xmath4 ( @xmath5 ) .
the interaction part of this hamiltonian has already been studied analytically in ref .
@xcite , within a canonical - ensemble approach to the equilibrium state of the corresponding model , where it was shown that if the two - body interaction is appropriately scaled , a universal thermodynamical behaviour is achieved for @xmath24 , e.g. , relations involving temperature @xmath25 , magnetization @xmath26 , and internal energy @xmath9 become @xmath6-independent in this interval . apart from this study of the equilibrium state , an investigation of the dynamical behaviour of the above - mentioned heisenberg model , and particularly , how the qsss may be affected by the exponent @xmath6 , has never been addressed in the literature , to our knowledge . in the present work
we study this model on a ring , i.e , @xmath27 with periodic boundary conditions , using molecular - dynamics simulations .
we perform a detailed analysis of the qsss behaviour by varying the energy , number of particles , and range of the interaction , i.e. , the exponent @xmath6 . in the next section
we define the model , the appropriate scaling for the interactions , the canonical - ensemble solution , equations of motion , and initial conditions to be used in the molecular - dynamics procedure . in section [ sec : results ] we present the results of our simulations , showing an agreement with some analytical results known in the literature for the equilibrium state , in some particular limits ; most importantly , we show the existence of qsss for energies @xmath9 below criticality ( @xmath10 ) .
such qsss , which were verified herein for initial conditions corresponding to zero magnetization , appear for values of @xmath6 in the range @xmath12 , being characterized by durations @xmath11 that increase for increasing values of @xmath0 , like those found numerically and analytically in previous works of the similar model of xy rotators @xcite .
finally , in section [ sec : conclusions ] we present our main conclusions .
first , let us define the model in terms of @xmath0 localized interacting classical rotators , on a @xmath28-dimensional hypercubic lattice , through the hamiltonian @xmath29 where @xmath5 and @xmath30 represents a vector with @xmath18 components , assigned to the rotator at site @xmath2 ( similar to a classical heisenberg spin variable ) , allowed to change its direction continuously inside a 3-dimensional sphere of unity radius , leading to the constraint @xmath31 .
moreover , @xmath32 depicts the corresponding angular momentum ( we are considering moments of inertia equal to unit , so that angular momenta and angular velocities are equivalent quantities ) .
the coupling constants may be identified as @xmath33 , where @xmath1 measures the ( dimensionless ) distance between rotators @xmath2 and @xmath3 , defined as the minimal one , given that periodic boundary conditions will be considered .
one should notice that the interaction term defined in eq .
( [ eq : hamiltonian_heisenberg ] ) is long - ranged for @xmath20 , in the sense that if @xmath34 the partition function does not admit a well - defined thermodynamic limit @xcite , e.g. , the internal energy per particle diverges in the limit @xmath35 , so that the system is said to be nonextensive @xcite .
consequently , in order to calculate the thermodynamic limit adequately whenever @xmath20 , the quantity @xmath36 has to be defined appropriately to ensure a total energy proportional to the system size @xmath0 . for @xmath7 this
is attained with @xmath37 , the so - called kac s prescription @xcite . for a general @xmath38
, the extensivity of the hamiltonian in ( [ eq : hamiltonian_heisenberg ] ) is achieved through the following choice for @xmath36 , @xmath39 this may be seen intuitively if we note that for @xmath0 large , @xmath40 , if @xmath41 ( hence , @xmath42 for @xmath7 ) , @xmath43 , if @xmath44 , and @xmath45 , if @xmath46 ; therefore , this proposal yields the interaction term in the hamiltonian ( [ eq : hamiltonian_heisenberg ] ) proportional to the system size @xmath0 , for all @xmath47 .
the need of the scaling in eq .
( [ eq : ntil_definicao ] ) , for systems characterized by long - range interactions , began to be realized within a generalized ferrofluid model @xcite ( see also @xcite ) .
this proposal was applied successfully to a controllable - range interaction ( @xmath6-dependent ) curie - weiss model , where it was shown numerically that by considering such a scaling , the magnetization per particle follows a thermodynamical equation of state , @xmath48 , that becomes independent of @xmath6 in the nonextensive regime @xcite .
moreover , applications of this scaling for lennard - jones - like systems were carried in ref .
in addition to the aforementioned works , the correctness of the conjectural scaling of eq .
( [ eq : ntil_definicao ] ) has been verified in several other systems , suggesting that it should be valid for a wide variety of systems with long - range interactions @xcite . a systematic analysis of eq .
has been particularly undertaken for the @xmath6-xy model @xcite , whose hamiltonian is identical to eq . , but
@xmath17 , i.e. , @xmath49 is treated as a two - dimensional classical rotator , allowed to change its direction continuously inside a circle of unit radius . for this system , it was shown both analytically @xcite and numerically @xcite that , for @xmath24 , the prescription ( [ eq : ntil_definicao ] ) associates to the @xmath6-xy model the same thermodynamical behaviour as the one previously known for the case @xmath7 @xcite .
furthermore , the work of ref .
@xcite has extended such a result to a general @xmath19-dimensional @xmath49 ( @xmath50 ) unit vector .
although the model of eq .
( [ eq : hamiltonian_heisenberg ] ) has been defined on a @xmath28-dimensional hypercubic lattice , from now on we will restrict ourselves to a ring , i.e. , a one - dimensional chain with periodic boundary conditions .
based on its successful use for the above - mentioned systems , we will consider the scaling of eq .
( [ eq : ntil_definicao ] ) for the present problem as well . within a canonical - ensemble solution , the kinetic term @xmath51 of the hamiltonian in eq .
( [ eq : hamiltonian_heisenberg ] ) does not bring difficulties in the evaluation of average values .
hence , the nontrivial part for the calculation of equilibrium properties of the hamiltonian ( [ eq : hamiltonian_heisenberg ] ) comes from the interaction term @xmath52 .
this term may become troublesome to deal analytically for dimensions @xmath53 , and to our knowledge , it has been investigated analytically only for @xmath27 , i.e. , the linear chain , in some particular regimes of @xmath6 , namely , @xmath54 and @xmath8 .
the model on an open linear chain with nearest - neighbour interactions ( limit @xmath55 ) was solved exactly in ref .
@xcite , whose solution is equivalent to the one considering periodic boundary conditions , in the thermodynamic limit @xcite ( see also refs .
@xcite ) . on the other hand ,
the fully - coupled limit ( @xmath7 ) corresponds to the case where the mean - field approach becomes exact , so that the model may be solved through a relatively easy calculation ( see , e.g. , ref .
@xcite ) . as mentioned above ,
using the scaling of eq .
( [ eq : ntil_definicao ] ) the solution for @xmath7 applies to all @xmath24 @xcite .
therefore , considering the model of hamiltonian ( [ eq : hamiltonian_heisenberg ] ) defined on a ring , i.e. , a one - dimensional chain with periodic boundary conditions , the solutions described above may be summarized , yielding for the internal energy and magnetization per particle respectively ( we work with units such that @xmath56 ) , @xmath57 } & ; \qquad \textrm{if}\hspace{0.3 cm } 0 \leq \alpha < 1~ , \\ \\
t \,+\ , \displaystyle{\frac{1}{2}}{\left[1-l{\left(1/2t\right ) } \right ] } & ; \qquad \textrm{if}\hspace{0.3 cm } \alpha \to \infty~ , \end{array}\r.\ ] ] @xmath58 where @xmath59 denotes the modulus of the magnetization vector , @xmath60 in these equations , @xmath61 represents the langevin function , @xmath62 and @xmath63 is the modified bessel function of the first kind .
the analytical results presented above are displayed as solid lines in fig .
[ fig : curvas_t_e_m ] , where one finds a continuous phase transition separating the ferromagnetic and paramagnetic states for values of @xmath6 in the interval @xmath64 ; for @xmath8 , there is no phase transition at a finite temperature , so that only the disordered phase with @xmath65 exists . in the first case , at sufficiently high temperatures ( or equivalently , high enough energies ) ,
the directions of the @xmath66 become randomly distributed , corresponding to the paramagnetic disordered phase , where one has the order parameter @xmath65 . at the fundamental state
all spins are parallel , corresponding to the ferromagnetically fully ordered case with @xmath67 .
hence , for all cases @xmath64 , in between these two regimes , a continuous phase transition occurs at a critical temperature @xmath68 , with an associated critical energy @xmath69 .
herein we work with cartesian components for the spin variables and angular momenta , written respectively as @xmath70 and @xmath71 , yielding @xmath72 variables for each rotator .
hence , the set of @xmath73 equations to be solved numerically can be cast in the following form @xcite , @xmath74}~ , \\
\nonumber \\ \label{eq : equacoes_movimentos } \dot{{\mathbf{s}}}_i & = & { \mathbf{l}}_i \times { \mathbf{s}}_i~,\end{aligned}\ ] ] for @xmath75 .
one notices that , in the particular case @xmath7 , the expression for @xmath76 used in previous works @xcite is recovered , namely , @xmath77 [ where @xmath78 represents the magnetization per particle of eq .
( [ eq : magnetizacao_definicao ] ) ] .
it is important to remind that eqs . and , written in the cartesian representation , are not canonical equations of motion , since @xmath79 and @xmath80 ( @xmath81 ) do not represent canonically conjugate pairs .
alternatively , one could also work along the line of refs .
@xcite , by using spherical coordinates in order to write the spin variable as @xmath82 and the squared angular momentum in the corresponding lagrangian as @xmath83 . in this representation
, one can derive equations of motion through the usual hamiltonian formalism , where each rotator is characterized by two angles , @xmath84 and @xmath85 , and their canonically conjugate momenta , @xmath86 and @xmath87 . although the use of spherical coordinates , instead of the cartesian ones of eqs .
( [ eq : equacoes_movimentol ] ) and ( [ eq : equacoes_movimentos ] ) , seems to be a natural way to handle the problem , the denominator @xmath88 that appears in @xmath89 becomes hard to be tackled numerically .
indeed , as @xmath90 approaches zero , one needs to decrease the time step used in the integrations of the equations of motion , in such a way that the computational time may increase substantially .
this difficulty has been discussed in the literature by several authors @xcite , where numerical techniques for circumventing it were proposed .
based on the above - mentioned reasons , herein we shall deal with the equations of motion expressed in terms of cartesian components , as in eqs .
( [ eq : equacoes_movimentol ] ) and ( [ eq : equacoes_movimentos ] ) .
it is straightforward to verify that , in addition to the total energy , the total angular momentum , @xmath91 , as well as the norm of each spin , @xmath92}^{1/2}$ ] , are also integrals of motion .
indeed , evaluating the time derivative of @xmath93 and taking into account that @xmath94 , we get @xmath95 similarly for @xmath96 , @xmath97 it should be stressed that the equation of motion ( [ eq : equacoes_movimentol ] ) corresponds precisely to the euler equation for a linear rigid body of unit length ( and unit inertial moments ) . in such system
the angular momentum @xmath98 is always perpendicular to @xmath49 ( the molecular " axis ) , yielding the constraint @xmath99 , to be incorporated in the initial state .
once this additional constraint is imposed , it will hold throughout the whole time evolution , since the product @xmath100 is also a constant of motion , as we can verify by using eqs .
( [ eq : equacoes_movimentol ] ) and ( [ eq : equacoes_movimentos ] ) , @xmath101 all our molecular - dynamical simulations were carried for a single copy of the system defined by eq .
( [ eq : hamiltonian_heisenberg ] ) , considering fixed values of the total number of rotators @xmath0 and energy per particle @xmath9 . since we are applying periodic boundary conditions , the model of eq .
( [ eq : hamiltonian_heisenberg ] ) becomes defined on a ring , with @xmath1 corresponding to the minimal dimensionless distance between rotators @xmath2 and @xmath3 , taking the values @xmath102 , for each rotator @xmath2 . to integrate the @xmath73 equations of motion [ eqs
( [ eq : equacoes_movimentol ] ) and ( [ eq : equacoes_movimentos ] ) ] we have used a standard fourth - order runge - kutta scheme , with an integration step chosen in such a way to yield conservation of the energy per particle within a relative fluctuation smaller than @xmath103 ; this was achieved with a time step typically and larger values of @xmath0 , by using a fast - fourier - transform algorithm . to implement this technique
we have used the fftw library http://www.fftw.org . ] of @xmath104 .
the initial conditions used were such that each angular momentum component @xmath80 ( @xmath81 ) was drawn at random from a uniform distribution , and then rescaled to yield zero total angular momentum for the whole system . in what concerns the variables
@xmath79 , the initial conditions corresponded to those of zero magnetization , achieved numerically as @xmath105 ( typically @xmath106 ) . for this , the components @xmath107 and @xmath108 were also drawn at random from a uniform distribution within the interval @xmath109}$ ] , whereas the components @xmath110 were set in such way to preserve the constraint @xmath111 @xmath112 , i.e. , through @xmath113)$ ] in these works , at the beginning of the simulations .
however , it was verified numerically that @xmath114 was also a constant of motion , in such a way that eq .
( [ eq : ddtlisi ] ) remained true @xmath112 . ]
this procedure leads to unnormalized spins , which are then normalized by dividing each component by @xmath116 ; finally , all @xmath80 s were rescaled again to obtain precisely the desired total energy @xmath9 . consistently with eqs .
( [ eq : solucao_u ] ) and ( [ eq : solucao_m ] ) , initial configurations with zero magnetization like the one just described above , only applies to a total energy @xmath117 .
although we are dealing with three cartesian components for the angular momenta ( @xmath118 and @xmath119 ) , due to the constraint @xmath111 only two of these components are independent .
let us then define the instantaneous kinetic temperature as given by @xmath120 . since we are working with a single copy of the system defined by eq .
( [ eq : hamiltonian_heisenberg ] ) , the time average at the equilibrium state should be identified with the thermodynamic temperature , @xmath121 , and in a similar way , one has for the equilibrium magnetization , @xmath122 . as it will be shown later on , such time averages should be computed after a sufficiently long time , where both @xmath123 and @xmath124 fluctuate around the values predicted by boltzmann - gibbs ( bg ) statistical mechanics . in this regime , where ergodicity is expected to hold , time averages and ensemble averages should coincide . in figure
[ fig : curvas_t_e_m ] we present results from analytical calculations ( full lines ) and data from numerical simulations in the long - time limit ( symbols ) , for the controllable - range inertial heisenberg model of eq .
( [ eq : hamiltonian_heisenberg ] ) . in figure
[ fig : curvas_t_e_m](a ) we represent the kinetic temperature versus the internal energy @xmath9 , whereas in fig . [ fig : curvas_t_e_m](b ) we do the same for the magnetization .
as mentioned above , at equilibrium , an average over the quantity @xmath123 is expected to coincide with the temperature @xmath25 ( obtained from the equipartition theorem ) , so that for the analytical results presented , the vertical axis of fig .
[ fig : curvas_t_e_m](a ) corresponds precisely to the equilibrium temperature @xmath25 , whereas for the numerical ones , it represents the quantity computed from a time average .
all data shown in fig . [ fig : curvas_t_e_m ] for both temperature and magnetization coincide with the analytical results , showing that the equilibrium state considered numerically should be the one predicted by bg theory .
the canonical - ensemble solution of the case @xmath7 ( see , e.g. , ref .
@xcite ) is represented by the full line in fig .
[ fig : curvas_t_e_m](a ) ( usually known as the caloric curve ) , presenting a discontinuity in its slope at the critical point @xmath125 , where the magnetization becomes zero . according to ref .
@xcite , the equilibrium analytical solution of the case @xmath7 should apply to any @xmath126 , so that the corresponding full lines in figs . [ fig : curvas_t_e_m](a ) and ( b ) hold for any @xmath6 in this range . such a universal behaviour was verified numerically for the @xmath6-xy model @xcite , and it is now established herein for the heisenberg model , as displayed in fig .
[ fig : curvas_t_e_m ] for typical values of @xmath126 . for completeness , in fig .
[ fig : curvas_t_e_m](a ) we also present the solution corresponding to the quasi - stationary state ( qss ) hat appears for @xmath10 , characterized by zero magnetization , leading to @xmath127 , and presenting a lower kinetic temperature than the corresponding equilibrium solution .
it was essentially within the energy range @xmath128 that qsss have been investigated within the present numerical investigation . in the limit
@xmath55 our model recovers the nearest - neighbour - interaction inertial heisenberg model on a linear chain , for which the interaction part was solved exactly in refs .
@xcite , and shown to present zero magnetization for all @xmath129 ; consequently , its caloric curve exhibits the smooth behaviour indicated in fig .
[ fig : curvas_t_e_m](a ) . in this case ,
our simulations for @xmath130 are in full agreement with the results of the analytical solutions . in figure
[ fig : t_vs_tempo_varios_u ] we exhibit the time evolution of @xmath131 ( conveniently rescaled in each case by the corresponding equilibrium temperature @xmath132 ) for typical values of energies , @xmath133 , and @xmath134 , and @xmath135 .
simulations were carried for a single copy of @xmath136 rotators , considering the above - defined initial conditions , for sufficiently long times ( up to times slightly larger than @xmath137 in some cases ) .
for all energies investigated , in the range @xmath138 , we have verified the existence of qsss and after these , a crossover is observed to a state whose temperature and magnetization coincide with those obtained analytically within bg statistical mechanics . for a given value of
@xmath6 one has that : ( i ) the lower energies have produced qsss with smaller durations @xmath11 , and one observes that @xmath11 increases substantially as one approaches the critical energy ; ( ii ) the gaps separating @xmath131 of these qsss and their corresponding values of equilibrium temperatures @xmath132 are larger for smaller energies , dropping to zero as @xmath139 .
on the other hand , for a fixed total energy , higher values of @xmath6 yield smaller @xmath11 , and it will be shown later on that these durations tend to zero as @xmath16 . therefore , the qsss disappear in both limits @xmath139 and @xmath16 .
the existence of qsss for energies below , but sufficiently close to @xmath140 , is well known to occur in the hmf and @xmath6-xy ( with @xmath141 ) models , and it has been found analytically and numerically by many authors @xcite . moreover ,
the most interesting characteristic of such states concerns an increase in their lifetime with the total number of rotators @xmath0 ; this property will also be verified for the present model . the qsss shown in fig .
[ fig : t_vs_tempo_varios_u ] appear throughout the whole range @xmath12 , as exhibited in fig .
[ fig : t_vs_tempo_varios_alphas ] for several values of @xmath6 .
each case corresponds to molecular - dynamics simulations of a single realization of @xmath136 heisenberg rotators defined by eq .
( [ eq : hamiltonian_heisenberg ] ) , with a total energy @xmath142 .
one sees that the lifetime of these qsss decrease for increasing values of @xmath6 in the range @xmath12 and that all qsss occur at a kinetic temperature corresponding to zero magnetization ( or slightly larger than zero ) , i.e. , @xmath143 . after these qsss ,
the system approaches the bg equilibrium temperature ( @xmath144 ) , associated to its energy through the caloric curve of fig .
[ fig : curvas_t_e_m ] . at @xmath145 one
notices a crossover from the existence of these metastable states to a smooth gradual increase in the kinetic temperature , such that for @xmath146 there is no clear evidence of qsss .
finally , for a sufficiently high value of @xmath6 ( e.g. , @xmath130 ) , the long - time behaviour of the kinetic temperature @xmath131 gradually approaches the bg equilibrium temperature of the corresponding @xmath55 model , i.e. , the nearest - neighbour - interaction model ( @xmath147 ) .
our numerical results however do not exclude a possible finite - size dependence of these short - lived qsss for @xmath148 , as already found in the @xmath6-xy case @xcite .
however , no dependence on @xmath0 is expected for @xmath149 , since the interaction potential becomes effectively short range in this limit .
an important question concerns the range of energies over which the above - mentioned qsss exist . in figure
[ fig : curva_t_u_qss ] we present part of the caloric curve for @xmath12 , corresponding to energies @xmath150 ( full line ) . numerical simulations for a single copy of @xmath136 rotators , in the cases @xmath134 , and @xmath135 , show that the qsss characterized by @xmath65 ( symbols along the dashed line ) exist within the range @xmath151 , in agreement with the results of fig .
[ fig : t_vs_tempo_varios_u ] . however , as one goes further below the critical point , roughly for @xmath152 , qsss are still found numerically , although characterized by finite values of magnetization , in spite of the initial conditions of zero magnetization considered . such states are understood due to the prevalence of the ferromagnetic couplings for sufficiently low values of @xmath9 , and they may be observed more clearly in the case @xmath153 , being sometimes difficult to be identified for higher values of @xmath6 ( e.g. , @xmath154 ) .
it should be noticed that within a small energy range ( typically @xmath155 ) , the qsss yield a negative microcanonical specific heat .
although this represents a well - known feature for the hmf model ( see discussion in ref .
@xcite ) , it is the first time that such a result is verified for a system of heisenberg rotators , to our knowledge .
the numerical data of fig .
[ fig : curva_t_u_qss ] suggests that a negative specific heat should occur for @xmath126 .
the most interesting feature of the qsss presented in figs .
[ fig : t_vs_tempo_varios_u ] and [ fig : t_vs_tempo_varios_alphas ] concerns an increase in @xmath11 for increasing @xmath0 , as shown in fig .
[ fig : t_vs_tempo_3_alphas ] , where we analyse the qsss for an energy @xmath142 in the cases @xmath134 , and @xmath135 , by varying the system sizes .
it should be mentioned that , in order to identify qsss clearly within the present approach , we had to simulate systems with sufficiently high values of @xmath0 ( essentially , @xmath156 or higher ) .
hence , in fig . [
fig : t_vs_tempo_3_alphas ] such a behaviour is presented for typical values of @xmath0 in the range from @xmath156 up to @xmath157 , for the smaller values of @xmath6 ( i.e. , @xmath153 and @xmath158 ) ; however , in the case @xmath159 the observation of qsss becomes more difficult , in such a way that one needs to simulate larger values of @xmath0 ( e.g. , @xmath160 or higher ) for this purpose .
the growth of @xmath11 for increasing @xmath0 represents a well - known feature in models of rotators , like the hmf and @xmath6-xy ( with @xmath141 ) models @xcite , as well as the inertial heisenberg model @xcite , yielding significant physical consequences , some of them directly related to the applicability of bg statistical mechanics .
more specifically , the order in which the thermodynamic limit ( @xmath22 ) and the infinite - time limit ( @xmath23 ) are considered becomes mostly relevant in the present case , in such a way that if the thermodynamic limit is carried first , one should remain in a qss forever .
this may be viewed as a breakdown of ergodicity , with the system being trapped in a qss , never reaching the bg equilibrium state ; consequently , time averages and ensemble averages do not coincide , violating a basic assumption of bg statistical mechanics .
the qss represents the final state in this case ; whether it should be interpreted as a truly equilibrium thermodynamical one represents an important question beyond the scope of the present work .
herein we have defined the duration @xmath11 of a given qss ( e.g. , those exhibited in fig .
[ fig : t_vs_tempo_3_alphas ] ) , as the time at which @xmath131 presents its halfway between the values at the qss and its corresponding @xmath132 .
the fact that we are computing this time interval starting from @xmath161 , including a short transient regime before reaching the qss , does not affect our final results , since this small transient becomes negligible in the final computation of @xmath11 , for sufficiently large values of @xmath0 . according to this ,
the growth of @xmath11 with the total number of rotators @xmath0 , for a given energy ( @xmath142 ) and three typical values of @xmath6 ( @xmath134 , and @xmath135 ) , is exhibited in fig .
[ fig : behavior_qss_n ] . in the three cases investigated , these behaviours are fitted by power laws , @xmath13 ( as shown in the respective inset , where the same data is presented in a log - log representation ) , with the exponent @xmath14 decreasing as @xmath6 increases , e.g. , @xmath162 , @xmath163 , and @xmath164 . in the first case ,
our estimate agrees with the one of ref .
@xcite ( see also @xcite ) , where @xmath165 was computed .
curiously , such an estimate has also been found numerically for qsss of the hmf model , with similar initial conditions ( zero magnetization ) @xcite , being supported through analytical approaches in ref .
one should mention that in the case @xmath159 the qsss are more difficult to be observed , appearing only for sufficiently high values of @xmath0 , such that data points for smaller @xmath0 are not exhibited . in spite of such difficulties ,
our simulations suggest that one should have @xmath15 in the limit @xmath16 .
although relevant , it is not possible to propose a particular scaling behaviour describing the decrease of @xmath14 with respect to @xmath6 , on the basis of the present numerical simulations .
such a scaling deserves further computational efforts , from which one could obtain a larger number of data points for an appropriate plot of @xmath14 versus @xmath6 ( see , e.g. , ref .
@xcite for the xy case ) .
the behaviours of the duration @xmath11 of the present qsss with respect to the total energy @xmath9 , as well as its decrease with the parameter @xmath6 are presented in fig .
[ fig : behavior_qss_ua ] . in figure
[ fig : behavior_qss_ua](a ) we exhibit data of simulations of @xmath136 rotators , for three values of @xmath6 ( @xmath134 , and @xmath135 ) , varying the energy in the range @xmath166 .
these results show that @xmath11 grows exponentially for increasing @xmath9 in this range , as can be seen more clearly through the log - linear representation in the inset .
although @xmath11 may become very large as @xmath167 , from the plots of fig .
[ fig : t_vs_tempo_varios_u ] one sees that the gap between the kinetic temperature of each qss and the equilibrium state , represented by @xmath132 , decreases as @xmath9 approaches @xmath140 .
hence , in this case , the qsss disappear by means of the difference between these two quantities , which goes to zero as @xmath167 , so that for @xmath168 one finds no qsss .
the dependence of @xmath11 on the interaction - range parameter @xmath6 ( @xmath169 ) is presented in fig .
[ fig : behavior_qss_ua](b ) , where we exhibit data of simulations for the energy @xmath142 and three values for the total number of rotators ( @xmath170 , and @xmath171 ) . in the log - linear representation
, one sees that these results do strongly depend on @xmath0 , although they are all suggestive that @xmath172 as @xmath16 ; more particularly , the linear - linear representation of the inset indicates this tendency more clearly in the case @xmath136 .
we have carried molecular - dynamics simulations on a one - dimensional hamiltonian system , composed by @xmath0 classical localized heisenberg rotators on a ring .
we have considered the two - body interaction characterized by a distance @xmath1 between rotators at sites @xmath2 and @xmath3 , decaying with the distance @xmath1 as a power law , @xmath4 ( @xmath5 ) . in this way
, one may control the range of the interactions by changing the parameter @xmath6 , and specially , two well - known cases may be recovered , namely , the fully - coupled ( i.e. , mean - field limit ) and nearest - neighbour - interaction models , in the limits @xmath7 and @xmath8 , respectively . for the first time in the literature , we have numerically established the validity of the correct scaling for spin dimensionality @xmath173 . considering this scaling , one obtains the same thermodynamical behaviour for any @xmath6 in the range @xmath174 , as analytically predicted in ref .
@xcite .
we have investigated the dynamics of the model by following the time evolution of a single copy of rotators , for energies @xmath9 below its critical value ( @xmath10 ) , considering initial conditions corresponding to zero magnetization .
analogously to what happens for the similar model of xy rotators , we have verified the existence of quasi - stationary states ( qsss ) for values of @xmath6 in the range @xmath12 .
but , in contrast to the xy rotators , the allowed number of initial conditions of the present model is substantially larger .
the particular types of initial conditions that will lead the system to robust qsss was discussed , representing a very relevant matter for further analytical investigations . for a given energy @xmath9 , our numerical analysis indicated that the durations @xmath11 grow by increasing @xmath0 , following a power law , @xmath13 , where the exponent @xmath14 gets reduced for increasing values of @xmath6 in the range @xmath12 ; particularly , our results suggest that @xmath15 as @xmath16 .
the relevance of establishing the growth of @xmath11 with @xmath0 accurately , achieved herein by considering sufficiently large values of @xmath0 , may be assessed if one takes into account the fact that in spite of the large amount of effort dedicated to the @xmath6-xy model , this matter is still controversial for this system ( see , e.g. , refs .
the particular scaling behaviour , describing the decrease of @xmath14 with respect to @xmath6 , requires further computational efforts and is left for future investigations .
moreover , for the initial conditions and energy range considered , we have shown that the duration of the qsss grows exponentially with the energy , @xmath175 , for @xmath0 and @xmath6 fixed .
our simulations have indicated that the gaps separating the kinetic temperature of these qsss and their corresponding equilibrium temperatures @xmath132 become larger for smaller energies , dropping to zero as @xmath139 . in this way , the qsss disappear by means of the difference between these two quantities , which goes to zero as @xmath167 , so that for @xmath168 one finds no qsss .
we have found qsss over a wide range of energies , typically for @xmath176 ( @xmath177 ) , although these states presented zero magnetization throughout a smaller interval , @xmath151 .
it should be mentioned that we have chosen to explore more deeply the qsss associated with a given value of energy ( @xmath142 ) , where such states could be observed easily through simulations of a single copy of rotators .
however , the results presented herein for this particular energy should be valid for any similar qss ( characterized by zero magnetization ) in the range @xmath151 . by going further below the critical point , roughly for @xmath152 , qsss were still found numerically , but in contrast to the above - mentioned ones , they are characterized by finite values of magnetization , in spite of the initial conditions of zero magnetization considered . within a small energy range ( typically @xmath155 ) , the numerical data of the kinetic temperature of these qsss versus @xmath9 yielded a negative microcanonical specific heat . although this represents a well - known feature for the hmf model , it is the first time that such a result has been verified for a system of heisenberg rotators , to our knowledge .
the particular case @xmath153 of the present analysis ( carried herein for the cartesian components of angular momenta and spin variables ) , coincides precisely with the one of ref .
@xcite , which have considered angular variables and their corresponding canonically - conjugated angular momenta .
herein , we have extended these results for power - law decaying interactions among spin variables , by investigating situations characterized by @xmath178 , showing that interesting qsss appear for any @xmath6 in the range @xmath12 . due to the wide diversity of possibilities for initial conditions in both spins and angular momenta in the present model , investigations for qsss under initial conditions different from the ones considered
herein would be highly desirable .
99 t. nakamura , _ statistical theory of hindered rotation in molecular crystals _ , http://dx.doi.org/10.1143/jpsj.7.264[j .
* 7 * , 264269 ] http://dx.doi.org/10.1143/jpsj.7.264[(1952 ) ] ; m. e. fisher , _ magnetism in one - dimensional systems the heisenberg model for infinite spin _ , http://dx.doi.org/10.1119/1.1970340[am .
j. phys .
* 32 * , 343346 ( 1964 ) . ]
g. s. joyce , _ classical heisenberg model _ , http://dx.doi.org/10.1103/physrev.155.478[phys .
rev . * 155 * , 478491 ( 1967 ) ] .
h. e. stanley , _ exact solution for a linear chain of isotropically interacting classical spins of arbitrary dimensionality _ , http://dx.doi.org/10.1103/physrev.179.570[phys .
rev . * 179 * , 570577 ( 1969 ) ] .
m. antoni and s. ruffo , _ clustering and relaxation in hamiltonian long - range dynamics _ , http://dx.doi.org/10.1103/physreve.52.2361[phys .
e ] http://dx.doi.org/10.1103/physreve.52.2361[*52 * , 23612374 ( 1995 ) ] . c. anteneodo and c. tsallis , _ breakdown of exponential sensitivity to initial conditions : role of the range of interactions _ , http://dx.doi.org/10.1103/physrevlett.80.5313[phys .
lett . * 80 * , 53135316 ( 1998 ) ] .
f. d. nobre and c. tsallis , _ classical infinite - range - interaction heisenberg ferromagnetic model : metastability and sensitivity to initial conditions _ , http://dx.doi.org/10.1103/physreve.68.036115[phys .
e * 68 * , 036115 ( 2003 ) ] .
y. y. yamaguchi , j. barr , f. bouchet , t. dauxois , and s. ruffo , _ stability criteria of the vlasov equation and quasi - stationary states of the hmf model _ , http://dx.doi.org/10.1016/j.physa.2004.01.041[physica a * 337 * , 3666 ( 2004 ) ] .
a. pluchino , v. latora , and a. rapisarda , _ glassy dynamics in the hmf model _ , http://dx.doi.org/10.1016/j.physa.2004.04.006[physica a * 340 * , 187195 ] http://dx.doi.org/10.1016/j.physa.2004.04.006[(2004 ) ] ; _ effective spin glass hamiltonian for the anomalous dynamics of the hmf model _ , http://dx.doi.org/10.1016/j.physa.2006.03.067[physica a ] http://dx.doi.org/10.1016/j.physa.2006.03.067[*370 * , 573584 ( 2006 ) ] ; _ metastable states , anomalous distributions and correlations in the hmf model _ , http://dx.doi.org/10.1016/j.physd.2004.01.029[physica d * 193 * , 315328 ( 2004 ) ] .
v. latora , a. rapisarda , and s. ruffo , _ lyapunov instability and finite size effects in a system with long - range forces _ , http://dx.doi.org/10.1103/physrevlett.80.692[phys .
lett . * 80 * , 692695 ( 1998 ) ] ; _ chaos and statistical mechanics in the hamiltonian mean field model _ , http://dx.doi.org/10.1016/s0167-2789(98)00217-6[physica d * 131 * , 3854 ( 1999 ) ] .
moyano and c. anteneodo , _ diffusive anomalies in a long - range hamiltonian system _ , http://dx.doi.org/10.1103/physreve.74.021118[phys
. rev . ] http://dx.doi.org/10.1103/physreve.74.021118[e * 74 * , 021118 ( 2006 ) ] .
a. pluchino , a. rapisarda , and c. tsallis , _ nonergodicity and central - limit behavior for long - range hamiltonians _ , http://dx.doi.org/10.1209/0295-5075/80/26002[europhys .
80 * , 26002 ( 2007 ) ] ; _ a closer look at the indications of @xmath179-generalized central limit theorem behavior in quasi - stationary states of the hmf model _ , http://dx.doi.org/10.1016/j.physa.2008.01.112[physica a * 387 * , 31213128 ( 2008 ) ] .
chavanis and a. campa , _ inhomogeneous tsallis distributions in the hmf model _ , http://dx.doi.org/10.1140/epjb/e2010-00243-x[eur .
j. ] http://dx.doi.org/10.1140/epjb/e2010-00243-x[b * 76 * , 581611 ( 2010 ) ] ; a. campa and p .- h .
chavanis , _ caloric curves fitted by polytropic distributions in the hmf model _ , http://dx.doi.org/10.1140/epjb/e2013-30947-0[eur .
j. b * 86 * , 170 ( 2013 ) ] .
w. ettoumi and m .- c .
firpo , action diffusion and lifetimes of quasistationary states in the hamiltonian mean - field model .
http://dx.doi.org/10.1103/physreve.87.030102[phys .
e * 87 * , 030102 ( 2013 ) ] .
a. campa , t. dauxois , and s. ruffo , _ statistical mechanics and dynamics of solvable models with long - range interactions _ , http://dx.doi.org/10.1016/j.physrep.2009.07.001[phys
. rep . * 480 * , 57159 ( 2009 ) ] . t. m. rocha filho , m. a. amato , a. e. santana , a. figueiredo and j. r. steiner , _ dynamics and physical interpretation of quasistationary states in systems with long - range interactions _ , http://dx.doi.org/10.1103/physreve.89.032116[phys . rev .
e ] http://dx.doi.org/10.1103/physreve.89.032116[*89 * 032116 ( 2014 ) ] ; t. m. rocha filho , a. e. santana , m. a. amato and a. figueiredo , _ scaling of the dynamics of homogeneous states of one - dimensional long - range interacting systems _ , http://dx.doi.org/10.1103/physreve.90.032133[phys .
e * 90 * ] http://dx.doi.org/10.1103/physreve.90.032133[032133 ( 2014 ) ] .
p. jund , s. g. kim , and c. tsallis , _ crossover from extensive to nonextensive behavior driven by long - range interactions _ , http://dx.doi.org/10.1103/physrevb.52.50[phys .
b * 52 * , 5053 ( 1995 ) ] .
a. campa , a. giansanti , d. moroni , and c. tsallis , _ classical spin systems with long - range interactions : universal reduction of mixing _ , http://dx.doi.org/10.1016/s0375-9601(01)00440-6[phys .
a * 286 * , 251256 ( 2001 ) ] . c. tsallis ,
_ introduction to nonextensive statistical mechanics approaching a complex world _ http://dx.doi.org/10.1007/978-0-387-85359-8[(springer , ] http://dx.doi.org/10.1007/978-0-387-85359-8[new york , 2009 ) ] .
l. j. l. cirto , v. r. v. assis , and c. tsallis , _ influence of the interaction range on the thermostatistics of a classical many - body system _ , http://dx.doi.org/10.1016/j.physa.2013.09.002[physica a * 393 * , 286296 ( 2014 ) ] .
firpo and s. ruffo , _ chaos suppression in the large size limit for long - range systems _ , http://dx.doi.org/10.1088/0305-4470/34/37/102[j .
a ] http://dx.doi.org/10.1088/0305-4470/34/37/102[*34 * , l511 ( 2001 ) ] ; c. anteneodo and r. o. vallejos , _ scaling laws for the largest lyapunov exponent in long - range systems : a random matrix approach _ , http://dx.doi.org/10.1103/physreve.65.016210[phys . rev .
e * 65 * , 016210 ( 2001 ) ] .
f. d. nobre and c. tsallis , _ metastable states of the classical inertial infinite - range - interaction heisenberg ferromagnet : role of initial conditions _ , http://dx.doi.org/10.1016/j.physa.2004.06.036[physica a * 344 * , 587594 ( 2004 ) ] .
a. campa , a. giansanti , and d. moroni , _ canonical solution of classical magnetic models with long - range couplings _ , http://dx.doi.org/10.1088/0305-4470/36/25/301[j .
a math . gen . * 36 * , 6897 ( 2003 ) ] .
a. giansanti , d. moroni , and a. campa , _ universal behavior in the static and dynamic properties of the @xmath6-xy model _ , http://dx.doi.org/10.1016/s0960-0779(01)00022-4[chaos soliton .
* 13 * , 407416 ( 2002 ) ] .
s. a. cannas and f. a. tamarit , _ long - range interactions and nonextensivity in ferromagnetic spin models _ , http://dx.doi.org/10.1103/physrevb.54.r12661[phys .
b * 54 * , r12661r12664 ( 1996 ) ] ; l. c. sampaio , m. p. de albuquerque , and f. s. de menezes , _ nonextensivity and tsallis statistics in magnetic systems _ , http://dx.doi.org/10.1103/physrevb.55.5611[phys .
b * 55 * , 56115614 ] http://dx.doi.org/10.1103/physrevb.55.5611[(1997 ) ]
. j. r. grigera , _ extensive and non - extensive thermodynamics . a molecular dynamic test
_ , http://dx.doi.org/10.1016/0375-9601(96)00276-9[phys .
a ] http://dx.doi.org/10.1016/0375-9601(96)00276-9[*217 * , 4751 ( 1996 ) ] ; s. curilef and c. tsallis , _ critical temperature and nonextensivity in long - range interacting lennard
jones - like fluids _ , http://dx.doi.org/10.1016/s0375-9601(99)00727-6[phys .
a * 264 * , 270275 ( 1999 ) ] .
j. d. parsons , _ linear chain of classical spins with arbitrary isotropic nearest - neighbor interaction _ , http://dx.doi.org/10.1103/physrevb.16.2311[phys . ] http://dx.doi.org/10.1103/physrevb.16.2311[rev .
b * 16 * , 23112312 ( 1977 ) ] .
j. barojas , d. levesque , and b. quentrec , _ simulation of diatomic homonuclear liquids _ , http://dx.doi.org/10.1103/physreva.7.1092[phys .
a ] http://dx.doi.org/10.1103/physreva.7.1092[*7 * , 10921105 ( 1973 ) ] .
d. j. evans , _ on the representation of orientation space _ , http://dx.doi.org/10.1080/00268977700101751[mol .
phys . * 34 * , 317325 ( 1977 ) ] ; d. j. evans and s. murad , _ singularity free algorithm for molecular dynamics simulation of rigid polyato _ , http://dx.doi.org/10.1080/00268977700101761[mol .
phys . ] http://dx.doi.org/10.1080/00268977700101761[*34 * , 327331 ( 1977 ) ] .
m. p. allen and d. j. tildesley ,
_ computer simulation of liquids _
( oxford university press , oxford , 1987 ) .
a. turchi , d. fanelli and x. leoncini , _ existence of quasi - stationary states at the long range threshold _ , http://dx.doi.org/10.1016/j.cnsns.2011.03.013[commun .
nonlinear sci .
simulat .
* 16 * 47184724 ( 2011 ) ] ; w. ettoumi and m .- c .
firpo , _ stochastic treatment of finite-@xmath0 effects in mean - field systems and its application to the lifetimes of coherent structures _ , http://dx.doi.org/10.1103/physreve.84.030103[phys .
e * 84 * 030103 ( 2011 ) ] .
s. gupta and d. mukamel , _ quasistationarity in a model of classical spins with long - range interactions _ , http://dx.doi.org/10.1088/1742-5468/2011/03/p03015[j .
: theor . exp .
p03015 ( 2011 ) ] .
r. bachelard and m. kastner , _ universal threshold for the dynamical behavior of lattice systems with long - range interactions _ , http://dx.doi.org/10.1103/physrevlett.110.170603[phys .
170603 * 110 * ( 2013 ) ] . | a numerical analysis of a one - dimensional hamiltonian system , composed by @xmath0 classical localized heisenberg rotators on a ring , is presented .
a distance @xmath1 between rotators at sites @xmath2 and @xmath3 is introduced , such that the corresponding two - body interaction decays with @xmath1 as a power - law , @xmath4 ( @xmath5 ) .
the index @xmath6 controls the range of the interactions , in such a way that one recovers both the fully - coupled ( i.e. , mean - field limit ) and nearest - neighbour - interaction models in the particular limits @xmath7 and @xmath8 , respectively .
the dynamics of the model is investigated for energies @xmath9 below its critical value ( @xmath10 ) , with initial conditions corresponding to zero magnetization .
the presence of quasi - stationary states ( qsss ) , whose durations @xmath11 increase for increasing values of @xmath0 , is verified for values of @xmath6 in the range @xmath12 , like the ones found for the similar model of xy rotators .
moreover , for a given energy @xmath9 , our numerical analysis indicates that @xmath13 , where the exponent @xmath14 decreases for increasing @xmath6 in the range @xmath12 , and particularly , our results suggest that @xmath15 as @xmath16 .
the growth of @xmath11 with @xmath0 could be interpreted as a breakdown of ergodicity , which is shown herein to occur for any value of @xmath6 in this interval . |
standard technique for estimating sparse signals is through the formulation of an inverse problem with the @xmath0 norm as convex proxy for sparsity . in particular , consider the problem of estimating a signal @xmath1 from a noisy observation @xmath2 , @xmath3where @xmath4 represents awgn .
we assume the underlying signal to be sparse with respect to an overcomplete tight frame @xmath5 , @xmath6 , which satisfies the tight frame condition , i.e. , @xmath7using an analysis - prior , we formulate the signal denoising problem as @xmath8_i ; a_i \right)\biggr\rbrace,\end{aligned}\ ] ] where @xmath9 are the regularization parameters , and @xmath10 is a non - smooth sparsity inducing penalty function .
the parameters @xmath11 control the non - convexity of @xmath12 in case it is non - convex .
the analysis prior is used in image processing and computer vision applications @xcite .
commonly , the @xmath0 norm is used to induce sparsity , i.e. , @xmath13 @xcite . in that case , problem is strictly convex and the global optimum can be reliably obtained .
the @xmath0 norm is not the tightest envelope of sparsity @xcite .
it under - estimates the non - zero values of the underlying signal @xcite .
non - zero values can be more accurately estimated using suitable non - convex regularizers .
non - convex regularization in an analysis model has been used for mri reconstruction @xcite , eeg signal reconstruction @xcite , and for computer vision problems @xcite .
however , the use of non - convex regularizers comes at a price : the objective function is generally non - convex .
consequently , several issues arise ( spurious local minima , a perturbation of the input data can change the solution unpredictably , convergence is guaranteed to the local minima only , etc . ) . in order to maintain convexity of the objective function while using non - convex regularizers ,
we propose to restrict the parameter @xmath11 of the non - convex regularizer @xmath12 . by controlling the degree of non - convexity of the regularizer we guarantee that the total objective function @xmath14 is convex .
this idea which dates to blake and zisserman @xcite and nikolova @xcite , has been applied to image restoration and reconstruction @xcite , total variation denoising @xcite , and wavelet denoising @xcite . in this letter
we provide a critical value of parameter @xmath15 to ensure @xmath14 in is strictly convex ( even though @xmath12 is non - convex ) .
in contrast to the above works , we consider transform domain regularization and prove that admm @xcite applied to the problem converges to the global optimum
. the convergence of admm is guaranteed , provided the augmented lagrangian parameter @xmath16 , satisfies @xmath17 .
in order to induce sparsity more strongly than the @xmath0 norm , we use non - convex penalty functions @xmath10 parameterized by the parameter @xmath18 . we make the following assumption of such penalty functions .
[ theorem::assumption 1 ] the non - convex penalty function @xmath19 satisfies the following 1 .
@xmath12 is continuous on @xmath20 , twice differentiable on @xmath21 and symmetric , i.e. , @xmath22 2 .
@xmath23 3 .
@xmath24 4 .
@xmath25 5 .
@xmath26 6 .
@xmath27 . and the function @xmath28 , @xmath29 . ] since @xmath27
, the @xmath0 norm is recovered as a special case of the penalty function @xmath12 .
the parameter @xmath15 controls the degree of non - convexity of @xmath12 .
note that the @xmath30 norm does not satisfy assumption [ theorem::assumption 1 ] .
the rational penalty function @xcite , @xmath31the logarithmic , and the arctangent penalty functions @xcite are examples that satisfy assumption [ theorem::assumption 1 ] .
the rational penalty @xmath12 for @xmath32 is shown in fig .
[ fig::phi minus absolute ] .
the proximity operator of @xmath12 @xcite , @xmath33 , is defined as @xmath34 for @xmath35 satisfying assumption [ theorem::assumption 1 ] , with @xmath36 , the proximity operator is a continuous non - linear threshold function with @xmath37 as the threshold value , i.e. , @xmath38 .
the proximity operator of the absolute value function is the soft - thresholding function .
there is a constant gap between the identity function and the soft - threshold function due to which the non - zero values are underestimated @xcite . on the other hand ,
non - convex penalty functions satisfying assumption [ theorem::assumption 1 ] are specifically designed so that the threshold function approaches identity asymptotically .
these non - convex penalty functions do not underestimate large values . in order to benefit from convex optimization principles in solving ,
we seek to ensure @xmath14 in is convex by controlling the parameter @xmath11 .
for later , we note the following lemma .
[ theorem::lemma 1 ] let @xmath10 satisfy assumption [ theorem::assumption 1 ] .
the function @xmath39 defined as @xmath40is twice continuously differentiable and concave with @xmath41 since @xmath12 and the absolute value function are twice continuously differentiable on @xmath42 , we need only show @xmath43 and @xmath44 . from assumption [ theorem::assumption 1 ] , we have @xmath45 , hence @xmath46 . again by assumption
[ theorem::assumption 1 ] we have @xmath47 , hence @xmath48 .
further , @xmath49 and @xmath50 .
thus the function @xmath51 is twice continuously differentiable .
the function @xmath51 is concave since @xmath52 . using assumption [ theorem::assumption 1 ]
it follows that @xmath53 . figure [ fig::phi minus absolute ] displays the function @xmath54 , which is twice continuously differentiable even though the penalty function @xmath12 is not differentiable .
the following theorem states the critical value of parameter @xmath11 to ensure the convexity of @xmath14 in .
[ theorem::theorem 1 ] let @xmath55 be a non - convex penalty function satisfying assumption [ theorem::assumption 1 ] and @xmath56 be a transform satisfying @xmath57 , @xmath58 .
the function @xmath59 defined in is strictly convex if @xmath60 consider the function @xmath61 defined as @xmath62_i ; a_i ) . \end{aligned}\ ] ] since @xmath63 is twice continuously differentiable ( using lemma [ theorem::lemma 1 ] ) , the hessian of @xmath63 is given by @xmath64where @xmath65_i;a_i\right)$ ] .
using , we write the hessian as @xmath66the transform @xmath56 has full column rank , from , hence @xmath67 is positive definite if @xmath68 thus , @xmath67 is positive definite if @xmath69_i;a_i ) > -\dfrac{1}{r\lambda_i}.\end{aligned}\ ] ] using lemma [ theorem::lemma 1 ] , we obtain the critical value of @xmath11 to ensure the convexity of @xmath63 , i.e. , @xmath70it is straightforward that @xmath71_i|.\end{aligned}\ ] ] thus , being a sum of a strictly convex function and a convex function , @xmath14 is strictly convex .
note that if @xmath72 , then the function @xmath73 is not convex , as the hessian of @xmath73 is not positive definite . as a result , @xmath74 is the critical value of @xmath11 to ensure the convexity of the function @xmath14 .
the following corollary provides a convexity condition for the situation where the same regularization parameter is applied to all coefficients .
[ theorem::corollary 1 ] for @xmath75 , the function @xmath14 in is strictly convex if @xmath76 @xmath77 we illustrate the convexity condition using a simple example with @xmath78 .
we set @xmath79 , \quad { \mathnormal{a}}^{t}{\mathnormal{a } } = 4{\mathnormal{i } } , \end{aligned}\]]and @xmath80 .
theorem [ theorem::theorem 1 ] states that the function @xmath63 defined in is convex for @xmath81 and non - convex for @xmath82 .
it can be seen in fig .
[ fig::convexity condition ] that the function @xmath63 is convex for @xmath83 , even though the penalty function is not convex .
however , when @xmath84 , the function @xmath63 ( hence @xmath14 ) is non - convex .
a benefit of ensuring convexity of the objective function is that we can utilize convex optimization approaches to obtain the solution . in particular , for @xmath13 , the widely used methods for solving are proximal methods @xcite and admm @xcite .
the convergence of admm to the optimum solution is guaranteed when the functions appearing in the objective function are convex @xcite .
the following theorem states that admm can be used to solve with guaranteed convergence , provided the augmented lagrangian parameter @xmath16 is appropriately set .
such a condition on @xmath16 was also given in @xcite .
note that @xmath16 does not affect the solution to which admm converges , rather the speed at which it converges .
.iterative algorithm for the solution to .
[ cols= " < " , ] [ table::algorithm ] let @xmath12 satisfy assumption [ theorem::assumption 1 ] and the transform @xmath56 satisfy the parseval frame condition .
let @xmath85 .
the iterative algorithm [ table::algorithm ] converges to the global minimum of the function @xmath14 in if @xmath86 we re - write the problem using variable splitting @xcite as [ eq::variable split ] @xmath87 the minimization is separable in @xmath88 and @xmath89 .
applying admm to yields the following iterative procedure with the augmented lagrangian parameter @xmath16 .
[ eq::iterative procedure ] @xmath90 the sub - problem for @xmath88 can be solved explicitly as @xmath91using .
the sub - problem for @xmath92 can be solved using @xmath93 , provided the function @xmath94 is convex .
consider the function @xmath95 defined as @xmath96 from lemma [ theorem::lemma 1 ] and the proof of theorem [ theorem::theorem 1 ] , @xmath97 is positive definite if @xmath98 since @xmath99 , it follows that @xmath100 is positive definite if @xmath17 .
hence @xmath101 is strictly convex for @xmath17 .
note that @xmath102 .
hence , the function @xmath94 , being the sum of a convex and a strictly convex function , is strictly convex . as
such , the minimization problem in is well - defined and its solution can be efficiently computed using the proximity operator of @xmath12 , i.e. , @xmath103_i ; \lambda_i/\mu_i , a_i \bigr).\end{aligned}\ ] ] since @xmath56 has full column rank , admm converges to a stationary point of the objective function ( despite having a non - convex function in the objective ) @xcite ; see also @xcite .
moreover , the function @xmath14 is strictly convex ( by theorem [ theorem::theorem 1 ] ) and the sub - problems of the admm are strictly convex for @xmath17 . as a result , the iterative procedure converges to the global minimum of @xmath14 . a globally convergent algorithm based on a different splitting
is presented in @xcite . in that approach , the objective function
is split into two functions , both of which are convex regardless of the auxillary parameter value . hence , no parameter constraint is required to ensure convergence .
we consider the problem of denoising a 1d signal that is sparse with respect to the undecimated wavelet transform ( udwt ) @xcite , which satisfies the condition with @xmath104 .
in particular , we use a 4-scale udwt with three vanishing moments .
the noisy signal is generated using wavelab ( http://www-stat.stanford.edu/%7ewavelab/ ) with awgn of @xmath105 .
we set the regularization parameters @xmath106 .
we use the same @xmath107 for all the coefficients in scale @xmath108 .
the value of @xmath109 is chosen to obtain the lowest rmse for convex and non - convex regularization respectively . to maximally
induce sparsity we set @xmath110 . for the 1d signal denoising example
, we use the non - convex arctangent penalty and its corresponding threshold function @xcite . for comparison
we use reweighted @xmath0 minimization @xcite , with @xmath109 chosen in order to obtain the lowest rmse . for the 1d signal denoising example . ]
figure [ fig::1d example ] shows that the denoised signal obtained using non - convex regularization has the lowest rmse and preserves the discontinuities .
further , the peaks are less attenuated using non - convex regularization in comparison with @xmath0 norm regularization . for further comparison
, we generate the noisy signal in fig .
[ fig::1d example ] for @xmath111 , and denoise it with non - convex and convex regularization .
we also denoise the noisy signal by direct non - linear thresholding of the noisy wavelet coefficients and by reweighted @xmath0 minimization .
we use the same @xmath109 values as in fig .
[ fig::1d example ] .
the value of @xmath109 for direct non - linear thresholding is also chosen to obtain the lowest rmse . as seen in fig . [ fig::rmse comparison ] , the non - convex regularization outperforms the three methods by giving the lowest rmse .
the rmse values are obtained by averaging over 15 realizations for each @xmath112 .
we consider the problem of denoising a 2d image corrupted with awgn .
we use the 2d dual - tree complex wavelet transform ( dt - cwt ) @xcite , which is 4-times expansive and satisfies with @xmath113 .
the noisy ` peppers ' image has peak signal - to - noise ratio ( psnr ) value of 14.6 db .
we use the same @xmath37 for all the sub - bands . as in the previous example
, we set the value of @xmath37 for each case ( convex and non - convex ) as a constant multiple of @xmath112 that gives the highest psnr .
norm regularization . ] .
( b ) psnr as a function of @xmath112 . ]
figure [ fig::2d example ] shows that the denoised image ( non - convex case ) contains fewer wavelet artifacts and has a higher psnr .
[ fig::psnr values](a ) shows the psnr values ( convex and non - convex ) for different values of @xmath37 . to further assess the performance of tight - frame non - convex regularization
, we realize several noisy ` peppers ' images with @xmath114 . as in the case of the 1d signal denoising , fig .
[ fig::psnr values ] shows that non - convex regularization offers higher psnr across different noise - levels .
this letter considers the problem of signal denoising using a sparse tight - frame analysis prior .
we propose the use of parameterized non - convex regularizers to maximally induce sparsity while maintaining the convexity of the total problem .
the convexity of the objective function is ensured by restricting the parameter @xmath15 of the non - convex regularizer .
we use admm to obtain the solution to the convex objective function ( consisting of a non - convex regularizer ) , and guarantee its convergence to the global optimum , provided the augmented lagrangian parameter @xmath16 , satisfies @xmath17 .
the proposed method outperforms the @xmath0 norm regularization and reweighted @xmath0 minimization methods for signal denoising .
s. boyd , n. parikh , e. chu , b. peleato , and j. eckstein , `` distributed optimization and statistical learning via the alternating direction method of multipliers , '' _ found .
trends mach .
_ , vol . 3 , no . 1 ,
pp . 1122 , 2010 .
p. l. combettes and j .- c .
pesquet , `` proximal splitting methods in signal processing , '' in h. h bauschke et al . , editors , _ fixed - point algorithms for inverse problems in science and engineering _ , springer - verlag , pp . 185212 , 2011 .
j. eckstein and d. p. bertsekas , `` on the douglas - rachford splitting method and the proximal point algorithm for maximal monotone operators , '' _ math
. program .
_ , vol .
55 , no . 1 - 3 , pp . 293318 , apr . 1992 .
m. hong , z .- q .
lo , and m. razaviyayn , `` convergence analysis of alternating direction method of multipliers for a family of nonconvex problems , '' _ proc .
speech signal process .
( icassp ) _ , pp . 15 , 2015 .
v. jojic , s. saria , and d. koller , `` convex envelopes of complexity controlling penalties : the case against premature envelopment , '' _ proc .
( aistats ) _ , vol .
15 , pp . 399406 , 2011 .
a. lanza , s. morigi , and f. sgallari , `` convex image denoising via non - convex regularization , '' in j .-
aujol , m. nikolova , and n. papadakis , editors , _ scale space and variational methods in computer vision _ , ser .
lecture notes in computer science , springer , vol .
9087 , pp .
666677 , 2015 .
s. magnusson , p. c. weeraddana , m. g. rabbat , and c. fischione , `` on the convergence of alternating direction lagrangian methods for nonconvex structured optimization problems , '' _
preprint _ , pp .
113 , 2014 .
a. majumdar and r. k. ward , `` non - convex row - sparse multiple measurement vector analysis prior formulation for eeg signal reconstruction , '' _ biomed . signal process .
control _ , vol . 13 , pp . 142147 , sep .
2014 .
m. nikolova , m. k. ng , and c. p. tam , `` fast nonconvex nonsmooth minimization methods for image restoration and reconstruction , '' _ ieee trans
. image process .
_ , vol . 19 , no . 12 , pp .
30733088 , dec . 2010 .
p. ochs , a. dosovitskiy , t. brox , and t. pock , `` on iteratively reweighted algorithms for nonsmooth nonconvex optimization in computer vision , '' _ siam j. imaging sci .
_ , vol . 8 , no . 1 , pp .
331372 , 2015 . | this letter considers the problem of signal denoising using a sparse tight - frame analysis prior .
the @xmath0 norm has been extensively used as a regularizer to promote sparsity ; however , it tends to under - estimate non - zero values of the underlying signal .
to more accurately estimate non - zero values , we propose the use of a non - convex regularizer , chosen so as to ensure convexity of the objective function .
the convexity of the objective function is ensured by constraining the parameter of the non - convex penalty .
we use admm to obtain a solution and show how to guarantee that admm converges to the global optimum of the objective function .
we illustrate the proposed method for 1d and 2d signal denoising . |
systems of ultracold bosonic atoms in two parallel atomic waveguides mutually coupled via quantum tunneling ( so - called extended bosonic josephson junctions ) have been a subject of intensive theoretical @xcite and experimental @xcite studies .
the finite spatial extension of these systems provides much richer physics compared to the case of a point - like bosonic josephson junction @xcite .
the novel features arise due to the enhanced role of noise and correlations in low - dimensional ultracold atomic systems . before discussing the effects of tunneling ,
we recall the basic properties of a bosonic system in an isolated waveguide @xcite .
this system is effectively one - dimensional ( 1d ) , if the interaction energy per atom ( we assume interatomic repulsion characterized by the effective 1d coupling strength @xmath0 ) and the temperature are well below the spacing between the discrete energy levels of the potential of tight radial confinement . in this case
quantum degeneracy does not lead to establishment of the long - range order ; instead , atoms form a quasicondensate , i.e. a system describable by a macroscopic wave function with strong phase fluctuations .
the characteristic length of the phase coherence in a quasicondensate at finite temperature @xmath1 is @xmath2 , where @xmath3 is the atomic mass , and @xmath4 is the mean linear density of atoms @xcite ( we assume an infinite system ; thermodynamic limit implies constant @xmath5 while both the atom number @xmath6 and the quantization length @xmath7 tend to infinity ) .
the power - law decrease of the single - particle correlation function takes place only at @xmath8 .
if two waveguides are tunnel - coupled , the system is described by the generalized hamiltonian @xmath9 } , \label{hmlt } \end{aligned}\ ] ] where @xmath10 is the atomic annihilation operator for the @xmath11th waveguide ( @xmath12 ) , @xmath13 is the chemical potential and @xmath14 is the tunnel splitting ( in frequency units ) , i.e. , the frequency interval between the two lowest eigenstates of the radial trapping hamiltonian ( the antisymmetric and symmetric superpositions of the single - atom states localized in either @xmath15 or @xmath16 wells of the double - well hamiltonian ) . in this case
the situation changes qualitatively : the tunnel coupling mutually locks phase fluctuations in the two quasicondensates @xcite .
phase locking ( as we shall quantify later , in sec .
[ sec.ii ] ) means that the distribution of the relative phase between the two quasicondensates becomes peaked around zero , while the local phase of an individual ( @xmath15 or 2 ) quasicondensate remains fully random ( the phase - density representation for quasicondensates will be discussed in sec .
[ sec.ii ] ) . in the spatial correlation of the local relative phase between two quasicondensates a new length parameter appears @xcite ,
@xmath17 the length @xmath18 sets the scale of restoration of the inter - waveguide coherence due to finite tunnel - coupling strength @xmath19 .
the tunnel - coupling strength is usually estimated from the single - particle energy ( kinetic and potential ) and the overlap in the potential barrier region of the wave functions for a particle localized in the 1st and 2nd waveguide .
however , it is also possible to take into account atomic interactions , see ref . @xcite and references therein .
experimentally , the interwell coherence can be observed by releasing the two quasicondensates from the trap and measuring locally the contrast and the phase of their interference pattern after time of flight @xcite . up to now
, only the theory based on linearization of the hamiltonian ( [ hmlt ] ) has been developed @xcite and applied to the analysis of the experimental data @xcite .
our work is aimed to develop a model of the steady - state thermal noise in tunnel - coupled quasicondensates beyond the harmonic approximation as well as to quantify the influence of the thermal noise to the macroscopic coherent dynamics of the system ( josephson oscillations ) .
our paper is organized as follows . in sec .
[ sec.ii ] we summarize the harmonic approach of ref .
@xcite . section [ stat ] is divided in two subsections . in subsection [ subiii.1 ]
we calculate the static correlation properties of our system beyond the harmonic approximation using the transfer operator technique in the classical limit .
the condition for negligibility of the quantum noise is also derived .
our way to model equilibrium state by numerical simulation of the system s relaxation to the equilibrium after a quench is explained in subsection [ subiii.2 ] .
section [ jxq ] deals with the noise - affected josephson oscillations .
we derive analitically the frequency of josephson oscillations modified by the thermal noise in our extended system .
we support our analysis by numerical simulations and also observe josephson oscillations washing out as a result of thermal noise for low enough tunnel coupling .
section [ concl ] contains our final remarks and conclusions .
explanations of the ways to derive the main equations of sec .
[ stat ] and sec .
[ jxq ] are presented in appendices [ a.a ] and [ a.b ] , respectively .
following the standard procedure @xcite , we represent our atomic field operators through the phase @xmath20 and density @xmath21 operators , obeying the commutation relation @xmath22=-i\delta ( z - z^\prime ) \delta _ { j\ , j^\prime } $ ] , as @xmath23 \sqrt { \hat \rho _ j(z ) } , \qquad j=1,2 .
\label{ph - d - repr}\ ] ] a discussion of the way to introduce the phase operator for quasicondensates by coarse graining a lattice model on length scales containing sufficiently many atoms can be found in ref .
the density operator can be represented as @xmath24 . since for quantum gases with repulsive atomic interactions density fluctuations
are suppressed , we can always consider the corresponding operator @xmath25 as a small correction .
however , the same is not always true for the phase fluctuations .
whitlock and bouchoule @xcite from the very beginning assumed the phase fluctuations to be small and thus linearized the hamiltonian eq .
( [ hmlt ] ) reducing it to @xmath26 , @xmath27 } .
\label{harmquant } \end{aligned}\ ] ] here the symmetric ( s ) and antisymmetric ( a ) variables are introduced via canonical transformation @xmath28/2 , \nonumber \\ \delta \ !
\hat \rho _ \mathrm{a}(z)&= & [ \delta \ ! \hat \rho _ 1 ( z)- \delta \ ! \hat \rho _ 2(z)]/2 , ~~ \hat \theta _ \mathrm{a } ( z)=\hat \theta _ 1 ( z)-\hat \theta _ 2 ( z ) .
\nonumber \end{aligned}\ ] ] diagonalization of the hamiltonian ( [ harmquant ] ) is based on the fourier transform @xmath29 , @xmath30 .
the frequencies @xmath31 of the symmetric and antisymmetric modes with the momentum @xmath32 are given by the dispersion relations @xmath33 correlations in two tunnel - coupled quasicondensates are experimentally accessible via the two - point correlation function @xmath34 . since the system described by the hamiltonian ( [ hmlt ] ) is translationally invariant , @xmath35 depends only on the difference of the two co - ordinates .
the symbol @xmath36 denotes the average of the normal ordered ( with respect to the atomic operators @xmath37 , @xmath38 ) form of the operator @xmath39 . in what follows , we omit the normal ordering notation , thus neglecting the atomic shot noise .
since the density fluctuations for @xmath40 , @xmath41 being the healing length , are suppressed by the atomic repulsion @xcite , the main contribution to this correlation function is given by the phase fluctuations , @xmath42\rangle $ ] .
the experimentally accessible length scale can not be shorter than the optical resolution length @xmath43 .
on this scale the shot noise yields the quantum uncertainty of the relative phase , coarse grained over the distance @xmath43 , of the order of @xmath44 . for @xmath45 and @xmath46 the shot - noise induced phase uncertainty does not exceed 0.075 rad .
this relatively small value can be always kept in mind when comparing theoretical predictions to measurement results . however , for the sake of simplicity , in what follows we assume @xmath47:\rangle \approx \langle \exp [ i\hat \theta _
\mathrm{a}(z^\prime ) - i\hat \theta _
\mathrm{a}(z)]\rangle $ ] and so on .
another point related to the use of the fully classical approximation is the substitution of the bose - einstein statistics of the elementary excitations by its classical limit , @xmath48 - 1}\approx \frac { k_\mathrm{b}t}{\hbar \omega _ \mathrm{a}(k)}. \label{eq.new5}\ ] ] one obtains strong deviations from eq .
( [ eq.new5 ] ) for @xmath49 , which corresponds , under typical experimental conditions , to the range of wave lengths shorter than @xmath43 , i.e. , not resolvable optically .
these considerations justify our method based on genuinely classical statistics . in the harmonic approximations fluctuations
are gaussian , hence , @xmath50\rangle = \exp \ { -\frac 12 \langle [ \hat \theta _
\mathrm{a}(z^\prime ) - \hat \theta _ \mathrm{a}(z)]^2\rangle \ } $ ] . expressing @xmath51 through creation and annihilation operators of the elementary excitations and calculating thermal populations of the elementary modes using eq .
( [ eq.new5 ] ) , whitlock and bouchoule obtained @xcite @xmath52\rangle = \exp \left [ -\frac { 2l_j}{\lambda _ t}(1-e^{-|z - z^\prime |/l_j})\right ] .
\label{cfb}\ ] ] from this expression we can see that tunnel coupling locks the relative phase between two quasicondensates .
this locking means that the relative - phase correlation function ( [ cfb ] ) does not decrease to zero , but even at @xmath53 has a finite value , corresponding to @xmath54 . on the contrary ,
the phase correlations in each of the waveguides are @xmath55\rangle = \langle \exp \ { i[\hat \theta _
\mathrm{s}(z^\prime ) \pm \frac 12\hat \theta _
\mathrm{a}(z^\prime ) -
\hat \theta _ \mathrm{s}(z ) \mp
\frac 12\hat \theta _ \mathrm{a}(z)]\ } \rangle $ ] , the upper and lower signs corresponding to @xmath15 and @xmath16 , respectively
. we can evaluate them using the statistical independence of noise in the symmetric and antisymmetric modes .
the result @xmath56\rangle = \exp \bigg { \ { } -\frac 12 \langle [ \hat \theta _
\mathrm{s}(z^\prime ) -\hat \theta _
\mathrm{s}(z ) ] ^2\rangle - \nonumber \\ & & \qquad \qquad \qquad \qquad \qquad \qquad \quad \frac 18 \langle [ \hat \theta _
\mathrm{a}(z^\prime ) -\hat \theta _
\mathrm{a}(z)]^2\rangle \bigg { \ } } \nonumber \\ & & \qquad = \exp \bigg { [ } - \frac { |z - z^\prime |}{2\lambda _
t}- \frac { l_j}{2\lambda _ t}(1-e^{-|z - z^\prime |/l_j})\bigg { ] } \label{eq.new6 } \end{aligned}\ ] ] decreases @xmath57 $ ] at @xmath53 because of the unlimited growth of the fluctuations of the symmetric component of the phase along the @xmath58-direction .
the correlation properties of the symmetric mode can be experimentally measured using the density - density correlations of the ultracold gas in a time - of - flight experiment @xcite , however , this subject is beyond the scope of our present paper .
the phase locking of the relative phase becomes most apparent if we treat the evolution of the relative phase along @xmath58 in the harmonic approximation as the ornstein - uhlenbeck stochastic process @xcite : while thermal excitations result in the relative phase diffusion , with the diffusion coefficient proportional to @xmath59 , the tunnel coupling gives rise to the friction " force that tends to restore a small ( ultimately zero ) local phase difference between the two quasicondensates .
note that anharmonic hamiltonian terms , which depend on the density fluctuations neglected in our present theory , do not affect much the static properties of the quasicondensate @xcite .
one needs to take them into account in the analysis @xcite of a slow process of the system s relaxation towards equilibrium starting from a non - equilibrium , pre - thermalized initial state @xcite , characterized by two different temperatures @xmath70 and @xmath71 for the symmetric and antisymmetric modes , respectively .
the applicability range of our fully classical approach can be determined as follows .
consider , for the sake of simplicity , distances shorter than @xmath18 .
the effects of tunnel coupling can be neglected at such short length scales , and the fully classical correlation function can be estimated @xcite as @xmath72\rangle \approx \exp ( -2|z - z^\prime |/\lambda _ t)$ ] .
we have to compare this result to the power - law decay of correlations due to _ quantum _ effects , which is obtained in the limit @xmath73 @xcite . neglecting , as previously , the contribution of the density fluctuations , we can write @xmath74\rangle \approx \lim _ { t\rightarrow 0 } \langle \hat{\psi } ^\dag _ 1 ( z^\prime ) \hat{\psi } _ 1 ( z ) \rangle \langle \hat{\psi } ^\dag _ 2 ( z ) \hat{\psi } _ 2 ( z^\prime ) \rangle $ ] and , finally , @xmath75\rangle \approx \left ( \frac { \lambda _ \mathrm{uv}}{|z - z^\prime | } \right ) ^{1/{\cal k } } , \label{eq.new1}\ ] ] where the quantum mechanical average over the ground state is taken , @xmath76 is the luttinger liquid parameter ( for quasicondensates , which are weakly interacting systems , @xmath77 ) , and @xmath78 is the ultraviolet cutoff of the theory .
( [ eq.new1 ] ) is valid if @xmath79 the estimation by popov @xcite yields @xmath80 .
we can fully neglect quantum fluctuations if their contribution to the decay of correlations is small , compared to the contribution of the thermal noise , on a given length scale .
the correlation decay is dominated by the thermal noise if the classical formula @xmath81 yields stronger decay of correlations than the quantum limit ( [ eq.new2 ] ) , i.e. , if @xmath82 the experimentally relevant range of @xmath83 is bound from below by @xmath43 , as we discussed in sec .
[ sec.ii ] , and @xmath84 in a typical experiment @xcite .
therefore the use of the fully classical approach is reasonable for @xmath85 we can evaluate the partition function ( [ eqz2 ] ) using the transfer operator technique @xcite .
first of all , we evaluate the phase - correlation function as ( see appendix [ a.a ] for the sketch of derivation ) @xmath86\rangle = \qquad \qquad \qquad \qquad & & \nonumber \\ \sum _ { n=0}^\infty \left| \langle n |e^{i\theta } |0\rangle \right| ^2 \exp [ -(\epsilon _ n-\epsilon _ 0 ) \label{cfg } \end{aligned}\ ] ] where @xmath87 @xmath88 is the eigenfunction ( normalized to 1 ) of the auxiliary schrdinger - type equation @xmath89 \psi _ n(\theta ) = \epsilon _ n\psi _ n(\theta ) , \label{aux2}\ ] ] and @xmath90 , @xmath91 , is the respective eigenvalue . for simplicity , we set periodic ( and not quasiperiodic ) boundary conditions to eq .
( [ aux2 ] ) with the period @xmath92 , thus neglecting the band structure of its spectrum , since the zero - quasimomentum solutions define all the system properties @xcite , which are relevant to our present work . in the limit of strong tunnel coupling , @xmath93 , the operator in the left - hand - side of eq .
( [ aux2 ] ) can be approximated by the harmonic oscillator hamiltonian ( in proper units ) , and @xmath94 , @xmath95 . in this limit
( [ cfg ] ) reproduces the result ( [ cfb ] ) that holds for small phase fluctuations . in the opposite limit , eq .
( [ aux2 ] ) can be solved perturbatively , and we obtain @xmath96\rangle \approx \left ( \frac { \lambda _ t^2}{8l_j^2}\right ) ^2 + \qquad \qquad \qquad & & \nonumber \\
\left [ 1- \left ( \frac { \lambda _ t^2}{8l_j^2}\right ) ^2 \right ] \exp \left ( -\frac { 2|z - z^\prime |}{\lambda _ t } \right ) , \quad { l_j}\gg { \lambda _ t}. \label{hight } \end{aligned}\ ] ] in what follows , we will be interested in calculating the value of @xmath97 which can be viewed as the mean interwell coherence .
this expression can be derived in different ways , e.g. , from eq .
( [ cfg ] ) by employing the statistical independence of phase fluctuations at two very distant points , @xmath98 , and recalling that @xmath99 . in a general case , eq .
( [ cosgen ] ) can be evaluated from the lowest - energy solution of the mathieu equation @xcite . in the two limiting cases
we obtain the asymptotics @xmath100 a possible physical explanation of the fact that the mean interwell coherence decreases at @xmath101 much slower that the harmonic approximation @xcite predicts , is the large probability of thermal excitation of a soliton in this limit .
each emerging soliton decreases the number of phononic states by 1 @xcite , and the phononic density of states is reduced mostly in the long - wavelength range ( for phonon momenta less than or of the order of @xmath102 ) , which gives the main contribution to the long - distance behavior of the correlation function ( [ cfg ] ) and , hence , to @xmath103 .
the results of sec .
[ subiii.1 ] are obtained at the equilibrium .
however , it is interesting to investigate also the process of equilibration in the system of two 1d quasicondensates after a quench .
the study of this dynamical problem is motivated by our recent numerical results @xcite related to thermalization in a single 1d quasicondensate . in ref .
@xcite we found that , despite the numerically confirmed integrability of the system , phononic ( low - momentum ) modes rapidly relaxed from their initial non - equilibrium state towards a final equilibrium state ; particle - like ( large - momentum ) excitations , on the contrary , exhibited almost no relaxation .
the equilibrium ensemble of phonons was different from the classical limit of equipartition of the thermal energy between all the degrees of freedom and was quite close to the bose - einstein distribution with the temperature @xmath104 determined by the total excitation energy of the initial non - equilibrium state .
observed fluctuations around this equilibrium state were due to the finite size of the system inherent to numerical modeling .
remarkably , the correlations observed at the length scales , which are large compared to the healing length to the healing length , as well as to the wavelength of an elementary excitation with the energy equal to @xmath105 , were well described by classical expressions .
note that the main contribution to the noise on these length scales stems from the low - energy excitations , which approximately exhibit classical equipartition of energy .
the need to extend the numerical approach of ref .
@xcite to tunnel - coupled 1d quasicondensates can also be seen from the following considerations .
our aim is to numerically check the theoretically predicted correlations of two tunnel - coupled quasicondensates at equilibrium .
this equilibrium state can be viewed as a result of the system s relaxation from its initial non - equilibrium state . moreover , the available analytic theory predicts only averages ; unlike the case of harmonic approximation , there is no way yet to generate individual realizations of the phase , obeying the necessary statistics , without simulating numerically the equilibration process .
the most obvious way to obtain numerically the equilibrium solution is to observe the numerical relaxation after a quench and wait until a steady - state regime establishes .
particular type of the quench and the corresponding initial conditions are , up to a certain degree , arbitrary , as long as the system exhibits true relaxational dynamics .
motivated by these considerations , we performed numerical modeling of the thermal equilibrium values of @xmath106 after the dynamical process of relaxation in our system after a quench .
we simulated the time evolution of two coupled gross - pitaevskii equations using the split - step method @xcite previously used by us @xcite to simulate the dynamics of a single quasicondensate and now extended to the case of tunnel - coupled systems .
as the initial conditions we took two independent quasicondensates with phonon modes populated randomly according to the bose - einstein thermal distribution . at @xmath107 we quenched the system by switching on the tunnel coupling between them .
we solved this coupled system for a time long enough to provide equilibration .
( color online ) mean interwell contrast as a function of the ratio of the length scales @xmath108 and @xmath18 .
solid line : exact theory given by eq .
( [ cosgen ] ) . dashed line : small - fluctuations approximation @xmath109 following from the linearized theory @xcite .
dots : results of the numerical simulations of the equilibration dynamics of two coupled condensates . units on the axes are dimensionless .
inset : magnified part of the main plot for small @xmath110 , illustrating the high - temperature asymptotics of eq .
( [ cosgen ] ) in comparison to the linearized theory result .
[ f:1 ] to juxtapose the input parameters of our numerical simulations to typical parameters of modern atom - chip experiments @xcite , we give the system parameters used in our simulations first in dimensional units , but later show them also in dimensionless form . the linear density for a single quasicondensate @xmath111 and the interaction constant @xmath112 with the radial trapping frequency @xmath113 khz and the @xmath114-wave scattering length @xmath115 nm for @xmath116rb yields the healing length @xmath117 m and the luttinger liquid parameter @xmath118 .
the periodic boundary conditions were set at an interval of the length @xmath119 .
the maximum integration time was @xmath120 s. after few hundreds milliseconds some kind of equilibrium was obtained .
the total energy of the system was conserved in our numerical simulations with a good ( @xmath121 ) accuracy , however , it was constantly redistributed in an oscillatory manner between different low - frequency elementary modes , including josephson oscillations . the nonlinear interaction between different modes ( see section [ jxq ] )
lead to excitation of josephson oscillations of the total number imbalance @xmath122 , where @xmath123 is the integral of the density in the @xmath11th quasicondensate over the whole length @xmath7 , i.e. , the number of atoms in this quasicondensate , @xmath124 . in general , the numerical stability of our split - step method was controlled using the criteria of ref .
the thermal coherence length was determined from the phase - correlation functions for each of the two quasicondensates taken _ separately _ by comparison of the numerically obtained value of @xmath125\rangle $ ] , @xmath12 , with its theoretical value @xmath126 for @xmath127 @xcite ( if we trace out the phase and density variables of one of the two tunnel - coupled quasicondensates , the properties of its remaining counterpart will be described by the same temperature as of the whole system at equilibrium ) .
the averaging is performed over statistically uncorrelated ( separated by sufficiently large distances ) intervals of the whole length @xmath7 for @xmath128 .
we never obtain complete equilibration . in each realization , the correlation length @xmath108 obtained in such a way oscillates around certain mean value , and so does the value of @xmath129 ( averaged over the length @xmath7 ) .
typically , @xmath130 m , which corresponds to @xmath131 nk .
we present the results of our numerical simulations in fig .
dots represent mean values of @xmath129 obtained by averaging over both the time ( on the quasi - equilibration stage of the system evolution ) and the ensemble of realizations .
the error bars in fig .
[ f:1 ] show the standard deviations of @xmath129 and @xmath108 .
these error bars indicate slow , quasiperiodic variations of @xmath129 and @xmath108 detected in our simulations . the range of @xmath110 shown in fig .
[ f:1 ] corresponds to @xmath19 increasing from @xmath132 hz up to @xmath133 hz . to summarize the results of the present section
, we can state that we developed a theory describing the static correlation properties more precisely than the harmonic model @xcite .
our approach is based on consideration of the classical partition function for the antisymmetric mode of our problem ( describable by the sine - gordon model ) and application of the well - known transfer operator technique @xcite . as one can see from fig .
[ f:1 ] , the difference between our results and those of ref .
@xcite is most apparent for intermediate and small values of @xmath110 ( intermediate and weak tunnel coupling ) .
the thermal noise effects considered in sec . [ stat ] reduce the frequency of josephson oscillations . consider the absolute number imbalance between two wells , @xmath134 , and its canonically conjugate variable , the overall phase difference @xmath135 between two quasicondensates . in the limit of the atomic repulsion energy dominating over the tunneling , @xmath136 , and for small - amplitude oscillations , @xmath137 , the evolution of these global " variables
is described by the set of equations ( see appendix [ a.b ] ) @xmath138 @xmath139 which is reduced , after elimination of the number - difference variable , to @xmath140 where @xmath141 is the frequency of the josephson oscillations for bosonic junction unaffected by thermal noise . at zero temperature , when the thermal noise is absent , and for @xmath142 , when the quantum noise can be neglected , spatial extension of the ultracold - atomic josephson junction plays no role and we can derive eq .
( [ j0 ] ) from the results of ref .
@xcite . in the case of small - amplitude josephson oscillations ,
the statistical properties of @xmath143 and @xmath144 do not differ significantly , in particular , @xmath145 , i.e. , the quadratic in @xmath135 correction is negligible , and eq .
( [ js3 ] ) reduces to @xmath146 \phi = \zeta ( t ) , \label{js4}\ ] ] where @xmath147 ( color online ) josephson oscillations for @xmath148 hz ( for other system parameters see section [ stat ] of the main text ) .
( a ) the relative imbalance as a function of time .
( b ) the power spectrum of the atom - number imbalance ( averaged over 7 realizations ) , peaked at theoretically predicted @xmath149 hz and broadened by thermal fluctuations .
[ f:2 ] ( color online ) the same as in fig .
[ f:2 ] , but for @xmath150 hz ( irregular behavior ) .
the spectral peak at theoretically predicted @xmath151 hz is smeared out .
@xmath152 at @xmath153 due to finite integration time . [ f:3 ] in eq .
( [ js4 ] ) we explicitly indicate the time argument of the random driving force @xmath154 and the term @xmath155 that describes fluctuations of the oscillation frequency due to the noise of @xmath67 caused by excitations with non - zero momenta .
( color online ) dots : the square of the frequency of numerically obtained josephson oscillations ( normalized to its value @xmath156 for zero thermal noise ) as a function of the mean interwell coherence @xmath129 . the straight line : theoretical prediction given by eq .
( [ oj2 ] ) .
units on the axes are dimensionless .
[ f:4 ] if we prepare some appreciable initial imbalance at @xmath107 , we obtain , to the first approximation , free josephson oscillations governed by the equation @xmath157 , i.e. , with the frequency reduced by @xmath158 compared to the noise - free case of eq .
( [ j0 ] ) .
the presence of the noise broadens the power spectrum of josephson oscillations @xmath159 where @xmath160 is the relative number imbalance .
the integration in eq .
( [ so ] ) is taken over the time interval @xmath161 when the system has already reached its nearly - equilibrium state ( typically , @xmath162
if @xmath163 is high enough , the theory @xcite predicts @xmath164 to be a peaked function , centered at @xmath163 and having the half - width at the half - maximum of the peak height @xmath165 \ ,
\mathrm{re } \int
_ 0^\infty dt^\prime \langle \zeta ( t)\zeta ( t+t^\prime )
\rangle \exp ( i\omega _ \mathrm{j}t^\prime ) $ ] .
the latter expression , roughly evaluated as @xmath166 , correctly describes the order of magnitude of the bandwidth @xmath167 hz of the numerically obtained spectra @xmath164 .
the presence of the random driving force is the source of excitation of josephson oscillations in the course of the system s evolution , even if initially at @xmath107 , @xmath168 and @xmath169 .
note , that all the elementary excitations with nonzero momenta in the antisymmetric mode have frequencies larger than than @xmath163 .
the energy transfer between nonzero - momentum excitations and josephson mode is thus an essentially nonlinear process . the nonlinear structure of the right - nand - side of eq .
( [ dforce ] ) provides the presence of the frequency @xmath163 in the spectrum @xmath170 of the driving force and thus ensures the parametric excitation of the josephson oscillations .
we confirmed our analytic estimations by the numerical simulations of two coupled 1d gross - pitaevskii equations already described in subsection [ subiii.2 ] .
an example of a sharp - peaked power spectrum of relative number imbalance is given in fig .
[ f:2 ] , together with an example of time dependence of @xmath171 .
if , on the contrary , @xmath172 , where @xmath173 is the typical time scale of fluctuations of @xmath174 , then the behavior of @xmath175 becomes irregular and @xmath164 does not exhibit a peak at @xmath176 any more ( see fig .
[ f:3 ] ) .
the results of numerical simulations shown in figs .
[ f:2 ] and [ f:3 ] demonstrate certain energy exchange , but no full equilibration between the josephson oscillations and phononic modes .
if we set @xmath177 and @xmath178 for @xmath179 hz ( or 0.1 hz ) , then at times @xmath180 between 650 ms and 1 s the mean energy of josephson oscillations is by an order of magnitude ( or by 1.5 orders of magnitude , respectively ) less than @xmath181 , where temperature @xmath1 is determined from the phase - correlation function for a single quasicondensate and is thus associated with the phononic modes .
this may indicate an extremely long thermalization time for josephson oscillations .
we selected our simulations that display a pronounced narrow peak of @xmath164 far from zero frequency ( which was the case for @xmath182 hz ) , estimated the josephson frequency @xmath163 and analyzed the dependence of @xmath183 on the mean interwell coherence .
the resulting values are in a good agreement with our theoretical prediction given by eq .
( [ oj2 ] ) , as can be seen from fig .
to conclude , we applied the transfer - operator technique to evaluate coherence and correlation properties of two tunnel - coupled 1d weakly - interacting , ultracold systems ( quasicondensates ) of bosonic atoms .
these properties are determined by the ratio of the two length scales : @xmath108 that describes the spatial scale of the loss of correlations between two points and @xmath18 that describes the scale for the phase - locking between two quasicondensates due to interwell tunneling . in the limit @xmath184 the fluctuations of the relative phase are small and we reproduce the results of the linearized theory of ref . @xcite . in the opposite case , we found the mean interwell coherence to decrease much slower ( @xmath185 ) than the exponential law predicted by the linearized theory .
we interprete such a behavior as a signature of thermal creation of sine - gordon solitons , which provide a shift of the relative phase by @xmath92 and thus do not contribute to the coherence loss , and the corresponding decrease of the density of states for phonons ( the excitations responsible for the coherence loss at large distances ) .
our analytic estimations are confirmed by numerical modeling of the equilibrium state as a final state of the system s relaxational evolution after a quench .
this task is solved by extending our numerical method @xcite to integration of two coupled 1d gross - pitaevskii equations .
we demonstrate , both analytically and numerically , that thermal fluctuations of the relative phase between two quasicondensates reduce the frequency of josephson oscillations in proportion to @xmath158 and broaden their spectrum . if the theoretically predicted value of @xmath163 is much less than the bandwidth of the thermal fluctuation ( which is of the order of the speed of sound divided by @xmath108 ) , regular josephson oscillations are not observed .
this work was supported by the the fwf ( project no .
p22590-n16 ) .
the authors thank t. berrada and j. schmiedmayer for helpful discussions .
we briefly recall here the basics of the transfer operator technique , following refs . @xcite .
we introduce a lattice with the step @xmath186 , @xmath187 being the number of sites .
we assume cyclic boundary conditions , @xmath188 then the partition function ( [ eqz2 ] ) can be written as @xmath189 , \label{a2}\end{aligned}\ ] ] where @xmath190 and integrals in our case are taken from @xmath191 to @xmath192 .
we omit the constant prefactor in eq .
( [ a2 ] ) for the sake of simplicity .
assume that eigenfunctions @xmath88 of the transfer operator @xmath193 form a set , which is complete , orthogonal , and normalized to unity , namely @xmath194 substituting eq .
( [ a6 ] ) into eq .
( [ a2 ] ) and using eq .
( [ a4 ] ) , we obtain @xmath195 the eigenvalues @xmath90 are positive ; in the thermodynamic limit the partition function ( [ a7 ] ) is dominated by the lowest eigenvalue @xmath196 , @xmath197 in the continuous limit @xmath198 eq .
( [ a4 ] ) is equivalent to the schrdinger - type equation ( [ aux2 ] ) .
strictly speaking , the spectrum of eq .
( [ a4 ] ) is shifted with respect to the spectrum of eq .
( [ aux2 ] ) by a common offset @xmath199 , which is related to normalization of the eigenfunctions .
since @xmath199 does not depend on @xmath200 , we neglect it in our calculations . to calculate correlation functions , in particular , eq .
( [ cfg ] ) , we note that @xmath201 and @xmath202 act on @xmath203 like quantum - mechanical perturbations , coupling @xmath203 to the whole spectrum of eigenfunctions with the matrix elements given by eq .
( [ aux1 ] ) . therefore the leading term for @xmath72\rangle $ ] in the limit of @xmath204
is the second - order perturbative correction to the propagator for the ground state ( with @xmath7 playing the role of imaginary time ) , and we obtain thus eq .
( [ cfg ] ) .
we begin with the lattice version of the classical sine - gordon hamiltonian that describes the dynamics of the antisymmetric mode of our system : @xmath205 } , \label{b1 } \end{aligned}\ ] ] where the @xmath11th generalized co - ordinate @xmath206 is the half - difference of the atomic numbers in the 1st and 2nd quasicondensates at the @xmath11th site , i.e. , the variable canonically conjugate to the local phase difference @xmath207 ( the @xmath11th generalized momentum ) . here
we neglect the nonlinear coupling between the symmetric and antisymmetric modes , like in eq .
( [ hfullnew ] ) in the continuous limit . for the sake of simplicity , we assume an odd number of sites in the lattice , @xmath208 , where @xmath209 is a positive integer
. then we do a canonical transformation @xmath210 where @xmath211 then the hamiltonian ( [ b1 ] ) reads @xmath212 \tilde{\theta } _ { \mathrm{a}}^2(\ell ) + \nonumber \\ & & \frac { g}{\delta z } \delta \ ! \tilde{n } _ { \mathrm{a}}^2(\ell ) \bigg{\ } } + \label{b4 } \\ & & { 2\hbar jn_\mathrm{1d}\delta z}\sum _ { j=1}^m\left \ { 1-\cos \left [ \sum _ { \ell = -m_0}^{m_0 } \tilde{\theta } _ { \mathrm{a}}(\ell ) \eta ( \ell , j)\right ] \right \ } .
\nonumber \end{aligned}\ ] ] from the hamiltonian equations @xmath213 we find , in particular , @xmath214 @xmath215 \ ! .\quad
\label{b7}\ ] ] in the limit of @xmath198 the sums over @xmath11 converge to integrals over @xmath58 . taking into account that @xmath216 , identifying the generalized momentum conjugate to @xmath217 as @xmath218 and recalling that @xmath219 , we obtain eqs .
( [ js1 ] , [ js2 ] ) .
the spatially fluctuating part of the phase is then @xmath220 . | we revisit the theory of tunnel - coupled atomic quasicondensates in double - well elongated traps at finite temperatures . using the functional integral approach , we calculate the relative phase correlation function beyond the harmonic limit of small fluctuations of the relative phase and its conjugate relative - density variable .
we show that the thermal fluctuations of the relative phase between the two quasicondensates decrease the frequency of josephson oscillations and even wash out these oscillations for small values of the tunnel coupling . |
the heavy - ion physics has attracted much attention during the last three decades @xcite . the behavior of nuclear matter under the extreme conditions of temperature , density , angular momentum etc .
, is a very important aspect of heavy - ion physics .
one of the important quantity which has been used extensively to study this hot and dense nuclear matter is the collective transverse in - plane flow @xcite .
this quantity has a beauty of vanishing at a certain incident energy .
this energy is dubbed as balance energy ( @xmath1 ) or the energy of vanishing flow ( evf ) @xcite .
this beauty is due to the counterbalancing of attractive mean - field at low incident energies and repulsive nucleon - nucleon ( nn ) collisions at higher incident energies .
the balance energy of the masses ranging from @xmath2 to @xmath3 at different colliding geometries was studied experimentally and theoretically and found to be sensitive with the composite mass of the system @xcite as well as with the impact parameter of a reaction @xcite .
+ with the passage of time , isospin degree of freedom in terms of symmetry energy and nn cross section is found to affect the balance energy or energy of vanishing flow and related phenomenon in heavy - ion collisions @xcite .
+ experimentally , pak _ et al . _ , studied the isospin effects on the collective flow and balance energy at central and peripheral collision geometries @xcite . on the other hand , theoretically , this effect is studied by using the isospin - dependent boltzmann uehling - uhlenbeck model ( ibuu ) @xcite , and isospin - dependent quantum molecular dynamics ( iqmd ) model @xcite . + as noted , balance energy is due to the counterbalancing of the attractive mean - field and repulsive nucleon- nucleon collisions .
the coulomb interaction in intermediate energy heavy - ion collisions is expected to play a dominant role in balance energy due to its repulsive nature .
these effects are supposed to be more pronounced in the presence of isospin effects @xcite . the comparative study which will show the shift in balance energy due to coulomb interactions in the presence of isospin effects by taking into account asymmetry of reaction in a controlled fashion is still missing in the literature .
the second point is the asymmetry of the reaction . in some of the studies ,
the asymmetry of a reaction is taken into care , but not in other , which is very important to study the isospin effects @xcite .
the asymmetry of the reaction can be defined by the parameter @xmath4 ; where @xmath5 and @xmath6 are the masses of target and projectile .
the @xmath7 = 0 corresponds to the symmetric reactions , whereas , non - zero value of @xmath7 define different asymmetry of the reaction .
it is worth mentioning that the reaction dynamics in a symmetric reaction ( @xmath7 = 0 ) can be quite different compared to asymmetric reaction ( @xmath8 ) @xcite .
this is due to the deposition of excitation energy in the form of compressional energy and thermal energy in symmetric and asymmetric reactions , respectively .
the effect of the asymmetry of a reaction on the multifragmentation is studied many times in the literature @xcite .
unfortunately , very little study is available for the asymmetry of the reaction in terms of transverse in - plane flow .
+ in this paper , we will perform the first ever study for the balance energy in terms of asymmetry of the reaction and then observe the effect of coulomb interactions , symmetry energy , equations of state as well as different frame of references . the iqmd model used for the present analysis
is explained in the sec .- ii .
the results are presented in sec .-
iii , leading to the conclusions in sec .- iv .
the isospin - dependent quantum molecular dynamics ( iqmd)@xcite model treats different charge states of nucleons , deltas and pions explicitly @xcite , as inherited from the vlasov - uehling - uhlenbeck ( vuu ) model @xcite . the iqmd model was used successfully in analyzing the large number of observables from low to relativistic energies @xcite .
one of its version ( qmd ) , has been very successful in explaining the subthreshold particle production @xcite , multi - fragmentation @xcite , collective flow @xcite , disappearance of flow @xcite , and density temperature reached in a reaction @xcite .
we shall not take relativistic effects into account , since in the energy domain we are interested , there is no relativistic effect @xcite .
the isospin degree of freedom enters into the calculations via both cross sections , mean field and coulomb interactions @xcite .
the details about the elastic and inelastic cross sections for proton - proton and neutron - neutron collisions can be found in refs.@xcite .
+ in this model , baryons are represented by gaussian - shaped density distributions + @xmath9 nucleons are initialized in a sphere with radius r = @xmath10 fm , in accordance with the liquid drop model .
each nucleon occupies a volume of @xmath11 so that phase space is uniformly filled .
the initial momenta are randomly chosen between 0 and fermi momentum @xmath12 .
the nucleons of the target and projectile interact via two and three - body skyrme forces and yukawa potential .
the isospin degrees of freedom is treated explicitly by employing a symmetry potential and explicit coulomb forces between protons of the colliding target and projectile .
this helps in achieving the correct distribution of protons and neutrons within the nucleus .
+ the hadrons propagate using hamilton equations of motion : + @xmath13 with + @xmath14 is the hamiltonian . @xmath15 the baryon - baryon potential @xmath16 , in the above relation , reads as + @xmath17 where @xmath18 , @xmath19 and @xmath20 .
the values of @xmath21 and @xmath22 depends on the values of @xmath23 , @xmath24 and @xmath25 @xcite . here
@xmath26 and @xmath27 denote the charges of the @xmath28 and @xmath29 baryon , and @xmath30 , @xmath31 are their respective @xmath32 components ( i.e. 1/2 for protons and -1/2 for neutrons ) .
the parameters @xmath33 and @xmath34 are adjusted to the real part of the nucleonic optical potential . for the density dependence of the nucleon optical potential ,
standard skyrme - type parameterizations is employed .
the skyrme energy density have been shown to be very successful at low incident energies where fusion is dominant channel @xcite .
the yukawa term is quite similar to the surface energy coefficient used in the calculations of nuclear potential for fusion @xcite .
the choice of the equation of state ( or compressibility ) is still a controversial one .
many studies advocates softer matter , whereas , much more believe the matter to be harder in nature @xcite .
we shall use both hard ( h ) and soft ( s ) equations of state that have compressibilities of 380 and 200 mev , respectively .
as discussed earlier , asymmetry of a reaction is found to affect the phenomena of muti - fragmentation in intermediate energy heavy - ion collisions @xcite . on the other hand , system mass dependence of balance energy
is studied many times in the literature @xcite . to check the effect of coulomb interactions and asymmetry of a reaction on the balance energy , we have fixed ( @xmath0 = @xmath35 = 152 ) and varied the asymmetry of the reaction just like this : @xmath36 ( @xmath37 ) , @xmath38 ( @xmath39 ) , @xmath40 ( @xmath41 ) , @xmath42 ( @xmath43 ) , @xmath44 ( @xmath45 ) , @xmath46 ( @xmath47 ) .
the asymmetry of a reaction with multi - fragmentation in this fashion is varied many times @xcite .
the whole reaction dynamics is studied at semi - central geometry by varying the incident energy between 50 and 250 mev / nucleon with an increment of 50 mev / nucleon by employing hard as well as soft equations of state .
we have checked the stability of the reacting nuclei in laboratory ( lab ) as well as in center of mass ( c.m . )
frame by taking into account the coulomb interactions .
our main purpose here is to understand the effect of equations of state and coulomb interactions on the energy of vanishing flow or alternatively , on the balance energy by taking into account the asymmetry of a reaction .
+ the directed transverse flow is calculated using @xmath48 @xcite + @xmath49 where y(i ) and @xmath50 are , respectively , the rapidity distribution and transverse momentum of the @xmath28 particle . + asymmetry dependence of directed flow in lab as well as center of mass frame in l.h.s , while , r.h.s for the relative effect of coulomb interactions .
the different lines in the figure are representing the effect of symmetry energy and coulomb interactions . ] to check the effect of frame of reference , we display in fig.1 , variation of the asymmetry @xmath7 on directed flow @xmath51 in lab . as well as center of mass frame at incident energy of e
= 50 mev / nucleon .
the top and bottom panels in the right hand side of the figure are representing the relative coulomb effect with respect to the asymmetry of the reaction .
this relative effect is calculated as : + @xmath52 as is evident from the figure , directed flow is found to increase in a systematic manner in c.m . as well as in lab frame at e = 50 mev / nucleon with asymmetry of the reaction .
the inclusion of coulomb interactions does not alter the conclusions .
note that the increase in the asymmetry is related with the increase in the n / z ratio . because the symmetry potential for the neutron rich systems is stronger compared to the neutron poor systems due to large relative neutron strength .
furthermore , the symmetry potential is repulsive for neutrons and attractive for protons . on the other hand ,
more negative value of directed flow ( dominating the mean field ) is observed in the absence of coulomb interactions in center of mass as well as in lab .
this is due to the enhancement of the chemical and mechanical instability domains in the absence of coulomb interactions @xcite .
similar type of study and conclusion was also performed for nuclear stopping in ref .
+ extensive study in the literature proves the stability of reactions in lab .
frame , but , keep in mind that the reaction under consideration in these studies were symmetric in nature . as is clear from the figure , asymmetric systems are found to be more stable in the center of mass frame compared to the lab frame .
moreover , if one consider lab frame , one is surely missing the effect of asymmetry . to further strengthen the stability of center of mass frame with asymmetry , the relative effect of coulomb interactions @xmath53
is studied in both frames .
the relative effect of the coulomb interactions is found to decrease with increase in the asymmetry of a reaction .
the systematic decrease can be seen in center of mass frame with asymmetry , while very weak dependence of coulomb interactions on the asymmetry is obtained in lab frame . for the further study
, we have opted the center of mass frame because we want to see the shift in the balance energy due to coulomb interactions , whose effect is clearly visible in center of mass frame .
+ the time evolution of directed flow at different incident energies in center of mass frame in the absence of coulomb interactions .
the left and right panels are for soft and hard equations of state , respectively . ] before we proceed further , let us check the time evolution of directed transverse flow . in fig .
[ fig2 ] , we display the time evolution of directed flow from e = 50 to 200 mev / nucleon in center of mass frame for soft ( l.h.s ) and hard ( r.h.s ) equations of state .
note that the compressibilities of soft and hard equations of state are 200 and 380 mev , respectively @xcite .
the time evolution is plotted in the absence of coulomb interactions to see the maximum effect of asymmetry of a reaction on the directed in - plane flow .
the figure reveals the following points : + a ) .
the quantity @xmath48 is observed to be constant throughout the whole distribution of time , while , the large variation is observed in the value at initial and final time steps when observed in the lab frame @xcite + b ) . with the increase in the incident energy , the directed flow is approaching towards more positive value .
this is due to the well known fact of increase in the frequent nn collisions with increase in the incident energy @xcite .
the behavior of the directed flow with asymmetry of a reaction follows the opposite trend at e = 50 mev / nucleon as compared to other high incident energies .
it has been discussed many times in literature and also clear from the present findings that attractive mean - field is dominating at e = 50 mev /nucleon compared to higher incident energies under consideration @xcite .
this is due to the different mechanisms contributing at low and high incident energies within isospin - dependent quantum molecular dynamics .
it was shown by us as well as by others @xcite that symmetry potential dominates the physics at low incident energy , while , nn cross sections one major driven force at higher incident energies .
furthermore , at low incident energies , symmetry potential is repulsive for neutrons and attractive for protons . with the increase in the asymmetry of a reaction ,
the number of neutrons increases and hence comparative repulsion due to neutrons also increases leading to less attractive value of the flow .
it is also clear from ref .
@xcite , that with the increase in asymmetry of a reaction , the participant zone decreases and spectator zone increases . at higher incident energies , where the effect of symmetry energy is negligible
, the contribution of nn collisions comes from the participant zone whereas mean - field contribution comes from the spectator zone .
hence directed flow is found to be less positive with increase in the asymmetry of the reaction .
the directed flow has less positive value with soft equation of state compared to hard equation of state .
the less positive means the dominance of mean - field .
this is due to the different compressibilities of hard ( 380 mev ) and soft ( 200 mev ) equations of state .
naturally , more is the compressibility , more are the number of collisions and hence more positive is the directed flow .
this is indicating that the directed flow is sensitive towards the equations of state .
+ excitation function of directed flow at different asymmetries with and without coulomb interactions at semi - central geometry . ] finally , in figs.[fig3 ] and [ fig4 ] , we display the excitation function of directed flow at different asymmetries from @xmath7 = 0.2 to 0.7 .
the value of abcissa at zero value of @xmath48 corresponds to the energy of vanishing flow ( evf ) or alternatively , the balance energy ( @xmath1 ) .
the fig . [ fig3 ] shows the shift in the balance energy due to coulomb interactions , while , fig .
[ fig4 ] is representing the shift in the balance energy due to different equations of state . in fig .
[ fig3 ] , one sees a linear enhancement in the nuclear flow with increase in the incident energy .
this increase in the transverse flow is sharp at smaller incident energies ( upto 200 mev / nucleon ) .
if one goes to higher incident energies , the value gets saturated as discussed in ref @xcite .
we have displayed here the results upto 250 mev / nucleon , since we are interested in and around balance energy . in the presence of coulomb interactions , more positive value of the flow is obtained .
this is due to the well known repulsive nature of coulomb interactions . at higher energies ,
the repulsion due to coulomb interactions is stronger during the early phase of the reaction and transverse momentum increases sharply .
the overall effect depends on the asymmetry of the reaction .
if one looks at the balance energy , the shift in the incident energy towards the higher value is obtained at @xmath54 with the asymmetry of the reaction .
this is showing that with increase in asymmetry of the reaction and in the absence of coulomb interactions , attractive mean - field is dominating the large region of incident energy .
the systematics of the balance energy with asymmetry of the reaction is discussed in fig .
. + excitation function of directed flow with hard and soft equations of state in the absence of coulomb interactions for different asymmetries . ] as we have seen in fig.[fig2 ] , that different equations of state show sensitivity towards the directed flow with respect to the asymmetry of a reaction . the detailed analysis with soft ( s ) and hard
( h ) equations of state is displayed in fig .
the excitation function of directed flow follows similar trend as explained in fig .
[ fig3 ] . for nearly symmetric systems ( @xmath7 = 0.2 )
, the balance energy is found to be independent of the equations of state , however , resonable differences are observed at higher incident energies .
more positive values of directed flow are obtained with hard equation of state compared to soft equation of state .
this is true at all asymmetries from ( @xmath7 = 0.3 to 0.7 ) .
this is due to different compressibilities of hard ( 380 mev ) and soft ( 200 mev ) equations of state . with an increase in the asymmetry of a reaction
, the shift in the balance energy towards the higher value of incident energy takes place with soft equation of state compared to hard one .
this is consistent with the findings in the literature @xcite .
+ power law dependence of balance energy with asymmetry of the reaction .
the lower panel is representing the relative @xmath55 effect of coulomb interactions on the balance energy . ] to sum up , in fig.[fig5 ] , we have displayed the asymmetry dependence of balance energy .
we displayed the results for hard and soft equations of state by switching off the coulomb interactions .
in addition , for a comparative study , the results in the presence of coulomb interactions with soft equation of state are also shown .
all the lines are fitted with power law of the form @xmath56 , where c and @xmath57 are the constants .
the values of @xmath57 in the absence of coulomb interactions for soft and hard equations of state are 0.375 and 0.282 , respectively , while in the presence of coulomb interactions for soft equation of state is @xmath57 = 0.06067 .
+ if we compare the asymmetry dependence of balance energy with mass dependence , the trend is opposite @xcite .
it clearly indicates that if one wants to study the isospin effect , then one has to consider the systems with @xmath58 , otherwise one is not able to get the exact information about the isospin effects .
it is also clear from the figure , that shift in the balance energy is observed due to coulomb interactions as well as due to equations of state with asymmetry of the reaction .
the shift is more due to coulomb interactions in comparison to equations of state , indicating the importance of coulomb interactions in intermediate energy heavy - ion collisions .
the higher balance energy is obtained with coulomb - off + soft equation of state followed by coulomb - off + hard equation of state and finally coulomb - on + soft equation of state . + for the further understanding
, the relative percentage difference in the balance energy is plotted in the lower panel denoted by the quantity @xmath59 given by + @xmath60 { \times}100\ ] ] the @xmath59 is found to increase with the increase in the asymmetry of a reaction .
this indicates shift of the nuclear matter towards the attractive mean - field region in the absence of coulomb interactions with the asymmetry of the reaction .
this difference @xmath61 = 30 mev / nucleon at @xmath37 , while it is 115 mev / nucleon at @xmath47 .
the difference of 90 mev / nucleon in the shift of balance energy with asymmetry can not be ignored .
this is the first ever study and experiments are called to verify the results .
our present aim was to understand the influence of coulomb interactions as well as equations of state on the dynamics of large asymmetric reactions in semi - central heavy - ion collisions . at low incident energies ,
the contribution of mean - field is more for the nearly symmetric systems , while at higher incident energies , opposite scenario is observed .
the balance energy is found to increase with the increase in the asymmetry of the reaction .
the balance energy is affected by the coulomb interactions compared to different equations of state .
this work has been supported by the grant from department of science and technology ( dst ) , government of india , new delhi , vide grant no.sr/wos-a/ps-10/2008 .
999 w. scheid _ et al . _ , phys . rev
. lett . * 32 * , 741 ( 1974 ) ; k. g. r. doss _
et al . _ ,
lett . * 57 * , 302 ( 1986 ) ; w. reisdorf _
et al . _ ,
nucl . part .
sci . * 47 * , 663 ( 1997 ) .
a. d. sood and r. k. puri , phys .
c * 69 * , 054612 ( 2004 ) ; a. d. sood _ et al .
b * 594 * , 260 ( 2004 ) ; a. d. sood _ et al .
j. a * 30 * , 57 ( 2006 ) ; r. chugh and r. k. puri , phys .
c * 82 * , 014603 ( 2010 ) . v. d. l. mota _
et al . _ ,
c * 46 * , 677 ( 1992 ) ; g. d. westfall , phys .
* 71 * , 1986 ( 1993 ) ; h. zhou _
et al . _ ,
c * 50 * , r2664 ( 1994 ) ; h. zhou _
et al . _ ,
a * 580 * , 627 ( 1994 ) ; d. j. magestro _ et al .
c * 61 * , 021602(r ) ( 2000 ) ; d. j. magestro _ et al . _ ,
c * 62 * , 041603(r ) ( 2000 ) .
f. s. zhang _ et al . _ , eur .
j. a * 9 * , 149 ( 2000 ) ; c. a. ogilvie _ et al .
lett . * 67 * , 1214 ( 1991 ) ; j. singh , s. kumar , and r. k. puri , phys .
c * 63 * , 054603 ( 2001 ) ; m. b. tsang _
et al . _ ,
* 71 * , 1502 ( 1993 ) ; a. schuttauf _
et al . _ ,
a * 607 * , 457 ( 1996 ) ; n. t. b. stone _
et al . _ ,
phys . rev . lett.*78 * , 2084 ( 1997 ) ; b. jackobsson _
et al . _ ,
nucl . phys .
a * 509 * , 195 ( 1990 ) ; h. feldmeier , nucl .
phys . * 515 * , 147 ( 1990 ) ; a. ono , h. horiuchi , t. maruyama , phys .
c * 48 * , 2946 ( 1993 ) ; _ ibid . _
* 47 * , 2652 ( 1993 ) ; p.
b. gossiaux , r. k. puri , ch .
hartnack , and j. aichelin , nucl .
a * 619 * , 379 ( 1997 ) ;
s. kumar and r. k. puri , phys . rev .
c * 58 * , 320 ( 1998 ) ; _ ibid . _ * 58 * , 2858 ( 1998 ) ; _ ibid . _
* 60 * , 054607 ( 1999 ) .
r. donangelo _ et al .
c * 52 * , 326 ( 1995 ) .
et al . _ ,
nucl . phys .
a * 531 * , 177 ( 1991 ) .
et al . _ ,
c * 68 * , 1618 ( 1998 ) .
y. k. vermani _
et al . _ , j. phys .
g * 36 * , 105103 ( 2009 ) .
v. kaur and s. kumar , phys . rev .
c * 81 * , 064610 ( 2010 ) . c. a. ogilivie _ et al .
lett . * 67 * , 1214 ( 1991 ) ; m.b .
_ , _ ibid .
_ * 71 * , 1502 ( 1993 ) ; r. t. de souza _ et al . _ ,
b * 268 * , 6 ( 1991 ) ; n. t. b. stone _
et al . _ ,
lett . * 78 * , 2084 ( 1997 ) ; w. j. llope _
et al . _ , phys .
c * 51 * , 1325 ( 1995 ) .
h. kruse , b. v. jacak , and h. stcker , phys .
lett . * 54 * , 289 ( 1985 ) ; j. j. molitoris and h. stcker , phys .
c * 32 * , r346 ( 1985 ) ; j. aichelin and g. bertsch , phys .
c * 31 * , 1730 ( 1985 ) ; c. hartnack _ et al .
lett . * 96 * , 012302 ( 2006 ) .
et al . _ ,
c * 58 * , 3494 ( 1998 ) ; s. kumar , s. kumar , and r. k. puri , phys .
c * 78 * , 064602 ( 2008 ) ; j. singh , s. kumar , and r. k. puri , phys .
c * 62 * , 044617 ( 2000 ) ; j. k. dhawan and r. k. puri , phys .
c * 75 * , 057601 ( 2007 ) ; j. k. dhawan and r. k. puri , phys .
c * 75 * , 057901 ( 2007 ) .
e. lehmann _
_ , z. phys . a * 355 * , 55 ( 1996 ) ; y. k. vermani , j. k. dhawan , s. goyal , r. k. puri , and j. aichelin , j. phys .
g : nucl . part .
* 37 * , 015105 ( 2010 ) ; y. k. vermani , s. goyal and r. k. puri , phys .
c * 79 * , 064613 ( 2009 ) ; y. k. vermani and r. k. puri , eur .
phys . lett . *
85 * , 62001 ( 2009 ) .
r. k. puri and n. dhiman , eur .
j a * 23 * , 429 ( 2005 ) ; r. arora , r. k. puri and r. k. gupta , eur . phys .
j a * 8 * , 103 ( 2000 ) ; r. k. puri , m. k. sharma and r. k. gupta , eur .
j a * 3 * , 277 ( 1998 ; r. k. puri and r. k. gupta , phys .
c * 51 * , 1568 ( 1995 ) ; r. k. puri , p. chattopadhyay and r. k. gupta , phys .
c * 43 * , 315 ( 1991 ) ; r. k. puri and r. k. gupta , phys .
c * 45 * , 1837 ( 1992 ) ; _ ibid
. _ j. phys .
g : nucl . part .
phys . * 18 * , 903 ( 1992 ) .
r. k. gupta _
et al . _ ,
c * 47 * , 561 ( 1993 ) ; r. k. gupta _
_ , j. phys .
g * 18 * , 1533 ( 1992 ) ; s. s. malik _
et al . _ , pramana j. phys . * 32 * , 419 ( 1989 ) ; r. k. puri , s. s. malik and r. k. gupta , eur .
* 9 * , 767 ( 1989 ) .
i. dutt and r. k. puri , phys .
c * 81 * , 047601 ( 2010 ) ; i. dutt and r. k. puri , phys .
c * 81 * , 044615 ( 2010 ) ; i. dutt and r. k. puri , phys
c * 81 * , 064608 ( 2010 ) ; i. dutt and r. k. puri , phys
c * 81 * , 064609 ( 2010 ) ; h. stcker and w. greiner , phys .
rep . * 137 * , 277 ( 1986 ) . | we aim to understand the role of coulomb interactions as well as of different equations of state on the disappearance of transverse flow for various asymmetric reactions leading to same total mass . for the present study ,
the total mass of the system is kept constant ( @xmath0 = 152 ) and asymmetry of the reaction is varied between 0.2 and 0.7 .
we find that the contribution of mean - field at low incident energies is more for nearly symmetric systems , while the trend is opposite at higher incident energies .
the coulomb interactions as well as different equations of state are found to affect the balance energy significantly for larger asymmetric reactions . |
the propagation and continuous flow of information are of utter importance for the development of stable economies throughout the world as they are a prerequisite for successful business transactions , short- and long - range communication , and so on @xcite .
often this information has to be encrypted in such a way that it can be safely transferred between the sender and recipient without allowing others to read the information that is present in such an encrypted message @xcite . on the other hand ,
malicious persons and organizations , but also governmental organizations , are continuously striving to break the key with which messages were encrypted because this might enable them to get those pieces of information that are needed to achieve their criminal , protective , or other goals @xcite .
it is probably due to the impact of turing s success in breaking the enigma that humanity became aware of the importance of cryptography in general , and the vulnerability of ciphers more in particular @xcite . since this major breakthrough , the functioning of the industrial , financial and public sector has become strongly dependent on the advances of cryptography .
for instance , while the availability of worldwide networks has enabled rapid dissemination of information , it has also stimulated cryptographic innovations because a significant share of this information may only be available to a few parties . in this manner , technological progress during the last decades has increased the need for secured communication and transactions , information shielding , and so on @xcite . in the last few decades , modern cryptography replaced mechanical schemes with new computing models .
this modern focus influenced the classical design of ciphers far beyond the original purpose .
nowadays , there are two class of cryptographic algorithms depending on the key : symmetric and asymmetric .
symmetric encryption algorithms use the same key for both encryption of plaintext and decryption of ciphertext .
this class of algorithm is also divided into two categories : stream ciphers and block ciphers .
block ciphers have gained wide popularity since the introduction of the first adopted encryption : the data encryption standard ( des ) @xcite , in the mid-1970s , yet nowadays this cipher is considered prone to brute force attacks . to overcome this shortcoming the international data encryption algorithm ( idea )
@xcite was designed in 1991 to replace des .
ever since , there has been a pursuit for the development of new algorithms that meet the rising security expectations . in 1997 , the national institute of standards and technology ( nist ) @xcite selected the official advanced encryption standard ( aes ) among many competitors , namely serpent @xcite , twofish @xcite , rc6 @xcite , rijndael @xcite , etc . to date , block ciphers are the most important elements in many cryptographic systems @xcite
. a block cipher breaks a message into blocks of elements ( bits ) and then encrypts one block ( _ plaintext _ ) at a time producing its corresponding output block ( _ ciphertext _ ) .
however , a block cipher by itself allows for the encryption of only one block , such that it is recommended to use a mode of operation in conjunction @xcite .
this mode of operation specifies a mechanism to improve the corresponding block cipher , while encrypting all of the blocks , one by one , as it goes along . motivated by the analogies between cryptographic and dynamical systems , on the one hand , and the lack of a means to discriminate between different modes of operation that can be used to encrypt a message with a block cipher using a single key , on the other hand , we demonstrate in this paper how lyapunov
exponents can be relied upon for tackling this problem . more specifically , by contemplating the whole of a cipher , ciphertext and key as an utter discrete dynamical system , _
i.e. , _ a cellular automaton ( ca ) , and by resorting to the notion of lyapunov exponents as they have been conceived for such systems @xcite , we show how these measures can be exploited to identify the mode of operation that was used during the encryption process .
although the cryptographic process of encrypting and decrypting information does not constitute a dynamical system as such , it has been reported that it is possible to draw parallels between cryptographic and dynamical systems @xcite .
hence , drawing upon such parallels , we have a means to exploit similar tools as the ones that have been conceived in the framework of dynamical systems in order to characterize cryptographic systems .
taking into account that the stability of a dynamical system is generally acknowledged as its main characteristic because it gives insight into its intrinsic nature @xcite , it is natural to verify whether the dynamical systems viewpoint of a cryptographic system allows for a similar notion in order to better understand the latter .
an exploration of this is further motivated by the fact that several researchers have noticed a close resemblance between a cryptographic system on the one hand , and a chaotic system , on the other hand @xcite , and the large number of chaos - based cryptosystems @xcite .
classically , the stability of a dynamical system is assessed by computing its so - called largest lyapunov exponent that quantifies how it behaves if it is evolved from two different but close initial conditions @xcite . either the corresponding phase space trajectories diverge or converge in which case we refer to the system as unstable or asymptotically stable , respectively , or the system is conservative , which means that the initial separation remains .
as the fields of cryptography and dynamical systems are not yet strongly interwoven , the basic definitions and concepts that relate to those systems and that are of interest within the framework of this paper are presented in section [ pre ] , while the dynamical systems viewpoint on cryptographic systems is presented in section [ method ] together with the proposed method for identifying the underlying mode of operation .
finally , the strengths of the proposed method are illustrated and discussed in section [ res ] by means of computer experiments .
in this section we introduce the specificities of both cryptographic and dynamical systems that are indispensable for a clear understanding this paper . classically , an encryption system encloses three major components , namely a cipher , a key , and finally , a ciphertext .
the former constitutes a sequence of instructions that must be executed in order to encrypt a given plaintext , which may be envisaged as a sequence of @xmath0 bits , such that it can be represented as a boolean vector @xmath1 of length @xmath0 .
the result of this encryption process using a key @xmath2 , which is a sequence of @xmath3 bits , is a so - called ciphertext , which may be represented in a similar fashion as a boolean vector @xmath4 of length @xmath0 @xcite .
of course , the real plaintext size varies and is mostly different from the length of the blocks for which a block cipher is designed . consequently
, common ciphers can not be applied directly for the encryption of arbitrary - length plaintext @xcite . in order to overcome this issue ,
so - called block ciphers have been designed and implemented .
a block cipher slices the plaintext of length @xmath0 into @xmath5 blocks of @xmath6 bits , after which each of these blocks is encrypted / decrypted by a block cipher , denoted as @xmath7 and @xmath8 , respectively .
mathematically , the encryption of a plaintext @xmath9 of length @xmath6 into a ciphertext of the same length can be formalized as @xmath10 , where @xmath11 . if the length @xmath0 of the plaintext is not a whole multiple of @xmath5 , additional bits are padded to the last block of the plaintext .
these block ciphers encrypt a plaintext in accordance with a well - defined procedure , which is commonly referred to as the mode of operation of a block cipher .
a block cipher encrypts one block at a time , and it is the mode of operation that allows a block cipher to encrypt blocks consecutively in a secure way .
most of them use an initialization vector ( iv ) , denoted @xmath12 , which adds randomness to the encryption process @xcite .
for instance , the counter mode uses a special method for generating counters in order to guarantee that each block in the sequence is different from every other block @xcite . from the many block ciphers currently available
, we will focus on the well - studied ones , being des @xcite ( @xmath13 , @xmath14 ) , idea @xcite , tea @xcite and xtea @xcite ( @xmath13 , @xmath15 ) , as well as aes @xcite , rc6 @xcite , twofish @xcite , serpent @xcite , seed @xcite and camellia @xcite ( @xmath16 , @xmath15 ) .
the specifications of these block ciphers and the modes of operation are public , so they can be implemented @xcite . in 2001
the nist compiled and recommended five modes of operation most of them were developed 30 years ago to be used with a block cipher , namely the electronic codebook ( ecb ) , cipher block chaining ( cbc ) , cipher feedback ( cfb ) , output feedback ( ofb ) , and counter ( ctr ) modes @xcite .
in addition to these nist modes , there are many others , among which the propagating cipher - block chaining ( pcbc ) @xcite mode will be considered in this paper .
mathematically , the encryption of an arbitrary - length plaintext by means of a block cipher in combination with the ecb mode of operation can be formulated as : @xmath17 where @xmath18 represents the encrypted counterpart of the @xmath19-th block of plaintext @xmath20 .
similarly , the formalism for the cbc mode is given by @xmath21 where @xmath22 is the mod 2 operator and @xmath23 .
for the ofb mode one may write : @xmath24 where @xmath25 with @xmath26 and @xmath12 is randomly selected from @xmath27 .
table [ modestab ] lists the formulas for the other modes of operation that are considered in this paper .
[ modestab ] in the remainder , a mode of operation of a block cipher is denoted as @xmath28 , which maps a given plaintext @xmath1 of length @xmath0 to a corresponding ciphertext @xmath4 of equal length , so that we may write @xmath29 . in order to clarify the functioning of a cryptographic system ,
we show in fig .
[ fig : ex1 ] a cryptographic system with @xmath30 blocks of @xmath31 bits that uses the cbc mode of operation .
we consider an exemplary plaintext @xmath32 , for which it holds that @xmath33 , @xmath34 and @xmath35 .
note that the outputs generated by @xmath36 are chosen only for illustration , _
we can see clearly that a single application of the function @xmath38 consists of @xmath5 applications of @xmath36 since the latter has to be applied to each of the plaintext blocks @xmath20 in order to construct the ciphertext @xmath4 .
@xmath39 { \node[inner sep=1pt , outer sep=0pt ] ( todotted ) { 101 } ; \draw[densely dotted ] ( todotted.south west ) -- ( todotted.south east ) ; } } ] \nonumber\\ \mathbf{p}&=&[010,001 ] \\
\end{aligned}\ ] ] @xmath40 { \node[inner sep=1pt , outer sep=0pt ] ( todotted ) { 101 } ; \draw[densely dotted ] ( todotted.south west ) -- ( todotted.south east ) ; } } \oplus 010 ) = \uline{100}\nonumber\\ c_2&=&e_k(\uline{100}\hspace{2 dd}\oplus 001 ) = \textbf{000}\nonumber\\ \mathbf{c}&=&m_{e_k}(\mathbf{p})=[100,000]\nonumber \end{aligned}\ ] ] [ cols="^,^",options="header " , ] in fig .
[ fig : ex2 ] we show an illustration of the cbc mode of operation by using eq . for the first two time steps in the evolution of its corresponding ca .
note that text styling has been added to make the effect of the mode of operation tractable , and also , that the same plaintext and @xmath12 values are used as the ones given in fig .
[ fig : ex1 ] . as the mode of operation
is applied iteratively , at every consecutive time step @xmath12 is put equal to @xmath41 because one has to select a new point in phase space that is the closest to the reference direction , which basically constitutes an orientation of the new initial configuration @xmath42 into the direction of @xmath41 . for reasons of uniformity , @xmath12 is the only parameter that can contain this initial configuration .
recently , sundry methods have been developed to unravel the dynamical properties of utter discrete dynamical systems such as ca @xcite .
as such , by relying on these for grasping the dynamics of a cryptographic system s corresponding ca , we might be able to gain deeper insight into the dynamics of the former .
more specifically , we will show in the remainder of this paper how lyapunov exponents of cas may be relied upon for identifying the mode of operation of the underlying cryptographic systems , in the same way as these measures have shown their usefulness for characterizing utter discrete dynamical systems @xcite .
suppose that the ca counterpart of a given cryptographic system @xmath43 is denoted as @xmath44 .
we can investigate the dynamics of the former in general , and its stability more in particular , by computing its so - called lyapunov exponent , which quantifies how the dynamical system behaves in the long run if it is evolved from two close initial conditions .
clearly , in our setting , this means that we will assess the sensitivity of the equivalent ca @xmath44 to a small perturbation of the plaintext since we put @xmath45 .
hence , this should yield insights into the behaviour of the cryptographic system if the plaintext is perturbed . taking into account that the smallest possible perturbation in such a two - state setting boils down to flipping one bit of the first block of the plaintext @xmath1
, @xmath46 constitutes the most influential block ( initial condition ) to the mode of operation .
besides , it holds that the initial damage vector @xmath47 , where @xmath22 is the mod 2 operator , contains only one non - zero element . in the remainder
, a cell @xmath48 for which holds that @xmath49 will be referred to as a * defective cell*. after pinning down a plaintext @xmath1 and its perturbed version @xmath50 that fulfills the latter criterion , we can evolve the equivalent ca @xmath44 from both @xmath45 and @xmath51 for one time step in order to obtain @xmath52 and @xmath53 . here
, it should be recalled that one time step in the evolution of a ca corresponds to one application of the mode of operation @xmath38 , which involves the update of several blocks . having updated all the blocks , we can compute the damage vector at the first time step @xmath54 consequently , the total number of defects at the first time step can be computed as @xmath55 at this point , the reader might think that the damage vector during subsequent time steps should be computed similarly , but this is certainly not the case because one would then neglect the fact that the defects can cancel out each other due to the discrete nature of the ca s state space @xcite .
the discrepancy between the number of defective cells and the number of defects is not yet clear after the first time step because every defective cell at @xmath56 traces back to the same initial defective cell .
however , as soon as the ca is evolved one more time step , a discrepancy emerges between these quantities .
this can be understood by explicitly tracking all possible pathways along which defects at @xmath56 may propagate and accumulate during one subsequent time step ( see fig .
[ propdef ] ) .
it is interesting to have a closer look at how nine defects can arise at the second time step in the evolution of the ca notwithstanding there are only five defective cells , which should be contributed to the fact that several of them may enclose multiple defects due to the existence of several pathways along which defects can propagate .
taking into consideration this reasoning , the correct number of defects at the @xmath57-th time step , denoted @xmath58 , should be computed in accordance with the following six - step procedure . 1 .
let the cryptographic system evolve for one time step , _
@xmath59 , and analogously for its perturbed version @xmath60 , both using the same initialization vector @xmath12 .
2 . calculate the damage vector @xmath61 given by eq . and set the initialization vector as the last block calculated for both versions , _
i.e. , _ @xmath62 and @xmath63 , as explained in section [ sec : analogies ] .
3 . for every @xmath48 for which @xmath64 , create a replica @xmath65 such that @xmath66 , where @xmath67 is the boolean complement of @xmath68 , and @xmath69 , for every @xmath70 . use the set @xmath71 to store these replicas .
let the cryptographic system evolve one more time step , _
i.e. , _ compute @xmath72 and @xmath73 , which boils down to evaluating @xmath74 with the same @xmath75 for all replicas @xmath76 .
calculate the total number of defects at the second time step as follows : @xmath77 and set the initialization vector as the last block calculated for both versions , _
@xmath78 and @xmath79 . 6 . for every @xmath48 and @xmath80 for which @xmath81 , create a replica @xmath65 such that @xmath82 and @xmath83 for every @xmath70 .
use a multiset @xmath84 to store these replicas . 7 .
repeat steps ( 4)(6 ) in every subsequent time step @xmath85 in order to assemble @xmath86 and @xmath87 . after computing the number of defects @xmath58 at every consecutive time step @xmath57
, the rate of divergence / convergence of initially close phase space trajectories @xmath88 of a ca @xmath44 can be obtained from : @xmath89 with its limit value @xmath90 generally referred to as the maximum lyapunov exponent ( mle ) of @xmath44 . in the framework of cryptographic systems
, @xmath88 quantifies how ciphertexts , which are obtained by iteratively encrypting two close plaintexts @xmath1 and @xmath50 , behave ( converge / diverge ) as the number of time steps grows . as indicated in papers on the le of cas
, one can derive a theoretical upper bound on these les , which has shown to depend on the number of neighbours @xcite .
calling to mind shannon s idea of diffusion , which is related to the avalanche effect and states that a slight change of the plaintext gives in worst case rise an entirely different ciphertext , we are able to derive a theoretical upper bound on the mle of cryptographic systems . for instance , consider the plaintext @xmath91 and its perturbed version @xmath92 , as well as their encryption @xmath93 and @xmath94 , respectively . in the worst case scenario ,
the number of differences may be at most 8 , which is the length of the plaintext @xmath0 .
this means that a defect in @xmath95 at the @xmath57-th time step can at most propagate to @xmath0 bits at the subsequent time step .
therefore , a mean - field approximation of the upper bound @xmath96 for the cryptographic systems becomes @xmath97 obviously , the higher the number of blocks @xmath5 , the higher the upper bound becomes . before turning to the experimental section of this paper ,
we illustrate in fig .
[ fig : ex3 ] the procedure by which the le of a mode of operation @xmath38 can be assessed .
note that we employed the same plaintext as in fig .
[ fig : ex2 ] . by moving along this table s rows
, we see the plaintexts that are encrypted by repeatedly applying the mode of operation @xmath38 .
note that some replicas are repeated as soon as the system is evolved for one time step , which is in agreement with the findings reported in fig .
[ propdef ] . in order to avoid having to keep track of all pathways individually
, one can optimize the algorithm by just counting every unique replica once and keeping track of its multiplicity .
for example , at @xmath98 the replica @xmath99 would be repeated twice , then at the subsequent time step this replica has two cell defects , which means four pathways in total at @xmath100 .
as such , by tracking the multiplicity of every defect rather than every defect itself , the efficiency of the algorithm increases substantially .
if the cryptographic system is sensitive to the plaintext from which it is evolved , the number of defects @xmath58 increases exponentially during its evolution , such that @xmath101 , and consequently , @xmath102 . on the other hand ,
the system should be referred to as insensitive to the underlying plaintext if the initial defect vanishes as the number of time steps increases .
such a situation implies that @xmath103 owing to the discrete nature of a ca s state space . since
cryptographic systems are designed in such a way that small changes in the plaintext give rise to substantially differing ciphertexts , it should not come as a surprise if @xmath104 would be positive for all the modes of operation at stake in this paper . still , the numerically assessed values of @xmath88 might differ significantly among different block ciphers and/or underlying modes of operation , such that they may still be used to discriminate between different modes of operation .
this usage will be illustrated in the following section .
to demonstrate the effectivity of the above methodology , we considered two groups of block ciphers on the basis of their block length @xmath6 .
the first group encloses the 64-block ciphers , namely des @xcite , idea @xcite , tea @xcite and xtea @xcite , which all use a @xmath105 key except for des that uses a key of length of @xmath14 bits .
the second group of block ciphers is composed of 128-block ciphers with key of @xmath15 bits : aes @xcite , rc6 @xcite , twofish @xcite , serpent @xcite , seed @xcite and camellia @xcite .
these ten block ciphers were adapted in such a way that they took the plaintext @xmath1 , the key @xmath2 and the initialization vector @xmath12 as input variables . moreover ,
these block ciphers were implemented with a predefined number of _ rounds _ , each of which consist of several inner steps in the course of an encryption process and they depend on the specific algorithm @xcite .
we considered eight rounds for idea , eighteen for des , and thirty - two rounds for both tea and xtea .
further , ten rounds for aes , sixteen for both twofish and seed , eighteen for camellia and thirty - two for serpent .
finally , we implemented the modes of operation in accordance with the formalism given by eqs .
for ecb , cbc and ofb , and in table [ modestab ] for cfb , ctr and pcbc
. to ensure the representativity of the computed lyapunov exponents , we computed the average @xmath104-values that were obtained when the cas were evolved from different plaintexts .
hence , the @xmath104-values reported in the remainder represent averages calculated over an ensemble @xmath106 of 200 randomly generated plaintexts @xmath107 and their perturbed versions @xmath108 , which were obtained by flipping only one bit in the first block of @xmath107 .
since such an ensemble was constructed for each of the concerned cryptographic systems , by mutually combining the underlying block cipher and the mode of operation , we considered a total of 60 combinations , _
i.e. , _ ten block ciphers and six modes of operation . consequently
, a total of 12000 plaintexts were generated randomly .
similarly , the key @xmath2 and the initialization vector @xmath12 were generated randomly in order to avoid both key repetitions and weak keys , which is a recommendation in cryptography @xcite . in the remainder
, we refer to the assembly of the 200 randomly generated plaintexts , keys and initialization vectors as the cryptographic dataset per cipher and mode of operation .
given that the block ciphers encrypt a plaintext by splitting it into a number of blocks and subsequently transcribing every block , the plaintexts were generated in such a way that they were composed of the same number of @xmath5 blocks , irrespective of the type of block cipher .
note that this implies a plaintext containing @xmath109 bits in case of the 64-block ciphers and @xmath110 bits in case of the 128-block ciphers .
thus , in order to compare @xmath104-values of 64- and 128-block ciphers , we normalized the numerically obtained @xmath104 with respect to @xmath96 obtained from eq . , we assessed @xmath88 for the ca counterpart of each of the concerned cryptographic systems .
for that purpose , the equivalent cas were evolved for @xmath111 time steps , which was sufficiently long because @xmath88 showed convergence in such a way that @xmath112 , being the maximum discrepancy between @xmath88 and @xmath113 at the end of the simulation . furthermore the consistency of the @xmath104-values across the members of the different ensembles is also demonstrated in fig . [ histodeltalog ] ( a ) , which depicts the frequency distribution of the standard deviation @xmath114 ( base 10 logarithm ) of @xmath115 calculated for the 60 cryptographic systems at stake .
this histogram shows the similarity of @xmath104 across the different plaintexts , as the standard deviation that comes along with the average mle for the different cryptographic systems is obviously small .
( base 10 logarithm ) of @xmath115 calculated over the 60 combinations of cryptographic systems .
( b ) frequency distribution of @xmath116.,scaledwidth=48.0%,scaledwidth=20.0% ] fig .
[ histodeltalog ] ( b ) shows a histogram of the discrepancies between the maximum and minimum observed mle among the members of the ensemble @xmath117 , denoted by @xmath118 .
the @xmath119-values obtained for the investigated datasets lie between @xmath120 and @xmath121 .
the maximum and minimum @xmath119 correspond , respectively , to the pairs of modes of operation with underlying block cipher : idea - ofb and serpent - cfb .
all together , these small discrepancies demonstrate once more that the @xmath104-values are highly consistent across the different plaintexts within the ensemble @xmath117 , such that we may draw conclusions on their average values . for simplicity , we mainly restrict our attention to idea and aes as representative members of the families of 64- and 128-block ciphers ( see fig .
[ fig : ideaaes ] ) , respectively , since similar results were obtained for the other block ciphers in the same families ( see fig . [
fig : others ] ) .
these figures depict the le curves for all the modes of operation versus the number of time steps note that the standard deviation of each curve is not shown since they are very small . taking into account eq . , the upper bound for the 64-block ciphers is @xmath122 , while it is @xmath123 for the 128-block ciphers were employed to normalize the @xmath104-values , respectively .
c + + from both figures , it is quite easy to discriminate between the modes of operation , except for the overlapping curves of ctr and ofb .
it is interesting to have a closer look at how their corresponding le increases almost exponentially during the first few time steps .
this behaviour can be explained by reconsidering the equations in table [ modestab ] .
indeed , ctr uses an inner pseudo - random number generator and ofb uses a new initialization vector at every time step , such that the number of defective cells will not only increase due to discrepancies naturally emerging between the plaintexts , but also due to the additional defects introduced through these random processes . in contrast , cfb displays the opposite effect , _
i.e. , _ its curve decays exponentially .
further , the lyapunov exponents of ecb and pcbc are constant , since the number of defective cells grows proportionally through time .
for instance , this behaviour can be explained by considering that ecb attains @xmath124 with the idea block cipher , which means that the number of defects @xmath125 equals approximately @xmath126 , being almost 50% of the block length .
the same behaviour occurs for the aes block cipher , where @xmath127 .
the original values of the exponents and standard deviation found at the end of the simulation are summarized in table [ tab : mle2 ] .
furthermore , we can discriminate between modes of operation in families of block ciphers not only by visual inspection of the graphs , but also by means of statistical tests to demonstrate it . paired
t - tests revealed that there are statistically significant differences between ecb , cbc , cfb and pcbc mutually at the 5% significance level .
however , no statistically significant difference was found between ctr and ofb .
these observations were found in both families of 64- and 128-block ciphers , which also allows the proposed method to discriminate between modes of operation of different families , for instance , idea - ctr can be distinguished from aes - ctr .
moreover , by means of the washer method @xcite , we analyzed statistically whether a curve ( for one of the ensemble s members ) for a given mode falls within the envelope of curves obtained for the entire ensemble for another mode of operation .
this procedure was repeated for all the curves within the given mode of operation at the 5% significance level , which indicated that at most 6.4% and 11.1% of the observations of the 64- and 128-block ciphers constituted outliers .
these outliers exist at the beginning of the simulation .
we can also notice that the cfb curve behaves completely different from the other ones ( non overlapping and not too close ) , which means that the cfb falls outside of the range of the le for the other modes of operation . in this section , we will examine and discuss the preceding results in more detail .
[ [ initial - conditions - mathbfp - and - mathbfp ] ] * initial conditions @xmath1 and @xmath128 * + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + recalling that an assessment of the lyapunov exponent involves tracking the ca evolution from two initial configurations for which it holds that @xmath47 and given the fact that we are dealing with booleans , the smallest possible perturbation in such a setting implies flipping the right - most bit of the plaintext .
however , we observed that the position of the flipped bit does not affect the numerical value of the lyapunov exponent , which can be understood by recalling the theoretical upper bound on the lyapunov exponent of cryptographic systems .
indeed , the encryption of the plaintext with a bit flipped at an arbitrary @xmath129 position at the first plaintext block ( @xmath130 ) may , in the worst case , affect all the bits of the ciphertext . [ [ cryptographic - signatures ] ] * cryptographic signatures * + + + + + + + + + + + + + + + + + + + + + + + + + + from figs .
[ fig : ideaaes][fig : others ] it should be noticed that the normalized @xmath104-values do not attain 1 , because none of the cryptographic systems at stake is attaining the worst case scenario mentioned in section [ method ] . this can be explained by the fact that this is an ideal and desirable `` scenario '' for cryptographic systems and that the block ciphers at stake have their limitations .
furthermore , we found an unexpected phenomenon with regard to the families of 64- and 128-block ciphers . a closer look at the curves tied up with these families shows that , in fact , we can also differentiate between block cipher families .
[ [ number - of - blocks ] ] * number of blocks * + + + + + + + + + + + + + + + + + + in order to investigate the importance of the number of blocks @xmath5 , we computed the lyapunov exponents for the same cryptographic systems but with @xmath131 blocks . in fig .
[ fig : comparemodes ] one can see the le curves for each mode of operation using the idea block cipher as a function of @xmath5 . from this figure it is clear that each curve is somehow lifted upwards with the number of blocks @xmath5 . from eq .
, it is expected to get higher @xmath104-values when the length of the plaintext increases .
in fact , this occurs for all the modes of operation at stake , except for ecb .
in particular , this indicates that irrespective of the number of blocks @xmath5 of the plaintext , the ecb s number of defects will attain approximately @xmath126 , which demonstrates that the first block is the only one affected by the initial conditions , while the other blocks do not spread defects .
furthermore , we obtained the same results with both groups of 64- and 128-block ciphers and the six modes of operation .
[ [ empirical - mle ] ] * empirical mle * + + + + + + + + + + + + + + + in section [ method ] we provided some mathematical foundations to obtain the theoretical upper bound @xmath96 for cryptographic systems
. however , the lack of an analytical upper bound for a specific mode of operation is an important limitation remaining .
perhaps , this limitation may seem contradictious to cryptography aims , because it indicates to find an analytical way to obtain the essence of a cryptographic system that is designed to avoid this gap . here
, we opted for an alternative approach , _
i.e. , _ an empirical estimation of the upper bound for the modes of operation as a function of the number of blocks , based on a multiple regression analysis , to gain a deeper insight into the behaviour of the le curves . in fig .
[ fig : upper64 ] we show the mle @xmath104 obtained as a function of @xmath5 for the idea and the corresponding regression curves ( see also fig .
[ fig : upper128 ] for the aes block cipher ) .
the coefficients of determination also demonstrate the strong influence by the number of blocks @xmath5 .
[ [ computational - considerations ] ] * computational considerations * + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + although the algorithm has been enhanced with different strategies to avoid tracking each individual defect , somehow the computational cost of the proposed methodology is very high , since the computational time dramatically increases as the number of blocks of the plaintext @xmath1 grows .
there is a strong relationship between cryptographic systems and discrete dynamical systems . in this work
we have outlined an approach to envisage a cryptographic system as an equivalent one - dimensional ca in order to assess its stability characteristics by computing the lyapunov exponent of the cryptographic system .
the proposed method was capable of distinguishing six cryptographic modes of operation , namely ecb , cbc , ofb , cfb , ctr and pcbc using two families of block ciphers des , idea , tea and xtea of 64 bits , as well as aes , rc6 , twofish , seed , serpent and camellia of 128 bits .
moreover , the proposed method is also capable of distinguishing between the two families of 64- and 128-block ciphers .
the results showed that the lyapunov exponent evolution pattern is maintained for each mode of operation and this is independent of the block cipher used .
we also provided a mathematical basis to obtain the theoretical upper bound @xmath96 for the cryptographic systems .
however , further work is required to theoretically analyze the upper bound on the le for each of the modes of operation . here
we only used an empirical assessment to fit the curves . finally , our results suggest that even modern and contemporary algorithms yield patterns that should be explored .
thus , our theoretical framework may offer a novel alternative to explore the weakness of these cryptographic systems and may ultimately lead to a classification of these systems according to their strength .
j. m. acknowledges support from fapesp ( the state of so paulo research foundation ) ( 2011/05461 - 0 ) .
o. m. b. acknowledges support from cnpq ( grant # 307797/2014 - 7 and # 484312/2013 - 8 ) and fapesp ( grant # 2011/01523 - 1 ) . this work has benefited from the cooperation agreement that has been established between universidade de so paulo and ghent university .
j. nechvatal , e. barker , l. bassham , w. burr , m. dworkin , j. foti , e. roback , report on the development of the advanced encryption standard ( aes ) , tech .
, national institute of standards and technology ( 2000 ) .
r. rivest , m. robshaw , r. sidney , y. l. yin , http://people.csail.mit.edu/rivest/rc6.pdf[the rc6 block cipher ] , tech .
, rsa laboratories and m.i.t .
laboratory for computer science ( 1998 ) .
http://dx.doi.org/http://people.csail.mit.edu/rivest/rc6.pdf [ ] .
j. machicao , a. marco , o. m. bruno , lyapunov exponent : a qualitative ranking of block cipher mode of operation , in : t. gilbert , m. kirkilionis , g. nicolis ( eds . ) , proceedings of the european conference on complex systems , springer proceedings in complexity , springer international publishing , bruxelles , belgium , 2012 , pp . 979985 .
r. schmitz , public key cryptography : a dynamical systems perspective , in : second international conference on emerging security information , systems and technologies , securware 08 , esterel , france , 2008 , pp .
209212 .
d. wheeler , r. needham , tea , a tiny encryption algorithm , in : lecture notes in computer science ( leuven , belgium : fast software encryption : second international workshop ) , vol .
1008 , 1994 , pp . 363366 .
k. aoki , t. ichikawa , m. kanda , m. matsui , s. moriai , j. nakajima , t. tokita , camellia : a 128-bit block cipher suitable for multiple platforms - design and analysis , in : proceedings of the 7th annual international workshop on selected areas in cryptography , sac 00 , springer - verlag , london , uk , 2000 , pp . | evidence of signatures associated with cryptographic modes of operation is established . motivated by some analogies between cryptographic and dynamical systems ,
in particular with chaos theory , we propose an algorithm based on lyapunov exponents of discrete dynamical systems to estimate the divergence among ciphertexts as the encryption algorithm is applied iteratively .
the results allow to distinguish among six modes of operation , namely ecb , cbc , ofb , cfb , ctr and pcbc using des , idea , tea and xtea block ciphers of 64 bits , as well as aes , rc6 , twofish , seed , serpent and camellia block ciphers of 128 bits .
furthermore , the proposed methodology enables a classification of modes of operation of cryptographic systems according to their strength . |
recall that an action of a group @xmath2 on a set @xmath3 is _
@xmath4-transitive _ if @xmath5 and for any two @xmath4-tuples of distinct elements of @xmath3 , @xmath6 and @xmath7 , there exists @xmath8 such that @xmath9 for @xmath10 .
the _ transitivity degree _ of a countable group @xmath2 , denoted @xmath11 , is the supremum of all @xmath12 such that @xmath2 admits a @xmath4-transitive faithful action . for finite groups ,
this notion is classical and fairly well understood .
it is easy to see that @xmath13 , @xmath14 , and it is a consequence of the classification of finite simple groups that any finite group @xmath2 other than @xmath15 or @xmath16 has @xmath17 .
moreover , if @xmath2 is not @xmath15 ,
@xmath16 , or one of the mathieu groups @xmath18 , @xmath19 , @xmath20 , @xmath21 , then @xmath22 ( see @xcite ) . for infinite groups , however , very little is known .
for example , we do not know the answer to the following basic question : does there exist an infinite countable group of transitivity degree @xmath4 for every @xmath12 ?
there are examples for @xmath23 , and @xmath1 , but the problem seems open even for @xmath24 .
there is also a new phenomenon , which does not occur in the finite world : highly transitive actions . recall that an action of a group is _ highly transitive _ if it is @xmath4-transitive for all @xmath12 .
we say that a group is _ highly transitive _ if it admits a highly transitive faithful action ; it is easy to see that a countably infinite group is highly transitive if and only if it embeds as a dense subgroup in the infinite symmetric group @xmath25 endowed with the topology of pointwise convergence .
obviously @xmath26 whenever @xmath2 is highly transitive , but we do not know if the converse is true .
yet another interesting question is whether there exists a reasonable classification of highly transitive groups ( or , more generally , groups of high transitivity degree ) .
the main goal of this paper is to address these questions in certain geometric and algebraic settings .
we begin by discussing known examples of highly transitive groups .
trivial examples of such groups are @xmath27 ( the group of all finitary permutations of @xmath28 ) and the infinite alternating group @xmath29 ( the group of all finitary permutations that act as even permutations on their supports ) . clearly every subgroup of @xmath25 that contains @xmath29 is also highly transitive .
a fairly elementary argument allows one to construct finitely generated groups of this sort ; we record the following .
[ ext ] for every finitely generated infinite group @xmath30 , there exists a finitely generated group @xmath2 such that @xmath31 and @xmath32 .
in particular , @xmath2 is highly transitive .
this proposition can be used , for instance , to construct finitely generated highly transitive torsion groups by taking @xmath30 to be finitely generated and torsion .
another interesting particular example of this sort is the family of houghton groups @xmath33 , @xmath34 , which correspond to the case @xmath35 .
these groups are elementary amenable and finitely presented for @xmath36 ( see example [ hough ] )
. on the other hand , there are highly transitive groups of a completely different nature .
the first such an example is @xmath37 , the free group of rank @xmath34 .
highly transitive faithful actions of @xmath37 were constructed by mcdonough @xcite , and it was shown by dixon @xcite that in some sense most finitely generated subgroups of @xmath38 are free and highly transitive .
other known examples of highly transitive groups include fundamental groups of closed surfaces of genus at least @xmath39 @xcite and , more generally , non - elementary hyperbolic groups without non - trivial finite normal subgroups @xcite , all free products of non - trivial groups except @xmath40 @xcite and many other groups acting on trees @xcite . garion and glassner @xcite used an interesting approach employing tarski monster groups constructed by olshanskii @xcite to show that @xmath41 is highly transitive for @xmath42 . motivated by similarity between @xmath41 and mapping class groups ,
they ask which mapping class groups are highly transitive .
another question from @xcite is whether @xmath43 is highly transitive
. the first main result of our paper is a theorem which simultaneously generalizes most results mentioned in the previous paragraph and allows us to answer the questions of garion and glassner . recall that an isometric action of a group @xmath2 on a metric space @xmath44 is _ acylindrical _ if for all @xmath45 , there exist @xmath46 such that for all @xmath47 with @xmath48 , there are at most @xmath49 elements @xmath8 satisfying @xmath50 and @xmath51 .
for example , it is easy to show that the action of a fundamental group of a graph of groups with finite edge groups on the associated bass - serre tree is acylindrical .
a group is called _ acylindrically hyperbolic _ if it is not virtually cyclic and admits an acylindrical action on a hyperbolic metric space with unbounded orbits @xcite .
the class of acylindrically hyperbolic groups includes many examples of interest ; for an extensive list we refer to @xcite . every acylindrically hyperbolic group @xmath2 has a maximal normal finite subgroup called the _ finite radical _ of @xmath2 and denoted @xmath52 @xcite .
[ main ] every countable acylindrically hyperbolic group admits a highly transitive action with finite kernel .
in particular , every countable acylindrically hyperbolic group with trivial finite radical is highly transitive .
it is well - known and easy to see that a group of transitivity degree at least @xmath39 can not have finite normal subgroups ( see lemma [ 2-trans ] ( b ) ) .
thus the kernel of a highly transitive action of a countable acylindrically hyperbolic group @xmath2 can not be smaller than @xmath52 .
in particular , the assumption about triviality of finite radical can not be dropped from the second sentence of the theorem .
our proof of theorem [ main ] is based on the notion of a small subgroup of an acylindrically hyperbolic group introduced in section 2 .
this notion seems to be of independent interest .
in addition , the proof of theorem [ main ] makes use of hyperbolically embedded subgroups introduced in @xcite and small cancellation theory in acylindrically hyperbolic groups developed in @xcite .
here we mention just few particular cases of our theorem considered in corollaries [ mcg0 ] , [ 3d0 ] , and [ raag][1-rel ] . by @xmath53
we denote a @xmath54 times punctured compact orientable surface of genus @xmath55 with @xmath56 boundary components .
[ maincor-1 ] 1 .
let @xmath2 be a countable group hyperbolic relative to a collection of proper subgroups .
then @xmath2 is highly transitive if and only if it is not virtually cyclic and has no non - trivial finite normal subgroups .
2 . for @xmath57 ,
the mapping class group @xmath58 is highly transitive if and only if @xmath59 , @xmath60 , and @xmath61 .
* theorem 1 ) ) @xmath41 is highly transitive if and only if @xmath36 .
4 . let @xmath62 be a compact irreducible @xmath63-manifold .
then @xmath64 is highly transitive if and only if it is not virtually solvable and @xmath62 is not seifert fibered .
a right angled artin group is highly transitive if and only if it is non - cyclic and directly indecomposable . 6 .
every @xmath0-relator group with at least @xmath63 generators is highly transitive .
recall that a free product of two non - trivial groups has no non - trivial finite normal subgroups .
thus part ( a ) of corollary [ maincor-1 ] covers the case of free products considered in @xcite as well as some results of @xcite ; it also covers the result about hyperbolic groups @xcite .
parts ( b ) and ( c ) answer the questions of garion and glassner mentioned above .
our contribution to ( c ) is the case @xmath65 .
theorem [ main ] also allows us to show that in various geometric and algebraic settings , transitivity degree of an infinite group can only take two values , namely @xmath0 and @xmath1 , and infinite transitivity degree is equivalent to being highly transitive .
more precisely , we consider three conditions for a group @xmath2 : 1 .
@xmath2 is highly transitive .
@xmath2 is acylindrically hyperbolic with trivial finite radical . in section 4 , we show that an infinite subgroup @xmath2 of a mapping class group is either acylindrically hyperbolic or satisfies certain algebraic conditions which imply @xmath66 .
a similar result for subgroups of @xmath63-manifold groups was obtained in @xcite . combining this with theorem [ main ] , we obtain the following .
let @xmath2 be an infinite subgroup of @xmath58 for some @xmath67 or an infinite subgroup of the fundamental group of a compact @xmath63-manifold .
then @xmath68 and conditions ( c@xmath69)(c@xmath70 ) are equivalent . a similar result can also be proved in some algebraic settings , e.g. , for subgroups of finite graph products .
let @xmath71 be a graph ( without loops or multiple edges ) with vertex set @xmath72 and let @xmath73 be a family of groups indexed by vertices of @xmath74 .
the _ graph product _ of @xmath73 with respect to @xmath71 , denoted @xmath75 , is the quotient group of the free product @xmath76 by the relations @xmath77=1 $ ] for all @xmath78 , @xmath79 whenever @xmath80 and @xmath81 are adjacent in @xmath74 .
graph products simultaneously generalize free and direct products of groups .
basic examples are right angled artin and coxeter groups , which are graph products of copies of @xmath82 and @xmath83 , respectively .
the study of graph products and their subgroups has gained additional importance in view of the recent breakthrough results of agol , haglund , wise , and their co - authors , which show that many groups can be virtually embedded into right angled artin groups ( see @xcite and references therein ) . [ maincor2 ] let @xmath2 be a countably infinite subgroup of a finite graph product @xmath75 .
suppose that @xmath2 is not isomorphic to a subgroup of one of the multiples .
then @xmath84 and conditions ( c@xmath69)(c@xmath70 ) are equivalent . if , in addition , every @xmath85 is residually finite , then @xmath68 .
to state our next result we need some preparation .
let @xmath37 denote the free group of rank @xmath56 and recall that a group @xmath2 satisfies a _ mixed identity _
@xmath86 for some @xmath87 if every homomorphism @xmath88 that is identical on @xmath2 sends @xmath89 to @xmath0 .
we say that the mixed identity @xmath86 is non - trivial if @xmath90 as an element of @xmath91 .
for the general theory of mixed identities and mixed varieties of groups we refer to @xcite .
we say that @xmath2 is _ mixed identity free _ ( or _ mif _ for brevity ) if it does not satisfy any non - trivial mixed identity .
the property of being mif is much stronger than being identity free and imposes strong restrictions on the algebraic structure of @xmath2 .
for example , if @xmath2 has a non - trivial center , then it satisfies the non - trivial mixed identity @xmath92=1 $ ] , where @xmath93 .
similarly , it is easy to show that an mif group has no finite normal subgroups , is directly indecomposable , has infinite girth , etc .
( see proposition [ mif - prop ] ) .
other examples of groups satisfying a non - trivial mixed identity are thompson s group @xmath94 @xcite or any subgroup of @xmath95 containing @xmath29 ( see theorem [ dichotomy ] below ) .
it is also worth noting that mif groups resemble free products from the model theoretic point of view .
more precisely , a countable group @xmath2 is mif if and only if @xmath2 and @xmath91 are universally equivalent as @xmath2-groups for all @xmath96 .
this means that a universal first order sentence with constants from @xmath2 holds true in @xmath2 if and only if it holds true in @xmath91 .
groups universally equivalent to @xmath2 as @xmath2-groups are exactly the coordinate groups of irreducible algebraic varieties over @xmath2 @xcite . for more details
we refer to section 5 and proposition [ mif ] .
+ in section 5 , we prove the following dichotomy for highly transitive countable groups , which captures the principal difference between examples provided by proposition [ ext ] and theorem [ main ] .
[ dichotomy ] let @xmath2 be a highly transitive countable group
. then exactly one of the following two mutually exclusive conditions holds . 1 .
@xmath2 contains a normal subgroup isomorphic to @xmath29 .
2 . @xmath2 is mif ( equivalently , @xmath2 and @xmath91 are universally equivalent as @xmath2-groups for every @xmath96 ) . it is known that infinite normal subgroups of acylindrically hyperbolic groups are also acylindrically hyperbolic and hence can not be torsion @xcite .
thus condition ( a ) from theorem [ dichotomy ] can not hold if @xmath2 is acylindrically hyperbolic . combining this with theorem [ main ]
we obtain the following .
let @xmath2 be an acylindrically hyperbolic group with trivial finite radical .
then @xmath2 is mif .
clearly @xmath2 is not mif whenever @xmath97 . in the particular case of non - cyclic torsion free hyperbolic groups ,
the above corollary was proved in @xcite by different methods .
theorem [ main ] and theorem [ dichotomy ] are summarized in fig .
[ classes ] . here by @xmath98
we denote the class of acylindrically hyperbolic groups with trivial finite radical . in section 5
we show that all inclusions are proper .
we mention two examples .
[ prop - ex ] 1 .
the group @xmath99=1\rangle$ ] is highly transitive and mif , but not acylindrically hyperbolic .
2 . there
exist finitely generated mif groups of transitivity degree @xmath0 .
the paper is organized as follows : in section 2 we review properties of acylindrically hyperbolic groups and hyperbolically embedded subgroups necessary for the proof of theorem [ main ] . in section 3
we prove theorem [ main ] , and in section 4 we apply this theorem to various classes of groups to obtain corollaries [ maincor-1]-[maincor2 ] . in section 5
we study the relationship between mixed identities and highly transitive actions and prove theorem [ dichotomy ] .
some open questions and relevant examples are discussed in section 6 .
[ [ acknowledgments . ] ] acknowledgments .
+ + + + + + + + + + + + + + + + we would like to thank ilya kapovich , olga kharlampovich , and alexander olshanskii for helpful discussions of various topics related to this paper .
we are especially grateful to dan margalit for answering our numerous questions on mapping class groups .
the second author was supported by the nsf grant dms-1308961 .
[ [ generating - alphabets - and - cayley - graphs . ] ] 2.1 .
generating alphabets and cayley graphs .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + when dealing with relative presentations of groups , we often need to represent the same element of a group @xmath2 by several distinct generators .
thus , instead of a generating set of @xmath2 , it is more convenient to work with an alphabet @xmath100 given together with a ( not necessarily injective ) map @xmath101 such that @xmath102 generates @xmath2 .
we begin by formalizing this approach .
let @xmath100 be a set , which we refer to as an _ alphabet_. let @xmath2 be a group and let @xmath101 be a ( not necessarily injective ) map .
we say that @xmath2 is _ generated _ by @xmath100 ( or @xmath100 is a _ generating alphabet _ of @xmath2 ) if @xmath2 is generated by @xmath102 .
note that a generating set @xmath103 can be considered as a generating alphabet with the obvious map @xmath104 . by the _ cayley graph _ of @xmath2 with respect to a generating alphabet @xmath100 , denoted @xmath105 , we mean a graph with vertex set @xmath2 and the set of edges defined as follows .
for every @xmath106 and every @xmath8 , there is an oriented edge @xmath107 in @xmath105 labelled by @xmath108 . given a ( combinatorial ) path @xmath54 in @xmath105 , we denote by @xmath109 its label .
note that if @xmath110 is not injective , @xmath105 may have multiple edges .
of course , the identity map on @xmath2 induces an isometry between vertex sets of the graphs @xmath105 and @xmath111 .
thus we only need to distinguish between @xmath105 and @xmath111 when dealing with labels ; in all purely metric considerations we do not need generating alphabets and can simply work with generating sets . given a generating set @xmath112 of @xmath2 and an element @xmath8 , let @xmath113 denote the word length of @xmath55 with respect to @xmath112 , that is the length of a shortest word in @xmath114 that represents @xmath55 . for @xmath115 , we define @xmath116 .
finally , let @xmath117 [ [ hyperbolically - embedded - subgroups . ] ] 2.2 .
hyperbolically embedded subgroups .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the typical situation when we apply the language described above is the following .
suppose that we have a group @xmath2 , a subgroup @xmath118 of @xmath2 , and a subset @xmath103 such that @xmath112 and @xmath118 together generate @xmath2 .
we think of @xmath112 and @xmath118 as abstract sets and consider the disjoint union @xmath119 and the map @xmath120 induced by the obvious maps @xmath121 and @xmath122 . by abuse of notation
, we do not distinguish bretween subsets @xmath112 and @xmath118 of @xmath2 and their preimages in @xmath100 .
this will not create any problems since the restrictions of @xmath123 on @xmath112 and @xmath118 are injective .
note , however , that @xmath110 is not necessarily injective . indeed if @xmath112 and subgroup @xmath118 intersect in @xmath2 , then every element of @xmath124 will have at least two preimages in @xmath100 : one in @xmath112 and another in @xmath118 ( since we use disjoint union in ( [ cala ] ) ) .
henceforth we always assume that generating sets and relative generating sets are symmetric .
that is , if @xmath125 , then @xmath126 . in particular
, every element of @xmath2 can be represented by a word in @xmath127 . in these settings
, we consider the cayley graphs @xmath128 and @xmath129 and naturally think of the latter as a subgraph of the former .
we introduce a _ relative metric _ @xmath130 $ ] as follows .
we say that a path @xmath54 in @xmath128 is _ admissible _ if it contains no edges of @xmath129 .
let @xmath131 be the length of a shortest admissible path in @xmath128 that connects @xmath132 to @xmath4 .
if no such a path exists , we set @xmath133 .
clearly @xmath134 satisfies the triangle inequality , where addition is extended to @xmath135 $ ] in the natural way .
[ hedefn ] a subgroup @xmath118 of @xmath2 is _ hyperbolically embedded in @xmath2 with respect to a subset @xmath103 _ , denoted @xmath136 , if the following conditions hold . 1 .
the group @xmath2 is generated by @xmath112 together with @xmath118 and the cayley graph @xmath128 is hyperbolic .
any ball ( of finite radius ) in @xmath118 with respect to the metric @xmath137 contains finitely many elements .
further we say that @xmath118 is _ hyperbolically embedded _ in @xmath2 and write @xmath138 if @xmath139 for some @xmath103 .
hyperbolically embedded subgroups were introduced and studied in @xcite as a generalizaton of relatively hyperbolic groups ; indeed , @xmath2 is hyperbolic relative to @xmath118 if and only if @xmath140 with @xmath141 ( * ? ? ?
* proposition 4.28 ) .
the following lemma is a particular case of this result .
[ heprod ] for any group @xmath2 , @xmath142 . in the next two lemmas
, we let @xmath2 be a group and @xmath118 a subgroup of @xmath2 . the following is a particular case of ( * ? ? ?
* proposition 4.35 ) . [ hehe ] suppose that @xmath140 for some @xmath103 and there is a subset @xmath143 and a subgroup @xmath144
. then @xmath145 .
given a group @xmath2 , a subgroup @xmath146 , and @xmath8 , we denote by @xmath147 the conjugate @xmath148 .
[ malnorm ] ( * ? ? ?
* proposition 2.8 ) if @xmath138 , then for all @xmath149 , @xmath150 . [ [ acylindrically - hyperbolic - groups . ] ] 2.3 .
acylindrically hyperbolic groups .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + recall that an isometric action of a group @xmath2 on a metric space @xmath151 is _ acylindrical _ if for every @xmath45 , there exist @xmath152 such that for every two points @xmath153 with @xmath154 , there are at most @xmath49 elements @xmath8 satisfying @xmath155 given a group @xmath2 acting on a hyperbolic space @xmath151 , an element @xmath8 is called _ loxodromic _ if the map @xmath156 defined by @xmath157 is a quasi - isometry for some ( equivalently , any ) @xmath158 .
every loxodromic element @xmath8 has exactly @xmath39 limit points @xmath159 on the gromov boundary @xmath160 .
loxodromic elements @xmath161 are called _ independent _ if the sets @xmath162 and @xmath163 are disjoint .
we will often use the following .
[ tri ] let @xmath2 be a group acting acylindrically on a hyperbolic space .
then exactly one of the following three conditions holds . 1
. @xmath2 is elliptic .
2 . @xmath2 contains a loxodromic element @xmath55 such that @xmath164 .
@xmath2 contains infinitely many loxodromic elements that are independent ( i.e. , have disjoint limit sets on the boundary ) .
[ df : acyl - gp ] a group @xmath2 is called _ acylindrically hyperbolic _ if it admits a non - elementary acylindrical action on a hyperbolic space . in the case of acylindrical actions on hyperbolic spaces being non - elementary is equivalent to the action having unbounded orbits and @xmath2 being not virtually cyclic by theorem [ tri ] .
it is easy to see from the definition that for every group @xmath2 we have @xmath165 and @xmath166 for every finite subgroup @xmath167 of @xmath2 .
following @xcite , we call a hyperbolically embedded subgroup @xmath138 _ non - degenerate _ if @xmath118 is infinite and @xmath168 .
the next result is a part of ( * ? ? ? * theorem 2.2 ) . in particular
, it allows us to apply all results from @xcite concerning groups with non - degenerate hyperbolically embedded subgroups to acylindrically hyperbolic groups . [ ah - hes ] a group @xmath2 is acylindrically hyperbolic if and only if @xmath2 contains a non - degenerate hyperbolically embedded subgroup .
if @xmath2 is acylindrically hyperbolic , then @xmath2 has a unique , maximal finite normal subgroup @xmath52 called the _ finite radical _ of @xmath2 ( * ? ? ?
* theorem 6.14 ) .
let @xmath2 be acylindrically hyperbolic .
then @xmath169 is acylindrically hyperbolic and has trivial finite radical .
[ [ small - subgroups - in - acylindrically - hyperbolic - groups . ] ] 2.4 .
small subgroups in acylindrically hyperbolic groups .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + given a group @xmath2 and a generating set ( or an alphabet ) @xmath100 , we say that the cayley graph @xmath170 is _ acylindrical _ if so is the action of @xmath2 on @xmath170 .
a hyperbolic space @xmath151 is called _ non - elementary _ if @xmath171 , where @xmath160 denotes the gromov boundary of @xmath151 .
we call a subgroup @xmath118 of a group @xmath2 _ small in @xmath2 _ ( or just _ small _ if @xmath2 is understood ) if there exists a generating set @xmath112 of @xmath2 such that @xmath172 and @xmath173 is hyperbolic , non - elementary , and acylindrical
. note that every group containing small subgroups is acylindrically hyperbolic .
[ faith ] assume that a group @xmath2 is acylindrically hyperbolic and @xmath174 .
then for every small subgroup @xmath146 , the action of @xmath2 on the coset space @xmath175 is faithful .
let @xmath49 denote the kernel of the action of @xmath2 on @xmath175 .
let @xmath112 be a generating set of @xmath2 such that @xmath172 and @xmath173 is acylindrical , hyperbolic , and non - elementary .
note that @xmath49 acts elliptically on @xmath173 as @xmath176 .
recall that for every acylindrical non - elementary action of a group on a hyperbolic space , every normal elliptic subgroup is finite ( * ? ? ?
* lemma 7.2 ) . applying this to the group @xmath49 acting on @xmath173
, we obtain that @xmath177 and consequently @xmath178 .
the proof of theorem [ main ] will make use of one particular example of small subgroups .
recall that given a generating set @xmath112 of a group @xmath2 , @xmath179 and @xmath180 denote the corresponding word length and metric on @xmath2 , respectively .
[ small ] let @xmath2 be a group , @xmath94 a subgroup of @xmath2 , @xmath112 a subset of @xmath2 such that @xmath181 .
suppose that a subgroup @xmath146 is generated by a subset @xmath182 such that @xmath183 and @xmath184 for all @xmath185 .
then @xmath118 is small .
moreover , for every subgroup @xmath186 that properly contains @xmath94 , there exists a subset @xmath187 such that @xmath188 is hyperbolic and acylindrical , @xmath189 , and the action of @xmath151 on @xmath188 is non - elementary .
the proof can be easily extracted from @xcite . in order to make our paper
self - contained , we recall necessary definitions and results from @xcite .
let @xmath190 and let @xmath54 be a ( combinatorial ) path in @xmath105 .
a non - trivial subpath @xmath191 of @xmath54 is called an _ @xmath94-component _
if @xmath192 is a word in the alphabet @xmath94 and @xmath191 is not contained in any bigger subpath of @xmath54 with this property .
if @xmath191 is an @xmath94-component of @xmath54 and @xmath193 are endpoints of @xmath191 , then all vertices of @xmath191 belong to the same coset @xmath194 ; in this case we say that @xmath54 _ penetrates _ @xmath195 and call the number @xmath196 the _ depth of the penetration_. given two elements @xmath197 , we say that a coset @xmath195 is _ @xmath198-separating _ if there exists a geodesic @xmath54 from @xmath199 to @xmath55 in @xmath170 such that @xmath54 penetrates @xmath195 and the depth of the penetration is greater than @xmath200 .
the term separating " is justified by the following result , which is a simplification of ( * ? ? ? * lemma 4.5 ) .
[ sep ] let @xmath2 be a group , @xmath94 a subgroup of @xmath2 , @xmath112 a subset of @xmath2 such that @xmath181 .
then for any constants @xmath201 and @xmath202 , there exists @xmath203 such that the following holds .
let @xmath197 and let @xmath195 be an @xmath204-separating coset . then every @xmath205-quasi - geodesic in @xmath206 that connects @xmath199 to @xmath55 penetrates @xmath195 .
it is observed in @xcite that if @xmath181 , then the cayley graph @xmath207 is not necessarily acylindrical .
however , the following lemma holds true .
its first claim is a particular case of theorem 5.4 in @xcite ; the second claim is not stated in that theorem itself , but is obvious from its proof .
[ y ] let @xmath2 be a group , @xmath94 a subgroup of @xmath2 , @xmath112 a subset of @xmath2 such that @xmath181 .
then there exists a set @xmath208 such that @xmath209 and @xmath210 is acylindrical .
more precisely , given a constant @xmath211 , let @xmath212 denote the set of all elements @xmath213 such that the set of all @xmath214-separating cosets is empty .
then for every large enough @xmath200 , the set @xmath215 satisfies the above requirements .
we are now ready to prove proposition [ small ] .
let @xmath216 .
for every element @xmath217 , let @xmath218 be a shortest path in @xmath219 connecting @xmath0 to @xmath132 such that @xmath220 is a word in the alphabet @xmath112 .
in particular , @xmath218 does not penetrate any left coset of @xmath94 .
it is straightforward to check that @xmath218 is @xmath205-quasi - geodesic for some @xmath221 and @xmath202 that only depend on @xmath167 and @xmath62 ( in fact , we can take @xmath222 and @xmath223 ) .
let @xmath200 be a constant such that the conclusion of lemma [ y ] holds and @xmath224 , where @xmath225 is given by lemma [ sep ] .
then for any @xmath217 , there are no @xmath226-separating cosets .
therefore , the set @xmath227 defined in lemma [ y ] contains @xmath118 . by lemma [ y ]
, @xmath210 is acylindrical .
it remains to show that the action of @xmath151 on @xmath228 is non - elementary . by theorem [ tri ]
, it suffices to show that @xmath151 has unbounded orbits and is not virtually cyclic .
let @xmath229 . by ( * ?
* corollary 6.12 ) , there exists @xmath230 such that @xmath231 is loxodromic with respect to the action on @xmath228 .
since @xmath232 , we obtain that the orbits of @xmath151 are unbounded .
finally assume that @xmath151 is elementary .
since @xmath94 is infinite , it must be of finite index in @xmath151 .
in particular , @xmath94 contains an infinite subgroup @xmath49 such that @xmath233 .
then for every @xmath234 , we obtain @xmath235 which contradicts lemma [ malnorm ] if @xmath236 .
thus the action of @xmath151 on @xmath228 is non - elementary .
[ [ outline - of - the - proof . ] ] 3.1 .
outline of the proof .
+ + + + + + + + + + + + + + + + + + + + + + + + + + the goal of this section is to prove theorem [ main ] .
we begin with a brief sketch of the proof .
we first note that by lemma [ finrad ] , it suffices to deal with the case of trivial finite radical .
assuming @xmath174 , we will construct a subgroup @xmath237 such that the action of @xmath2 on the left cosets of @xmath238 will be highly transitive .
@xmath238 will be defined as the union of a sequence of subgroups @xmath239 given @xmath240 and two tuples of pairwise distinct cosets @xmath241 and @xmath242 , we define @xmath243 by choosing an element @xmath244 and setting @xmath245 then clearly @xmath246 , so after enumerating all tuples of elements of @xmath2 and repeating this process inductively we will get that the action of @xmath2 on the set @xmath247 of left cosets of @xmath238 is highly transitive , provided @xmath248 . to the best of our knowledge
, this natural construction first appeared in @xcite .
it was also used by chaynikov @xcite to prove theorem [ main ] in the particular case when @xmath2 is hyperbolic .
the main difficulty is to ensure that that the action of @xmath2 on @xmath3 is faithful .
note that if the elements @xmath249 is not chosen carefully , then we are likely get @xmath250 . in chainikov
s paper @xcite , the proof of faithfulness of the action is built on the well - developed theory of quasi - convex subgroups of hyperbolic groups . currently there is no analogue of this theory in the more general context of acylindrically hyperbolic groups , so we choose another approach based on the notion of a small subgroup introduced in the previous section and small cancellation theory in acylindrically hyperbolic groups developed in @xcite . more precisely , we prove in lemma [ indstep ] that if @xmath240 is small for some @xmath251 , then by choosing the element @xmath249 carefully ( small cancellation theory is employed here ) one can ensure that the subgroup @xmath243 defined by ( [ hi+1 ] ) is still small .
in addition , we show that is @xmath252 is any finite subset of @xmath2 disjoint from @xmath240 , we can keep it disjoint from @xmath243 .
recall also that the action of @xmath2 on the space of left cosets @xmath175 is faithful whenever @xmath118 is a small subgroup and @xmath253 ( lemma [ faith ] ) .
these results allow us to iterate the above construction in such a way that the resulting action of @xmath2 on @xmath254 is faithful .
[ [ the - inductive - step . ] ] 3.2 . the inductive step . + + + + + + + + + + + + + + + + + + + + + + + + the main goal of this subsection is to prove the following proposition , which takes care of the inductive step in the proof of theorem [ main ] . given a group @xmath2 , a subgroup @xmath118 , and two collections of elements @xmath255 and @xmath256 of @xmath2
, we say that the triple @xmath257 is _ admissible _ if @xmath258 for all @xmath259 .
[ indstep ] let @xmath2 be a group , @xmath118 a small subgroup of @xmath2 , @xmath260 and @xmath261 elements of @xmath262 .
assume that the triple @xmath257 is admissible .
let also @xmath263 be a finite subset disjoint from @xmath118 .
then there exists @xmath8 and a subgroup @xmath264 such that 1 .
2 . @xmath167 is small in @xmath2 .
@xmath266 for all @xmath267 .
4 . @xmath268 .
we first recall some well - known results and terminology concerning free products .
* chapt iv , theorem 1.2)[nform ] let @xmath269 , @xmath270 be arbitrary groups .
each element of a free product @xmath271 can be uniquely expressed as @xmath272 where @xmath273 , each @xmath249 is a non - trivial element of one of the factors and successive elements @xmath274 belong to different factors .
the expression @xmath272 as in the above theorem representing an element @xmath275 is called the _ normal form _ of the element @xmath55 .
the following corollary is an immediate consequence of the uniqueness of normal forms .
[ flength ] let @xmath275 and let @xmath276 be a generating set for @xmath277 .
let @xmath272 be the normal form for @xmath55 , where @xmath278 .
then @xmath279 in particular , the embeddings of metric spaces @xmath280 are isometric . [ undisfp ]
let @xmath2 be a group with a generating set @xmath281 , @xmath118 a subgroup of @xmath2 such that @xmath282 .
let @xmath255 , @xmath283 be elements of @xmath262 .
assume that the triple @xmath257 is admissible and let @xmath284 , @xmath285 and @xmath286 then @xmath287 and the natural map @xmath288 is lipschitz .
first , if @xmath289 , then @xmath290 . assume now that @xmath291 .
let @xmath292 for @xmath293 and let @xmath294 be a minimal word in @xmath182 representing @xmath132 .
then @xmath295 for some @xmath296 , where each @xmath297 is a ( possibly trivial ) element of @xmath118 and @xmath298 .
then @xmath299 , and note that each @xmath300 .
let @xmath301 , where @xmath302 for @xmath303 and @xmath304 for @xmath305 .
we also define @xmath306 .
then @xmath307 we show now that consecutive @xmath308-letters in this expression do not cancel , which means that ( [ normform ] ) becomes a normal form for @xmath132 in the free product @xmath309 after possibly removing the trivial @xmath310 and combining adjacent powers of @xmath308 .
suppose for some @xmath311 , @xmath312 and @xmath313 . then @xmath314 .
if @xmath315 , then @xmath316 since @xmath317 . if @xmath318 , then @xmath319 , for otherwise @xmath320 and @xmath321 would freely cancel in the word @xmath294 .
hence @xmath316 since it is a conjugate of @xmath297 in this case . by a similar argument
, it follows that if @xmath322 and @xmath323 , then @xmath316 .
thus , if @xmath324 , then @xmath325 and hence the corresponding @xmath308-letters can not cancel with each other .
since the normal form for @xmath132 contains at least one @xmath308-letter , theorem [ nform ] gives that @xmath326 and hence @xmath287 .
furthermore , by corollary [ flength ] , @xmath327 , it follows that @xmath328 by definition , a group is acylindrically hyperbolic if it admits a non - elementary acylindrical action on some hyperbolic metric space . in @xcite
it is shown that an acylindrically hyperbolical group g always has such an action on a hyperbolic space of the form @xmath329 where @xmath281 is a generating set of @xmath2 .
let @xmath281 be a generating set of @xmath2 such that @xmath329 is hyperbolic and acylindrical .
a subgroup @xmath330 is called _ suitable with respect to a _ if the action of @xmath151 on @xmath329 is non - elementary and @xmath151 does not normalize any non - trivial finite subgroup of @xmath2 .
the following is a modification of ( * ? ? ?
* theorem 7.1 ) ; note that condition ( a ) is a simplification of ( * ? ? ?
* theorem 7.1 ( d ) ) and condition ( b ) is an immediate consequence of the proof of ( * ? ? ?
* theorem 7.1 ) . recall that @xmath331 denotes the ball of radius @xmath49 in @xmath2 with respect to the metric @xmath332 .
[ scthm ] let @xmath281 be a generating set of @xmath2 such that @xmath329 is hyperbolic and acylindrical , and let @xmath330 be suitable with respect to @xmath281 .
let @xmath333 and @xmath334 .
then there exists a group @xmath335 and a surjective homomorphism @xmath336 such that 1 .
there exists a generating set @xmath337 of @xmath335 such that @xmath338 and @xmath339 is hyperbolic , acylindrical , and non - elementary .
@xmath340 is injective .
@xmath341 for some @xmath342 .
in particular , @xmath343 .
we are now ready to prove the main result of this subsection .
since @xmath118 is small , there exists a generating set @xmath103 such that @xmath172 and @xmath173 is hyperbolic , acylindrical , and non - elementary . by theorem [ tri ]
, there exists @xmath8 loxodromic with respect to the action on @xmath173
. then @xmath344 ( ( * ? ? ?
* corollary 3.16 ) ) and hence @xmath345 by lemma [ heprod ] and lemma [ hehe ] .
let @xmath346 and let @xmath347 , @xmath348 , and @xmath182 be as in lemma [ undisfp ] .
obviously @xmath182 is bounded with respect to the metric @xmath349 . by lemmas [ undisfp ] and
[ small ] applied to the group @xmath347 , hyperbolically embedded subgroup @xmath350 , generating set @xmath351 , and subgroups @xmath352 and @xmath353 , we can chose a generating set @xmath215 of @xmath347 such that @xmath354 , @xmath355 is acylindrical , hyperbolic , and the action of @xmath2 on @xmath355 is non - elementary .
since @xmath253 , @xmath2 does not normalize any non - trivial finite normal subgroup of @xmath347 and hence @xmath2 is suitable with respect to @xmath215 . since @xmath252 is finite
, we can choose @xmath333 such that @xmath356 .
applying theorem [ scthm ] gives a group @xmath357 and a surjective homomorphism @xmath358 . by condition ( c ) of theorem [ scthm ] and the obvious tietze transformation , @xmath359 .
in particular , the composition of the natural embedding @xmath360 and @xmath361 is the identity map on @xmath2 .
let @xmath362 and @xmath363 . since @xmath364 , @xmath365 . by construction @xmath366 for @xmath293
, so the definition of @xmath55 and @xmath167 gives that @xmath266 for @xmath293 .
since @xmath354 , @xmath367 and condition ( a ) of theorem [ scthm ] gives that @xmath167 is small in @xmath2 .
since @xmath263 and @xmath252 is disjoint from @xmath118 , @xmath252 is also disjoint from @xmath348 by lemma [ undisfp ] .
finally , @xmath368 , so condition ( c ) of theorem [ scthm ] gives that @xmath369 .
[ [ proof - of - theorem - main . ] ] 3.3 .
proof of theorem [ main ] .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + below we employ the standard notation @xmath370 for elements @xmath153 of a group . by lemma [ finrad ] it suffices to prove the theorem in the case @xmath253 .
let @xmath371 if @xmath372 , @xmath373 , and @xmath8 , we write @xmath374 to mean @xmath375 for all @xmath267 .
we also enumerate all elements of @xmath2 : @xmath376 we will construct a sequence of subgroups @xmath377 of @xmath2 and a sequence of elements @xmath378 of @xmath2 such that the following conditions hold for every @xmath379 . 1 .
@xmath240 is small in @xmath2 .
2 . for each @xmath380 , we have @xmath381 .
if @xmath382 and the triple @xmath383 is admissible , then there exists @xmath384 such that @xmath385 .
we proceed by induction .
set @xmath386 .
assume now that we have already constructed subgroups @xmath387 satisfying ( a)(c ) for some @xmath388 . the action of @xmath2 on the coset space @xmath389 is faithful by ( a ) and lemma [ faith ]
. hence there exists @xmath390 such that @xmath391 .
let @xmath392 note that @xmath252 is disjoint from @xmath240 by part ( b ) of the inductive assumption and the choice of @xmath393 .
let @xmath394 and @xmath395 . if the triple @xmath396 is not admissible , we set @xmath397 .
all inductive assumptions obviously hold in this case .
henceforth we assume that @xmath396 is admissible
. then we can apply proposition [ indstep ] to the small subgroup @xmath398 and let @xmath399 .
conditions ( a)(c ) for @xmath243 follow immediately from the proposition .
let @xmath400 , and let @xmath401 denote the corresponding coset space .
we first notice that the action of @xmath2 on @xmath247 is faithful .
indeed assume that some element @xmath310 acts trivially .
then all conjugates of @xmath310 by elements of @xmath2 belong to @xmath402 . in particular , @xmath403 and
hence @xmath404 for all sufficiently large @xmath251 .
however this contradicts ( b ) .
further , for any two collections of pairwise distinct cosets @xmath405 and @xmath406 , there exists @xmath251 such that @xmath407 , @xmath408 .
since @xmath409 , the triple @xmath410 is admissible .
then by ( c ) we have @xmath411 for all @xmath412 , which obviously implies @xmath413 .
thus the action of @xmath2 on @xmath3 is highly transitive .
[ [ preliminaries - on - permutation - groups . ] ] 4.1 .
preliminaries on permutation groups .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we begin by reviewing some basic concepts and facts about multiply transitive groups used in this section . for more details
we refer the reader to @xcite . by countable " we mean finite or countably infinite .
let @xmath3 denote a countable set and let @xmath414 denote the symmetric group on @xmath3 .
we will also use the notation @xmath38 for the symmetric group on a countable infinite set .
a permutation group @xmath415 is _ @xmath4-transitive _ if the action of @xmath2 on @xmath247 is @xmath4-transitive ; @xmath2 is _ primitive _ if it is transitive and does not preserve any non - trivial equivalence relation ( or partition ) on @xmath3 . here an equivalence relation on @xmath247 is called _ trivial _ , if it is equality or if all elements of @xmath247 are equivalent . in the following two lemmas ,
we collect some well - known properties of primitive and @xmath39-transitive groups which will be used many times throughout this section . since the proofs are short and fairly elementary , we provide them for convenience of the reader .
[ prim ] let @xmath415 be a primitive permutation group . 1 .
every non - trivial normal subgroup of @xmath2 is transitive . in particular ,
if @xmath2 is infinite then it has no non - trivial finite normal subgroups .
2 . for every @xmath416 , @xmath417 is a maximal subgroup of @xmath2 .
let @xmath2 be primitive and @xmath418 .
the collection of @xmath49-orbits defines a @xmath2-invariant partition of @xmath3 , hence by primitivity this partition is trivial .
this means that either @xmath49 acts transitively on @xmath3 or @xmath419 .
further , let @xmath420 .
then the @xmath2-translates of the orbit @xmath421 form a @xmath2-invariant partition of @xmath3 .
by primitivity either @xmath422 or @xmath423 . in the former case we have @xmath424 ; in the latter case @xmath425 .
recall that a group @xmath2 has _
infinite conjugacy classes _ , abbreviated _ icc _ , if every nontrivial conjugacy class of @xmath2 is infinite .
note that icc groups can not have non - trivial center or non - trivial finite normal subgroups .
[ 2-trans ] let @xmath415 be a @xmath39-transitive permutation group . 1 .
@xmath2 is primitive .
2 . if @xmath2 is infinite , then it is icc .
part ( a ) is obvious . to prove ( b ) , let @xmath426 and let @xmath427 be distinct elements of @xmath3 .
( note that @xmath247 is infinite since so is @xmath2 . )
let also @xmath428 be an element of @xmath3 such that @xmath429 . since @xmath2 is 2-transitive , for all @xmath430 there exists @xmath431 such that @xmath432 and @xmath433 .
then @xmath434,which means that @xmath435 are all distinct conjugates of @xmath55 .
hence @xmath2 is icc .
the next lemma is also quite elementary .
it will be used to bound transitivity degree of various groups from above .
[ min ] let @xmath436 be an infinite primitive permutation group , @xmath437 a transitive subgroup of @xmath347 , and @xmath438 a non - trivial normal subgroup of @xmath347 .
suppose that @xmath439=1 $ ] .
then @xmath438 is an infinite minimal normal subgroup of @xmath347 .
let @xmath440 be a non - trivial subgroup of @xmath438 such that @xmath441 .
consider any @xmath442 and fix some @xmath443 . by part ( a ) of lemma
[ prim ] , @xmath440 is transitive
. hence there exists @xmath444 such that @xmath445 .
that is , @xmath446 for @xmath447 . since @xmath437 is transitive , for every @xmath448 there is @xmath449 such that @xmath450 .
since @xmath439=1 $ ] we obtain @xmath451 thus @xmath452 fixes @xmath247 pointwise , i.e. , @xmath453 , which in turn implies @xmath454 . as
this is true for every @xmath442 , we obtain @xmath455 . it remains to note that @xmath438 is infinite by lemma [ prim ] ( a ) .
we say that a group @xmath2 is _ product - like _ if @xmath2 contains non - trivial normal subgroups @xmath456 such that @xmath457 .
the term product - like " reflects the fact that such groups often occur as subgroups of direct products having non - trivial intersections with multiples . a group @xmath2 is said to _ virtually _ have some property @xmath458 if a finite index subgroup of @xmath2 has @xmath458 .
[ p - like ] let @xmath2 be a countably infinite virtually product - like residually finite group
. then @xmath66 .
suppose that there exists a faithful , @xmath39-transitive action of @xmath2 on a set @xmath3 . by our assumption
, @xmath2 contains a finite index subgroup @xmath459 and non - trivial subgroups @xmath460 such that @xmath457 . by (
* theorem 7.2d ) , a finite index subgroup in an infinite @xmath39-transitive permutation group is primitive .
thus @xmath459 acts primitively on @xmath3 . by lemma [ prim ] ( a )
, @xmath252 is transitive .
as @xmath281 and @xmath252 are normal in @xmath459 , we have @xmath461 \le a\cap b=\ { 1\}$ ] . applying lemma [ min ] for @xmath462 , @xmath463 , and @xmath464
, we obtain that @xmath281 is an infinite minimal normal subgroup of @xmath459 .
however existence of such a subgroup obviously contradicts residual finiteness of @xmath459 .
note that there are many @xmath39-transitive solvable groups ( e.g. , affine groups of fields , see the discussion of affine groups after question [ q1 ] ) .
however , such groups can not be residually finite .
[ sol ] let @xmath2 be an infinite residually finite virtually solvable group .
then @xmath66 .
let @xmath459 be a solvable normal subgroup of @xmath2 of finite index and let @xmath281 be the maximal abelian normal subgroup of @xmath459 . since @xmath459 is solvable @xmath465 . since @xmath281 is characteristic in @xmath459 , it is normal in @xmath2 .
assume that @xmath2 acts faithfully and @xmath39-transitively on an infinite set @xmath3 .
then @xmath281 is transitive by lemma [ prim ] ( a ) .
since @xmath281 is abelian , we can apply lemma [ min ] to @xmath466 and @xmath467 . as above , we conclude that @xmath281 is an infinite minimal normal subgroup of @xmath2 , which contradicts residual finiteness of @xmath2 .
[ [ mapping - class - groups ] ] 4.2 .
mapping class groups + + + + + + + + + + + + + + + + + + + + + + + + + our main reference for mapping class groups is @xcite .
we review some basic definitions and necessary facts below . given @xmath468 , let @xmath469 be a @xmath54 times punctured compact , connected , orientable surface of genus @xmath55 with @xmath56 boundary components .
we also write @xmath470 for @xmath471 . by
the the _ mapping class group _ of @xmath151 , denoted @xmath472 , we mean the group of isotopy classes of orientation preserving homeomorphisms of @xmath151 ; both homeomorphisms and isotopies are required to restrict to the identity on @xmath160 ( however we do allow homeomorphisms to permute the punctures ) . given some @xmath473 with @xmath474 , we can glue a once punctured closed disc to each boundary component of @xmath151 .
let @xmath475 be the resulting surface .
assume , in addition , that @xmath151 is not a disc nor an annulus .
then there is a _ capping homomorphism _
@xmath476 whose kernel is generated by dehn twists corresponding to components of @xmath160 ( * ? ? ?
* sections 3.6.1 , 3.6.2 ) .
recall that a _ dehn twist _ @xmath477 around a curve @xmath108 commutes with every @xmath478 fixing @xmath108 .
in particular , this implies that @xmath479 is a central subgroup of @xmath472 .
we say that a surface @xmath480 is _ exceptional _ if @xmath481 .
thus exceptional surfaces are spheres with at most @xmath482 punctures and tori with at most @xmath0 puncture .
the following is a complete list of mapping class groups of exceptional surfaces ( see ( * ? ? ?
* chapter 2 ) ) : @xmath483 @xmath484 and @xmath485 mapping class groups of non - exceptional surfaces are acylindrically hyperbolic since they act acylindrically on the corresponding curve complexes , which are hyperbolic @xcite ; mapping class groups of exceptional surfaces are either finite or non - virtually cyclic and hyperbolic . hence we obtain the following ( see section 8 of @xcite ) . [ mod - ah ] @xmath486 is acylindrically hyperbolic for all @xmath487 . thus to describe highly transitive mapping class groups we need to know which mapping class groups contain non - trivial finite normal subgroups .
the following lemma is probably known to specialists .
however we were unable to find a reference , so we prove it here .
the idea of the proof was suggested by dan margalit .
[ finsub ] if @xmath60 and @xmath61 , then @xmath486 has no non - trivial finite normal subgroups . for basic properties of dehn twists used in this proof we refer to (
* chapter 3 ) .
recall also that under our assumptions the action of @xmath488 on the curve complex is faithful @xcite .
assume that an element @xmath489 belongs to a finite normal subgroup @xmath49 and let @xmath490 be an essential simple closed curve on @xmath491 . by the classification of subgroups generated by two distinct dehn twists ( see @xcite ) , the subgroup @xmath492 is isomorphic to one of the groups @xmath82 , @xmath493 , or the free group of rank @xmath39 .
note that @xmath494 since neither of the groups from the above list has torsion . on the other hand , @xmath495=t_c^2ft_c^{-2}f^{-1}=t_c^2t_{f(c)}^{-2}\in h\cap n$ ] since @xmath49 is normal .
therefore @xmath495=1 $ ] , which implies that @xmath496 ( see ( * ? ? ?
* section 3.3 ) ) . since this is true for every essential simple closed curve @xmath490 , @xmath199 acts trivially on the curve complex of @xmath470 and consequently @xmath497 .
[ mcg0 ] @xmath58 admits a highly transitive faithful action on a countably infinite set if and only if @xmath59 , @xmath60 , and @xmath61 .
let @xmath498 . if @xmath474 , then either @xmath2 is trivial ( if @xmath499 is a closed disk or a closed annulus ) or has a non - trivial central subgroup ( the kernel of the capping homomorphism ) and hence @xmath66 .
similarly @xmath66 if @xmath500 since @xmath501 in this case .
further if @xmath59 and @xmath499 is exceptional , then @xmath2 is either trivial or has a non - trivial finite normal subgroup , so it can not have a highly transitive faithful action on a countably infinite set again . in all remaining cases ,
@xmath2 is acylindrically hyperbolic and has trivial finite radical by the previous lemma , so the result follows from theorem [ main ] .
let us now discuss subgroups of mapping class groups .
the following observation will be used several times .
[ g+n ] let @xmath473 . by gluing @xmath502 to every boundary component of @xmath151 , we obtain an embedding @xmath503 .
thus , if we want if we want to prove certain claim for subgroups of @xmath504 , it suffices to prove it for @xmath59 .
[ pl - or - ah ] let @xmath505 .
assume that @xmath2 is non - trivial and icc .
then @xmath2 is either virtually product - like or acylindrically hyperbolic with trivial finite radical . by remark [ g+n ]
, we can assume that @xmath506 .
we consider two cases .
_ if @xmath2 contains a pseudo - anosov element @xmath108 , then it is either virtually cyclic or acylindrically hyperbolic . indeed in the non - exceptional case it follows from the fact that @xmath108 is contained in a virtually cyclic hyperbolically embedded subgroup of @xmath2 ( * ? ? ?
* theorem 2.19 ) and theorem [ ah - hes ] .
since @xmath2 is icc , it can not be virtually cyclic and can not contain non - trivial finite normal subgroups .
thus @xmath2 is acylindrically hyperbolic with trivial finite radical .
_ now assume that @xmath2 does not contain a pseudo - anosov element .
then by ivanov s theorem @xcite @xmath2 is reducible , i.e. , it fixes a multicurve .
moreover , in this case there is a multicurve @xmath490 and a finite index subgroup @xmath507 with the following properties .
let @xmath508 be the connected components of @xmath509 .
we naturally think of @xmath276 as a punctured closed surface ( cuts along components of @xmath490 give rise to new punctures ) .
then @xmath459 fixes every @xmath276 setwise and the restriction of @xmath459 to @xmath276 is either trivial or infinite irreducible .
this gives a homomorphism @xmath510 whose kernel is generated by dehn twists around components of @xmath490 ( * ? ? ?
* proposition 3.20 ) .
in particular , @xmath511 is central in @xmath459 .
since @xmath2 is icc , so is @xmath459 and hence @xmath512 is injective .
let @xmath513 and @xmath514 . if both @xmath269 and @xmath270 are non - trivial , then @xmath2 is product - like .
if @xmath515 , then the composition of @xmath516 and the projection map to the first factor in ( [ reducible ] ) is injective and its image is either trivial or irreducible ; in the latter case we complete the proof as in the first case . if @xmath517 , then the composition of @xmath512 with the projection to @xmath518 is injective and we complete the proof by induction .
we will also need the following result .
it is well - known to specialists ( see , e.g. , the remark in the end of section 6.4.3 of @xcite ) , but we were unable to locate the proof in the literature .
we are grateful to dan margalit for explaining the argument to us .
[ rf ] the group @xmath504 is residually finite for all @xmath519 . since a subgroup of a residually finite group is residually finite , we can assume that @xmath59 by remark [ g+n ] .
for @xmath520 , residual finiteness of @xmath486 is a classical result of grossman @xcite .
now assume that @xmath521 .
it is easy to see that any ( residually finite)-by - finite group is residually finite .
thus it suffices to prove residual finiteness of @xmath522 , the subgroup of all elements of @xmath486 that fix the punctures .
the latter group has a faithful representation by automorphisms of @xmath523 @xcite .
since @xmath523 is residually finite and the automorphism group of a residually finite finitely generated group is residually finite @xcite , we obtain that @xmath522 is residually finite , and hence so is @xmath486 .
the next corollary refers to conditions ( c@xmath69)-(c@xmath70 ) stated in the introduction .
[ mcg1 ] let @xmath2 be an infinite subgroup of @xmath504 .
then @xmath68 and conditions ( c@xmath69)-(c@xmath70 ) are equivalent
. moreover , ( c@xmath69)-(c@xmath70 ) are also equivalent to 1 .
@xmath2 is non - trivial , icc , and not virtually product - like .
we first note that ( c@xmath524 ) @xmath525 ( c@xmath70 ) by lemma [ pl - or - ah ] .
further , ( c@xmath70 ) @xmath525 ( c@xmath526 ) by theorem [ main ] and the implication ( c@xmath526 ) @xmath525 ( c@xmath69 ) is obvious .
assume now that @xmath2 does not satisfy ( c@xmath524 ) . if @xmath2 is not icc , then @xmath66 by part ( b ) of lemma [ 2-trans ] . if @xmath2 is virtually product - like , then @xmath66 by lemma [ rf ] and corollary [ p - like ] .
thus @xmath66 whenever @xmath2 does not satisfy ( c@xmath524 ) .
in particular , we obtain ( c@xmath69 ) @xmath525 ( c@xmath524 ) .
this completes the proof of the equivalence of ( c@xmath69)-(c@xmath70 ) and also shows that @xmath68 .
[ [ manifold - groups . ] ] 4.3 .
@xmath63-manifold groups .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let @xmath2 be a subgroup of the fundamental group of a compact @xmath63-manifold .
it was proved in ( * ? ? ?
* theorem 5.6 ) that one of the following three ( mutually exclusive ) conditions holds : 1 .
@xmath2 is acylindrically hyperbolic with trivial finite radical .
2 . @xmath2 contains an infinite cyclic normal subgroup @xmath182 and @xmath527 is acylindrically hyperbolic .
@xmath2 is virtually polycyclic .
moreover , if @xmath2 is itself a fundamental group of a compact irreducible @xmath63-manifold @xmath62 ( this is always the case if @xmath2 is finitely generated by the scott s core theorem ) , then we can replace condition ( ii ) with a more geometric one . 1 .
@xmath62 is seifert fibered .
[ 3d0 ] let @xmath2 be an infinite subgroup of the fundamental group of a compact @xmath63-manifold .
then @xmath68 and conditions ( c@xmath69)(c@xmath70 ) are equivalent .
this follows immediately from theorem [ main ] and the above trichotomy . indeed if @xmath2 satisfies ( i ) , then @xmath26 by theorem [ main ] . if @xmath2 satisfies ( ii ) , it is not icc ( note that the conjugacy class of any element @xmath528 is contained in @xmath529 ) and hence @xmath66 by part ( b ) of lemma [ 2-trans ] . finally if @xmath2 satisfies ( iii ) , then @xmath2 can not act primitively on a countable set by lemma [ sol ] and hence @xmath66 .
thus conditions ( c@xmath69)(c@xmath70 ) hold simultaneously in case ( i ) and fail simultaneously in cases ( ii ) and ( iii ) . similarly , if @xmath2 is itself a fundamental group of a compact irreducible @xmath63-manifold @xmath62 , we obtain the following .
[ 3d1 ] let @xmath62 be a compact irreducible @xmath63-manifold .
then @xmath64 admits a highly transitive faithful action on a countably infinite set iff it is not virtually solvable and @xmath62 is not seifert fibered .
[ [ graph - products . ] ] 4.4 . graph products .
+ + + + + + + + + + + + + + + + + + + + it is proved in @xcite that for every subgroup @xmath2 of a finite graph product @xmath75 , one of the following conditions holds .
* there is a short exact sequence @xmath530 , where @xmath167 is finite and @xmath151 is isomorphic to a subgroup of some @xmath85 . * @xmath2 is virtually cyclic ; * @xmath2 contains two infinite normal subgroups @xmath531 such that @xmath532 . * @xmath2 is acylindrically hyperbolic . we will also need a result of cameron .
[ cameron ] suppose that @xmath533 is infinite and @xmath4-transitive for some @xmath534 .
let @xmath49 be a non - trivial normal subgroup of @xmath2 .
then either @xmath49 is @xmath535-transitive or @xmath536 and @xmath49 is an elementary abelian @xmath39-group ( i.e. , a direct sum of copies of @xmath537 ) .
similar to the previous cases , we obtain the following . [ gp ] let @xmath538 be a countably infinite subgroup of a graph product of a finite family of groups @xmath85 , @xmath539 .
suppose that @xmath2 is not isomorphic to a subgroup of some @xmath85 .
then @xmath84 and conditions ( c@xmath69)(c@xmath70 ) are equivalent . if , in addition , every @xmath85 is residually finite , then @xmath68 .
if @xmath2 satisfies ( a ) , then @xmath167 must be non - trivial and hence @xmath66 by part ( b ) of lemma [ 2-trans ] .
similarly @xmath66 in the case ( b ) as non - trivial virtually cyclic groups are not icc .
suppose ( c ) holds .
if @xmath540 , then @xmath2 contains a non - trivial finite normal subgroup and we obtain @xmath541 as above .
thus it remains to consider the case @xmath542 .
we are going to prove that @xmath543 in this case .
arguing by contradiction , suppose that @xmath544 .
let @xmath49 denote the ( normal ) subgroup of @xmath2 generated by @xmath545 and @xmath546 .
clearly @xmath547 .
let @xmath2 act faithfully and @xmath63-transitively on a set @xmath3 . according to lemma [ cameron ] , there are two possibilities .
_ neither of the subgroups @xmath545 , @xmath546 is @xmath39-transitive .
then @xmath49 is an elementary abelian @xmath39-group , so we can apply lemma [ min ] to @xmath548 and conclude that @xmath49 is a minimal normal subgroup of @xmath2 .
however @xmath549 , @xmath550 are proper subgroups of @xmath49 . contradiction .
_ assume now that one of the subgroups @xmath545 , @xmath546 , say @xmath546 , is @xmath39-transitive .
let @xmath551 be distinct elements .
@xmath545 is transitive by lemma [ prim ] .
hence there exists @xmath552 such that @xmath553 .
since @xmath546 is @xmath39-transitive , there exists @xmath554 such that @xmath555 and @xmath556 .
then @xmath557 while @xmath558 .
this contradicts @xmath559\le n_1\cap n_2=\ { 1\}$ ] .
notice that by a theorem of green @xcite ( see also @xcite ) , a finite graph product of residually finite groups is residually finite .
thus if all multiples are residually finite , so is @xmath2 and instead of considering cases 1,2 above we can immediately conclude that @xmath66 by corollary [ p - like ] . finally , if @xmath2 satisfies ( d )
, the conclusion follows from theorem [ main ] and part ( b ) of lemma [ 2-trans ] .
note that the case @xmath560 in the above corollary does realize .
the following example is well - known .
[ 2cc ] let @xmath561 be a countably infinite group where all non - trivial elements are conjugate .
first examples of countable groups of this kind were constructed by g. higman , b. neumann and h.neumann ( see @xcite ) ; finitely generated examples can be found in @xcite .
let @xmath562 .
then @xmath2 naturally acts on @xmath561 by the rule @xmath563 note that this action is isomorphic to the action of @xmath2 on cosets @xmath564 , where @xmath565 .
it is easy to see that this action is @xmath39-transitive .
indeed let @xmath566 be two pairs of distinct elements of @xmath561 .
since all nontrivial elements of @xmath561 are conjugate , there exists @xmath567 such that @xmath568 .
let also @xmath569 .
obviously we have @xmath570 and @xmath571 finally we observe that if @xmath572 for every @xmath573 , then substituting @xmath574 we obtain @xmath575 .
hence @xmath576 for all @xmath573 , i.e. , @xmath577 is a conjugacy class .
clearly this implies @xmath578 and consequently @xmath579 .
thus the action is faithful and we have @xmath580 . on the other hand , if @xmath544 , then both multiples of @xmath562 act @xmath39-transitively by lemma [ cameron ] and we get a contradiction arguing as as in case 2 of the proof of corollary [ gp ] . thus @xmath560 .
if @xmath2 is a finite graph product of copies of @xmath82 , i.e. , a right angled artin group , then it is either @xmath82 , of decomposes as a direct product of two non - trivial groups , or is acylindrically hyperbolic @xcite .
since @xmath2 is torsion free , its finite radical is trivial and we obtain the following .
[ raag ] a right angled artin group is highly transitive if and only if it is non - cyclic and directly indecomposable . [ [ miscellaneous - results . ] ] 4.5 . miscellaneous results .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + here we discuss applications of our main theorem to some other classes of acylindrically hyperbolic groups .
let @xmath2 be a group hyperbolic relative to a collection of proper subgroups .
then @xmath2 is either acylindrically hyperbolic or virtually cyclic .
let @xmath2 be hyperbolic with respect to a collection of proper subgroups @xmath581 .
we accept the definition of a relatively hyperbolic group from @xcite .
in particular , we do not assume that @xmath2 is countable .
withoul loss of generality we can assume that all @xmath582 are non - trivial .
we consider two cases .
_ there exists an infinite @xmath583 . by (
* proposition 4.28 ) we have @xmath138 . since @xmath118 is infinite and @xmath168 , it is non - degenerate and hence @xmath2 is acylindrically hyperbolic by theorem [ ah - hes ] .
_ suppose now that all @xmath584 are finite .
again there are two possibilities .
first assume that @xmath585 is finite .
then by ( * ? ? ?
* theorem 2.40 ) @xmath2 is an ( ordinary ) hyperbolic group .
hence @xmath2 is either virtually cyclic or acylindrically hyperbolic .
finally assume that @xmath586 is infinite .
then by part 1 ) of ( * ? ? ?
* theorem 2.44 ) @xmath2 decomposes as a non - trivial free product of infinitely many non - trivial groups .
it is easy to see that the action on the associated bass - serre tree is acylindrical and non - elementary .
thus @xmath2 is acylindrically hyperbolic in this case . from this lemma , theorem [ main ] and lemma [ 2-trans](b )
we obtain the following .
let @xmath2 be a countable group hyperbolic relative to a collection of proper subgroups .
then @xmath2 is highly transitive if and only if it is not virtually cyclic and has no non - trivial finite normal subgroups . [ outfn ] @xmath41 is highly transitive if and only if @xmath36 .
recall that @xmath41 is acylindrically hyperbolic for @xmath34 @xcite .
@xmath587 has non - trivial finite radical , hence @xmath588 by lemma [ 2-trans ] ( b ) . on the other hand , for @xmath36 ,
@xmath41 has trivial finite ( and even amenable ) radical .
this follows immediately , for example , from @xmath589-simplicity of @xmath590 for @xmath36 proved in @xcite .
this fact can also be derived from results of @xcite .
[ 1-rel ] every @xmath0-relator group with at least @xmath63 generators is highly transitive .
let @xmath2 be a @xmath0-relator group @xmath2 with at least @xmath63 generators .
it was proved in @xcite that @xmath2 is acylindrically hyperbolic .
thus we only need to show that @xmath2 has trivial finite radical .
it is shown in @xcite ( see the proof of theorem iii there ) that every @xmath0-relator group with at least @xmath63 generators splits as an hnn - extension @xmath591 , where @xmath456 are proper subgroups of @xmath118 , and there exists @xmath592 such that @xmath593 . let @xmath437 be the bass - serre tree associated to this decomposition and let @xmath594 be a finite normal subgroup . recall that every finite group acting on a tree without inversions fixes a vertex .
hence @xmath49 fixes a vertex of @xmath437 .
but since @xmath49 is normal and the action of @xmath2 is transitive , @xmath49 fixes all vertices of @xmath437 . in particular , @xmath595 .
[ [ mixed - identity - free - groups . ] ] 5.1 .
mixed identity free groups .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we begin by introducing notation , which will be used throughout the rest of the paper .
let @xmath37 denote the free group of rank @xmath56 with basis @xmath596 . given a group @xmath2 ,
we identify elements of the free product @xmath91 with their normal forms ( see theorem [ nform ] ) . thus we think of elements of @xmath91 as words in the alphabet @xmath597 . given @xmath87 and any elements @xmath598 , we denote by @xmath599 the word obtained from @xmath89 by substituting @xmath600 for @xmath601 for all @xmath602 .
alternatively , we can describe @xmath599 as the image of @xmath89 under the homomorphism @xmath603 that is identical on @xmath2 and sends @xmath601 to @xmath600 , @xmath602 .
recall that a group @xmath2 is said to satisfy a _
mixed identity _
@xmath86 for some @xmath87 if @xmath604 for all @xmath605 .
we say that a mixed identity @xmath86 is _ non - trivial _ if @xmath90 as an element of @xmath91 .
[ 1-vs - n ] it is easy to see that if @xmath606 , then for every @xmath96 there is an embedding @xmath607 which is identical on @xmath2 .
( e.g. , using normal forms in free products it is easy to show that the subgroup of @xmath608 generated by @xmath2 , @xmath609 ,
, @xmath610 is naturally isomorphic to @xmath91 for every @xmath611 . )
thus @xmath2 satisfies some non - trivial mixed identity @xmath87 if and only if it satisfies a non - trivial mixed identity @xmath612 ( i.e. , with one variable ) . a group @xmath2 is called _ mixed identity free _ ( or mif , for brevity ) if it does not satisfy any non - trivial mixed identity .
recall that , for a given group @xmath2 , a _
@xmath2-group _ is a group @xmath118 with a fixed embedding @xmath613 .
let @xmath614 denote the first order language of @xmath2-groups .
thus the alphabet of @xmath614 consists of the standard logical symbols , the group operations ( multiplication and taking inverse ) , and constants corresponding to elements of @xmath2 .
let @xmath615 denote the universal theory of @xmath118 as a @xmath2-group .
that is , @xmath615 is the set of all universal sentences in @xmath614 that hold true in @xmath118 .
two @xmath2-groups @xmath616 , @xmath617 are _ universally equivalent _ if @xmath618 . in general , being universally equivalent as @xmath2-groups is stronger that being universally equivalent in the usual sense ( i.e. , as @xmath619-groups ) . for details on @xmath2-groups and
their elementary theory we refer to @xcite .
[ mif ] a countable group @xmath2 is mif if and only if @xmath2 and @xmath91 are universally equivalent as @xmath2-groups for all @xmath96 . the `
if ' part is obvious since satisfying a non - trivial mixed identity can be encoded by a universal sentence in @xmath614 .
thus we only need to prove the ` only if ' part .
assume that @xmath2 is mif .
fix any @xmath96 .
let @xmath620 .
let @xmath621 be the natural embedding .
let @xmath622 and for @xmath623 , let @xmath624,\ ] ] where the commutator is defined inductively by @xmath625=a_1^{-1}a_2^{-1}a_1a_2\;\;\ ; { \rm and } \;\;\ ; [ a_1 , \ldots , a_k , a_{k+1}]=[[a_1 , \ldots , a_k ] , a_{k+1}].\ ] ] using normal forms , it is easy to see that @xmath626 is a non - trivial element of @xmath627 . since @xmath2 is mif , for every @xmath251 , there exist @xmath628 such that @xmath629 . in particular , this implies that @xmath630 for all @xmath631 .
let @xmath632 be a non - principal ultrafilter on @xmath28 and let @xmath633 denote the corresponding ultrapower of @xmath2 .
let @xmath310 denote the element of @xmath633 represented by the sequence @xmath634 , @xmath635 .
we think of @xmath633 as a @xmath2-group , where the fixed embedding of @xmath2 is the diagonal one .
let @xmath636 be the homomorphism whose restriction on @xmath2 is the diagonal embedding and such that @xmath637 is mapped to @xmath310 for @xmath635 . by the choice of elements @xmath638
, we obtain that for every @xmath639 , @xmath630 for all but finitely many @xmath251 .
since @xmath632 is non - principal , this implies @xmath640 for every @xmath639 .
hence @xmath641 is injective .
thus we obtain a sequence of embeddings of @xmath2-groups : @xmath642 . by the os theorem , @xmath2 and
@xmath633 have the same elementary ( in particular , universal ) theory as @xmath2-groups .
since universal properties of algebraic structures are inherited by substructures , we obtain @xmath643 consequently , @xmath644 .
the next proposition shows that being mif is a much stronger property than not satisfying any nontrivial identity .
recall that the _ girth _ of a finitely generated group @xmath2 is the supremum of all @xmath96 such that for every finite generating set @xmath112 of @xmath2 , the cayley graph @xmath173 contains a cycle of length at most @xmath56 without self - intersections @xcite .
[ mif - prop ] for any mif group , the following conditions hold . 1 .
@xmath2 is icc .
in particular , @xmath2 has trivial center and contains no non - trivial finite normal subgroups .
@xmath2 is not decomposable as direct products of two non - trivial groups .
every non - trivial subnormal subgroup of @xmath2 is mif .
if @xmath2 is finitely generated , then @xmath2 has infinite girth .
\(a ) let @xmath645 be an element whose conjugacy class in @xmath2 consists of @xmath56 elements
. then @xmath2 satisfies the mixed identity @xmath646=1 $ ] .
\(b ) let @xmath647 .
fix any elements @xmath648 , @xmath649 .
then @xmath2 satisfies the mixed identity @xmath650 , [ b , x]]=1 $ ] , which is obviously non - trivial provided @xmath645 and @xmath651 .
\(c ) clearly it suffices to prove that every nontrivial normal subgroup @xmath594 is mif .
let @xmath652 .
assume that @xmath49 satisfies an identity @xmath653 for some @xmath654 .
since @xmath49 is normal , @xmath2 satisfies the identity @xmath655)=1 $ ] . since @xmath2 is mif , @xmath655)=1 $ ] as an element of @xmath608 . by using normal forms
, it is easy to see that @xmath656 is isomorphic to the subgroup of @xmath656 generated by @xmath49 and @xmath92 $ ] via the map that takes @xmath657 to @xmath92 $ ] and is identical on @xmath49 .
this map takes @xmath655)$ ] to @xmath658 and hence @xmath653 as an element of @xmath656 .
thus @xmath49 can only satisfy the trivial identity .
\(d ) suppose that @xmath2 has finite girth @xmath4 .
then for every finite generating set @xmath659 of @xmath2 , there exists a non - trivial element @xmath660 of length at most @xmath4 such that @xmath604 in @xmath2 . consider any generating set @xmath661 of @xmath608 and let @xmath662 denote the image of @xmath600 under the epimorphism of @xmath608 that is identical on @xmath2 and sends @xmath657 to an element @xmath663 . since @xmath664 generate @xmath2 for every @xmath8 , the group @xmath2 satisfies the universal sentence @xmath665 where @xmath666 is the list of all non - trivial elements of @xmath37 of length at most @xmath4 . by proposition [ mif ]
we obtain that the same sentence holds true in @xmath608 .
in particular , it is true for @xmath667 . since @xmath668
, we obtain that there is a non - trivial relation of length at most @xmath4 between @xmath661 .
this is true for any generating set @xmath661 of @xmath608 , so we obtain that @xmath608 has finite girth .
however this is well - known to be false ( for example , see ( * ? ? ?
* example 2.15 ) ) .
the proof of proposition [ mif - prop](d ) actually shows that for all @xmath4 , @xmath2 has a generating set with @xmath56 elements and no simple cycles of length @xmath669 , where @xmath56 is independent of @xmath4 .
this is equivalent to @xmath2 _ preforming _ the free group @xmath37 in the language of @xcite and to @xmath2 being without _ almost - identities _ in the language of @xcite .
however , it is unclear if this is actually stronger then the property of infinite girth ( see ( * ? ? ?
* question 8.5 ) ) .
[ [ a - dichotomy - for - highly - transitive - groups ] ] 5.2 . a dichotomy for highly transitive groups + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + [ non - rel ]
let @xmath670 be an arbitrary permutation , @xmath112 a finite subset of @xmath28 .
then for any permutations @xmath671 with infinite supports and any @xmath672 , there exists @xmath673 such that @xmath674 and @xmath675 for some @xmath96 .
the idea of the proof is to construct a sequence of pairwise distinct numbers @xmath676 and a permutation @xmath673 such that @xmath674 and the elements @xmath677 and @xmath600 act as follows : @xmath678 then clearly @xmath679 .
we will first construct the desired sequence @xmath680 by induction .
let @xmath681 be any element of @xmath682 .
further for @xmath683 , we let @xmath684 and chose any @xmath685 such a choice is always possible since @xmath686 is finite while @xmath687 is infinite by our assumption . we let
note that ( [ ni ] ) guarantees that @xmath689 and @xmath690 is disjoint from @xmath686 .
this completes the inductive step .
clearly the resulting sequence @xmath680 consists of pairwise distinct numbers and does not contain any elements from @xmath691 .
further , for each @xmath692 , we fix arbitrary natural numbers @xmath693 , @xmath694 , such that all @xmath693 are pairwise distinct and are not contained in @xmath695 .
again this is possible since @xmath695 is finite . for @xmath696 , we define a cycle @xmath697 ( if @xmath698 , then @xmath699 )
. we also denote by @xmath700 any permutation from @xmath38 with support @xmath701 such that @xmath702 for all @xmath125 .
we now define @xmath703 note that the sets @xmath704 , and @xmath112 are pairwise disjoint by the choice of elements @xmath705 , @xmath706 , and @xmath693 .
hence we have @xmath707 for every @xmath125 , and @xmath708 for @xmath692 .
thus @xmath677 and @xmath600 act as prescribed by the diagram ( [ tai ] ) .
recall that a subset of a topological space is _ comeagre _ if it is the intersection of countably many sets with dense interiors .
since @xmath38 is a polish space with respect to the topology of pointwise convergence , every comeagre subset of @xmath38 is dense ( and , in particular , non - empty ) by the baire category theorem .
it is customary to think of comeagre subsets as subsets including most " elements of the ambient space .
[ gen ] we say that a certain property _ holds for a generic element _ of a topological space @xmath437 if it holds for all elements of a comeagre subset of @xmath437 .
[ fp ] let @xmath2 be a highly transitive subgroup of @xmath38 .
then either @xmath2 contains @xmath29 or for a generic element @xmath673 , the subgroup generated by @xmath2 and @xmath308 is naturally isomorphic to @xmath709 .
recall that by the wielandt theorem ( * ? ? ?
* theorem 5.2 ) every infinite primitive subgroup of @xmath38 containing a non - trivial permutation with finite support contains @xmath29 .
thus we can assume that every non - trivial element of @xmath2 has infinite support .
let @xmath89 be a non - trivial element of the free product @xmath608 .
recall that for @xmath8 , @xmath710 denotes the image of @xmath89 under the homomorphism @xmath711 that is identical on @xmath2 and sends @xmath657 to @xmath55 .
note that the set of solutions of any equation in a topological group is closed and hence the set @xmath712 is open .
by lemma [ non - rel ] , @xmath713 is also dense ; indeed the lemma claims that we can find an element from @xmath713 in any open neighborhood of any element @xmath714 .
hence the set @xmath715 is comeager .
clearly @xmath716 for all @xmath717 .
we are now ready to prove the main result of this section .
[ dichoto ] let @xmath2 be a highly transitive countable group .
then one of the following two mutually exclusive conditions holds . 1 .
@xmath2 contains a normal subgroup isomorphic to @xmath29 .
g is mif .
we identify @xmath2 with a dense subgroup of @xmath38 . by corollary [ fp ] , either @xmath2 contains a normal subgroup isomorphic to @xmath29 or there exists @xmath673 such that the subgroup generated by @xmath2 and @xmath308 is naturally isomorphic to @xmath709 . in the latter case @xmath2 can not satisfy any non - trivial mixed identity .
indeed , by remark [ 1-vs - n ] it suffices to show that @xmath2 does not satisfy any non - trivial mixed identity with one variable .
let @xmath718 and suppose that @xmath719 for all @xmath8 .
then @xmath719 for all @xmath720 since the set of solutions of any equation in a topological group is closed and @xmath2 is dense in @xmath38 .
however @xmath721 , a contradiction . it remains to prove that conditions ( a ) and ( b ) are mutually exclusive .
assume that @xmath2 contains @xmath29 as a normal subgroup .
let @xmath722 , @xmath723 , @xmath724 , @xmath725 .
then for every @xmath726 , the elements @xmath727 , @xmath728 , @xmath729 , and @xmath730 are @xmath63-cycles with disjoint supports .
hence the support of at least one of them is disjoint from @xmath731 .
this means that at least one of the commutators @xmath732 $ ] , @xmath733 $ ] , @xmath734 $ ] , @xmath735 $ ] is trivial .
thus @xmath736 satisfies the universal sentence @xmath737=1\vee [ b^g , a]=1 \vee [ c^g , a]=1\vee [ d^g , a]=1),\ ] ] which obviously fails in @xmath738 for @xmath667 . by proposition [ mif ] , this implies that @xmath29 is not mif .
( in this particular case we can easily prove this fact directly by taking the iterated commutator of the four commutators in ( [ comm ] ) . ) finally , by part ( c ) of proposition [ mif - prop ] , we conclude that @xmath2 is not mif either .
it is known that any acylindrically hyperbolic group contains non - cyclic free subgroups and hence does not satisfy any non - trivial identity @xcite .
it is also known that a non - cyclic torsion free hyperbolic group does not satisfy any non - trivial mixed identity @xcite .
we can generalizes these results as follows .
[ ah - mif ] let @xmath2 be an acylindrically hyperbolic group .
then @xmath169 is mif . in particular , if @xmath174 , then @xmath2 is mif .
we first note that the quotient group @xmath169 is acylindrically hyperbolic by lemma [ finrad ] , so it suffices to deal with the case @xmath174 . in this case
@xmath2 is highly transitive by theorem [ main ] .
further , it is proved in ( * ? ? ?
* corollary 1.5 ) that every infinite normal subgroup of @xmath2 is acylindrically hyperbolic .
@xmath29 is not acylindrically hyperbolic since every acylindrically hyperbolic group contains infinite order elements by theorem [ tri ] .
hence @xmath2 can not contain @xmath29 as a normal subgroup . to complete the proof it remains to refer to theorem [ dichoto ] . a natural question arising from comparing corollary
[ ah - mif ] and theorem [ dichoto ] is whether every highly transitive mif group is acylindrically hyperbolic .
we will show that the answer is negative by making use of a result from @xcite .
let @xmath281 be a subgroup of a group @xmath2 .
recall that for a subset @xmath739 , the core of @xmath281 with respect to @xmath151 is defined by the formula @xmath740 .
further , @xmath281 is called _ highly core free _ in @xmath2 if for every finite @xmath741 , every @xmath96 , and every subsets @xmath742 such that @xmath743 there exists @xmath744 such that @xmath745 . the following is theorem 2.2 in @xcite .
[ fms ] let @xmath118 be an hnn - extension of a countable group @xmath2 with associated subgroups @xmath281 and @xmath252 .
assume that @xmath281 and @xmath252 are highly core free in @xmath2 .
then @xmath118 is highly transitive .
[ bs12 ] let @xmath746=1\rangle.\ ] ] then @xmath118 is highly transitive and mif , but not acylindrically hyperbolic .
it is easy to see that the subgroup @xmath747 is highly core free in @xmath748 . indeed for every finite @xmath741 and every decomposition ( [ core ] ) , there exists at least one @xmath4 such that the natural projection of @xmath749 to @xmath750 contains infinitely many positive powers of @xmath202 .
obviously this implies @xmath745 .
hence @xmath118 is highly transitive by lemma [ fms ] . since @xmath2 is torsion free so is @xmath118 .
thus @xmath118 does not contain a subgroup isomorphic to @xmath29 , so @xmath118 is mif by theorem [ dichoto ] .
observe that @xmath751 is an @xmath452-normal subgroup of @xmath118 ; that is , @xmath752 is infinite for every @xmath217 . by
* corollary 1.5 ) , any @xmath452-normal subgroup of an acylindrically hyperbolic group is acylindrically hyperbolic .
this implies that @xmath118 is not acylindrically hyperbolic . finally , we record the following explicit construction of finitely generated highly transitive subgroups satisfying condition ( a ) of theorem [ dichoto ] .
[ ht - a ] for every finitely generated infinite group @xmath30 , there exists a finitely generated group @xmath2 such that @xmath31 and @xmath32 .
in particular , @xmath2 is highly transitive .
we identify @xmath38 and @xmath753 .
let @xmath754 be a generating set of @xmath30 and let @xmath755 be the embedding induced by the left regular action .
let @xmath600 denote the transposition @xmath756 and let @xmath757 .
let us show that @xmath2 contains @xmath758 .
let @xmath759 , @xmath760 , be a transposition .
assume first that @xmath761 for some @xmath251 .
then @xmath762 . similarly if @xmath763 , then @xmath764 . in both cases @xmath765 . in the general case ,
there exist elements @xmath766 , where @xmath767 and @xmath768 for some @xmath769 .
then @xmath770 .
thus @xmath765 for every @xmath760 and therefore @xmath771 .
clearly @xmath758 is normal in @xmath2 since it is normal in @xmath772 .
note also that @xmath773 .
indeed every element of @xmath774 has single ( infinite ) orbit since the action of @xmath30 on itself is the left regular one , while every element of @xmath775 moves only finitely many elements .
thus @xmath776 .
[ torsion ] there exist infinite finitely generated torsion groups admitting highly transitive faithful actions .
take a finitely generated infinite torsion group @xmath30 and apply the proposition .
[ [ mixed - identities - in - limits - of - acylindrically - hyperbolic - groups . ] ] 5.3 .
mixed identities in limits of acylindrically hyperbolic groups .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + our next goal is to show how corollary [ ah - mif ] can be used to construct examples of mif groups which are very far from being acylindrically hyperbolic .
we first recall the definition of the space of marked group presentations .
it is a particular case of a more general construction due to chabauty ; for groups it was first studied by grigorchuk @xcite .
let @xmath777 be the free group with basis @xmath778 and let @xmath779 denote the set of all normal subgroups of @xmath777 .
given @xmath780 , let @xmath781 it is easy to see that @xmath782 is a compact hausdorff totally disconnected ( ultra)metric space @xcite .
one can naturally identify @xmath779 with the set of all _ marked @xmath4-generated groups _
, i.e. , pairs @xmath783 , where @xmath2 is a group and @xmath784 is a generating @xmath4-tuple of @xmath2 .
for this reason the space @xmath779 with the metric defined above is called the _ space of marked groups with @xmath4 generators_. let @xmath785 denote the set of all marked @xmath4-generated acylindrically hyperbolic groups with trivial finite radical and let @xmath786 denote its closure in @xmath779 . in the proposition below
, the word ` generic ' is used in the sense of definition [ gen ] .
[ lim - mif ] let @xmath787 be a subset of @xmath779 for some @xmath534 such that @xmath788 is dense in @xmath787 .
then a generic presentation from @xmath787 defines a mif group .
let @xmath789 . if @xmath2 is a marked @xmath4-generated group ( i.e. , a quotient group of @xmath777 ) , we denote by @xmath790 the natural image of @xmath791 in @xmath608 .
let @xmath792 be the set of all marked @xmath4-generated groups @xmath2 such that @xmath793 in @xmath608 or @xmath794 in @xmath2 for some @xmath8 .
it is easy to see that if a certain marked @xmath4-generated group @xmath2 belongs to @xmath792 , then all marked @xmath4-generated groups from some open neighborhood of @xmath2 belong to @xmath792 .
thus @xmath792 is open .
further , let @xmath795 .
clearly @xmath796 consists of mif groups ( see remark [ 1-vs - n ] ) . by corollary [ ah - mif ] , @xmath797 .
thus @xmath796 is an intersection of countably many open dense subsets of @xmath787 , i.e. , it is comeagre in @xmath787 . by proposition [ mif ] ,
every countable mif group is universally equivalent to an acylindrically hyperbolic group ( namely , the free product @xmath91 ) .
the following corollary of proposition [ lim - mif ] shows that universal equivalence can not be replaced by elementary equivalence and even with @xmath798-equivalence here .
[ 2cc ] there exist a finitely generated mif group with exactly @xmath39 conjugacy classes .
in particular , there exists a mif group that is not elementary equivalent ( not even @xmath798-equivalent ) to an acylindrically hyperbolic group .
let @xmath799 denote the set of all marked presentations of @xmath4-generated torsion free acylindrically hyperbolic groups .
clearly @xmath800 .
it was proved in ( * ? ? ?
* corollary 1.13 ) that a generic presentation in @xmath801 defines a group with @xmath39 conjugacy classes . on the other hand ,
a generic presentation in @xmath801 defines a mif group by proposition [ lim - mif ] .
since intersection of two comeagre sets is comeagre , a generic presentation in @xmath801 defines a group satisfying both properties .
it remains to note that @xmath801 is compact being a closed subset of a compact space , and hence comeagre subsets of @xmath801 are non - empty by the baire category theorem .
finally we note that groups with exactly @xmath39 conjugacy classes can be characterized by the @xmath798-formula @xmath802 note also that acylindrically hyperbolic groups have infinitely many conjugacy classes ( and , moreover , exponential conjugacy growth , see @xcite ) .
this implies the second claim of the corollary .
our next result provides examples of mif groups which are not highly transitive .
[ mif - tor ] there exists mif torsion groups of transitivity degree @xmath0 .
let @xmath803 be the class of all marked @xmath4-generated hyperbolic groups @xmath118 such that 1 .
@xmath118 is not virtually cyclic .
for any @xmath804 , @xmath805=1 $ ] implies @xmath806=1 $ ] .
olshanskii showed that for every @xmath534 , the closure @xmath807 contains a dense subset consisting of torsion groups ( this is an immediate consequence of ( * ? ? ?
* theorems 3 and 4 ) ) .
let @xmath808 .
for every @xmath379 , let @xmath809 denote the subset of @xmath779 consisting of marked groups @xmath2 such that @xmath791 has finite order in @xmath2 .
obviously @xmath810 is open and the set @xmath811 consists of torsion groups . combining this with the olshanskii s result and proposition [ lim - mif ] , we obtain that the set of mif torsion groups is comeagre in @xmath812 .
in particular , there exists a torsion mif marked @xmath4-generated group @xmath813 .
notice that every marked group from @xmath807 satisfies ( * * ) .
indeed suppose that ( * * ) fails in some marked group @xmath814 .
then it fails in any marked group in some open neighborhood of @xmath2 ( for the same elements @xmath657 and @xmath815 as in @xmath2 )
. hence ( * * ) fails in some group from @xmath803 , a contradiction . in particular , @xmath437 satisfies ( * * ) and therefore every involution in @xmath437 is central .
thus if @xmath437 has an involution , it has @xmath816 by lemma [ 2-trans ] ( b ) .
if @xmath437 does not have involutions , then every element of @xmath437 has odd order and hence no subgroup of @xmath437 can surject on @xmath83 . on the other hand , in any @xmath39-transitive permutation group acting on a set @xmath3 ,
the quotient of the setwise stabilizer of any pair of elements of @xmath3 modulo the pointwise stabilizer of the same elements is isomorphic @xmath83 .
thus , in any case , we obtain @xmath816 .
in this section we discuss some open questions about transitivity degree of countable groups .
we begin with two problems mentioned in the introduction .
it is well - known and easy to prove that for every field @xmath94 , the natural action of the @xmath0-dimensional affine group @xmath819 on @xmath94 is @xmath39-transitive .
recall that @xmath819 is the group of transformations of @xmath820 generated by the linear maps @xmath821 and @xmath822 for @xmath823 and @xmath824 .
note also that if @xmath94 is infinite , then @xmath819 does not admit any @xmath63-transitive faithful action .
indeed , it is easy to see that @xmath825 .
if @xmath819 has a faithful @xmath63-transitive action on a set @xmath3 and @xmath826 , then the action of @xmath827 on @xmath3 must be @xmath39-transitive by the cameron s theorem ( lemma [ cameron ] ) .
however this contradicts lemma [ 2-trans ] ( b ) .
thus @xmath828 .
another example of groups of transitivity degree @xmath39 is given in example [ 2cc ] . to construct a countable group @xmath2 of transitivity degree @xmath63 , we slightly modify an example from @xcite showing that part ( b ) of lemma [ cameron ] does realize .
let @xmath72 be the direct sum of countably many copies of @xmath83 .
we think of @xmath72 as a vector space over @xmath829 with basis @xmath830 .
let @xmath281 be the group of automorphisms of @xmath72 with finite support .
let @xmath831 be the corresponding split extension .
the action of @xmath72 on itself by translations and the action of @xmath281 by automorphisms induces a faithful ( affine ) action of @xmath2 on @xmath72 .
let @xmath832 be any distinct vectors of @xmath72 . using the translation action of @xmath72
, we can shift this triple to @xmath833 , where @xmath834 , @xmath835 .
note that @xmath836 and both vectors are non - zero ; since the field of scalars is @xmath829 , @xmath80 and @xmath81 are linearly independent .
hence there is an automorphism of @xmath72 with finite support , which maps the triple @xmath833 to @xmath837 .
this shows that @xmath544 . on the other hand ,
if @xmath838 , then the abelian group @xmath72 must be @xmath63-transitive by lemma [ cameron ] , which contradicts lemma [ 2-trans ] ( b ) . in general , it is rather difficult to show that a certain group with @xmath26 _ does not _ admit a highly transitive action . to do this
, one can try to find a property of groups which can hold in a @xmath4-transitive group for any @xmath4 , but never holds for a highly transitive group .
we wonder if satisfying a non - trivial mixed identity is such a property . in other words , we would like to know if every countable group with @xmath26 is mif .
compute transitivity degrees of some interesting particular groups .
concrete examples include @xmath839 for a countably infinite field @xmath94 , free burnside groups , thompson s group @xmath94 , baumslag - solitar groups , etc .
it is well - known that the action of @xmath839 on the corresponding projective space is @xmath63-transitive .
thus @xmath840 .
the action of @xmath839 on the projective space is not @xmath482-transitive if @xmath94 is infinite , but we can not exclude the possibility that there exists another faithful action which is @xmath482-transitive .
thus we do not know what @xmath841 is .
we do not even know whether @xmath839 is highly transitive or not .
note that @xmath839 is not acylindrically hyperbolic .
indeed it is either finite or simple by the jordan - dixon theorem , while acylindrically hyperbolic groups are never simple @xcite .
it is easy to see that transitivity degree of any solvable baumslag - solitar group is @xmath0 ( e.g. , by corollary [ sol ] ) .
the same holds for @xmath842 ; indeed in this case the conjugacy class of @xmath843 is finite and we can apply lemma [ 2-trans ] ( b ) .
the first interesting case is @xmath844 .
we do not even know if @xmath845 is highly transitive .
alexander olshanskii ( private communication ) noted that transitivity degree of a free burnside group of exponent @xmath56 is finite and does not exceed the maximal number @xmath4 such that @xmath56 is divisible by all positive integers @xmath846 . indeed if @xmath2 acts on @xmath28 faithfully and @xmath4-transitively , then there is a subgroup @xmath146 ( namely , the setwise stabilizer of @xmath847 ) that admits an epimorphism onto the symmetric group @xmath749 . hence for every positive integer @xmath846 , @xmath118 must contain an element whose order is divisible by @xmath54 . in particular ,
transitivity degree of a free burnside group of odd exponent is @xmath0 .
it is quite possible that this is also true for all infinite free burnside groups of sufficiently large exponent .
[ hough ] fix some @xmath96 and let @xmath850 be the disjoint union of @xmath56 copies @xmath851 of @xmath28 .
let @xmath33 be the group of all permutations @xmath452 of @xmath850 such that @xmath452 acts as a translation along each @xmath549 outside a finite set .
these groups , commonly called houghton groups , were introduced in @xcite .
it is known that @xmath33 if finitely generated for @xmath34 and finitely presented for @xmath36 .
further it is not hard to show that for @xmath34 , the derived subgroup of @xmath33 coincides with @xmath852 .
thus @xmath33 is ( locally finite)-by - abelian ( in particular , it is elementary amenable ) and highly transitive . for details , see @xcite .
it is likely that lacunary hyperbolic amenable groups constructed in @xcite are mif .
recall that a group @xmath2 is _ lacunary hyperbolic _ if it is finitely generated and at least one asymptotic cone of @xmath2 is an @xmath853-tree .
in fact , we conjecture the following
. this should follow from the following fact : every non - trivial mixed identity @xmath854 in a @xmath855-hyperbolic group @xmath2 with trivial finite radical can be violated by an element @xmath8 of length at most @xmath856 , where @xmath857 is the sum of lengths ( in @xmath2 ) of elements @xmath858 involved in the normal form of @xmath89 .
it seems plausible that this result can be proved by taking @xmath55 to be a high power of an infinite order element @xmath859 such that no @xmath600 belongs to @xmath860 ( i.e. , to the maximal virtually cyclic subgroup of @xmath2 containing @xmath132 ) .
however verifying details of this proof is outside of the scope of our paper .
lacunary hyperbolic groups can be thought of as infinitely presented analogues of hyperbolic groups .
we note , however , that chainikov s theorem about highly transitive actions of hyperbolic groups does not extend to lacunary hyperbolic ones .
there exists a non - virtually cyclic torsion free lacunary hyperbolic group @xmath2 such that @xmath861 .
indeed let @xmath2 be a non - cyclic torsion free lacunary hyperbolic group such that every proper subgroup of @xmath2 is cyclic .
such groups were constructed in @xcite .
assume that @xmath2 acts @xmath39-transitively on a set @xmath112 and let @xmath118 be the stabilizer of some @xmath125 .
then @xmath118 is infinite cyclic .
however mazurov @xcite proved that there are no @xmath39-transitive permutation groups with infinite cyclic point stabilizers .
v. v. chaynikov , properties of hyperbolic groups : free normal subgroups , quasiconvex subgroups and actions of maximal growth , ph.d .
thesis , vanderbilt university , ( 2012 ) , available at http://etd.library.vanderbilt.edu/available/etd-06212012-172048/unrestricted/chaynikov.pdf a. olshanskii , m. sapir , on @xmath777-like groups , algebra logika * 48 * ( 2009 ) , no . 2 , 245 - 257 , 284 , 286287 , doi 10.1007/s10469 - 009 - 9044 - 2 ( russian , with english and russian summaries ) ; english transl . , algebra logic * 48 * ( 2009 ) , no . 2 , 140 - 146 . | we prove that every countable acylindrically hyperbolic group admits a highly transitive action with finite kernel .
this theorem uniformly generalizes many previously known results and allows us to answer a question of garion and glassner on the existence of highly transitive faithful actions of mapping class groups .
it also implies that in various geometric and algebraic settings , the transitivity degree of an infinite group can only take two values , namely @xmath0 and @xmath1 . here by _
transitivity degree _ of a group we mean the supremum of transitivity degrees of its faithful permutation representations .
further , for any countable group @xmath2 admitting a highly transitive faithful action , we prove the following dichotomy : either @xmath2 contains a normal subgroup isomorphic to the infinite alternating group or @xmath2 resembles a free product from the model theoretic point of view .
we apply this theorem to obtain new results about universal theory and mixed identities of acylindrically hyperbolic groups .
finally , we discuss some open problems . |
ngc 6791 is an extreme galactic star cluster with an old age of @xmath68 gyr @xcite , a high metallicity [ fe / h][email protected] @xcite , and an unusual orbit that periodically brings it close to the bulge of the milky way @xcite .
@xcite was the first to provide an estimate for the total mass of ngc 6791 : @xmath63700@xmath12 down to @xmath1320 , confirmed by @xcite .
it is widely acknowledged that ngc 6791 is one of the most massive old open clusters in our galaxy . as indicated by the discovery of several extremely blue subdwarfs @xcite and the presence of a prominent red clump , morphology of the color - magnitude diagram ( cmd ) for ngc 6791 is complex .
the proposed enhanced mass loss along the red giant branch ( rgb ) appears to explain the presence of hot subdwarfs and the abnormally young 2.4 gyr white dwarf cooling age @xcite .
recently , @xcite reported that the cmd in the inner part of ngc 6791 ( @xmath14 ) is somewhat different from that in its outer part ( @xmath15 ) .
specifically , there is a dichotomy near the main - sequence turnoff , which the authors interpret as a result of protracted star formation .
this cluster is critically important to understanding a number of issues , such as stellar evolution and population synthesis at high metallicity in the milky way and other galaxies , specifically , in ellipticals and in spirals with bulges . due to its relative proximity and its abundance of members , ngc 6791
provides perhaps the best opportunity for detailed studies of old and metal - rich populations . despite a rich array of extant photometric and spectroscopic studies ,
however , the basic knowledge of cluster membership by kinematic means has been limited until now to radial velocities of a few dozen bright stars ( e.g. , * ? ? ?
* ; * ? ? ?
* ) and preliminary proper motions for a few thousand stars @xcite .
ngc 6791 is located in the field of view ( fov ) of nasa s _ kepler _ mission @xcite and is one of the targets of the kepler open cluster study @xcite .
it is also one of the clusters in the wiyn open cluster study @xcite again , largely owing to its extreme properties .
this letter presents the highlights of a new and comprehensive proper - motion study of ngc 6791 , based upon the selection of exquisite deep photographic plates in combination with a decade of ccd imaging .
the resulting proper motions provide nearly definite cluster membership for stars brighter than @xmath122 and , thus , unveil most of the cluster s population with unprecedented clarity .
ngc 6791 is a rich but faint star cluster with a main - sequence turnoff at @xmath1617.5 ( * ? ? ?
* hereafter sbg ) , thus requiring relatively long exposures even with 34 m class telescopes . fortunately , there are deep photographic plates dating back to 1961 .
we selected the best 33 plates ( assessed by the depth and sharpness of images ) taken with the lick 3 m and the kitt peak national observatory ( kpno ) 4 m telescopes .
these plates were digitised using the space telescope science institute s gamma ii multi - channel scanning microdensitometer .
the second epoch includes 66 frames obtained with the noao ccd mosaic imager at the kpno 4 m telescope in 1999 - 2007 .
the finest 24 ccd mosaic frames in @xmath0 filters ( 1 , 30 , and 300 s exposures ) were obtained in 2009 november with megacam at the 3.6 m canada - france - hawaii telescope .
these frames served for construction of a master catalog consisting of 128,771 objects down to @xmath125 and over an fov of 1 deg@xmath17 .
we put positions on the system of the ucac3 catalog @xcite and calculated proper motions on this system using the iterative central - plate - overlap algorithm ( e.g. , * ? ? ?
the final catalog contains a total of 58,901 objects down to @xmath123.8 over a smaller area of @xmath60.8 deg@xmath18 , defined by a circular fov ( with radius 30 ) of the kpno photographic plates .
the highest precision of our proper motions is 0.08 mas yr@xmath3 .
cluster membership probabilities are calculated using the frequency functions for cluster and field stars , obtained in selected magnitude and spatial ranges .
discrimination between cluster members and field stars is excellent down to @xmath122 .
details of the astrometric reductions and the description of the catalog itself will appear elsewhere ( platais et al .
, in preparation ) . for photometry
, we used only the megacam images in @xmath0 filters .
the central part of ngc 6791 is fairly crowded , often resulting in blended images .
we note that all megacam images are preprocessed with elixir which , besides the standard steps , also normalizes the instrumental photometric zeropoint between the chips ( see * ? ? ?
* ; * ? ? ?
* ) . the short , 1 s exposures obtained in subarcsecond seeing were instrumental in constructing an initial catalog of positions and aperture photometry ( mag_auto magnitudes from sextractor ) .
magnitudes from longer exposures were added only if they matched specified tolerances on position and magnitude . in this way it was possible to minimize the deleterious effects of image blending .
nevertheless , image blending and ccd chip edge effects may occasionally yield stellar colors that are significantly off .
our instrumental photometry was put into the standard @xmath0 system by employing the extant secondary photometric sequence in ngc 6791 @xcite .
the rms error of the calibrating fits is 0.02 mag . on average , a calibrated @xmath19 magnitude near the main - sequence turnoff of ngc 6791
is @xmath60.41 mag fainter than that in @xmath20 .
one of the most comprehensive photometric surveys of ngc 6791 is that by sbg , who also provide extensive lists of bright stars , red giants , and faint blue stars in this cluster all possible or likely members .
now , we are in a position to provide nearly definite star - by - star kinematic membership , with a caveat that stars with blended images may not appear in our proper motion catalog or , if they do , may be biased .
an example of such a case is the eclipsing binary v20 a definite cluster member @xcite but incorrectly classed as a field star in our catalog .
its measured proper motion is apparently corrupted by the adjacent star ( @xmath21=2@xmath22 ) in the amount of @xmath66@xmath23 .
the cmd in figure 1 shows the cleanest and the most complete census of ngc 6791 to date .
it contains all stars with cluster membership probability @xmath2419% , as well as stars with @xmath25% but located on the main sequence , all within a radius @xmath26 from the cluster center very close to that adopted by @xcite .
the cluster membership probabilities for peripheral members of ngc 6791 ( @xmath27 ) are less reliable because of the overwhelming number of field stars and numerical instabilities in finding the cluster population .
hence , in this region the selection of cluster members is limited to stars that are on the main sequence and have @xmath282% . from the combined cmd we deleted all stars below the main sequence and with @xmath29 , except for a few very blue stars with @xmath30% .
likewise , all stars with @xmath31 and located well above the main sequence and to the right of the rgb were eliminated .
this leaves us with a total of 5699 probable cluster members . the most prominent part of the horizontal branch ( hb)the red clump contains 26 now kinematically - confirmed stars , consistent with the photometric selection by @xcite .
more exciting is the presence of extreme horizontal branch ( ehb ) stars in ngc 6791 , first recognized by @xcite .
the following ehb stars are kinematic cluster members : b1, ... ,b7 , b9 ( names from * ? ? ?
* ; * ? ? ?
a much fainter star , b8 , has a marginal @xmath32=10% ; however , its astrometry has large uncertainties and therefore we consider it to be a possible cluster member .
in addition , star b16 ( @xmath32=59% ) appears to be a cluster member and the following sbg stars are likely cluster members as well : s3472 ( @xmath32=98% ) , s13881 ( @xmath32=60% ) . the brightest uv - source , b10
, @xcite is a definite field star . in table 1
we provide two new ehb stars in ngc 6791 , located at @xmath33 .
besides new ehb candidates , table 1 lists a number of probable cluster members that we argue are classical hb stars .
the bluest of these is the well - studied star 2 - 17 @xcite , which is frequently assigned to the ehb .
an intriguing feature of the hb here is an upturn at its red end .
this could be the effect of duplicity among the red clump stars , although that may not be the only interpretation .
we stress that proving the reality of the hb in ngc 6791 depends critically upon the ability to distinguish hb stars from blue stragglers .
there is one very bright star , s12158 ( @xmath1311.96 @xmath341.24 ) , which is formally a probable cluster member ( @xmath3591% ) but defies classification .
based on its possibly abnormally high luminosity , we consider it provisionally to be a field star .
m67 is the first open cluster where the so - called sub - subgiants were discovered @xcite .
these apparent binary stars occupy an area of the cmd below the subgiant branch , which is not easy to populate with any combination of two normal cluster stars . owing to the low frequency of field stars in this part of the cmd , we are confident that at least five likely sub - subgiants are present in ngc 6791 . in the notation of sbg
, these are : s83 , s746 , s3626 , s13753 , and s15561 .
three of these are also the known variable stars v17 , v59 , and 01431_10 listed by @xcite .
only star v17 is in common with the alternative list of sub - subgiants ( red stragglers ) given by @xcite .
the depth and spatial extent of our survey allow the derivation of a reliable luminosity function ( lf ) for the entire cluster down to @xmath36 .
the only omissions are secondaries in binary stars and severe blends .
we have counted all stars from the cropped cmd ( figure 1 ) in 0.5 mag bins , in two ways : star - by - star and by weighting each star with its membership probability ( figure 2 ) . to translate @xmath19 into @xmath37 ,
an apparent distance modulus of 13.56 was applied .
the peak at @xmath38 and the smooth downward trend toward fainter magnitudes is unmistakable .
it is consistent with the appearance of the lf in @xcite , although we can not replicate the flattening of their lf at @xmath39even less so , the flat luminosity functions provided in @xcite and @xcite .
a total of @xmath61000 cluster members per 1 mag bin around the main - sequence turnoff ( @xmath40 ) in our present sample is a record high among the known old galactic open clusters .
we constructed a ( number ) surface - density profile for ngc 6791 using the 4830 stars in figure 1 brighter than @xmath41 ( where the proper - motion discrimination between cluster members and field stars is best ) .
this is shown in figure 3 .
the filled circles in this plot show the number of stars per unit area , as a function of projected cluster - centric radius , defined by adding up stars weighted by their membership probabilities in a series of concentric circular annuli .
the open circles result from adding up the total numbers of the same stars directly , without weighting by @xmath42 .
we fitted a single - mass , isotropic @xcite model to the @xmath42-weighted density profile in figure 3 , using only the data inside @xmath43 to constrain the fit .
this implies a rather low central concentration for the cluster : @xmath44 , corresponding to a dimensionless central king potential , @xmath45 .
the best - fitting king scale radius is @xmath46 , which , given the low fitted concentration , implies a projected core ( half - power ) radius of @xmath47 .
the associated tidal radius is @xmath48 , and the effective ( projected `` half - mass '' ) radius is @xmath49 ( it is normally the case that estimates of @xmath50 are more robust than those of other characteristic radii ) . for a distance of 4.0 kpc to ngc 6791 @xcite , @xmath51 ; thus , we have that the projected core radius of the cluster is @xmath52 pc ; the effective radius is @xmath53 pc ; and the fitted tidal radius is @xmath54 pc .
fitting to various other number density profiles defined by both brighter and fainter magnitude cuts , and both weighted and not weighted by @xmath42 , yields results for @xmath55 , @xmath50 , and @xmath56 that are generally compatible with the range of values suggested by the numbers just given for the @xmath57 case .
the same is true for fits to luminosity surface - density profiles defined by adding up the luminosities of all the stars in our series of annuli . in exploring these various definitions of the density profile
, we saw a clear tendency for the inferred tidal radius ( and thus the concentration , @xmath58 ) to increase systematically along with @xmath50 , to a lesser extent as stars with fainter and fainter magnitudes were included to define the observed density profile .
this is primarily because stars with @xmath5920 in our sample are somewhat depleted , relative to the brighter stars , in the innermost @xmath60@xmath61 of ngc 6791 .
however , it is not clear whether this is a physical effect ( e.g. , due to mass segregation ) or an artifact of crowding and image blending in the core of the cluster .
the total luminosity of all stars with @xmath57 in our catalogue is @xmath62@xmath63 ( assuming @xmath64 kpc , corrected for @xmath65 mag of extinction , and depending on whether or not the luminosities are weighted by @xmath42 ) . adopting a theoretical @xmath66 stellar mass - luminosity relation from the padova
isochrones @xcite implies a total mass of @xmath7 for the observed stars .
this is a lower limit to the cluster mass because we have not attempted corrections for stellar binarity or incompleteness .
we note that , if a @xcite model with @xmath67 and @xmath68 pc accurately describes the internal mass profile of ngc 6791 , then the projected , one - dimensional velocity dispersion averaged within the effective radius should be @xmath69 .
this corresponds to an intrinsic proper - motion dispersion of @xmath70 , which is much smaller than the formal uncertainties in our proper motions .
it is also significantly lower than the radial - velocity dispersion of @xmath71 reported for about a dozen red giants by @xcite , suggesting either that the @xcite dispersion may be spuriously high , or that ngc 6791 may not be in virial equilibrium .
perhaps related to the second option , we note that a 5000 @xmath72 cluster in a galactic orbit with a pericenter of @xmath73 kpc and an apocenter of @xmath74 kpc @xcite is expected to have a tidal radius of @xmath75 at pericenter , and @xmath76 at apocenter as against a value of @xmath77 suggested by our king - model fitting of ngc 6791 .
deciding the true dynamical state of this cluster will require much more comprehensive and higher - precision proper - motion and radial - velocity surveys , to obtain the tightest possible constraints on its orbit and to delineate accurately its internal kinematics .
even a casual inspection of the cluster s cmd reveals an unusually broad rgb on the order of @xmath780.1 mag , while the formal uncertainties in the colors of these stars do not exceed @xmath60.02 mag .
there is a variety of potential sources of increased scatter in the cmd , such as instrumental effects including the limitations of aperture photometry in crowded fields , presence of field stars , differential reddening , binarity of stars , metallicity variations , and possible effects on color / luminosity of rgb stars by variable mass loss suggested by @xcite .
our proper - motion - vetted cmd offers a better look at the morphology of the rgb , which is more sensitive to differential reddening than the main sequence ( apart from its turnoff region ) .
in addition , our sample covers the largest area ever considered , thus enhancing the chances of finding the signs of differential reddening .
unfortunately , we can not take advantage of significantly reduced reddening effects in the near infrared .
there is a poorly understood scatter in the @xmath79 color of our rgb stars from either 2mass or @xcite , which prevents us from using this color as a nearly reddening - free reference .
instead , we examined the morphology of the rgb branch in @xmath80 cmd using approximately equal numbers of inner and outer probable cluster members by dividing the cluster at @xmath815@xmath2 .
then , we draw a ridge line for the inner sample , encompassing subgiants and rgb stars ; see figure 4 .
the majority of rgb stars at @xmath82 are bluer than this ridge line but , surprisingly , nearly all rgb stars at @xmath83 are redder than the ridge line of the inner sample .
the shift in rgb color between the inner and outer samples is @xmath84 mag , whereas the age difference of 1 gyr suggested by @xcite would produce only a @xmath60.01 mag color shift in @xmath85 across the rgb region @xcite .
we conjecture that this color shift , at least in part , is due to differential reddening . among the effects listed above
, it is the only physical effect known to have an undeniable spatial dependency .
our data provide evidence that differential reddening may not be ignored across the face of ngc 6791 .
further advancements in our understanding of such a small effect require accurate near - infrared photometry and reliable identification of binaries in the upper part of the cmd .
this study provides a comprehensive and clean list of cluster members in one of the astrophysically most important galactic star clusters .
we hope that our findings in ngc 6791 and excellent cluster membership will stimulate in - depth future studies of individual cluster members .
we thank d. crawford , j. cummings , e. hubbard , t. kinman , b. lynds , s. majewski , r. michie ( deceased ) , k. mighell , a. sarajedini , and m. sosey , whose time and effort at the telescope made this project possible .
we thank the referee barbara anthony - twarog for pointing out the limitations of our data .
this work has been supported in part by nsf grant ast 09 - 08114 to jhu ( i.p . ) and by nasa grant nnx09ah18a ( the kepler open cluster study ) .
astrophysics at keele university is supported by an stfc rolling grant .
this study is based on observations obtained with megaprime / megacam , a joint project of canada - france - hawaii telescope ( cfht ) and cea / dapnia , at the cfht which is operated by the nrc of canada , the institute national des sciences de luniverse of the cnrs , and the university of hawaii .
bedin , l.r . ,
piotto , g. , carraro , g. , king , i. r. , & anderson , j. , 2006 , , 460 , l27 boesgaard , a. m. , jensen , e. e. c. , & deliyannis , c. p. 2009 , , 137 , 4949 borucki , w. j. , et al . , 2010 , science , 327 , 977 carney , b. w. , lee , j .- w . , & dodson , b. , 2005 , , 129 , 656 carraro , g. , villanova , s. , demarque , p. , mcswain , m. v. , & bedin , l. r. , 2006 , , 643 , 1151 clem , j. l. , vandenberg , d. a. , stetson , p. b. , 2008 , , 135 , 682 cudworth , k. m. , & anthony - twarog , b. j. , 1993 , , 25 , 1454 de marchi , f. , et al . , 2007 , , 471 , 515 girardi , l. , grebel , e. k. , odenkirchen , m. , & chiosi , c. , 2004 , , 422 , 205 grundahl , f. , clausen , j. v. , hardis , s. & frandsen s. 2008 , , 492 , 171 jones , b. f. , & walker m. f. , 1988 , , 95 , 1755 kalirai , j. s. , bergeron , p. , hansen , b. m. s. , kelson , d. d. , reitzel , d. b. , rich , r. m. , & richer , h. b. , 2007 , , 671 , 748 kaluzny , j. , 2003 , , 53 , 51 kaluzny , j. , & rucinski , s. m. , 1995 , , 114 , 1 kaluzny , j. , & udalski , a. , 1992 , , 42 , 29 king , i. r. , 1966 , , 71 , 64 king , i. r. , bedin , l. r. , pittto , g. , cassisi , s. , & anderson , j. , 2005 , , 130 , 626 kinman , t. d. , 1965 , , 142 , 655 landsman , w. , bohlin , r. c. , neff , s. g. , oconnell , r. w. , roberts , m. s. , smith , a. m. , & stecher , t. p. , 1998
, , 116 , 789 magnier , e. a. , & cuillandre , j .- c . , 2004 , , 116 , 449 marigo , p. , girardi , l. , bressan , a. , groenewegen , m. a. t. , silva , l. , & granato , g. l. 2008 , , 482 , 883 mathieu , r. d. , 2000 , in asp conf .
198 , star clusters and associations : convection , rotation , and dynamos , r. pallavicini , g. micela , & s. sciortino , san francisco : asp , 517 mathieu , r. d. , van den berg , m. , torres , g. , latham , d. , verbunt , f. , & stassun , k. , 2003 , , 125 , 246 meibom , s. , & kepler team , 2010 , , 42 , 284 peterson , r. c. , & green , e. m. , 1998 , , 502 , l39 scott , j. e. , friel , e. d. , & janes , k. a. , 1995 , , 109 , 1706 stetson , p. b. , bruntt , h. , grundahl , f. , 2003 , , 115 , 413 twarog , b. a. , carraro , g. , & anthony - twarog , b. j. , 2011 , , 727 , l7 zacharias , n. , et al . , 2010 , , 139 , 2184 41698 & @xmath86 & @xmath87 & 20.291 & @xmath880.140 & 46 & ehb candidate + 58783 & 290.4511108 & 37.7326775 & 17.667 & 0.032 & 56 & ehb candidate + 45106 & 290.1308289 & 37.6360130 & 14.999 & 0.508 & 27 & s2101 + 52460 & 290.0726929 & 37.6899757 & 15.116 & 0.845 & 42 & s440 + 58857 & 290.2606201 & 37.7331924 & 15.186 & 0.530 & 99 & 3 - 22 ; rv=@xmath882 var + 60456 & 290.2799683 & 37.7429771 & 15.186 & 1.017 & 99 & 3 - 27 + 64067 & 290.1457214 & 37.7645378 & 15.253 & 0.387 & 99 & s2746 + 64589 & 290.2497864 & 37.7675743 & 15.155 & 0.278 & 99 & 2 - 17 ; rv=@xmath8848 + 65895 & 290.2140503 & 37.7751503 & 14.785 & 1.075 & 99 & nw-20 ; rv=@xmath8854 + 69734 & 290.2836304 & 37.7970619 & 14.568 & 1.096 & 99 & 3 - 33 ; rv=@xmath8820 + 69976 & 290.1916809 & 37.7985649 & 14.984 & 1.045 & 95 & 2 - 45 ; rv=@xmath8853 + 82982 & 290.3112488 & 37.8854256 & 15.276 & 0.901 & 53 & s14140 + 89107 & 290.1376343 & 37.9322052 & 14.215 & 1.101 & 41 & s2382 + 93625 & 290.4263306 & 37.9681969 & 15.165 & 0.590 & 58 & + | we present comprehensive cluster membership and @xmath0 photometry of the prototypical old , metal - rich galactic star cluster ngc 6791 .
the proper - motion catalog contains 58,901 objects down to @xmath124 , limited to a circular area of radius 30@xmath2 .
the highest precision of the proper motions is 0.08 mas yr@xmath3 .
our proper motions confirm cluster membership of all main and also some rare constituents of ngc 6791 .
the total number of probable cluster members down to @xmath4 ( @xmath58 ) is @xmath64800 , corresponding to @xmath7 .
new findings include an extended horizontal branch in this cluster .
the angular radius of ngc 6791 is at least 15@xmath2 ( the effective radius is @xmath8 while the tidal radius is @xmath9 ) .
the luminosity function of the cluster peaks at @xmath104.5 and then steadily declines toward fainter magnitudes .
our data provide evidence that differential reddening may not be ignored in ngc 6791 . |
with the wide application , the charged particles acceleration based on the laser - plasma interaction have been rapidly concerned by many simulations and experiments @xcite . in recent years , the proton acceleration generated by an ultraintense laser irradiating a solid target has been investigated extensively due to the potential applications in ions fast ignition ( fi ) of inertial confinement fusion ( icf ) @xcite , compact proton sources for cancer therapy @xcite , laboratory astrophysics @xcite and so on .
for instance , the fi by proton beams to transport a few of meters to the dense core usually requires a high beam quality @xcite .
however , the coupling efficiency from laser to protons is too low to achieve the enough ignition energy . therefore , how to gain a high quality beam with high energy , low energy spread and small size is a challenge topic currently . in the past years
, several acceleration schemes have been proposed and studied for obtaining higher quality proton beams by many researchers . according to the parameters variety of the laser and targets
, the acceleration approaches are usually classified into target normal sheath acceleration ( tnsa ) @xcite , breakout afterburner acceleration ( boa ) @xcite , shock wave acceleration @xcite , and radiation pressure acceleration ( rpa ) @xcite , etc .. the tnsa , which has been studied rapidly and implemented in the experiments , is that protons can be accelerated to high energy through the electrostatic sheath field created by the hot electrons expanding into vacuum at the rear side of the target . nevertheless , the practical application is limited because of the obtained protons possessing a large divergence , large energy spread and low number density . moreover , in order to obtain a dense and monoenergetic beam , the rpa is proposed to efficiently accelerate the protons .
this scheme is mainly associated to the intense space - charge field , which is created by the radiation pressure of the laser . yet
, multi dimensional simulations show that the rpa is not enough stable due to the excited transverse instability , such as weibel instability and rayleigh - taylor - like instability @xcite . in order to suppress the instability ,
many solutions are presented and studied in detail like of with a properly tailored laser @xcite or / and with a shaped foil target @xcite , etc .. recently zou _ et al . _
@xcite proposed first a cone target used to focus laser and guide fast electrons in fi @xcite .
while the cone target can suppress the transverse expansion by a co - moving focusing electric field and achieve a better beam quality , however , the proton beams will diffuse following the laser field when they fly out from the right exit of the cone .
this deficiency makes the protons not be accelerated availably to a farther distance . in this work , a gold cone with a capillary attached to the behind
is proposed and involved problems are studied by using the two - dimensional particle - in - cell ( pic ) simulations .
the results show that this cone - capillary target can availably collimate and guide the protons , then achieve a better beam quality with a density @xmath0 , monoenergetic peak @xmath1 , spatial emittance @xmath6 with divergence angle @xmath3 and diameter @xmath4 in a farther distance .
the enhanced reasons are mainly attributed to the focus by the transverse electric field @xmath7 generated in the cone as well as the capillary . besides , in the process of exploring the capillary parameters , one finds that the capillary lengths act as a important role in the protons acceleration .
this paper is organized as follows .
section ii outlines the target configurations , simulation parameters and results . besides
, the enhanced reasons of protons acceleration are also discussed and analyzed in detail .
the capillary parameters are examined in section iii , and the optimal capillary length is given .
finally a brief summary is presented in sec .
for the purpose to demonstrate the enhancement of proton acceleration , we have designed two kinds of targets by using the gold cone and gold cone - capillary .
we have investigated the created protons quality by them via the two - dimensional pic simulations .
two different cases of targets are illustrated in fig .
[ fig:1 ] whose the case ii by fig .
[ fig:1](b ) represents the gold cone with a capillary attached behind the cone and as a comparison the case i by fig .
[ fig:1](a ) stands for the gold cone only .
.,width=566 ] the simulation is performed with the pic code epoch @xcite .
the simulation box is @xmath8 and the grid size is @xmath9 with @xmath10 grid cells .
the preionized cone or / and cone - capillary consist of electrons and au ions whose density is established to be @xmath11 .
the cone is located from @xmath12 to @xmath13 with the cone - angle of @xmath14 from horizon .
the thickness of its wall is @xmath15 and the diameter of the left opening is @xmath16 . a proton - rich foil , which is full ionized with density @xmath17 and thick @xmath18 , is situated at a distance @xmath19 away from the left boundary of the cone .
a capillary with length @xmath20 is attached behind the cone . here
@xmath21 is the laser wavelength and @xmath22 ( @xmath23 laser frequency , @xmath24 electron mass and @xmath25 electron charge ) is the critical electron plasma density .
besides , each cell consists of @xmath26 weighted particles per species in our simulations . a circular polarized ( cp ) laser pulse is normally incident from the left boundary of the box .
the laser has a peak intensity of @xmath27 , and rises to the maximum value in @xmath28 then remains constant for @xmath29 , where @xmath28 is the laser period .
the intensity profile of the laser is gaussian with spot size of @xmath30 [ full width at half maximum ( fwhm ) ] .
besides , we use the absorption boundary in @xmath31 direction and @xmath32 direction for both fields and particles , respectively . and the case ii [ ( b ) and ( d ) ] at @xmath33 [ ( a ) and ( b ) ] and @xmath34 [ ( c ) and ( d ) ] . here the proton density is normalized by @xmath35.,width=566 ] [ ( a ) , ( c ) ] and @xmath36 [ ( b ) , ( d ) ] .
blue and red lines represent the case i and the case ii , respectively.,width=566 ] now let us to see the results of proton acceleration in the two different kinds of targets .
figure [ fig:2 ] shows the distributions of the proton - rich foil density in both cases . at @xmath33 , the transverse expansion of the proton - rich foil
is suppressed greatly in both cases , which is good for the transverse collimation in the acceleration process , as shown in fig .
[ fig:2](a ) and [ fig:2](b ) .
these results are in accordance with the ref .
meanwhile in comparison with the case i and ii the proton density is more compacter for the case i. but as time goes on the situation is reversed since proton density becomes more and more compact for the case ii .
this point can be seen at @xmath36 in fig . [
fig:2](c ) and ( d ) that a dense proton source can be gained with a density @xmath0 and bunch transverse diameter @xmath37 in case ii , but in contrast the proton density decreases significantly and bunch also diffuses seriously in the transverse direction in the case i. to demonstrate the enhanced acceleration effect by using the cone - capillary , the energy spectra and angular distribution of the protons are shown in fig .
[ fig:3 ] . for the energy spectra
they are plotted in fig . [ fig:3](a ) and [ fig:3](b ) at different time .
it shows that there are a monoenergetic proton bunches in both cases at @xmath33 .
similarly as in the density the better quality seems to be got in case i at this relative earlier time .
for example , see fig .
[ fig:3](a ) , the bunch quality of a peak energy @xmath38 with energy spread @xmath39 in the case i is indeed better than that of a peak energy @xmath40 with energy spread @xmath41 in the case ii .
however with time increasing , the situation is reversed again for the energy spectra .
in fact the monoenergetic peak structure disappears evenly in the case i at @xmath36 while it remains a good quality by a peak energy @xmath1 with energy spread @xmath42 in the case ii , see fig .
[ fig:3](a ) , even if the protons have moved out from the capillary .
on the other hand the protons number at the monoenergetic peak in the case ii is almost three times higher than that in the case i. for the angular distribution they are plotted in fig . [
fig:3](c ) and [ fig:3](d ) .
the momentum angle of protons is defined by @xmath43 , where @xmath44 and @xmath45 are the momentum of the protons in the transverse @xmath32-axis and longitudinal @xmath31-axis , respectively . in the case
ii , it is surprising to see that the fwhm of the proton divergence distribution maintains @xmath5 so that the spatial emittance is about @xmath46 for the bunch of @xmath47 diameter at time of @xmath48 , which is greatly beneficial for the practical applications . in the case
i , however , the proton divergence angle is large , which decreases slightly from about @xmath49 at @xmath50 to @xmath51 at @xmath34 .
evenly the corresponding spatial emittance increases from @xmath52 at @xmath50 to @xmath53 at @xmath34 .
these results obviously indicate that the case ii is benefit to accelerate the protons to a farther distance and get a improved high quality protons bunch .
thus it is necessary to clarify what is behind this improvement .
we will make some analysis and discussion on the electric and magnetic fields for the both cases in the next subsection , which will show their important roles , in particular the role by the transverse electric field , on the enhancement of proton quality .
for the case i [ ( a ) and ( c ) ] and case ii [ ( b ) and ( d ) ] at @xmath50 [ ( a ) and ( b ) ] and @xmath34 [ ( c ) and ( d ) ]
. here the electric field is normalized by @xmath54.,width=566 ] at @xmath50 ( a ) and @xmath34 ( b ) in the position @xmath55 , respectively .
blue and red lines represent the case i and the case ii , respectively . here the electric field is normalized by @xmath54.,width=566 ] in the above subsection we have exhibited the enhanced acceleration effect of protons by using the cone - capillary .
now let us to see how this enhancement is affected mainly by the electric and magnetic fields in different target structures which plays a key role to the improved collimation of produced protons bunch .
the longitudinal electric fields @xmath56 at @xmath33 and @xmath34 are plotted respectively in fig .
[ fig:4 ] . at @xmath33 [ fig .
[ fig:4](a ) and [ fig:4](b ) ] , in the region of about @xmath57 , one can note that the @xmath56 for the case i is obviously stronger than that for the case ii .
thus , the protons can be accelerated to higher energy , corresponding to a higher monoenergetic peak structure in the case i. but as time goes on [ fig . [ fig:4](c ) and [ fig:4](d ) ] , the @xmath56 diffuses along transverse and becomes quite weak in the region of about @xmath58 .
however what is interesting for the case ii , a stronger @xmath56 is built in a farther distance due to the focus of the capillary at @xmath36 , which can provide driving force for the proton acceleration in long time . in order to show this feature clearer ,
the slices of the @xmath56 in position of @xmath55 are plotted in fig .
[ fig:5 ] for both cases .
one can see that , for the case i , there exists a stronger @xmath56 to accelerate the protons forward at early time , while decreases significantly at later time , see the blue lines in fig . [
fig:5](a ) and [ fig:5](b ) .
for the case ii , however , the @xmath56 becomes stronger and stronger with time increasing , see the red lines in fig .
[ fig:5](a ) and [ fig:5](b ) , which is prevailing to the protons acceleration in longer time because of the focus by the capillary . and the case ii [ ( b ) and ( d ) ] at @xmath33 [ ( a ) and ( b ) ] and @xmath34 [ ( c ) and ( d ) ] . here the electron density is normalized by @xmath35.,width=566 ] for the case i [ ( a ) and ( c ) ] and case ii [ ( b ) and ( d ) ] at @xmath50 [ ( a ) and ( b ) ] and @xmath34 [ ( c ) and ( d ) ] . here the electric field is normalized by @xmath54.,width=566 ] for the case i [ ( a ) and ( c ) ] and case ii [ ( b ) and ( d ) ] at @xmath50 [ ( a ) and ( b ) ] and @xmath34 [ ( c ) and ( d ) ] . here the magnetic field is normalized by @xmath54.,width=566 ] when the laser propagates into the cone or / and cone - capillary , surface electrons are first pulled out into the vacuum by laser electric fields , then they are accelerated forward by the lorentz force of @xmath59 , as shown in fig .
[ fig:6 ] .
the electrons expanded into the vacuum continue to propagate along the cone wall and induce a strong transverse electric field @xmath7 , shown in fig .
[ fig:7 ] , and meanwhile an intense quasistatic magnetic field @xmath60 , shown in fig .
[ fig:8 ] .
the induced quasistatic magnetic field plays a role of driving the electrons out from the cone surface , while the @xmath7 acts to draw electrons back into the inside of the cone .
moreover , it is worth to note that the moving @xmath7 can also play the part of focusing the protons in the transverse , which has been proposed by zou _
this results in a balance between the electric field and the magnetic field @xcite that leads to the collimation and guidance of the electrons along the cone or / and cone - capillary . at @xmath33
, the protons can collimated by the transverse electric field @xmath7 for both cases , as shown in fig .
[ fig:7](a ) and [ fig:7](b ) .
however , as time goes on , the @xmath7 disappears in the region of @xmath61 for the case i. while it is still quite strong due to the existence of the capillary for the case ii , which ensures to focus sustainedly the protons in a farther distance .
from the preceding discussion , we have researched how the protons acceleration is enhanced by the cone - capillary .
one can conclude that the @xmath7 generated by the capillary acts as the crucial role in guiding and collimating the protons .
accordingly , the length of capillary is the key parameter in dominating the beam quality .
this reminds us to make more simulations to see whether there exists an optimal capillary length for protons quality when the other parameters are fixed .
( black curve ) and the number density locating the peak ( red curve ) as the function of the length @xmath62 of the capillary , respectively.,width=566 ] , ( b ) @xmath63 , ( c ) @xmath64 and ( d ) @xmath36 , respectively.,width=566 ] the proton divergence angle @xmath65 and number density of the peak energy are plotted as the function of the capillary length @xmath62 , as shown in fig .
[ fig:9 ] .
obviously the length @xmath66 represents the case i. it shows that the @xmath65 decreases with @xmath62 , see the black curve in fig .
[ fig:9 ] , and the number density around of the peak energy increases with @xmath62 , see the red curve in fig .
[ fig:9 ] .
certainly these results are mainly attributed to the collimation of the @xmath7 generated by the capillary . on the other hand
, it should be emphasized that the number density increases appreciably when @xmath67 .
we believe the reason is that the bulk protons have not escaped out from the right opening of the capillary
. moreover , fig .
[ fig:10 ] exhibits the proton energy spectra with different capillary lengths at different moments .
as time goes on , it shows that the shorter the length of the capillary , the earlier the monoenergetic peak of the proton beam disappears .
thus this can be summed up that the shorter capillary , for example @xmath68 , @xmath69 and @xmath70 , is not conducive to accelerating protons to a farther distance .
we can also note that for the longer capillary such as @xmath20 and @xmath71 , the monoenergetic peak are approximately coincident at early time , however , at later time , see fig .
[ fig:10](d ) , comparing the cases of @xmath20 with the length @xmath67 , the monoenergetic peak in both of energy value and number density are better for the former case .
this is not surprising because that the longer capillary would make the longitudinal electric field @xmath56 decreasing , which would result in the decrease of the total conversion efficiency from the laser to protons .
overall , the extensive simulation results indicate an optimal proper capillary length existence , e.g. @xmath20 shown in fig .
[ fig:10 ] , by which one can achieve a high quality proton source even in longer time .
in summary , we have investigated the enhanced proton acceleration in the cone - capillary by using the 2d3v pic simulations performance .
compared with the cone without the capillary , our results show that the protons can be accelerated and guided to a farther distance by using the cone - capillary . this enhancement has been analyzed and discussed in detail .
first , when the protons enter into the cone , they can be collimated in transverse direction by the @xmath7 generated in the cone for both cases .
afterwards , they can be continually accelerated and guided by the @xmath7 generated in the capillary , which results in a higher quality proton source in longer time for the case ii .
yet it decreases significantly and diffuses along transverse direction for the case i. as a result , a dense proton source can be achieved with a density @xmath0 , monoenergetic peak @xmath1 , spatial emittance @xmath2 with divergence angle @xmath3 and diameter @xmath72 by the cone - capillary .
lastly , the capillary lengths are also optimized and analyzed .
these results may have many important implications such as the ions fast ignition , the medical applications as well as the laboratory astrophysics and so on .
this work was supported by the national natural science foundation of china ( nsfc ) under grant nos . 11475026 , 11305010 , 11365020 and the nsaf of china under grant no .
the computation was carried out at the hscc of the beijing normal university .
the authors are particularly grateful to cfsa at university of warwick for allowing us to use the epoch .
m. roth , t. e. cowan , m. h. key , s. p. hatchett , c. brown , w. fountain , j. johnson , d. m. pennington , r. a. snavely , s. c. wilks , k. yasuike , h. ruhl , f. pegoraro , s. v. bulanov , e. m. campbell , m. d. perry , and h. powell , phys .
86 * , 436 ( 2001 ) .
r. a. snavely , m. h. key , s. p. hatchett , t. e. cowan , m. roth , t. w. phillips , m. a. stoyer , e. a. henry , t. c. sangster , m. s. singh , s. c. wilks , a. mackinnon , a. offenberger , d. m. pennington , k. yasuike , a. b. langdon , b. f. lasinski , j. johnson , m. d. perry , and e. m. campbell , phys .
lett . * 85 * , 2945 ( 2000 ) .
j. fuchs , p. antici , e. dhumieres , e. lefebvre , m. borghesi , e. brambrink , c. a. cecchetti , m. kaluza , v. malka , m. manclossi , s. meyroneinc , p. mora , j. schreiber , t. toncian , h. ppin , and p. audebert , nat .
* 2 * , 48 ( 2006 ) .
m. s. wei , s. p. d. mangles , z. najmudin , b. walton , a. gopal , m. tatarakis , a. e. dangor , e. l. clark , r. g. evans , s. fritzler , r. j. clarke , c. hernandez - gomez , d. neely , w. mori , m. tzoufras , and k. krushelnick , phys .
93 * , 155003 ( 2004 ) .
x. r. hong , w. j. zhou , b. s. xie , y. yang , l. wang , j. m. tian , r. a. tang and w. s. duan , _ realization of the radiation pressure acceleration in the near critical density regime _ , phys .
plasmas ( under review 2016 ) . c. a. j. palmer , j. schreiber , s. r. nagel , n. p. dover , c. bellei , f. n. beg , s. bott , r. j. clarke , a. e. dangor , s. m. hassan , p. hilz , d. jung , s. kneip , s. p. d. mangles , k. l. lancaster , a. rehman , a. p. l. robinson , c. spindloe , j. szerypo , m. tatarakis , m. yeung , m. zepf , and z. najmudin phys .
lett . * 108 * , 225002 ( 2012 ) .
r. kodama , p. a. norreys , k. mima , a. e. dangor , r. g. evans , h. fujita , y. kitagawa , k. krushelnick , t. miyakoshi , n. miyanaga , t. norimatsu , s. j. rose , t. shozaki , k. shigemori , a. sunahara , m. tampo , k. a. tanaka , y. toyama , t. yamanaka , and m. zepf , nature ( london ) * 412 * , 798 ( 2001 ) .
t. d. arber , k. bennett , c. s. brady , a. lawrence - douglas , m. g. ramsay , n. j. sircombe , p. gillies , r. g. evans , h. schmitz , a. r. bell , and c. p. ridgers , plasma phys .
fusion * 57 * , 113001 ( 2015 ) .
r. kodama , y. sentoku , z. l. chen , g. r. kumar , s. p. hatchett , y. toyama , t. e. cowan , r. r. freeman , j. fuchs , y. izawa , m. h. key , y. kitagawa , k. kondo , t. matsuoka , h. nakamura , m. nakatsutsumi , p. a. norreys , t. norimatsu , r. a. snavely , r. b. stephens , m. tampo , k. a. tanaka , and t. yabuuchi , nature ( london ) * 432 * , 1005 ( 2004 ) . | a scheme with gold cone - capillary is proposed to improve the protons acceleration and involved problems are investigated by using the two - dimensional particle - in - cell simulations .
it is demonstrated that the cone - capillary can efficiently guide and collimate the protons to a longer distance and lead to a better beam quality with a dense density @xmath0 , monoenergetic peak energy @xmath1 , spatial emittance @xmath2 with divergence angle @xmath3 and diameter @xmath4 .
the enhancement is mainly attributed to the focusing effect by the transverse electric field generated by the cone as well as the capillary , which can prevent greatly the protons from expanding in the transverse direction .
comparable to without the capillary , the protons energy spectra have a stable monoenergetic peak and divergence angle near to @xmath5 in longer time .
besides , the efficiency of acceleration depending on the capillary length is explored , and the optimal capillary length is also achieved . such a target may be benefit to many applications such as ions fast ignition in inertial fusion , proton therapy in medicine and so on . |
where knot theory studies embeddings of circles into @xmath0 , spatial graph theory studies embeddings of arbitrary graphs .
let @xmath1 be a finite abstract graph .
let @xmath2 be an embedding of @xmath1 into the @xmath3-sphere .
we will call @xmath4 a _ spatial embedding of @xmath1 _ or simply a _
spatial graph_. since it is the map @xmath2 that matters in this construction , we will often refer to @xmath2 as a spatial embedding .
we will often use the notation @xmath5 to emphasize that we are thinking of the spatial graph as a subcomplex of @xmath0 .
several analogues of link homotopy have been proposed for spatial graphs , such as the edge homotopy and vertex homotopy of taniyama @xcite .
two spatial embeddings of a graph @xmath1 are called _ edge - homotopic _ if they are transformed into each other by _ self crossing changes _ and ambient isotopies , where a self crossing change is a crossing change between two arcs of the same edge , and _ vertex - homotopic _ if they are transformed into each other by crossing changes between arcs of two adjacent edges and ambient isotopies .
each of these notions reduces to link homotopy in the case when the graph is a disjoint union of circles , and invariants of these relations have been produced @xcite .
classification of the embeddings of some graphs under these relations is possible .
for example , spatial embeddings of @xmath6 up to edge - homotopy have been completely analyzed by nikkuni @xcite .
milnor invariants are sufficient to classify two and three component links up to link homotopy @xcite and contributed to the eventual classification of all links @xcite
. so far , no analogue of these invariants exists for the spatial graph relations listed above .
since milnor invariants are a useful tool for studying link homotopy , our goal is to introduce a generalization of link homotopy for spatial graphs that allows a reasonable generalization of milnor s link homotopy invariants .
the equivalence relation we will use is called _
component homotopy_. let @xmath7 denote the set of all spatial embeddings of an abstract graph @xmath1 .
two embeddings @xmath8 are called component homotopic if @xmath9 can be transformed to @xmath10 by ambient isotopy and crossing changes between edges that belong to the same component of @xmath1 .
note that if @xmath1 is connected , then any two embeddings are component homotopic .
this is the analogue of the fact that any knot is link homotopic to the unknot , though for most graphs there is no canonical choice of embedding to call the `` unknot . ''
however , up to component homotopy , there is a clear choice for the analogue of the unlink , which we will call a _ completely split _ embedding .
a spatial embedding @xmath9 of @xmath1 is completely split if each component of @xmath9 can be separated from all others by an embedded @xmath11 .
note that any two completely split spatial embeddings @xmath12 are component homotopic , and thus up to component homotopy , there is a canonical `` unlink . ''
if milnor s link homotopy invariants vanish for @xmath13 , then @xmath13 is link homotopic to the unlink @xcite .
the vanishing of milnor s link homotopy invariants is equivalent to the milnor group of the link , @xmath14 , being isomorphic to the free milnor group @xmath15 . choosing the correct analogue of these groups , which we will call @xmath16 and @xmath17 , we are able to prove the following theorems , which are direct generalizations of theorem 8 and corollary 2 of milnor s original paper @xcite
. * theorem [ mainthm ] * _ let @xmath9 be an embedding of a graph @xmath1 into @xmath0 . then @xmath18 where the isomorphism preserves generators ( up to conjugacy ) if and only if @xmath9 is component homotopic to a completely split embedding .
_ * theorem [ itriv ] * _ let @xmath9 be an embedding of a graph @xmath1 into @xmath0 with components @xmath19 .
if the map @xmath20 from @xmath17 to @xmath16 sends generators to conjugates of generators , and @xmath21 is the induced map on the groups for @xmath22 , then @xmath23 if and only if @xmath9 is component homotopic to an embedding where @xmath24 can be separated from the rest of @xmath9 by an embedded @xmath11 . _ a spatial subgraph of @xmath9 that is homeomorphic to @xmath25
is called a _
constituent link_. the following theorem shows that the link homotopy classes of certain constituent links of @xmath9 determine whether @xmath9 is component homotopic to a completely split embedding .
note that this is in contrast to edge homotopy and vertex homotopy , where there are infinite families of examples of non - split embeddings , all of whose constituent links are link homotopic to trivial links @xcite .
* theorem [ splitthm ] * _ let @xmath13 be a constituent link of @xmath9 , where at most one component of @xmath13 is contained in each component of @xmath9 .
every such constituent link of @xmath9 is link homotopic to the trivial link if and only if @xmath9 is component homotopic to a completely split embedding . _
it is also possible to extract numerical invariants from the @xmath16 by looking at successive nilpotent quotients .
milnor s numbers can be interpreted as arising from the elements in the kernel of the map @xmath26 , and so by examining the kernel of the map @xmath27 we will be able to produce similar ( though less subtle ) numerical invariants .
naturally , any invariant of component homotopy is an invariant of edge homotopy and vertex homotopy , so these numerical invariants are invariants of those relations as well .
we would like to thank peter teichner for helpful conversations , and akira yasuhara for an engaging discussion that led to example [ e : akiraeg ] .
in this chapter , we will study embeddings of graphs up to component homotopy .
that is , up to colored edge homotopy , where every edge in a given component of a graph is labeled with the same color , and each component of the graph is given a distinct color .
notice that for the case @xmath28 , this is simply link homotopy . in an analogue to the colored milnor group @xmath29 of @xcite , for an embedding @xmath30 we can define a `` free '' group @xmath17 associated to @xmath1 , and a group @xmath16 associated to the embedding .
the meridians of the edges of @xmath9 form a normally generating set for @xmath31\gg$ ] , where the generator @xmath32 is the meridian of the @xmath33th edge in the @xmath34th component , the @xmath35 are elements of @xmath36 , and @xmath37 denotes the subgroup normally generated by the set @xmath38 .
in fact , we need only one generator per loop in @xmath9 , but by contracting a spanning tree , we may think of these generators as meridians for a subset of the edges .
let @xmath17 denote the free colored milnor group corresponding to @xmath1 , defined in the following way .
fix a spanning tree @xmath39 of @xmath1 .
label the edges of component @xmath34 in @xmath40 by @xmath41 .
define @xmath42 , where @xmath43 is the free group with generators @xmath41 , and @xmath44\gg $ ] , with @xmath45 arbitrary words in the @xmath41 .
we will often refer to the relations induced by @xmath46 as the _ milnor relations_. when @xmath17 is generated by @xmath41 with @xmath47 and @xmath48 then @xmath49 is generated by meridians @xmath32 with @xmath47 and @xmath48 , for the same values of @xmath50 and @xmath51 . for an edge @xmath41 in @xmath1 , there is a unique minimal path @xmath52 in @xmath39 connecting the end points of @xmath41 .
let @xmath53 .
note that the free colored milnor group @xmath17 is a colored milnor group in the sense of freedman and teichner , and hence is nilpotent by lemma 3.1 of @xcite . as @xmath16 is a quotient of @xmath17 , it is nilpotent as well .
this is important , as it allows us to make use of the following theorem of stallings @xcite .
let @xmath54 and @xmath55 be nilpotent groups , and @xmath56 a homomorphism .
if the induced map @xmath57 is an isomorphism , and the induced map @xmath58 is surjective , then @xmath59 is an isomorphism .
[ stallingsthm ] with stallings theorem at our disposal , it is possible to show that the group @xmath16 is invariant under component homotopy on the spatial graph @xmath9 in the same way as for the colored milnor group in @xcite . roughly , the idea is this : given @xmath60 a spatial embedding of @xmath1 , and @xmath61 be a component homotopy from @xmath60 to @xmath62 , let @xmath63 be the track of this homotopy in @xmath64 , and @xmath65 .
consider the maps @xmath66 induced by inclusion .
the inclusions carry meridians of @xmath24 to meridians of @xmath63 , so the @xmath67 clearly induce isomorphisms @xmath68 . by using finger moves to introduce self intersections on @xmath63
, we can arrange @xmath69 . then by alexander duality ,
@xmath70 is generated by the handelbodies dual to @xmath24 , and the linking tori at the self intersection points .
these linking tori simply realize the relations in @xmath46 , so @xmath71 is surjective
. thus , by stallings theorem , @xmath72 and @xmath73 are isomorphisms , and hence @xmath74 .
it will be necessary later to understand the structure of the groups @xmath16 in more detail . given a spatial embedding @xmath9 , a regular neighborhood of @xmath24 is a handlebody , and the 2-cell of this handlebody induces the relation @xmath75 = 1 $ ] in @xmath36 and hence in @xmath16 .
we will call this relation the _ surface relation _ induced by @xmath24 , and the element @xmath76 $ ] the _ surface element_. just as is shown for the classical milnor group in @xcite , it is possible to show that @xmath16 has the presentation below .
the fact that the only relations in @xmath16 are the surface relations and the milnor relations will be important in later arguments .
the group @xmath16 has the presentation @xmath77 , where the @xmath32 are meridians to the edges @xmath41 , @xmath51 is the surface relation given by @xmath24 , and @xmath78 denotes the relations arising from @xmath46 .
[ presentation ] we begin with a wirtinger presentation for @xmath79 , which is calculated from a projection of the spatial graph . before projecting , we may first contract a spanning tree of @xmath9 , as this does not affect @xmath36 .
given a projection , choose an arbitrary orientation for each edge , and label the generators corresponding to each arc of that edge in the diagram by @xmath80 , @xmath81 in the order the arcs are met as the edge is traversed .
at each vertex we have the relation that the product of the meridians to all the arcs meeting that vertex is trivial . in a fixed component ,
each edge begins and ends at the vertex , so we may choose a diagram such that the relation induced by that vertex is @xmath82 .
the relations in the wirtinger presentation for this diagram are those coming from the vertices and those of the form @xmath83 induced by the crossings , where the @xmath84 are arbitrary words in the @xmath85 . using the milnor relations , the relations from the crossings , and a double induction as in @xcite
, we can eliminate the relations from the crossings and reduce the set of generators to only the @xmath86 .
the only remaining relations are those induced by the vertices , which are of the form @xmath87 = 1 $ ] .
these are precisely the surface relations discussed above . recall that a spatial embedding @xmath9 of @xmath1 is completely split if each component of @xmath9 can be separated from all others by an embedded @xmath11 .
we may now formulate an analogue of theorem 8 of @xcite .
let @xmath9 be an embedding of a graph @xmath1 into @xmath0 .
then @xmath18 where the isomorphism preserves meridinal generators ( up to conjugacy ) if and only if @xmath9 is component homotopic to a completely split embedding .
[ mainthm ] suppose that @xmath9 is component homotopic to a completely split embedding . since @xmath16 is invariant under component homotopy , we may assume that @xmath9 is completely split .
the group @xmath16 is a free group modulo the set @xmath46 and the surface relations generated by the handlebodies forming the boundaries of neighborhoods of each component of @xmath9 .
the surface relations in @xmath16 are of the form @xmath88 = 1 $ ] for fixed @xmath34 .
since each component is separated from all others , for a fixed @xmath34 , each @xmath89 is a word in the @xmath32 .
since @xmath32 commutes with all other generators with index @xmath34 , @xmath90 = 1 $ ] for each pair @xmath91 .
thus , in @xmath16 the surface relations are trivial , so we have only the relations introduced by @xmath46 , and the map @xmath92 is an isomorphism @xmath27 .
suppose that an isomorphism @xmath93 exists .
we will induct on the number of edges in the final component .
clearly if @xmath9 has one component , it is component homotopic to a split embedding , so the base case is trivial .
suppose @xmath94 is a spatial graph of @xmath50 components , and the first betti number @xmath95 .
then there are @xmath96 generators of @xmath16 arising from @xmath97 , specifically @xmath98 , @xmath99 .
if deleting an edge of @xmath97 does not change @xmath100 , that edge is contractible and so the argument is trivial .
so suppose that deleting an edge @xmath101 does change @xmath100 .
we may choose the spanning tree @xmath39 to avoid @xmath101 , so let @xmath102 be the generator of @xmath16 represented by a meridian of @xmath101 .
let @xmath103 and @xmath10 be the restriction of @xmath2 to @xmath104 .
there is an obvious surjection @xmath105 given by @xmath106 , and the kernel of this map is clearly @xmath107 .
let @xmath108 be the restriction of @xmath20 to this kernel and @xmath109 be the map induced by @xmath20 .
we want to show that @xmath110 is an isomorphism . by the 5 lemma ,
this is equivalent to showing that @xmath108 is an isomorphism .
@xmath111_{\theta_{e } } \ar@{^{(}->}[r ] & cmf(g ) \ar[d]_{\theta}^{\cong } \ar@{->>}[r]^{x_{nr}=1 } & cmf(g ' ) \ar[d]_{\theta ' } \ar[r ] & 1 \ar[d]\\ \ll m_{nr } \gg \ar@{^{(}->}[r ] & cmf(\phi ) \ar@{->>}[r]^{m_{nr}=1 } & cm(\phi ' ) \ar[r ] & 1}\ ] ] since @xmath20 sends generators to conjugates of generators , @xmath112 and so @xmath113 .
clearly , as @xmath108 is the restriction of an isomorphism it is injective , so it suffices to show that it is surjective . given @xmath114
, we have @xmath115 .
thus @xmath108 is an isomorphism , and so is @xmath110 .
as @xmath10 has fewer edges , and @xmath110 is an isomorphism carrying generators to conjugates of generators , we may apply the induction hypothesis and use component homotopy to convert @xmath10 to a completely split embedding .
if we perform this component homotopy on @xmath9 simply carrying the extra edge @xmath101 along , we arrive at the embedding shown in figure [ setup ] .
( 360,144 ) after the inductive hypothesis . the edge labeled @xmath101 winds through some or all of the components .
] since the embedding @xmath10 is completely split , in @xmath116 the loop @xmath117 is a word in the @xmath98 .
thus , in @xmath16 , @xmath117 is a word in @xmath98 , @xmath118 and @xmath119 for any @xmath120 .
now , since @xmath121 = 1 $ ] in @xmath16 , we know that @xmath122=1 \in cm(\phi)$ ] for all @xmath123 .
we have the relation @xmath124 = 1 $ ] in @xmath16 from the 2-cell of the neighborhood of @xmath97 , but as we just saw that @xmath122=1 $ ] , @xmath123 , we may conclude that @xmath125 = 1 $ ] in @xmath16 . using the isomorphism @xmath20
, we have that @xmath126 = 1 $ ] in @xmath17 . in @xmath17 ,
the only elements that commute with a generator @xmath41 are conjugates of generators of the same color @xcite .
thus , we know @xmath127 for some @xmath128 , and hence has the same form in @xmath16 .
this means that @xmath129 $ ] in @xmath130 .
we may eliminate the commutators through component homotopy as shown in figure [ linkcancel ] .
to eliminate the terms of @xmath131 , we use the component homotopy move as in figure [ graphcancel ] .
we may reduce to @xmath132 in @xmath130 .
thus the loop @xmath133 is contractible and may be isotoped into the sphere containing @xmath97 , so @xmath9 is component homotopic to a completely split embedding .
( 360,144 ) $ ] . ]
( 360,144 ) . ]
let @xmath134 denote the colored milnor group of @xmath135 , the spatial graph obtained from @xmath9 by removing the @xmath34th component @xmath24 .
that is @xmath136 for a fixed @xmath34 . given a map @xmath93 , we have the sequence @xmath137 & cmf(g ) \ar[r]^{\theta } & cm(\phi)}\ ] ] let @xmath138 , and let @xmath21 denote the restriction of the map @xmath20 to @xmath139 .
this induces the sequence @xmath140 & cmf(g^{i } ) \ar[r]^{\theta^{i } } & cm(\phi^{i})}\ ] ] we are now able to produce a theorem analogous to milnor s corollary 2 of @xcite .
let @xmath9 be an embedding of a graph @xmath1 into @xmath0 .
suppose the map @xmath141 sends generators to conjugates of generators .
then @xmath142 if and only if @xmath9 is component homotopic to an embedding where @xmath24 can be separated from the rest of @xmath9 by an embedded @xmath11 .
[ itriv ] this proof is similar to that of theorem [ mainthm ] .
recall that the only relations in @xmath16 are those induced by @xmath46 and the surface relations . under @xmath20 ,
the set @xmath46 in @xmath16 is hit only by elements of @xmath46 in @xmath17 , so @xmath143 is generated by the preimages of the surface relations .
suppose that @xmath9 is component homotopic to an embedding where @xmath62 can be separated from the other components by an embedded @xmath11 .
splitting along this @xmath11 , van kampen s theorem implies that @xmath144 , where @xmath145 .
as the surface element @xmath146 $ ] lies in @xmath147 and the surface elements @xmath148 $ ] lie in @xmath130 for @xmath149 , the surface relation given by @xmath62 is trivial in @xmath16 , and the surface elements given by the @xmath24 , @xmath150 can be written without the generators @xmath151 . thus , @xmath152 under the map taking @xmath153 to 1 and @xmath41 to @xmath41 for @xmath150 .
now suppose that the component of interest is @xmath97 and @xmath154 .
we now induct on the number of edges in @xmath97 as before . since @xmath155 , we have that the surface relation from @xmath97 is trivial , that is @xmath156 = 1 $ ] in @xmath17
. however , by induction we know that @xmath156 = [ x_{nr},l_{nr } ] = 1 $ ] in @xmath17 .
we may now use the same techniques as in the proof of theorem [ mainthm ] to alter @xmath9 by component homotopy so that @xmath97 can be separated from the other components .
recall that a _ constituent link _ is a subgraph of @xmath9 that is homeomorphic to a disjoint union of circles .
let @xmath13 be a constituent link of @xmath9 , where at most one component of @xmath13 is contained in each component of @xmath9 .
every such constituent link of @xmath9 is link homotopic to the trivial link if and only if @xmath9 is component homotopic to a completely split embedding .
[ splitthm ] let @xmath9 be a spatial graph component homotopic to a completely split embedding @xmath10 , and let @xmath13 be a constituent sublink of @xmath9 , with at most one component of @xmath13 contained in each component of @xmath9 .
the component homotopy from @xmath9 to @xmath10 induces a link homotopy carrying @xmath13 to @xmath157 .
each component of @xmath10 can be separated from the others by an embedded @xmath11 .
each separating 2-sphere contains a single component of @xmath157 , which by additional component homotopy moves ( contained in that sphere ) can be unknotted .
thus , @xmath13 is link homotopic to the unlink .
assume that all the relavent constituent links of @xmath9 are link homotopically trivial .
we will work by induction on the number of components in @xmath9 .
for a one component graph , all spatial embeddings trivially satisfy the condition to be completely split .
suppose @xmath158 is an @xmath50 component graph and @xmath5 is not split up to component homotopy .
then by theorem [ mainthm ] , no map from @xmath17 to @xmath16 preserving generators ( up to conjugacy ) is an isomorphism .
so given the map @xmath159 , there must exist a nontrivial element in the kernel .
relabeling the components of @xmath9 if necessary , it must be of the form @xmath160 $ ] .
so some @xmath161 \neq 1 $ ] in @xmath17 .
delete all edges in @xmath54 except for the edges forming @xmath162 to produce @xmath163 , and form the embedding @xmath164 by restricting @xmath2 to @xmath165 .
we have reduced @xmath62 to a circle , so we may now think of @xmath162 as the zero framed pushoff .
suppose @xmath161 = 1 $ ] in @xmath166 , then @xmath167 in @xmath166 by @xcite .
the map @xmath168 is given by @xmath169 for @xmath170 , so the kernel is normally generated by those @xmath153 .
so , in @xmath17 , @xmath171 , and thus @xmath161 = [ x_{1j_{1 } } , \pi x_{1j}^{g_{j } } ] = 1 $ ] in @xmath17 .
this is a contradiction , so @xmath161 \neq 1 $ ] in @xmath166 , and thus @xmath172 given by the restriction of @xmath20 is not an isomorphism .
so by theorem [ mainthm ] , @xmath164 is not completely split .
suppose @xmath173 = 1 $ ] in @xmath166 for all @xmath33 .
then as discussed in the proof of theorem [ mainthm ] @xmath174 , and so @xmath175 $ ] in @xmath176 .
we may now do component homotopy as in figure [ linkcancel ] and figure [ graphcancel ] to separate @xmath177 from the rest of @xmath164 .
let @xmath178 , and @xmath179 . then , @xmath164 is component homotopic to @xmath180 .
the component homotopy induces a link homotopy on the relevant constituent links , so by assumption , these constituent links of @xmath181 are link homotopic to trivial links .
the spatial graph @xmath181 has @xmath182 components , so by induction , it is completely split up to component homotopy , and hence so is @xmath164 .
this is a contradiction .
thus , there exists some @xmath183 with @xmath184 \neq 1 $ ] in @xmath166 .
now , as before , delete edges of @xmath55 to reduce it to the circle defined by @xmath185 , forming @xmath186 . again , restricting @xmath2 , we obtain an embedding @xmath187 , and @xmath188 \neq 1 $ ] in @xmath189 .
repeat the argument as above until each component of @xmath1 has been reduced to a circle , forming @xmath190 .
form @xmath10 by restricting @xmath2 to @xmath104 .
notice that @xmath191 \neq 1 $ ] in @xmath192 .
now , @xmath10 is an @xmath50 component link @xmath13 where each component of @xmath13 came from a distinct component of @xmath1 .
thus @xmath116 is @xmath14 , and the map @xmath193 induces a map @xmath194 , where @xmath15 is the free milnor group on @xmath50 generators .
the commutator @xmath191 \neq 1 $ ] in @xmath15 , but is an element of the kernel of @xmath110 . since the map @xmath195 takes generators to conjugates of generators but is not an isomorphism , by theorem [ mainthm ] @xmath13 is not link homotopically trivial . to determine when a spatial graph @xmath9 is completely split up to component homotopy , theorem [ splitthm ] implies that it is sufficient to check that the constituent links ( of a certain type ) are all split instead of checking the conditions of theorem [ mainthm ] .
note that a similar statement is not true for theorem [ itriv ] .
as the following example shows , it is possible for a every constituent link of @xmath9 to be @xmath34-trivial , but @xmath9 to not be component homotopic to @xmath196 .
[ e : akiraeg ] let @xmath9 be the spatial graph shown in figure [ akiraeg ] .
then clearly every constituent link is @xmath3-trivial .
however , we have the relation @xmath197 = 1 \in cm(\phi)$ ] , but @xmath198 \neq 1 \in cm(g)$ ] .
thus , for the map @xmath199 , we have @xmath200 , and so by theorem [ itriv ] the embedding @xmath9 is not component homotopic to @xmath201 .
teichner and freedman introduced the notion of colored link homotopy in @xcite . in this notion , each component of link is assigned a color , and crossing changes are allowed between components of the same color .
thereom [ splitthm ] can be easily extended to the statement that @xmath9 is component homotopic to a completely split embedding if and only if every constituent link of @xmath9 is colored link homotopic to the unlink , where two components of the link have the same color if and only if they are contained in the same component of @xmath9 .
this shows that in some sense , the group @xmath16 contains information about colored link homotopy classes of all the constituent links .
it is well known that the milnor group of a link is nilpotent , and the group @xmath16 is also nilpotent .
the kernel of the map @xmath27 is generated by surface elements , which are products of commutators of elements .
this suggests that examining successive quotients by all commutators of a fixed length may be productive .
let @xmath63 be a subgraph of @xmath1 , and @xmath5 a spatial embedding .
let @xmath202 be the spatial embedding of @xmath203 that is induced by @xmath2 .
let @xmath204 denote the colored milnor group of @xmath202 , given a map @xmath205 , we have the sequence below : since the groups @xmath16 and @xmath204 are invariant under component homotopy , so is @xmath209 . here
we consider @xmath211 as a subgroup of @xmath17 , and @xmath212 $ ] denotes all commutators of length @xmath50 whose entries are elements of @xmath17 .
notice that if @xmath28 , then @xmath213 is the length of the index of the shortest nonvanishing milnor homotopy invariant whose index contains @xmath34 .
the first nonvanishing milnor invariant of the borromean rings is @xmath214 . if @xmath13 is the borromean rings , then the kernel of the map @xmath215 is generated by the element @xmath216 $ ] .
when the first component is removed , however , the resulting link is the unlink , so @xmath217 is trivial .
thus , @xmath218 .
notice that while the surface relation given by @xmath223 contains multiple commutators , only the shortest matters for the computation of @xmath224 . also , the presence of the length two commutator involving @xmath223 and @xmath225 does not affect the computation of @xmath226 .
in fact , by the proof of theorem [ splitthm ] , when @xmath227 the value of @xmath209 will be the length of the shortest non - vanishing ( colored ) milnor invariant of any link that contains a cycle from @xmath228 . | link homotopy has been an active area of research for knot theorists since its introduction by milnor in the 1950s .
we introduce a new equivalence relation on spatial graphs called component homotopy , which reduces to link homotopy in the classical case .
unlike previous attempts at generalizing link homotopy to spatial graphs , our new relation allows analogues of some standard link homotopy results and invariants . in particular we can define a type of milnor group for a spatial graph under component homotopy , and this group determines whether or not the spatial graph is splittable .
more surprisingly , we will also show that whether the spatial graph is splittable up to component homotopy depends only on the link homotopy class of the links contained within it .
numerical invariants of the relation will also be produced . |
Subsets and Splits
Data Mining Abstracts
This query searches for and retrieves 2000 documents containing the phrase 'data mining' in the abstract, providing limited but specific document samples.